Geological Repository Systems for Safe Disposal of Spent Nuclear Fuels and Radioactive Waste (Woodhead Publishing Series in Energy)

Geological repository systems for safe disposal of spent nuclear fuels and radioactive waste

© Woodhead Publishing Limited, 2010

Related titles: Understanding and mitigating ageing in nuclear power plants: Materials and operational aspects of plant life management (PLiM) (ISBN 978-1-84569-511-8) Plant life management (PLiM) is a safety-based methodology for the management of nuclear power plants over their entire lifetime. It is used by plant operators and regulators to assess the condition of nuclear power plants and to establish the technical and economic requirements for safe, long-term operation. This book critically reviews the fundamental ageing–degradation mechanisms that affect nuclear power plant structures, systems and components (SSC), along with relevant analysis and modelling methods and mitigation paths. Coverage of plant maintenance and replacement routes is extended through chapters on the development of advanced materials and components, as well as through reactortype specific PLiM practices Handbook of advanced radioactive waste conditioning technologies (ISBN 978-1-84569-626-9) Conditioning technology is essential to the nuclear fuel cycle for the immobilisation and encapsulation of spent nuclear fuels and radioactive wastes, forming the initial engineered barrier required for radioactive waste transportation, storage and disposal. Long term radiological safety is ensured through the performance and durability of conditioned waste, which advanced conditioning technologies and techniques continue to improve upon. This book provides a comprehensive and systematic reference source on advanced radioactive waste conditioning technologies. Coverage includes fundamental science and safety criteria, treatment processes and immobilisation technologies, the development of specific types of waste form materials and systems, and performance and safety analysis. Developments and innovation in carbon dioxide (CO2) capture and storage technology: Volume 2 (ISBN 978-1-84569-797-6) Carbon dioxide (CO2) capture and storage (CCS) is the one advanced technology that conventional power generation cannot do without. CCS technology reduces the carbon footprint of power plants by capturing and storing the CO2 emissions from burning fossil-fuels and biomass. Capture technology ranges from post- and pre-combustion capture to combustion-based capture. Storage options range from geological sequestration in deep saline aquifers and utilisation of CO2 for enhanced oil and gas recovery, to mineral carbonation and biofixation of CO2. This volume critically reviews carbon storage and utilisation, covering all the main geological, terrestrial and ocean sequestration options and their environmental impacts, as well as other advanced concepts such as utilisation and photocatalytic reduction. Details of these and other Woodhead Publishing books can be obtained by: . .

visiting our web site at www.woodheadpublishing.com contacting Customer Services (e-mail: [email protected]; fax: +44 (0) 1223 893694; tel.: +44 (0) 1223 891358 ext. 130; address: Woodhead Publishing Limited, Abington Hall, Granta Park, Great Abington, Cambridge CB21 6AH, UK)

If you would like to receive information on forthcoming titles, please send your address details to: Francis Dodds (address, tel. and fax as above; e-mail: francis. [email protected]). Please confirm which subject areas you are interested in.

© Woodhead Publishing Limited, 2010

Woodhead Publishing Series in Energy: Number 9

Geological repository systems for safe disposal of spent nuclear fuels and radioactive waste Edited by Joonhong Ahn and Michael J. Apted

CRC Press Boca Raton Boston New York Washington, DC

WOODHEAD

PUBLISHING LIMITED Oxford Cambridge New Delhi

© Woodhead Publishing Limited, 2010

Published by Woodhead Publishing Limited, Abington Hall, Granta Park, Great Abington, Cambridge CB21 6AH, UK www.woodheadpublishing.com Woodhead Publishing India Private Limited, G-2, Vardaan House, 7/28 Ansari Road, Daryaganj, New Delhi – 110002, India www.woodheadpublishingindia.com Published in North America by CRC Press LLC, 6000 Broken Sound Parkway, NW, Suite 300, Boca Raton, FL 33487, USA First published 2010, Woodhead Publishing Limited and CRC Press LLC # Woodhead Publishing Limited, 2010 The authors have asserted their moral rights. This book contains information obtained from authentic and highly regarded sources. Reprinted material is quoted with permission, and sources are indicated. Reasonable efforts have been made to publish reliable data and information, but the authors and the publishers cannot assume responsibility for the validity of all materials. Neither the authors nor the publishers, nor anyone else associated with this publication, shall be liable for any loss, damage or liability directly or indirectly caused or alleged to be caused by this book. Neither this book nor any part may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, microfilming and recording, or by any information storage or retrieval system, without permission in writing from Woodhead Publishing Limited. The consent of Woodhead Publishing Limited does not extend to copying for general distribution, for promotion, for creating new works, or for resale. Specific permission must be obtained in writing from Woodhead Publishing Limited for such copying. Trademark notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation, without intent to infringe. British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library. Library of Congress Cataloging in Publication Data A catalog record for this book is available from the Library of Congress. Woodhead Publishing ISBN 978-1-84569-542-2 (book) Woodhead Publishing ISBN 978-1-84569-978-9 (e-book) CRC Press ISBN 978-1-4398-3109-0 CRC Press order number: N10189 The publishers’ policy is to use permanent paper from mills that operate a sustainable forestry policy, and which has been manufactured from pulp which is processed using acid-free and elemental chlorine-free practices. Furthermore, the publishers ensure that the text paper and cover board used have met acceptable environmental accreditation standards. Typeset by Data Standards Limited, Frome, Somerset, UK Printed by TJ International Limited, Padstow, Cornwall, UK

© Woodhead Publishing Limited, 2010

Contents

Contributor contact details

xv

Woodhead Publishing Series in Energy

xx

Preface

xxiii

Part I

Part I Introduction to geological disposal of spent nuclear fuels and radioactive waste

1

Multiple-barrier geological repository design and operation strategies for safe disposal of radioactive materials M. APTED, INTERA Inc., USA; and J. AHN, University of California, Berkeley, USA

3

1.1 1.2 1.3 1.4 1.5 1.6 1.7

Introduction 4 Multiple-barrier geological repository for radioactive materials 5 Basic disposal strategies for radioactive materials 7 Containment of radioactive materials 7 Constraints on concentration of radioactive materials 12 Summary 24 References 25

2

Spent nuclear fuel recycling practices, technologies and impact on geological repository systems 29 M. S. Y. CHU, M. S. Chu & Associates, LLC, USA

2.1 2.2 2.3 2.4 2.5 2.6

Background and introduction Current spent nuclear fuel reprocessing technologies Advanced spent nuclear fuel recycling technologies Impacts of spent nuclear fuel recycling on geological disposal Future trends References

© Woodhead Publishing Limited, 2010

29 32 33 36 41 42

vi

Contents

3

Near-surface, intermediate depth and borehole disposal of low-level and short-lived intermediate-level radioactive waste 43 I. G. CROSSLAND, Crossland Consulting, UK

3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8 3.9 3.10

Introduction Outline of the sections Safety requirements for near-surface disposal Safety of disposal facilities Styles of near-surface disposal Designing for safety Future trends Sources of further information and advice IAEA requirements for geological disposal References

43 46 46 48 53 66 72 74 75 79

4

Underground research facilities and rock laboratories for the development of geological disposal concepts and repository systems 82 I. BLECHSCHMIDT and S. VOMVORIS, National Cooperative for the Disposal of Radioactive Waste–NAGRA, Switzerland

4.1 4.2 4.3 4.4 4.5 4.6 4.7

Introduction Types of URLs and their role in the staged development of geological repositories Planning and designing an underground research facility: basic considerations Public outreach and the role of URLs as training platforms Case studies Future trends References

82 85 98 101 102 115 117

Part II Geological repository systems: characterisation, site surveying and construction 5

Crystalline geological repository systems: characterisation, site surveying and construction technologies and techniques

121

A. J. HOOPER, Alan Hooper Consulting Ltd, UK 5.1 5.2 5.3 5.4 5.5

Introduction Lithologies Geological structure Rock mechanics and geotechnical properties Hydrogeology

© Woodhead Publishing Limited, 2010

121 123 125 127 128

Contents

vii

5.6 5.7 5.8 5.9 5.10 5.11 5.12 5.13

Geochemistry Radionuclide transport Disturbance by excavation or waste emplacement Stability Feasibility of construction Future trends Sources of further information References

133 138 140 144 146 148 150 150

6

Clay geological repository systems: characterisation and site surveying technologies and techniques 153 J. DELAY, National Radioactive Waste Management Agency, France

6.1 6.2 6.3 6.4 6.5 6.6 6.7 6.8 6.9 6.10 6.11 6.12 6.13 6.14 6.15

Foreword Specific features of a clay site survey Survey tools Survey strategy Technologies Geological mapping Geophysical seismic surveys Drilling Underground structures surveys Core lab analysis Integration of results Future trends Sources of further information Acknowledgements References

153 155 159 162 166 166 167 168 178 178 179 180 181 182 182

7

Assessing the long-term stability of geological environments for safe disposal of radioactive waste

188

K. J. WILSON and K. R. BERRYMAN, GNS Science, New Zealand 7.1 7.2 7.3 7.4 7.5

Introduction 188 Long-term volcano-tectonic stability issues for safe disposal of radioactive waste 189 Geochemical stability issues for safe disposal of radioactive waste 193 Potential climate change issues for safe disposal of radioactive waste 194 Using geological, geophysical and geochemical techniques for quantifying stability for safe disposal of radioactive waste 196

© Woodhead Publishing Limited, 2010

viii

Contents

7.6

Modelling long-term stability for safe disposal of radioactive waste 209 Future trends 210 Summary 212 Sources of further information and advice 213 Acknowledgements 213 References 213

7.7 7.8 7.9 7.10 7.11 8

Far-field process analysis and radionuclide transport modelling in geological repository systems

222

M. MAZUREK, University of Bern, Switzerland 8.1 8.2 8.3 8.4 8.5 8.6

Framework Transport and retardation in argillaceous sedimentary formations Transport and retardation in crystalline-basement environments Quantifying radionuclide transport: two case studies Emerging trends References

222 227 232 238 249 252

Part III Engineered barrier systems for geological repositories: containment materials and technology 9

Immobilisation of spent nuclear fuel and high-level radioactive waste for safe disposal in geological repository systems

261

E. R. VANCE and B. D. BEGG, Australian Nuclear Science and Technology Organisation, Australia 9.1 9.2 9.3 9.4 9.5 9.6 9.7 9.8

Generation of high-level waste from nuclear fuel Historical waste form development for processing Candidate waste forms and disposition schemes Inert matrix fuels Geological disposal Conclusions Acknowledgements References

© Woodhead Publishing Limited, 2010

261 265 271 278 280 283 284 284

Contents 10

Development and application of low-pH concretes for structural purposes in geological repository systems

ix

286

M. C. ALONSO , J. L. GARCI´A CALVO and A. HIDALGO, Eduardo Torroja Institute for Construction Sciences, Spain; and L. FERNA´NDEZ LUCO, INTECIN–Universidad de Buenos Aires, Argentina 10.1 10.2 10.3 10.4 10.5 10.6 10.7 11

Introduction Functional cementitious material requirements for geological disposal Design and properties of low-pH cements Development and production of low-pH concretes: shotcrete plug application Long-term durability Sources of further information and advice References Development and application of smectitic buffer and backfill materials in geological repository systems

286 290 294 304 311 316 317 323

R. PUSCH, SWECO AB/Geodevelopment International AB, Sweden 11.1 11.2 11.3 11.4 11.5 11.6

Introduction Types, properties and fabrication of the buffer Design and performance of the buffer Types, properties and fabrication of backfill Long-term performance References

323 326 331 341 344 351

12

Near-field processes, evolution and performance assessment in geological repository systems

353

W. ZHOU, Rensselaer Polytechnic Institute, USA; and R. ARTHUR, INTERA Inc., USA 12.1 12.2 12.3 12.4 12.5 12.6 12.7

Introduction Near-field component: engineered barrier system (EBS) Near-field component: host rock Summary description of near-field containment and isolation Overview of near-field process modeling Future trends in near-field analysis References

© Woodhead Publishing Limited, 2010

353 355 359 365 366 374 376

x

Contents

13

Nuclear waste canister materials, corrosion behaviour and long-term performance in geological repository systems

379

F. KING, Integrity Corrosion Consulting Ltd, Canada; and D. W. SHOESMITH, University of Western Ontario, Canada 13.1 13.2 13.3 13.4 13.5 13.6 13.7 13.8 14

Introduction Environmental aspects important for nuclear waste canister performance Selection of nuclear waste canister materials Corrosion behaviour of candidate nuclear waste canister materials Long-term performance of nuclear waste canisters Future trends Sources of further information and advice References

379 380 389 391 410 412 413 414

Post-containment performance of geological repository systems: source-term release and radionuclide migration in the near- and far-field environments 421 Ch. POINSSOT and C. FILLET, CEA, Commissariat a` l’Energie Atomique et aux Energies Alternatives, France; and J.-M. GRAS, JMG Consulting, France

14.1 14.2 14.3 14.4 14.5 14.6 14.7 14.8 14.9 14.10 14.11 14.12 14.13

Introduction Waste form degradation Long-term nuclear glass performance Long-term behaviour of spent nuclear fuel (SNF) Brief overview of low- and intermediate-level waste (L/ILW) performance Conclusion on the relative performance of the different waste forms Radionuclide fate after release Role of aqueous chemical processes in defining the relevant aqueous chemical species Significance of retention processes as a net retardation effect Coupling with transport processes Summary of the behaviour of the main radionuclides Conclusion References

© Woodhead Publishing Limited, 2010

421 423 427 437 452 458 458 460 466 471 477 478 479

Contents

xi

Part IV Performance assessment, expert judgement and knowledge management for geological repository systems 15

Safety assessment for deep geological disposal of high-level radioactive waste in geological repository systems

497

P. N. SWIFT, Sandia National Laboratories, USA 15.1 15.2 15.3 15.4 15.5 15.6 15.7 15.8 15.9

Introduction Goals of a safety assessment Steps in a typical safety assessment Acknowledging uncertainty in a safety assessment Applications of safety assessment Future trends Sources of further information and advice Acknowledgments References

497 499 501 509 513 516 518 518 519

16

Safety assessment for near-surface disposal of low- and intermediate-level radioactive waste 522 M. W. KOZAK, INTERA Inc., USA

16.1 16.2 16.3

16.4 16.5 16.6 16.7 16.8 17

Introduction 522 Definition and performance measures for near-surface disposal of low- and intermediate-level radioactive waste 523 Key issues and development of safety assessment for nearsurface disposal of low- and intermediate-level radioactive waste 525 Safety assessment methodology for near-surface disposal of low- and intermediate-level radioactive waste 528 Application of safety assessment for near-surface disposal of low- and intermediate-level radioactive waste 531 Future trends 539 Sources of further information and advice 543 References 544 Treatment of uncertainty in performance assessments for the geological disposal of radioactive waste

547

J. C. HELTON, Arizona State University, USA; and C. J. SALLABERRY, Sandia National Laboratories, USA 17.1 17.2 17.3

Introduction Conceptual structure of a performance assessment Propagation of uncertainty

© Woodhead Publishing Limited, 2010

547 549 557

xii

Contents

17.4 17.5 17.6 17.7 17.8

Computational design of a performance assessment Sensitivity analysis Concluding discussion Acknowledgments References

565 569 572 574 575

18

Assessment of expert judgments for safety analyses and performance assessment of geological repository systems

580

K. E. JENNI, Insight Decisions, USA; and A. VAN LUIK, US Department of Energy, USA 18.1 18.2 18.3 18.4 18.5 18.6 18.7 18.8 18.9 18.10 19

Introduction 580 Quantifying uncertainties for decision analyses 584 Biases in judgments and the development of formal probability elicitation protocols 587 Assessments with multiple experts 593 Aggregating assessments from multiple experts 598 Degrees of rigor and formality in assessing expert judgments 600 Future trends 602 Sources of further information and advice 605 Acknowledgments 605 References 606 Application of knowledge management systems for safe geological disposal of radioactive waste 610 H. UMEKI, Japan Atomic Energy Agency, Japan; and H. TAKASE, Quintessa K.K, Japan

19.1 19.2 19.3 19.4 19.5 19.6 19.7 19.8 19.9

Introduction Knowledge management: definitions and nomenclature Disposal programme structures and knowledge flows Identification of critical problems and development of solutions Japan Atomic Energy Agency (JAEA) knowledge management system (KMS): the basic concept JAEA KMS: demonstration of application to safety case development Assessment of knowledge engineering and advanced information technology Constructing and visualising safety case arguments for geological disposal of radioactive waste Compiling, synthesising and organising knowledge

© Woodhead Publishing Limited, 2010

610 611 612 614 615 618 624 626 628

Contents 19.10 19.11 19.12 19.13 19.14

Facilitation of communication, multidisciplinary collaboration and efficient use of resources Future trends Sources of further information and advice Acknowledgements References

xiii

632 633 635 636 636

Part V Radiation protection, regulatory methodologies environmental monitoring and social engagement for geological repository systems 20

Radiation protection principles and development of standards for geological repository systems

641

M. JENSEN, Swedish Radiation Safety Authority, Sweden 20.1 20.2 20.3 20.4 20.5 20.6 20.7 20.8 20.9 21

Introduction Understanding safety of geological disposal Dose and/or risk in geological repository systems Probability and risk in geological repository systems Assessment of probability for scenarios Time scales in geological repository systems Optimization and best available technology (BAT) in geological repository systems Future trends References Development of risk-informed, performance-based regulations for geological repository systems

641 642 646 649 653 654 657 659 660 663

T. MCCARTIN and J. KOTRA, US Nuclear Regulatory Commission, USA; and G. WITTMEYER, Center for Nuclear Waste Regulatory Analyses, USA 21.1 21.2 21.3 21.4 21.5 21.6

Introduction Regulatory principles and methodologies for safe geological disposal Development and application of methodologies Future trends Disclaimer References

© Woodhead Publishing Limited, 2010

663 664 668 675 675 676

xiv

Contents

22

Environmental monitoring programs and public engagement for siting and operation of geological repository systems: experience at the Waste Isolation Pilot Plant (WIPP)

678

J. CONCA and T. KIRCHNER, New Mexico State University, USA 22.1 22.2 22.3 22.4

22.5 22.6 22.7 22.8 22.9 22.10 23

Introduction History of salt and site selection of the Waste Isolation Pilot Plant (WIPP) History and current status of CEMRC Survey of factors related to contaminant exposure and perceptions of environmental risks in the region around the Waste Isolation Pilot Plant (WIPP) Internal dosimetry and whole body monitoring of area citizens Air monitoring of geological repository systems Future trends Conclusions Acknowledgments References Methods for social dialogue in the establishment of radioactive waste management programmes

678 682 689

693 696 702 715 716 716 717 719

K. ANDERSSON, Karita Research AB, Sweden 23.1 23.2 23.3 23.4 23.5 23.6 23.7 23.8

Introduction The emergence of participation in nuclear waste management Rationales for participation in nuclear waste management programmes The Swedish dialogue and transparency process Public participation processes in nuclear waste management programmes The context of social dialogue in nuclear waste management programmes Conclusions References Index

719 721 723 725 727 730 735 737 741

© Woodhead Publishing Limited, 2010

Contributor contact details

(* = main contact)

Chapter 2

Editors

M. S. Y. Chu M. S. Chu & Associates, LLC 1333 Camino Cerrito SE Albuquerque, NM 87123 USA Email: [email protected]

J. Ahn* Department of Nuclear Engineering University of California, Berkeley Berkeley, CA 94720-1730 USA Email: [email protected] M. J. Apted INTERA Incorporated 3900 S. Wadsworth Blvd, Suite 555 Denver, CO 80235 USA Email: [email protected]

Chapter 3 I. G. Crossland Crossland Consulting Nympsfield Gloucestershire GL10 3UB UK Email: i.g.crossland@btinternet. com

Chapter 1 M. J. Apted* INTERA Incorporated 3900 S. Wadsworth Blvd, Suite 555 Denver, CO 80235 USA Email: [email protected] J. Ahn Department of Nuclear Engineering University of California, Berkeley Berkeley, CA 94720-1730 USA Email: [email protected]

Chapter 4 I. Blechschmidt* and S. Vomvoris Department of International Services and Projects National Cooperative for the Disposal of Radioactive Waste– NAGRA Hardstrasse 73 CH-5430 Wettingen Switzerland Email: ingo.blechschmidt@nagra. ch; [email protected]

© Woodhead Publishing Limited, 2010

xvi

Contributor contact details

Chapter 5

Chapter 9

A. J. Hooper Alan Hooper Consulting Ltd 5 Wickwar Road Kingswood Wotton-under-Edge Gloucestershire GL12 8RF UK Email: [email protected]

E. R. Vance* and B. D. Begg Australian Nuclear Science and Technology Organisation PMB 1 Menai NSW 2234 Australia Email: [email protected]; [email protected]

Chapter 6

Chapter 10

J. Delay Meuse/Haute-Marne Centre Andra, National Radioactive Waste Management Agency RD 960-55290 Bure France Email: [email protected]

Dr M. C. Alonso*, J. L. Garcı´ a Calvo and Dr A. Hidalgo Department of Construction Materials: Physics and Chemistry Eduardo Torroja Institute for Construction Sciences CSIC Serrano Galvache, 4 28033 Madrid Spain Email: [email protected]; [email protected]; [email protected]

Chapter 7 K. J. Wilson* and K. R. Berryman GNS Science PO Box 30368 Lower Hutt New Zealand Email: [email protected]; [email protected]

Chapter 8 M. Mazurek Rock–Water Interaction Institute of Geological Sciences University of Bern Baltzerstr. 3 CH-3012 Bern Switzerland Email: [email protected]

Dr L. Ferna´ndez Luco Facultad de Ingenierı´ a–INTECIN– Universidad de Buenos Aires Laboratorio de Materiales y Estructuras Av. Las Heras, 2214 1117 Cuidad Auto´noma de Buenos Aires Argentina Email: lfernandez@fi.uba.ar

Chapter 11 R. Pusch SWECO AB/Geodevelopment International AB SE-22370 Lund Sweden

© Woodhead Publishing Limited, 2010

Contributor contact details Email: pusch@geodevelopment. ideon.se

Chapter 12 W. Zhou* Rensselaer Polytechnic Institute Department of Mechanical, Aerospace and Nuclear Engineering, 110 8th St Troy, NY 12180 USA Email: [email protected] R. Arthur INTERA Inc. 3900 S. Wadsworth Blvd Suite 555 Denver, CO 80228 USA Email: [email protected]

Chapter 13 F. King* Integrity Corrosion Consulting Ltd Nanaimo British Columbia Canada V9T 1K2 Email: [email protected] D. Shoesmith Chemistry Building University of Western Ontario London Ontario Canada N6A 5B7 Email: [email protected]

Chapter 14 C. Poinssot* Radio Chemistry and Processes Department Nuclear Energy Division

xvii

Commissariat a` l’Energie Atomique et aux Energies Alternatives Marcoule, BP 17171 30207 Bagnols-sur-Ce`ze France Email: [email protected] C. Fillet Nuclear Energy Division Commissariat a` l’Energie Atomique et aux Energies Alternatives Saclay 91191 Gif sur Yvette France Email: catherine.fi[email protected] J.-M. Gras JMG Consulting 77920 Samois-sur-Seine France Email: [email protected]

Chapter 15 P. N. Swift Sandia National Laboratories Mail Stop 1369 Albuquerque, NM 87185-1369 USA Email: [email protected]

Chapter 16 M. W. Kozak INTERA Inc. 3900 South Wadsworth Blvd Suite 555 Denver, CO 87110 USA Email: [email protected]

© Woodhead Publishing Limited, 2010

xviii

Contributor contact details

Chapter 17 J. C. Helton* Department 1545 Sandia National Laboratories Albuquerque, NM 87185-0748 USA Email: [email protected] C. J. Sallaberry Department 6783 Sandia National Laboratories Albuquerque, NM 87185-1370 USA Email: [email protected]

Chapter 20 Mikael Jensen Swedish Radiation Safety Authority Dept of Radioactive Materials SE-171 16 Stockholm Sweden Email: [email protected]

Chapter 21

Chapter 18 K. E. Jenni* Insight Decisions, LLC 1616 17th Street, Suite 268 Denver, CO 80202 USA Email: kjenni@insightdecisions. com A. van Luik US Department of Energy Office of Civilian Radioactive Waste Management 1551 Hillshire Dr. Las Vegas, NV 89134 USA Email: [email protected]

Chapter 19

T. McCartin* and J. Kotra Mail stop: E2-B2 US Nuclear Regulatory Commission Washington, DC 20555-0001 USA Email: Timothy.McCartin@nrc. gov; [email protected] G. Wittmeyer Center for Nuclear Waste Regulatory Analyses Southwest Research Institute 6220 Culebra Road San Antonio, TX 78238-5166 USA Email: [email protected]

Chapter 22

H. Umeki* Japan Atomic Energy Agency 2-1-8 Uchisaiwaicho, Chiyoda-ku, Tokyo 100-8577 Japan Email: [email protected] H. Takase Quintessa K.K.

A7-707 2-3-1 Minatomirai, Nishi-ku Yokohama 220-6007 Japan Email: [email protected]

J. Conca* and T. Kirchner Carlsbad Environmental Monitoring and Research Center New Mexico State University 1400 University Drive Carlsbad, NM 1-575-706-0214 USA Email: [email protected]

© Woodhead Publishing Limited, 2010

Contributor contact details

Chapter 23 K. Andersson Karita Research AB Box 6048 SE-187 06 TA¨BY Sweden Email: [email protected]

© Woodhead Publishing Limited, 2010

xix

Woodhead Publishing Series in Energy

1

Generating power at high efficiency: Combined cycle technology for sustainable energy production Eric Jeffs

2

Advanced separation techniques for nuclear fuel reprocessing and radioactive waste treatment Edited by Kenneth L. Nash and Gregg J. Lumetta

3

Bioalcohol production: Biochemical conversion of lignocellulosic biomass Edited by K. W. Waldron

4

Understanding and mitigating ageing in nuclear power plants: Materials and operational aspects of plant life management (PLiM) Edited by Philip G. Tipping

5

Advanced power plant materials, design and technology Edited by Dermot Roddy

6

Stand-alone and hybrid wind energy systems: Technology, energy storage and applications Edited by J. K. Kaldellis

7

Biodiesel science and technology: From soil to oil Jan C. J. Bart, Natale Palmeri and Stefano Cavallaro

8

Developments and innovation in carbon dioxide (CO2) capture and storage technology Volume 1: Carbon dioxide (CO2) capture, transport and industrial applications Edited by M. Mercedes Maroto-Valer

9

Geological repository systems for safe disposal of spent nuclear fuels and radioactive waste Edited by Joonhong Ahn and Michael J. Apted

10 Wind energy systems: Optimising design and construction for safe and reliable operation Edited by John D. Sørensen and Jens N. Sørensen

© Woodhead Publishing Limited, 2010

Woodhead Publishing Series in Energy

xxi

11 Solid oxide fuel cell technology: Principles, performance and operations Kevin Huang and John Bannister Goodenough 12 Handbook of advanced radioactive waste conditioning technologies Edited by Michael I. Ojovan 13 Nuclear reactor safety systems Edited by Dan Gabriel Cacuci 14 Materials for energy efficiency and thermal comfort in buildings Edited by Matthew R. Hall 15 Handbook of biofuels production: Processes and technology Edited by Rafael Luque, Juan Campelo and James Clark 16 Developments and innovation in carbon dioxide (CO2) capture and storage technology Volume 2: Carbon dioxide (CO2) storage and utilisation Edited by M. Mercedes Maroto-Valer 17 Oxy-fuel combustion for power generation and carbon dioxide (CO2) capture Edited by Ligang Zheng 18 Small and micro combined heat and power (CHP) systems: Advanced design, performance, materials and applications Edited by Robert Beith 19 Hydrocarbon fuel conversion technology: Advanced processes for clean fuel production Edited by M. Rashid Khan 20 Modern gas turbine systems: High efficiency, low emission, fuel flexible power generation Edited by Peter Jansohn 21 Concentrating solar power (CSP) technology: Developments and applications Edited by Keith Lovegrove and Wes Stein 22 Nuclear corrosion science and engineering Edited by Damien Fe´ron 23 Power plant life management and performance improvement Edited by John Oakey 24 Direct-drive wind and marine energy systems Edited by Markus Mueller 25 Advanced membrane science and technology for sustainable energy and environmental applications Edited by Angelo Basile and Suzana Nunes

© Woodhead Publishing Limited, 2010

© Woodhead Publishing Limited, 2010

Preface

Since the mid-20th century, nuclear power generation, radioactive materials utilization in medicine and industry, and nuclear weapons production have produced significant amounts of radioactive wastes at various different levels of radioactivity concentrations. Some radionuclides have half-lives much longer than the use of nuclear and radiation technologies. Therefore, a reliable technological and societal system is necessary for management of these radioactive wastes that assure safety, for both current and future generations. Geological disposal has been selected, developed, and implemented over the past half century by virtually all nuclear countries as the safest and most effective final disposition method for radioactive materials. The goal of geological disposal is simple – to keep hazardous radioactive wastes away from people. Thanks to spontaneous radioactive decay, the radiological hazard of such wastes decreases over time, unlike other types of hazardous wastes produced by modern societies. Thus, if we can successfully isolate the materials of concern for a ‘sufficiently’ long time, then we can achieve the goal of limiting the amount of radiation from radioactive materials to future populations. Achieving this seemingly straightforward goal, however, requires the consideration and integration of diverse technological, regulatory, and social factors. How can important, safety-related questions be collected from concerned stakeholders and addressed? Technologically, how can we find and characterize suitable geological disposal sites and then design repository systems that can limit any eventual radiological releases to acceptably safe low levels? How can we assure that the multiple engineered and natural barriers of such repository systems function as we expect far into the future? More fundamentally, how is long-term safety to be evaluated and independently confirmed? The concept of geological disposal started as a relatively simple measure, but the history of its development shows that the issue of geological disposal has triggered fundamental

© Woodhead Publishing Limited, 2010

xxiv

Preface

questions regarding the principles, standards, and approaches to successful implementation of final disposal concepts. While many countries that have generated nuclear waste are just beginning the process toward geological disposal, several important milestones were achieved in the last decade of the 20th century and the first decade of the 21st century. The Waste Isolation Pilot Plant (WIPP) was commissioned in 1999 in Carlsbad, New Mexico, as a deep geological repository for the disposal of US defense-generated transuranic wastes. Other disposal systems for low-level and intermediate-level wastes have also been implemented in Sweden, Finland, Japan, and elsewhere. License applications for deep disposal of used nuclear fuel and reprocessed highlevel waste are at, or near, the point of submittal in the US, Sweden, and Finland. Significant progress toward final disposal of nuclear wastes is also occurring in France, Switzerland, and Belgium. The last fifty years in waste disposal programs has also seen some setbacks and adjustments in response to scientific, regulatory, and societal concerns. The 1960s Project Salt Vault in Lyons, Kansas in the US was abandoned after local public objections. Early siting attempts for low- and intermediate-level wastes in Switzerland were rejected by popular local votes. In the mid-1990s the initial environmental impact statement for a deep geological repository in Canada for the disposal of used CANDU fuel was found to be technically sound but not socially supported. A public inquiry into potential disposal of intermediate-level waste in the Sellafield area of the United Kingdom also led to the concept being rejected. Decisions on progress toward nuclear waste disposal in Germany have fluctuated during decades of social and political debate centered on nuclear power. In the US, the Yucca Mountain Repository Program submitted a license application to the Nuclear Regulatory Commission in 2008, but at the time of writing the fate of this program remains uncertain. It is notable, however, that following on from these setbacks, national repository programs have learned and applied important lessons on the types of technological, social, and regulatory factors that require open consideration, transparent documentation, and close integration. Successful geological disposal program re-starts are being made in many countries, in part because nuclear wastes from power generation, medical applications, and industrial uses already exist, and have so far been held in temporary storage facilities. There is an acknowledged need for these wastes to be managed and disposed of safely to protect current and future generations. As the issue of long-term safety is universal, active, and mutually beneficial, international discussions and collaborations have played important roles since the 1970s, and have been particularly helpful and effective in developing national geological disposal programs. The results of this collaboration and information sharing include the core part of

© Woodhead Publishing Limited, 2010

Preface

xxv

conceptual design and philosophy, regulatory frameworks, and approaches to safety assessment by nuclear countries. Thus, it is now understood that the geological repository is not just a large-scale civil engineering project that constructs a bunch of tunnels containing radioactive waste canisters, but also a systematic process that is supported by a set of scientific principles, independent regulatory reviews, and methodologies on information exchange and decision-making with the general public in order to be successful. Based on this recent progress and development, the Editors of this book thought that it would be timely to compile a book about geological disposal that reviews the state-of-the-art of the technology, regulatory philosophy, and social-interaction framework that are contributing to sustained progress toward assuring long-term safety of geological disposal. Because geological disposal encompasses a wide range of ‘-ologies’, we asked experts from many diverse fields to contribute chapters; while standard topics such as materials science, design, and geoscience are addressed, we purposefully broadened our topics to include regulatory and societal areas as well. The book consists of five parts. The first part, containing Chapters 1 to 4, is an introduction to geological disposal, and explains design strategy, relation with fuel cycle, and conceptual designs for low-level and high-level waste repositories. The second part, including Chapters 5 to 8, discusses site characterization, site surveying, and construction for various types of host rocks, such as crystalline rock, clay, and salt. The third part, from Chapters 9 to 14, deals with engineered-barrier technologies, ranging from wastesolidification materials to buffer/backfill materials, and modeling of their interaction to evaluate radionuclide release rates from the barriers. The fourth part, from Chapters 15 to 19, covers the performance assessment for the geological disposal system, which is a methodology developed for evaluation of the long-term performance measure to be compared with the regulatory guidelines for safety judgment. The last part, from Chapters 20 to 23, discusses how we can interpret the results of performance assessment and communicate with the public for regulatory decision-making. The Editors wish to thank the authors for their excellent contributions of these chapters, which broadly cover the state-of-the-art in subjects pertinent to geological disposal. The reference list at the end of each chapter will also be useful for the reader who would like to learn more. The Editors also stress that continued international progress is being made worldwide on geological disposal and that interested readers can use the authors and their affiliations as an aid to keep pace with such future developments in this field. In developing this book the Editors have been blessed with assistance from Ian Borthwick, an excellent commissioning editor, and the team at Woodhead Publishing Limited, UK. Without Ian’s patience and stimulating comments and suggestions, this book could not have been realized.

© Woodhead Publishing Limited, 2010

xxvi

Preface

Last but not least, the Editors would like to give special thanks to Professor Thomas H. Pigford and Professor Paul L. Chambre´, both emeriti of the University of California at Berkeley. Their guidance, mentoring, and formal instruction have been a primary source of education and insights that have helped to guide our careers since the 1980s. Joonhong Ahn Michael J. Apted

© Woodhead Publishing Limited, 2010

1 Multiple-barrier geological repository design and operation strategies for safe disposal of radioactive materials M . A P T E D , INTERA Inc., USA; J . A H N , University California, Berkeley, USA

Abstract: This introductory chapter provides basic insights and guiding principles for establishing and evaluating the long-term safe isolation of radioactive wastes in geological repositories. The intended audience is new technical researchers and reviewers, interested in understanding how their specific expertise is integrated into a multi-discipline safety assessment. The focus is on deep geological disposal, appropriate for the disposal of spent nuclear fuel (SNF), reprocessed high-level waste (HLW), and long-lived, intermediate level waste (LL/ILW). Many of the principles discussed here, however, equally apply to near-surface disposal of lower activity wastes. Two basic types of processes affecting the long-term safe containment and isolation of radioactive waste in deep geological repositories are examined; (1) delay-and-decay processes and (2) concentration-attenuation processes. The robustness of different types of isolation processes, based on their effectiveness and reliability, are discussed. A ‘top-down’ safety assessment of an integrated, multiple-barrier repository system is vital in order to identify and prioritize safety-important barriers and processes, and to use such safety-importance insights to guide an efficient and effective research, development, and design program. Key words: containment, geological repository, high-level waste (HLW), mass transfer, multiple barriers, performance assessment, radioactive waste, repository design, safety assessment, safety importance, solubility, sorption, spent nuclear fuel (SNF), transport.

3 © Woodhead Publishing Limited, 2010

4

Geological repository systems for safe disposal

1.1

Introduction

Geological disposal has been the recommended approach for the permanent disposal of radioactive wastes since the seminal US National Academy of Sciences/National Research Council’s 1957 report The Disposal of Radioactive Wastes on Land (NAS, 1957). In the NAS report, disposal of liquid high-level wastes from weapons production was recommended. In this regard, the concept of the geologic disposal at the beginning was quite different from present ones, which are based on solid waste forms. The bases for the contemporary concept and principles of geologic disposal and the method for safety assessment were considered to be established in the 1970s, (e.g. see NEA, 1977, 1991; KBS, 1983). As of today, over 30 nations with nuclear power plants (NPPs) are engaged at various stages in exploring the feasibility, conducting site selection, characterization and design programs, and/or licensing of geological repositories (Witherspoon and Bodvarsson, 2006). There are also programs exploring the possibility of international repositories for a volunteer consortium of nations (e.g. IAEA, 2004). While interesting options involving advanced fuel cycle transmutation, reprocessing, and even extended storage are also being evaluated in some countries (e.g. NEA, 2002), it has become a well-accepted consensus that all of these options eventually require geological disposal as part of the overall solution for safe management and permanent disposal of radioactive wastes (NAS, 2001; Chapman and McCombie, 2003). While confidence has been well established in the engineering community, it is still recognized as a contentious regulatory, social, and scientific issue. Discussions include scientific points, such as long-term safety confirmation, so-called unknown ‘unknowns’ issue (MacFarlane and Ewing, 2006), as well as social points, such as equity between generations and equity between the repository-host community and the rest of the country (NAS, 2003). On the regulatory side, especially in the United States, there have been active discussions involving regulatory agencies, academia, and law makers in order to reach consensus for safety standards for long-term safety of geologic disposal (NAS, 1990, 1995). These discussions have affected, and have been affected by, the site selection processes and development of a repository concept in each country. Often, discussions have included various options of nuclear fuel cycles, hoping that some type of fuel cycles could reduce difficulties significantly. As mentioned in the previous paragraph, however, after three decades of discussion, we reached an understanding that geologic disposal of radioactive wastes is necessary for any type of fuel cycle. The purpose of this introductory chapter is to provide some basic insights and guiding principles for establishing and evaluating the long-term safe

© Woodhead Publishing Limited, 2010

Multiple-barrier geological repository design/operation strategies

5

isolation of radioactive wastes in geological repositories. The focus is on deep geological disposal, appropriate for the disposal of spent nuclear fuel (SNF), reprocessed high-level waste (HLW), and long-lived intermediate level waste (LL/ILW), as advocated by the International Atomic Energy Agency (IAEA, 2009). Many of the principles discussed here, however, equally apply to near-surface, trench-type disposal of lower activity wastes, as discussed in other chapters of this book. The intended audience for this chapter are workers just beginning in the field of geological disposal of radioactive wastes, experts from traditional engineering and scientific disciplines who may be called upon to review multi-disciplinary geological repository programs, and any interested reader with college-level mathematical and technical training. For more experienced workers, there are numerous sources on advanced research, design, and development studies that are being conducted internationally; indeed, the other chapters in this volume provide an excellent window into many of these important, on-going research, development, and deployment (RD&D) areas. By contrast, this chapter’s objective is to provide a short primer on the basis for repository concepts and a simplified context by which to understand better the relevance and safety importance of current RD&D studies.

1.2

Multiple-barrier geological repository for radioactive materials

To understand basic strategies for assuring long-term, safe disposal of radioactive waste, it is first necessary to introduce the concept of a system of multiple barriers. Figure 1.1 shows a representative deep-geological repository design with multiple barriers that include (moving from the inside to the outside): . . . .

a waste form (also called a ‘waste matrix’), a combination of metal canisters (also called ‘container,’ ‘package’, or ‘overpack’), an encompassing buffer and backfill, and the host rock (geosphere) of the repository site.

The main objective of this multiple-barrier system is to secure as long a time as possible for radionuclides contained in the waste form to reach the human environment, so that radiological impacts of disposed wastes would be reduced to an acceptable level. Placing high-level nuclear waste to stable rock formation at > 300 m depth by itself provides important radiationshield, anti-intrusion, and psychological barriers that greatly promote public safety and security compared to temporary surface storage.

© Woodhead Publishing Limited, 2010

6

Geological repository systems for safe disposal

1.1 Representative illustration of a multiple-barrier deep geological repository system.

When radioactive waste is emplaced within such a conceptual repository design, as shown in Fig. 1.1, two basic periods of analysis regarding the long-term isolation of that nuclear waste can be identified: . .

containment period, during which groundwater is initially prevented from contacting the waste form by the canister, and after failure of the canister due to corrosion; low-release period, during which groundwater can contact the waste form, allowing the release and transport of dissolved radionuclides through the multiple barriers of the repository eventually to reach the accessible surface environment.

Repository concepts are therefore developed around a combination of both engineered and natural barriers that can favorably affect the containment and low-release behavior of a repository. The number, type, and assigned safety functions of these various multiple barriers varies among concepts, depending on factors such as the type of waste form, radionuclide inventory in the waste form, the type of host rock, the geological, hydrological, and geochemical settings, the required regulatory time scale for safety analysis, etc. In order to develop insights into specific repository concepts, it is useful to identify and understand basic disposal strategies and how they are successful.

© Woodhead Publishing Limited, 2010

Multiple-barrier geological repository design/operation strategies

1.3

7

Basic disposal strategies for radioactive materials

Two basic strategies, or principles, for assuring the long-term safe disposal of nuclear waste by a geological repository can be broadly identified: . .

containment (sometimes called ‘delay and decay’) and constraining concentration (sometimes called ‘dilute and disperse’).

There are, as discussed below, many different ways in which these strategies may be implemented into a repository concept.

1.4

Containment of radioactive materials

A safety-important characteristic of radioactive waste is that radioactive decay acts to reduce the total radiological hazard over time. Each radionuclide has a characteristic half-life (t1/2), in which the initial mass inventory/activity of a radionuclide (A0) will decrease by one-half for every time-period (t) equal to its half-life. Mathematically, this is expressed as AðtÞ ¼ A0 2t=t1=2 . As a heuristic rule-of-thumb, it can be assumed that the initial activity A0 (initial mass inventory) of a given radionuclide may be considered to have decayed to insignificance (reduced by an arbitrary factor of 1024) over a period of 10 half-lives. Therefore, containment of a radionuclide with a halflife of t1/2 anywhere within the multiple-barrier repository system (from the engineered barrier system (EBS) to the geosphere) for a time period of 10t1/2, will effectively eliminate that specific radionuclide from safety considerations.

1.4.1 Canister containment To explore the effectiveness of ‘delay and decay’, the impact of the canister containment can first be considered. For certain repository concepts, the targeted containment time (tc) for a canister is 1000 years (e.g. JNC, 2000), whereas for other repository concepts the targeted containment time for the canister is on the order of 1 000 000 years (e.g. SKB, 2006; US DOE, 2008). Figure 1.2 shows the impact of a 1000-year canister for representative radionuclides occurring in nuclear waste.* The 1000-year canister effectively eliminates short-lived radionuclides with half-lives less than 100 years (i.e. *

For heat-producing waste forms such as SNF and HLW, there is an initial rise in temperature due to radioactive decay of relatively short-lived radionuclides, such as Co-60 (t1/2 = 5.27 years), Sr-90 (t1/2 = 29.1 years), Cs-137 (t1/2 = 30 years), and Am-241 (t1/2 = 432 years). A 1000-year (or greater) containment time effectively allows the dissipation of such radiogenic heating, so the need for analysis of the effects of elevated temperature and elevated temperature gradients on the post-containment release of radionuclides can be obviated.

© Woodhead Publishing Limited, 2010

8

Geological repository systems for safe disposal

1.2 Impact of canister containment on the reduction and elimination of the initial inventory as a function of the radionuclide half-life. A containment time (tc) of 1000 years is assumed for illustration purposes.

less than 0.1tc), such as Cs-137 and Sr-90, and partially attenuates the initial inventory of Am-241 with a 432-year half-life. If the canister lifetime is extended to 1 000 000 years, radionuclides with half-lives up to 100 000 years can be eliminated. Note, however, that there are numerous radionuclides present in SNF/HLW/ILW with half-lives greater than 100 000 years for which a 1 000 000-year container will have no significant ‘delay and decay’ impact on reducing the initial inventory. In addition to the canister containment time, we can consider additional residence time in the waste form until radionuclides are released by wastematrix dissolution. Furthermore, hardly soluble nuclides such as actinides would form precipitates in the vicinity of the waste-matrix dissolution location. Thus, in geochemical conditions, radionuclides would stay in the vicinity of the waste form for a substantially long time even after the canister failure. This effect, however, should be categorized as the concentration constraint, and is discussed in the following section.

1.4.2 Transport time Another example of ‘delay and decay’ by a repository system is the transport time of radionuclides to migrate through various barriers of the repository system. Following the same formulation as for canister impacts, if the transport time (tt) is equal to or greater than 10 times the half-life of a given

© Woodhead Publishing Limited, 2010

Multiple-barrier geological repository design/operation strategies

9

radionuclide, then that radionuclide will effectively decay to insignificance during that transport. Diffusion and advection are the two limiting processes for the aqueous transport of dissolved radionuclides through both natural and engineered barriers. The actual transport time of radionuclides is a function of the transport rate and solute–solid interaction processes that act to retard that transport (NEA, 1993). The most commonly cited of these retardation processes is ‘sorption’, for which a sorption coefficient Kd (in units of m3/ kg) can be measured.* Diffusive transport within a repository can be achieved either by selection of a host rock site with extremely low permeability, such as clay or mudstone (e.g. Verstricht and DeBruyn, 2000; Nagra, 2002; Andra, 2005), or the inclusion of an engineered ‘buffer’ barrier (Apted, 1995; JNC, 2000; SKB, 2006). Conversely, groundwater may advectively flow through porous or fractured repository host rock, and formulations for modeling such flow depend on a number of rock-specific and site-specific features, including degree fracturing, transmissivity of fractures, interconnected porosity, regional hydraulic head, hydraulic conductivity, and anisotropy/layering of rock (Domenico and Schwartz, 1998). As an illustration of the impact of transport time, Fig. 1.3 extends Fig. 1.2 and shows as a solid line where transport time (tt) is equal to half-life (t1/2), with the inclined dashed line showing where transport time equals 10 times half-life ðtt ¼ 10t1=2 Þ. Any radionuclides with transport times lying in the upper, left-hand region above this inclined dashed line would effectively decay to insignificance during the transport. As an example of the effectiveness of transport as a ‘containment’ or ‘delay and decay’ barrier, Fig. 1.3 shows as larger, darker-colored circles the calculated diffusive-transport time (tdt) for a representative set of radionuclides, assuming diffusion through a 0.7-m thick (b) buffer with a porosity (ε) of 0.4 and a density (r) of 1800 kg/m3, with an effective (i.e. unretarded) diffusion coefficient (De) of 3.15610–2 m2/s and using representative buffer sorption coefficients (JNC, 2000). A characteristic diffusive-transport time for a sorbing radionuclide can be calculated based on this simple formulation for Fick’s law:  b2 ðe þ rKd tdt ¼ De For a non-sorbing radionuclide with Kd = 0, the diffusive-transport time through a 0.7-m thick buffer would be about 6 years, while for a strongly sorbing radionuclide with Kd = 1 m3/ kg, the diffusive-transport time *

‘Sorption’ can entail and combine a number of different types of solute–solid interactions, depending on how it is measured and normalized. See NEA (1993) for a basic review.

© Woodhead Publishing Limited, 2010

10

Geological repository systems for safe disposal

1.3 Impact of transport time on the reduction and elimination of the initial inventory as a function of the radionuclide half-life. The larger, darker-colored circles are calculated transport times for specific radionuclides diffusing across a 0.7-meter thick buffer. The smaller, lighter-colored cirles are the calculated combined transport times for radionuclides diffusing across the same buffer plus advective transport through 100 meters of rock at a flow rate of 1 meter per year.

through the same thickness buffer would be about 28 000 years. Figure 1.3 shows that diffusive transport through a 0.7-m thick buffer would eliminate many radionuclides with short- to intermediate-length half-lives, although longer-lived radionuclides with half-lives greater than 50 000 years would not experience a significant reduction in their initial inventory. If the diffusion coefficient can be reduced, the transport time can be increased. For example, with the 1-m thick Richard barrier buffer (Apted, 1995; US DOE, 1995), suitable for use in unsaturated host rock, with De = 3.15610–5 m2/ year, the diffusive-transport time for a non-sorbing radionuclide with Kd = 0, would be about 127 000 years, and over 200 million years for a sorbing radionuclide with Kd = 1 m3/ kg. Now consider the additional ‘delay and decay’ impacts attributable to advective transport through a host rock of a repository once radionuclides have been released from the buffer. Figure 1.3 shows as smaller, lightercolored circles the combination of calculated advective-transport time (tat) added to the previously calculated diffusive-transport times through the buffer (tdt) for selected radionuclides. A simplistic advective-transport model is used to calculate transport time, assuming retarded transport of the same set of representative radionuclides through a generic host rock (JNC,

© Woodhead Publishing Limited, 2010

Multiple-barrier geological repository design/operation strategies

11

2000), further assuming a 100 m path length (L) from the edge of the buffer to some subvertical major water-bearing fracture zone, a rock porosity (ε) of 0.01, a bulk density (r) of 2800 kg/m3 for the rock, and a flow rate (U) 1 m/ year: tat ¼

Lf1 þ ½ð1  eÞ=erKd g U

For a non-sorbing radionuclide with Kd = 0, the advective-transport time through 100 m of rock would be about 10 000 years, while for a strongly sorbing radionuclide with Kd = 1 m3/ kg, the advective-transport time through 100 m of rock would be about 38 000 years. The difference in position between the larger, darker-colored circles and the smaller, lighter colored circles in Fig. 1.3 represents the additional ‘delay and decay’ contribution attributable to retarded advective transport through this generic host rock (JNC, 2000). Note that while there is some contribution toward reducing the inventories of radionuclides with short and intermediate half-lives due to advective transport, it can be readily seen in Fig. 1.3 that even combined diffusive and advective transport cannot, by themselves, lead to effective delay and decay reductions in the inventory of all radionuclides, and especially radionuclides with half-lives greater than about 50 000 years. Of course, for different repository concepts and host rocks, input parameters affecting both diffusive- and advective-transport times can be different, possibly leading to more (or less) significant ‘containment’ capabilities for buffer and host rock. For fractured host rock there are additional processes (e.g. matrix diffusion (Neretnieks, 1980; Ahn, 1988)) that can retard even chemically non-sorbing radionuclides such as I-129 and act to further delay advective transport. For hydrologically saturated, lowpermeability clay and mudstone host rock in which only diffusive transport occurs, the entire thickness of the host rock formation could provide an extremely wide ‘diffusive transport barrier’. For example, the diffusivetransport time through a 100 m thick clay formation for a non-sorbing radionuclide with Kd = 0 would be about 127 000 years, with a diffusivetransport time of over 200 million years for a radionuclide with Kd = 1 m3/ kg.

1.4.3 Additional issues One concern with ‘delay and decay’ approaches is that the decay of activity (mass inventory) for a given radionuclide leads to an increase in the activity (mass inventory) for its decay-daughter radionuclide, and this daughter radionuclide (or other, subsequent, grand-daughter nuclides) may present a greater radiological hazard than the parent radionuclide. For example,

© Woodhead Publishing Limited, 2010

12

Geological repository systems for safe disposal

ORIGEN-2 calculations of radioactive decay in a 60 000 MWd/MTU pressurized water reactor spent fuel show that the initial activity of Ra226 increases approximately by a factor of 26105 after 1000 years of containment, and by a factor of about 76107 after 100 000 years of containment (Roddy et al., 1986). Recent safety assessments for the disposal of spent fuel in long-lived copper canisters emplaced in fractured granite (SKB, 2006; Posiva, 2007) show that uranium-series daughter radionuclides (Th-230 and Ra-226) and neptunium-series daughter radionuclides (Th-229) can actually become significant contributors to the calculated dose rate for such repository concepts, despite the fact of the trivial initial abundances of these daughter radionuclides in spent fuel (Roddy et al., 1986). There are additional mitigating factors to be considered in evaluating the safety impacts of ‘delay and decay’ barriers and processes. If, for example, the regulatory period for safety assessment is 100 000 years, then a mean canister lifetime of greater than 100 000 years would, by itself, provide assurance of regulatory compliance. Furthermore, there are physical processes, such as matrix diffusion, that can retard and delay the advective migration of even non-sorbing radionuclides such as I-129, as will be discussed in the next section. Also, an extended mean time of canister failure may imply a broadening in the overtemporal distribution of canister failure times, and extended path length of advective transport through the host rock may lead to significant lateral dispersion (hence, dilution) of transported radionuclides. These latter processes, however, can more correctly be placed into the second major strategy for waste isolation, constraints on concentration.

1.5

Constraints on concentration of radioactive materials

As argued in the previous section, containment (‘delay and decay’) strategies, via extended canister lifetimes or extended transport times through repository barriers, can be effective in reducing and even eliminating the initial inventory (hence, any potential release) of radionuclides with short and intermediate half-lives. The effectiveness of a containment-only strategy, however, is more problematic for longer-lived radionuclides. A second, post-containment/low-release period strategy based on additional processes is therefore needed to ensure successfully the safe disposal of nuclear waste containing long-lived radionuclides. During the low-release period, radionuclides will be released as dissolved species into groundwater contacting waste forms, followed by aqueous transport through the set of engineered and natural barriers of a repository system, potentially leading to radiological doses to future humans. Such

© Woodhead Publishing Limited, 2010

Multiple-barrier geological repository design/operation strategies

13

potential doses will be related to the concentration of dissolved radionuclides that may eventually reach the biosphere. Therefore, this second isolation strategy can be grouped under the collective term of ‘constraints on concentration’, also sometimes referred to as ‘dilute and disperse’. In the context of this overview, ‘constraints on concentration’ refers to any process or barrier that acts to reduce the concentrations of radionuclides as they migrate from one location to another location within the repository system.

1.5.1 Waste-form dissolution and radioelement solubility The most obvious and arguably the most important constraints on concentration arise when groundwater first contacts a waste form at the time of containment failure. Waste-form dissolution is driven by the fact that the groundwater is undersaturated* with respect to the waste-form components. The dissolution of the waste form leads to an increase in the concentration of radioactive and non-radioactive components in the groundwater contacting the waste-form surface, as schematically shown in Fig. 1.4. The dissolution rate (Rd) can be related to the time-dependent ion activity product† of dissolved components, Q(t), and the theoretical equilibrium constant for the dissolving waste form, Keq (Aagaard and Helgeson, 1982; NEA, 1993):   QðtÞ Rd ¼ kþ 1  Keq where k+ is the initial dissolution rate of the waste form when Q(t) = 0. Note that both k+ and Rd are typically expressed in units of mass of waste form released per unit surface area of exposed waste form per unit time. In an open system, where dissolved components are immediately transported away from the surface of a dissolving waste form, the dissolution rate remains constant, Rd = k+ (point A in Fig. 1.4). In a closed system, the concentrations (i.e. thermodynamic activities) of dissolved components of the waste form would theoretically increase * It can be somewhat confusing in repository-related literature that both hydrologists and geochemists use the term ‘saturation’, although with completely different meanings. Hydrologically, ‘saturation’ refers to the degree in which open pores of the rock are filled with water, whereas geochemically, ‘saturation’ refers to the necessary concentration of dissolved chemical species to cause the precipitation of a specific solid phase containing those species. † In thermodynamic analyses, chemical reactions are written in terms of effective concentrations or activities of dissolved species (ai), rather than their concentrations (ci). By definition, the extent that activity differs from concentration is expressed by the activity coefficient (γi), such that ai ¼ gi ci . For example, see Aagaard and Helgeson (1982) and Langmuir (1997) for more detailed discussions.

© Woodhead Publishing Limited, 2010

14

Geological repository systems for safe disposal

1.4 Schematic diagram of the evoluation of concentration of a representative radioelement i at the surface of a dissolving nuclear waste form under expected repository conditions.

ðQðtÞ ? Keq ; as t ? ?Þ until an equilibrium saturation (or ‘solubility limit’) is reached between the contacting water and the waste form (point C in Fig. 1.4). Note that at equilibrium, while the net change in concentration over time is zero, there is actually a dynamic balance in which the forward (dissolution) and reverse (precipitation) rates are equal but non-zero (NEA, 1993). Thus, there are two limiting rate processes potentially imposing constraints on concentration of released radionuclides: . .

rate of dissolution (Rd) of species from the waste-form surface (surfacereaction control) and rate of transport (Rt) of dissolved species away from the waste-form surface (near-field transport control).

Such limiting rate processes have long been recognized and applied in natural geochemical and diagenetic systems (e.g. Berner, 1978), as well as in chemical engineering systems (Chambre´ et al., 1982). However, which of these two processes is the dominant constraint or control on radionuclide concentrations released from waste forms in a repository? Detailed mass-transfer analyses of the relative importance of

© Woodhead Publishing Limited, 2010

Multiple-barrier geological repository design/operation strategies

15

surface reaction or near-field transport with respect to conditions of repository systems have been conducted (Chambre´ et al., 1982, 1988; KBS, 1983; NAS, 1983). These analyses incorporate the time-dependent dissolution rate of a waste form under repository conditions as a complex function of the waste-package geometry, initial waste-form dissolution rate, diffusion in the buffer*, advection in the host rock, and environmental conditions of the near field. In such analyses, it was found that if the dissolution rate is relatively greater than the diffusive transport rate away from the waste-form surface, the concentration of radioelement i will continue to increase at the wasteform surface until the solubility limit Cs is reached. Thereafter, it will be Cs that constrains the concentration of i at the waste-form surface, even though there may be continued dissolution of the waste form. Conversely, if dissolution is relatively slower than the diffusive transport away from the waste-form surface, then the concentration of a dissolved radioelement at the waste-form surface approaches a value of αCs, where α (in this case < 1) is the factor determined by the mass transfer around the waste form, inventories of radionuclide i and the waste form, and the dissolution rate of the waste form (Chambre´ et al., 1988). For most radioelements in waste forms in most repository conditions, the former case is observed (NAS, 1983; Chambre´ et al., 1988). This means that solubility limits imposed by radioelement-bearing solids will constrain concentration, and the dissolution rate of the waste-form matrix is not a long-term constraint. Figure 1.4 schematically shows the time-dependent evolution of concentration for a representative radionuclide i at the surface of a waste form under repository conditions. Initially at time tA there is a high dissolution rate (Rd = k+) because the contacting water is assumed to be totally undersaturated with respect to radionuclide i, causing a rapid increase in concentration of radionuclide i. At the longer time of tB, the dissolution rate decreases because of the gradual increase in dissolved wasteform components (i.e. QðtÞ ? Keq ; as t ? ?). Eventually, the solubility limit for a phase C containing radionuclide i is reached at time tC, and phase C will precipitate. Thereafter, the solubilitylimit of that phase C will set a fixed boundary condition for the concentration of radionuclide i at the waste-form surface. Note, however, that even after the solid phase that set solubility-limited concentration for radioelement i precipitates, the initial waste form will continue to dissolve. Also note that the time-dependent decrease in the dissolution rate shown in Fig. 1.4 illustrates why it would be excessively and unreasonably * For most deep geological repository concepts (Witherspoon and Bodvarsson, 2006), a lowpermeability buffer is emplaced around the waste canister, to ensure diffusion-limited transport conditions within the engineered barriers system (EBS).

© Woodhead Publishing Limited, 2010

16

Geological repository systems for safe disposal

conservative to extrapolate short-term (days to years) laboratory dissolution, or ‘leach rate’, data to predict radioelement-release behavior over repository-relevant time scales of hundreds of years or more (NAS, 1983).

1.5.2 Additional waste-form considerations Because radioelement solubility and the waste-form dissolution rate provide such fundamental constraints on radioelement concentrations, it is important to note additional considerations and alternative bounding cases: Metastability In the previous discussion of Fig. 1.4, it was assumed that phase C was the only stable solid phase that could precipitate from solution. Waste forms, such as high-level waste glasses, are thermodynamically unstable in water, and cannot re-precipitate from solution. Indeed, glass is thermodynamically unstable with respect to the potential formation of many different, compositionally related crystalline (mineral) solids that could precipitate when glass dissolves under natural conditions (Dibble and Tiller, 1981). As an illustration, consider the dissolution of borosilicate glass in which numerous polymorphic phases of SiO2(±H2O) could precipitate, each phase with a different solubility limit for the dissolved species H4SiO4. Which of these many silica polymorphs will precipitate to impose a solubility limit for H4SiO4? In natural, low-temperature (<100 8C) geochemical systems analogous to a deep geological repository, it is usually the thermodynamically least stable (hence, highest solubility) solids that precipitate first, in accordance with the Ostwald step-rule (Dibble and Tiller, 1981). The implication of metastability is illustrated in Fig. 1.4, where the initially precipitated phase C may be metastable, and if so, it might convert to more thermodynamically stable (hence, a lower solubility limit for radioelement i) phase D and then phase E over time; however, the time scale for such conversion is often not well known. Therefore, multiple lines of evidence, especially verification from natural and laboratory studies, should be considered in selecting and justifying specific solubility-limiting solids for radioelements. Shared solubility for radioelements It is technically correct to speak of ‘radioelement solubility’ rather than ‘radionuclide solubility’ because there may be multiple isotopes (both stable and radioactive) of a given radioelement present in the repository, with each radioactive isotope having a different, characteristic half-life (Ahn and Suzuki, 1993). A solubility-limiting solid phase for Am, for example, will

© Woodhead Publishing Limited, 2010

Multiple-barrier geological repository design/operation strategies

17

incorporate Am-241 (t1/2 = 241 years) and Am-243 (t1/2 = 7950 years). Thus, the elemental solubility limit for Am must be shared among all Am isotopes according to the individual isotope mass fraction in the solubilitylimiting solid phase. Note, however, that individual isotope mass fractions change over time because of their different half-lives, with the result that the effective solubility-limited concentration will increase over time for longerlived isotopes (in this case, Am-243), even though the overall Am solubilitylimited concentration is assumed to be time-independent. Low-solubility waste form Some waste forms under certain repository conditions may be characterized by extremely low dissolution rates, with fractional dissolution rates of 10–6/ year or lower (SKB, 2006), leading to the congruent release of radionuclides. Under simulated conditions of a repository in the reducing environment, for example, the dissolution rate, Rd, of the UO2 matrix of spent fuel is almost undetectable over laboratory time scales (e.g. Spahiu et al., 2004). Masstransfer analyses (Garisto and Garisto, 1988; Apted and Engel, 1988) have previously shown that for such exceptionally low dissolution rates, the concentration of radioelements at the surface of the dissolving UO2 matrix will be constrained, not by the solubilities of separate solubility-limiting solid phases but by the extremely low solubility of the UO2 matrix, normalized by the time-dependent mass fraction of each radionuclide contained in the matrix (Ahn, 2008). Inventory-limited release of radioelements The so-called ‘instant release fraction’ (IRF) is the small proportion of relatively volatile radioelements that migrate to and condense at UO2 grain boundaries and the gap between the UO2 matrix and cladding of spent nuclear fuel during reactor operations (Johnson and Tait, 1997). Upon canister failure, radioelements in the IRF (typically a few percentage of the overall inventory for such radioelements, the remaining inventory residing within the UO2 matrix) are assumed to dissolve instantaneously when contacted by groundwater, without consideration of any formation of solubility-limiting phases. Similar inventory-limited release behavior can also be ascribed to the so-called ‘wash-off’ fraction of radionuclides present in some low-level waste forms. Pulsed-release and spikes in radionuclide concentrations that arise for instant-release radionuclides can have important implications in evaluating overall repository performance.

© Woodhead Publishing Limited, 2010

18

Geological repository systems for safe disposal

High-solubility radioelements Some radioelements, especially those that form anions or anionic complexes in solution, have exceptionally high solubilities. For such high-solubility radioelements, no solubility limits are typically applied in repository performance assessments. Many of the IRF radionuclides, such as C-14, Cl-36, and I-129, have high solubilities under repository conditions. It is possible, however, that high-solubility radioelements may co-precipitate into solid phases formed by non-radioactive, major chemical components (see next item). Trace-element behavior and co-precipitation Essentially all radioelements other than U and possibly Pu are present at trace-element abundances (Langmuir, 1997). In nature (hence, under expected repository conditions), trace elements typically do not each form their own, discrete, solubility-limiting solid phase. Instead, trace elements are more likely to be co-precipitated within oxides, hydroxides, aluminosilicates, and carbonates composed of the non-radioactive, major elements present in natural groundwater systems (Bruno and Sandino, 1987; Langmuir and Apted, 1992; SKB, 2006). The discussion above has focused on the generalized release behavior and constraints on radionuclide concentrations for a single waste package. The next sets of concentration constraints consider the entire repository system, with ensembles of 100s to 10 000s of waste packages.

1.5.3 Temporally distributed containment failure Perhaps surprisingly, the first ensemble factor to be considered in attenuating concentration of released radionuclides is containment, or, more specifically, the temporal distribution of containment failures. As noted previously, upon containment failure there is an expected rapid release of the so-called instant release fraction (IRF) from spent fuel (Johnson and Tait, 1997). Highly soluble radionuclides such as C-14, Cl-36, and I-129 in the IRF typically display a ‘spike’ release upon the failure of a single canister. If all spent fuel canisters of a repository were to fail at exactly the same time, then these individual ‘spike’ releases of C-14, Cl-36, and I-129 would occur together and be additive. Because such highly soluble radionuclides also show little or no sorption behavior toward engineered or natural barriers through which they migrate, the ‘spike’ IRF release from spent fuel may be largely unretarded during transport. This could lead to an exceptionally high peak dose rate from such IRF radionuclides, if the highly

© Woodhead Publishing Limited, 2010

Multiple-barrier geological repository design/operation strategies

19

unlikely case of simultaneous failure of all spent fuel canisters and funneling of released radionuclides into the same transport path is assumed. In the seminal 1983 KBS-3 report (KBS, 1983) for Sweden, it was argued that because of uncertainties in parameters affecting the corrosion of copper canisters, there would be a distribution of canister failures over time. In the KBS report (1983) it was assumed that canister failures would be uniformly distributed, starting at a conservatively low value of 100 000 years and continuing up to the expected containment time of 1 000 000 years. Thus, the individual ‘spike’ releases of soluble, non-sorbing IRF radionuclides were likewise uniformly distributed over this same time interval, leading to a significant reduction in the calculated peak dose from such radionuclides (Apted and Engel, 1988). This approach to time-distributed containment failure leading to an attenuation in the concentrations of released IRF radionuclides was greatly advanced by the Canadian repository program. AECL (1994) showed that variation in the temperature (hence, corrosion rate of canisters) naturally arose from geometric consideration of the different regions of a panel of emplaced waste canisters. The corners of a rectilinear panel of waste canisters would reach a lower peak temperature and cool more quickly than canisters located along the edge of the panel, while canisters in the center portion of the repository panel would reach a higher peak temperature sustained for a longer period of time than canisters along the edge of the panel. The time-dependent distribution in canister failures was governed, therefore, by the natural variation in temperature–time histories of waste canisters, rather than from uncertainty. More recently spatial variation in groundwater composition has also been included as a defensible basis in deriving a distribution in containment failure for a repository system (EPRI, 2006; US DOE, 2008). These same analyses show that complex co-variations of temperature, relative humidity, and stress conditions naturally lead to different canister failure modes, and hence, distributed times of containment failures for the same canister material.

1.5.4 Spatially distributed containment failures Another ensemble process attenuating concentration of released radionuclide is consideration of the spatial distribution of failing waste packages. It has been shown (Ahn et al., 2002; Kawasaki et al., 2004) that the release plume of radionuclides from earlier failed canisters can act to reduce the release rate of radionuclides from adjoining waste canisters that fail at a later time. This attenuation arises because the concentration of radionuclides released from the earlier failed canister can overlap with an adjoining waste canister (or canisters) that fails subsequently. Thus, the pre-

© Woodhead Publishing Limited, 2010

20

Geological repository systems for safe disposal

existing concentration of released radionuclides reduces the concentration gradient controlling diffusive release of radionuclides from an adjoining waste canister when it fails at a later time. The consideration of the spatially distributed nature of canister failures and the impact on the overall radionuclide release is perhaps one of the most neglected aspects in repository safety assessment. Many repository safety assessments calculate the release from a typical failed waste canister and then scale that release proportionally by the total number of waste canisters in the repository. Even if temporal distributions in canister failures are considered, this simple approach of proportional scaling of the release behavior of a single waste canister to the entire ensemble of waste canisters is extremely conservative. This is because this simple scale-up approach considers each waste canister to be at the edge of the repository (i.e. the same distance from any subvertical fracture zone connecting the repository geosphere to the biosphere). In reality, waste canisters within wasteemplacement panels may have a wide distribution of far-field path lengths. While such natural variability in path length is not expected to provide any concentration constraints or lowering of the release of long-lived, solubilitylimited radionuclides, such path-length variability could lead to a significant lowering in the time-dependent release to the biosphere of long-lived, nonsolubility-limited IRF radionuclides that display ‘spike’ release from individual waste canisters. In addition to the wide distribution of the path length, spatial distribution of the canisters in a repository would result in greater cross-sectional areas of the radionuclide transport path in the geosphere. It would be unlikely that radionuclides released from the repository are funneled into a constrained path. In the conventional simple proportional scaling approach, even with the temporal distribution taken into account, this effect is not evident. While for the conservative safety assessment purpose, where a bounding case is considered, this approach can still provide useful information, it wipes out important sensitivity in the repository performance with the size (or capacity) of the repository, and is not suitable for a repository design and optimization exercise (Murakami and Ahn, 2008).

1.5.5 Far-field transport Transport of radionuclides through the near-field buffer and far-field geosphere, including sorptive retardation, was previously discussed under ‘delay and decay’ and found to effective only for short and intermediate half-life radionuclides. For long-lived radionuclides that are solubility controlled, sorption in the far field only delays the time to attain a steadystate release of such radionuclides from the repository, but does not affect the peak release. This is because eventually an equilibrium between

© Woodhead Publishing Limited, 2010

Multiple-barrier geological repository design/operation strategies

21

sorption/ desorption is reached throughout the repository system before there is significant radioactive decay of long-lived radionuclides. For ‘spike’ releases of IRF radionuclides, however, the characteristic sharp concentration fronts are naturally dispersed by advective and diffusive transport (Domenico and Schwartz, 1998), leading to a lowering of the peak concentration that is progressively more effective with longer transport times and path lengths. Natural dispersion of spike releases can also be enhanced by chemical sorption, although many of the key, long-lived IRF radionuclides form anionic solution species that have extremely low or zero sorption behavior. Matrix diffusion (Ahn, 1988; Domenico and Schwartz, 1998) is an additional retardation process that can occur in fractured host rocks. Dissolved radionuclides diffuse into the water-filled micro-cracks of the rock matrix, located along the primary fracture-flow pathways and driven by chemical concentration gradients. Because matrix diffusion is a physically based process (i.e. does not rely on chemical bonding), it can be effective in both delaying the transport time of even anionic species such as C-14, Cl-36, and I-129 and further enhancing natural dispersion and lowering the concentration of IRF radionuclides before they reach the biosphere. Further constraints on concentration of radionuclides prior to their release into the biosphere can occur from possible mixing and dilution within overlying aquifers (JNC, 2000) and the isotopic dilution due to the presence of naturally occurring, stable, and radioactive isotopes for elements such as C, Cl, Ni, Se, Sn, Cs, I, U, and Th. The need for and allowance of such factors such as aquifer mixing and isotopic dilution depends, however, on the formulations and requirements embedded in safety standards for deep geological repositories, which vary among different countries.

1.5.6 Cumulative effect of constraints on concentration Figure 1.5 schematically shows the sequential and cumulative effects that various constraints on concentration can have on the eventual dose rate received to a future population during the low-release period of a deep geological repository. While dose rates are evaluated in a total system performance assessment with the radionuclide concentrations arising at the geosphere–biosphere boundary, it is possible to apply dose models conceptually to the concentrations arising at any portion of the pathway for radionuclide release shown in Fig. 1.5, to observe effects of each barrier. For example, in the Kristallin-I safety assessment (Nagra, 1994), a so-called ‘robust scenario’ was calculated by applying dose modeling to the release of radionuclides from the EBS, conservatively ignoring all of the isolation contributions arising from far-field transport through the fractured granite geosphere. This ‘robust scenario’ analysis showed that radiological dose

© Woodhead Publishing Limited, 2010

22

Geological repository systems for safe disposal

1.5 Schematic representation of sequential and cumulative effects of multiple constraints on radioelement concentrations that commonly occur for a deep geological repository.

limits were met at the EBS–host rock boundary, illustrating both the immense isolation capability of the EBS as well as the significant reserves of isolation performance and protection that are attributable to the combined EBS and natural barrier system of the repository. This robust repository performance is schematically represented in Fig. 1.5, showing compliance with regulatory safety standards occurring at the point of radionuclide release from the EBS to the geosphere. Another way to illustrate the impact of constraints on concentration on the performance of a repository is to examine the distribution of mass of dose-contributing radionuclides over time and space. Figure 1.6 (Nagra, 2002) presents an analysis of long-lived Np-237, which is often a key contributor to peak dose rates calculated for many repository systems (e.g. SKB, 2006; Posiva, 2007; US DOE, 2008). In Fig. 1.6, the release and migration of Np-237 from a single waste package failing 1000 years after emplacement is presented. The mass of Np-237 located within several spatial compartments of the repository is calculated as a function of time, and

© Woodhead Publishing Limited, 2010

Multiple-barrier geological repository design/operation strategies

23

1.6 Calculated distribution of Np-237 as a function of time and space in a deep geological repository (adapted from Nagra, 1994).

normalized by the initial mass of Np-237 (Nagra, 2002). Several key points are worth observing. First, precipitation of the solubility-limiting phase controlling the longterm concentration of Np-237 at the waste-form surface occurs rapidly. Although the initial borosilicate glass waste form does not completely dissolve until approximately 100 000 years after emplacement, the dissolution behavior of the borosilicate glass during this time period has absolutely no impact on controlling the release of Np-237 from the EBS once the precipitate is formed. Second, the growth in the mass of Np-237 in the precipitate confirms that as the initial mass of Np-237 is released from the dissolving waste form, this mass of Np-237 remains trapped in the newly formed precipitated solid phase. Indeed, essentially all of the initial Np-237 remains trapped within the EBS as a precipitate up until the time that Np-237 effectively decays to insignificance (about 21 million years after emplacement, equal to 10 times the 2.146106 years half-life). Third, Fig. 1.6 shows that about 0.05 to 0.1 % of the initial Np-237 mass remains sorbed on to the bentonite (i.e. within the buffer portion of the EBS) over the entire low-release period and approximately 0.005 % of the initial Np-237 mass (or activity) remains within the geosphere. Ultimately, only 10–5 of the initial Np-237 mass (activity) is released to the biosphere

© Woodhead Publishing Limited, 2010

24

Geological repository systems for safe disposal

many million years in the future and the release of that mass of Np-237 is distributed over a 20-million year period (i.e. an average annual release rate on the order of 5610–13 of the initial mass/year). Over 99.999 % of Np-237 is never released from the repository system. Clearly, the 1000-year containment time has no applicable effect on long-lived Np-237; this enormous degree of isolation performance is attributable to the various constraints on concentration that occur throughout the repository system. Similar observations were also made by an analysis with a multiple-canister repository configuration (Ahn et al., 2002).

1.6

Summary

A brief review of isolation strategies based on containment and constraints on concentration relevant to conceptual repository design and safety assessment have been presented. Conceptual repository designs are typically based on the inclusion of several barriers and processes representative of these two strategies. The actual selection of which barriers and processes to rely upon are guided by several considerations: . . . . . . .

type and radionuclide inventory of nuclear waste to be disposed, regulatory requirements, especially with respect to the time period and metric for assessing safety, type of geological host rock and hydrological setting, effectiveness of a given barrier/process in achieving long-term safe isolation, reliability of a given barrier/process in assuring long-term performance, feasibility of constructing, emplacing waste, and safety monitoring for the conceptual repository design, and natural-event phenomena that are likely to occur at the repository site (e.g. climate change/glaciation, seismicity, volcanism, uplift/subsidence, etc.).

The initial context for developing repository concepts is based on the relative radiological hazard of the radioactive waste to be disposed, as well as the legal framework and regulatory requirements for implementing a repository program. Regulatory requirements, in particular, often define the process by which a geological repository is to be developed, assessed, and licensed as a nuclear facility with respect to assuring the protection of public health and safety. Building on this initial context, the type of host rock will have a significant influence on the type, number, and assigned safety functions of engineered barriers. Certain geological formations can be called highisolation sites, because of either the absence of water to provide a medium for radionuclide transport (e.g. bedded salt, salt domes) or an exceedingly

© Woodhead Publishing Limited, 2010

Multiple-barrier geological repository design/operation strategies

25

low rock-matrix permeability and the absence of fractures (e.g. bedded clay, mudstone), which assures slow, diffusive transport over reasonably long (> 50 meters) distances to the near-surface environment. Such high-isolation sites typically do not require or greatly rely upon engineered barriers to assure long-term isolation. Other geological formations can be called hydrological flow sites, in which groundwater (hence, dissolved radionuclides) at the repository depth moves through open pores and/or fractures at slow rates over long distances before reaching the near-surface environment (e.g. fractured granite, fractured tuff, sedimentary rock formations). For such hydrological flow sites, the relative roles and contributions of host rock and engineered barriers have been summarized (SKB, 1992): The . . . safety assessment shows that a repository constructed deep down in . . . crystalline basement with engineered barriers possessing long-term stability fulfills the safety requirements proposed by the [regulatory] authorities with ample margin. The safety of such a repository is only slightly dependent on the ability of the surrounding rock to retard and sorb leaking radioactive materials. The primary function of the rock is to provide stable mechanical and chemical conditions over a long time period so that the long-term performance of the engineered barriers is not jeopardized. As in all engineering enterprises, the attributes and isolation capabilities of the man-made and natural components of a repository system must be assessed and appropriately balanced. The design must also be feasible to construct, emplace waste, and monitor safety during the operational period. It is worth noting that mature repository programs in countries such as Sweden, Finland, and the US now entering into a licensing phase have experienced evolution in their design concepts. Such evolution has been an iterative process motivated by a number of factors, including the collection of new site characterization data, revisions to regulatory standards, adaptation of designs to pre-closure and post-closure issues, and continuing technical oversight and regulatory review regarding the likelihood and possible consequences of natural-event scenarios.

1.7

References

Aagaard P and Helgeson H (1982), ‘Thermodynamics and kinetic constraints on reaction rates among minerals and aqueous solutions: I. Theoretical considerations’, American Journal of Science, 282, 237–285. AECL (1994), ‘The disposal of Canada’s nuclear fuel waste: the Vault Model for postclosure assessment’, AECL-10714, COG-93-4, Atomic Energy of Canada

© Woodhead Publishing Limited, 2010

26

Geological repository systems for safe disposal

Limited, Whiteshell Nuclear Research Establishment, Pinawa, Manitoba, Canada. Ahn J (1988), ‘Mass transfer and transport of radionuclides in fractured porous rock’, PhD dissertation, University of California, Berkeley. Ahn J (2008), ‘Effects of repository conditions on environmental impact reduction by recycling’, in 10th OECD/NEA Information Exchange Meeting on Actinide and Fission Product Partitioning and Transmutation, 6–10 October 2008, Mito, Japan, OECD/NEA. Ahn J and Suzuki A (1993), ‘Diffusion of the 241Am → 237Np decay chain limited by their elemental solubilities in artificial barriers of high-level radioactive waste repositories’, Nuclear Technology, 101(1), 79–91. Ahn J, Kawasaki D and Chambre´. P (2002), ‘Relationship among performance of geological repositories, canister-array configuration, and radionuclide mass in waste’, Nuclear Technology, 140, 94–112. Andra (2005), ‘Synthesis of an evaluation of the feasibility of a geological repository in an Argillaceous Formation’, in Dossier 2005 Argile, National Agency for Radioactive Waste Management, Chaˆtenay-Malabry, France. Apted M (1995), ‘Robust EBS design and source-term analysis for the partially saturated Yucca Mountain Site’, in Proceedings of the 4th Annual International Conference on High-level Radioactive Waste Management, vol. 2, Las Vegas, Nevada, pp. 485–490. Apted M and Engel D (1988), ‘Analysis of congruent matrix release, precipitation and time-dependent containment failure on spent fuel performance,’ in Scientific Basis for Nuclear Waste Management XI, edited by M Apted and R Westerman, Material Research Society, Pittsburgh, Pennsylvania, pp. 303– 312. Berner R (1978), Early Diagenesis: A Theoretical Approach, Princeton University Press, Princeton, New Jersey. Bruno J and Sandino A (1987), ‘Radionuclide co-precipitation’, SKB-TR-87-23, Swedish Spent Fuel and Nuclear Waste Management Company, Stockholm, Sweden. Chambre´ P, Pigford T H, Fujita A, Kanki T, Kobayashi A, Lung H, Ting D, Sato Y and Zavoshy S J (1982), ‘Analytical performance models for geologic repositories’, LBL-14842, Lawrence Berkeley Laboratory, California. Chambre´ P, Kang C, Lee W and Pigford T (1988), ‘The role of chemical reaction in waste-form performance’, in Scientific Basis for Nuclear Waste Management XI, edited by M Apted and R Westerman, Material Research Society, Pittsburgh, Pennsylvania, pp. 285–291. Chapman N and McCombie C (2003), Principles and Standards for the Disposal of Long-lived Radioactive Wastes, Pergamon Press, London, UK. Dibble W and Tiller W (1981), ‘Kinetic model of zeolite paragenesis in tuffaceous sediments’, Clay and Clay Minerals, 29, 323–330. Domenico P and Schwartz F (1998), Physical and Chemical Hydrology, John Wiley & Sons, Inc., New York. EPRI (2006), ‘Program on technology innovation: EPRT Yucca Mountain spent fuel repository evaluation – 2006 Progress Report’, Report 1013445, Electric Power Research Institute, Palo Alto, California. Garisto N and Garisto F (1988), ‘Mass-transport precipitation coupling in finite

© Woodhead Publishing Limited, 2010

Multiple-barrier geological repository design/operation strategies

27

systems’, AECL-9562, Atomic Energy of Canada Limited, Whiteshell Nuclear Research Establishment, Pinawa, Manitoba, Canada. IAEA (2005), ‘Multilateral approaches to the nuclear fuel cycle’, Expert Group Report to the Director General of the International Atomic Energy Agency, International Atomic Energy Agency, Vienna. IAEA (2009), ‘Classification of radioactive waste: general safety guide’, IAEA Safety Series GSG-1, International Atomic Energy Agency, Vienna. JNC (2000), ‘H12: Project to establish the scientific and technical basis for HLW disposal in Japan’, Japan Nuclear Cycle Development Institute, Tokai-Mura, Naka-Gun, Ibaraki, Japan. Johnson L and Tait J (1997), ‘Release of segregated nuclides from spent fuel’, SKBTR-97-18, Swedish Spent Fuel and Nuclear Waste Management Company, Stockholm, Sweden. Kawasaki D, Ahn J, Chambre´ P L and Halsey W G (2004), ‘Congruent release of long-lived radionuclides from multiple canister arrays’, Nuclear Technology, 148, 181–193. KBS (1983), ‘Final storage of spent nuclear fuel: KBS-3, SKBF/KBS Report KBS-3, Swedish Nuclear Fuel Supply Company, Stockholm, Sweden. Langmuir D (1997), Aqueous Environmental Geochemistry, Prentice Hall, Upper Saddle River, New Jersey. Langmuir D and Apted M (1992), ‘Backfill modifications using geochemical principles to optimize high-level nuclear waste isolation’, in Scientific Basis for Nuclear Waste Management XV, edited by C. Sombret, Material Research Society, Pittsburgh, Pennsylvania, pp. 13–24. MacFarlane A M and Ewing R C (eds) (2006), Uncertainty Underground: Yucca Mountain and the Nation’s High-Level Nuclear Waste, ISBN-10: 0-262-13462-4, ISBN-13: 978-0-262-13462-0. Murakami H and Ahn J (2008), ‘Development of compartment models for radionuclide transport in repository region’, in 12th International High-Level Radioactive Waste Management Conference (IHLRWM), 7–11 September 2008, Las Vegas, Nevada, American Nuclear Society. Nagra (1994), ‘Kristallin-I: conclusions from the Regional Investigation Programme for siting a HLW repository in crystalline basement of Northern Switzerland’, Nagra NTB 93-09E, Wettingen, Switzerland. Nagra (2002), Project Opalinus Clay: Safety Report: demonstration of disposal feasibility for spent fuel, vitrified high-level waste and long-lived intermediatelevel waste’, Nagra NTB 02-05, Wettingen, Switzerland. NAS (1957), The Disposal of Radioactive Wastes on Land, US National Academy of Sciences/National Research Council, National Academy Press, Washington, DC. NAS (1983), A Study of the Isolation System for Geologic Disposal of Radioactive Waste, US National Academy of Sciences/National Research Council, National Academy Press, Washington, DC. NAS (1990), Rethinking High-Level Radioactive Waste Disposal, US National Academy of Sciences/National Research Council, National Academy Press, Washington, DC. NAS (1995), Technical Bases for Yucca Mountain Standards, US National Academy

© Woodhead Publishing Limited, 2010

28

Geological repository systems for safe disposal

of Sciences/National Research Council, National Academy Press, Washington, DC. NAS (2001), Disposition of High-Level Waste and Spent Nuclear Fuel, The Continuing Societal and Technical Challenges, US National Academy of Sciences/National Research Council, National Academy Press, Washington, DC. NAS (2003), One Step at a Time: The Staged Development of Geologic Repositories for High-Level Radioactive Waste, US National Academy of Sciences/National Research Council, National Academy Press, Washington, DC. NEA (1977), Objectives, Concepts and Strategies for the Management of Radioactive Waste Arising from Nuclear Power Programmes, Nuclear Energy Agency, Organization for Economic Co-operation and Development, Paris, France. NEA (1991), Disposal of Radioactive Waste: Review of Safety Assessment Methods, Nuclear Energy Agency, Organization for Economic Co-operation and Development, Paris, France. NEA (M Apted, ed.) (1993), The Status of Near-field Modelling: Proceedings of a Technical Workshop – Cadarache, France, 11–13 May 1993, Nuclear Energy Agency, Organization for Economic Co-operation and Development, Paris, France. NEA (2002), Accelerator-Driven Systems (ADS) and Fast Reactors (FR) in Advanced Nuclear Fuel Cycles, A Comparative Study, Nuclear Energy Agency, Organization for Economic Co-operation and Development, Paris, France. Neretnieks I (1980), ‘Diffusion in the rock matrix: an important factor in radionuclide retardation?’, Journal of Geophysical Research, 85, 4379. Posiva (2007), ‘Safety assessment for a KBS-3H spent nuclear fuel repository at Olkiluoto: Summary Report’, Posiva 2007-06, Posiva Oy, Eurajoki, Finland. Roddy J, Claiborne H, Ashline R, Johnson P and Rhyne B (1986), ‘Physical and decay characteristics of commercial LWR spent fuel’, ORNL/TM-9591, Oak Ridge National Laboratory, Oak Ridge, Tennessee. SKB (1992), ‘SKB’91: final disposal of spent nuclear fuel: importance of the bedrock for safety’, SKB-TR-92-20, Swedish Spent Fuel and Nuclear Waste Management Company, Stockholm, Sweden. SKB (2006), ‘Long-term safety for KBS-3 repositories at Forsmark and Laxemar – a first evaluation’, SKB-TR-06-09, Swedish Spent Fuel and Nuclear Waste Management Company, Stockholm, Sweden. Spahiu K, Cui D and Lundstro¨m M (2004), ‘The fate of radiolytic oxidants during spent fuel leaching in the presence of dissolved near-field hydrogen’, Radiochimica Acta, 92, 625–629. US DOE (1995), Total System Performance Assessment. US DOE (2008), License Application. Verstricht J and DeBruyn D (2000), ‘Belgian concept for HLW disposal: development and demonstration’, in Waste Management 2000, Tucson, Arizona. Witherspoon P and Bodvarsson G (2006), ‘Geological challenges in radioactive waste isolation: Fourth Worldwide Review’, LBNL-59808, Lawrence Berkeley National Laboratory, Berkeley, California.

© Woodhead Publishing Limited, 2010

2 Spent nuclear fuel recycling practices, technologies and impact on geological repository systems M . S . Y . C H U , M. S. Chu & Associates, LLC, USA

Abstract: This chapter discusses the potential impacts of spent nuclear fuel recycling on geological repositories. This chapter first provides background information on spent nuclear fuel management practices in the US and internationally, and then discusses the various advanced recycling technologies that are being used or considered and finally the potential impacts of these recycling technologies on the disposal of high-level wastes in geological repositories. Key words: spent nuclear fuel, waste management, recycling technologies, disposal, geologic repository.

2.1

Background and introduction

Essential to recycling of spent nuclear fuel is the separation of fissionable material from the spent nuclear fuel to reuse it in new reactor fuel. In the United States, spent nuclear fuel and high-level radioactive waste have been produced and accumulated since the 1950s. Nuclear fuel reprocessing technology was developed by the US government during the Manhattan project while building atomic bombs. Large-scale reprocessing plants began at nuclear weapons production sites in the Hanford site, Washington, and the Savannah River site in South Carolina. In 1956, the US Atomic Energy Commission (AEC), the predecessor agency of the US Department of Energy (DOE) and the US Nuclear Regulatory Commission (NRC) announced a program to encourage private industry to begin reprocessing spent nuclear fuel. Commercial reprocessing first started at the West Valley plant in New York in 1966. Around the same period of time, several other commercial entities attempted to reprocess waste; however, for a variety of reasons (economical as well as political) all 29 © Woodhead Publishing Limited, 2010

30

Geological repository systems for safe disposal

commercial reprocessing activities were terminated by the early 1970s. In 1977, President Carter announced that the United States would defer commercial reprocessing indefinitely due to proliferation risks associated with separating plutonium from spent nuclear fuel. Although President Reagan lifted this indefinite ban on commercial reprocessing in 1981, no commercial reprocessing facilities have been developed since the 1970s due to economic as well as political reasons. Past reprocessing in the US generated millions of gallons of highly radioactive liquid waste. These liquid high-level wastes have been stored in tanks in the Hanford and Savannah sites, and efforts have been under way to vitrify the waste into glass logs in preparation for the final disposal in a geological repository. Currently, nuclear energy supplies about 20% of the United States’ electricity needs and spent nuclear fuel and high-level wastes are stored temporarily at 121 commercial and government sites in 39 states in the US. A commercial spent nuclear fuel inventory, which includes spent nuclear fuel generated by 104 operating reactors and 14 reactors that are no longer in operation, is currently estimated to be 58 000 metric tons of heavy metals (MTHM). This inventory is estimated to be increasing by approximately 2000 MTHM per year from the current operating reactors. As of January 2007, 47 license extensions have been granted to the existing nuclear reactors for 20-year license extensions. It is estimated that the cumulative inventory of spent nuclear fuel will be 109 300 MTHM at 2040. If all 104 operating reactors received license extensions, then the inventory at 2050 would be about 130 000 MTHM. Recently, the US Nuclear Regulatory Commission (NRC) has indicated that between 2007 and 2010, 23 new license applications (34 total numbers of units) from nuclear industries are expected. With this new wave of potential reactors, the inventory of spent nuclear fuel will grow significantly in the next few decades. The issues of the ultimate disposal of spent nuclear fuel and radioactive waste have been the subject of scientific studies and political debates for many decades around the world. In the US, the Nuclear Waste Policy Act (NWPA) was enacted by Congress in 1982 and establishes a process for the siting, construction, and operation of geological repository for the permanent disposal of spent nuclear fuel and high-level waste. The NWPA also designated the Federal government with the responsibility for the disposal of commercial spent nuclear fuel. Between 1982 and 1987, a suite of potential sites around the country was selected and studied. In 1987, Congress amended the NWPA and selected the Yucca Mountain site, Nevada, as the only site for further study and characterization. Since then, extensive site characterization, experimental and modeling investigations, and engineering designs have been performed at the Yucca Mountain site.

© Woodhead Publishing Limited, 2010

Spent nuclear fuel recycling

31

In 2002, Congress passed a joint resolution that approved the DOE to prepare for a license application for Yucca Mountain as the disposal site. On 3 June 2008, the Department of Energy submitted the license application to the NRC, seeking construction authorization for the repository at Yucca Mountain (DOE, 2008a). The NWPA has a provision that sets a statutory limit on the amount of spent nuclear fuel and high-level waste that can be disposed of at Yucca Mountain to be 70 000 MTHM. DOE estimates that with the current and projected inventory of spent nuclear fuel, the limit for the Yucca Mountain repository will be reached by approximately 2010. It is apparent, given the future trend of nuclear energy, that the demand for disposal capacity will far exceed the 70 000 MTHM limit by law. Recently DOE submitted a report to Congress expressing the need to either develop a second repository or to remove the statuary disposal limit of 70 000 MTHM (DOE, 2008b). Although the US has not been reprocessing commercial spent nuclear fuel, the government has been conducting research activities in the recycling of spent nuclear fuel. In particular, DOE initiated the Advanced Fuel Cycle Initiative (AFCI) in 2002 to advance recycling technologies in separating and transmutating long-lived radionuclides to reduce radiotoxicity in spent nuclear fuel. Furthermore, in 2006, as part of President Bush’s Advanced Energy Initiative, DOE launched the Global Nuclear Energy Partnership (GNEP) program (DOE, 2007). The goal of the GNEP is to recycle used nuclear fuel to assure maximum energy recovery, to reduce proliferation concerns, to allow developing countries safely and securely to deploy nuclear power, and to reduce the volume and toxicity of wastes that require deep geological repository disposal. It is a long-standing US national security policy objective to reduce proliferation risks throughout the nuclear fuel cycle through comprehensive efforts to prevent the risks of nuclear weapons materials. An important element of the GNEP program is to create a safe, orderly system to support commercial nuclear power internationally without adding to the dangers of weapons proliferation. As a result, the GNEP emphasizes advanced reprocessing technologies that aim to reduce the proliferation potential associated with the weapons-usable materials inherent in the nuclear fuel cycle, while acquiring nuclear energy economically. In the studies of nuclear fuel cycle options, open or closed, it is recognized that residual nuclear waste will be generated that requires final disposal. The National Research Council in 1957 recommended that deep geological isolation would be a suitable approach for disposal (NRC, 1957). Other nations have adopted the same policy since then. However, no nation yet has a fully functioning geological disposal repository for high-level waste. In the US, the Waste Isolation Pilot Plant in New Mexico has an operating

© Woodhead Publishing Limited, 2010

32

Geological repository systems for safe disposal

geological repository in salt for the disposal of defense transuranic (TRU) waste.

2.2

Current spent nuclear fuel reprocessing technologies

The PUREX (plutonium–uranium extraction) process has been the reprocessing technology for the past 40 years. It was first developed in the US for separating pure plutonium for military purposes. It is a solvent extraction process using the extractant tributylphosphate (TBP) mixed in a largely inert hydrocarbon solvent. Since the opening of the first PUREX plant at the Savannah River site in 1954, the PUREX process has been utilized in a variety of flowsheets and is still being used in all commercial reprocessing plants currently operating in the world. This technology has been improved rapidly to adapt continuously to the evolving characteristics of spent nuclear fuel and other constraints such as regulatory requirements. For example, national and international regulations are more stringent regarding safety and security, and waste released and public dose exposure limits have been lowered. Commercial PUREX reprocessing in general consists of four main technological operations: fuel handling and shearing, fuel dissolution, materials separation and purification, and waste treatment and conditioning (IAEA, 2005). At present PUREX reprocessing produces two types of waste: . .

waste from the process itself in the form of a liquid solution of fission products and actinides, and waste comprising hulls and end fittings from the structure of the fuel, insoluble fission products, and other operational residues.

The first type of liquid waste is usually concentrated and then vitrified, and is classified as high-level waste ready to be disposed of in a deep geological repository. The second type of waste can be compacted into containers and usually stored for further disposition. Several countries currently have operating reprocessing plants. The La Hague reprocessing plants UP2 and UP3 in France have been reprocessing spent nuclear fuel for decades using the PUREX technology. La Hague has a capacity of 1700 MTFM per year. The Thermal Oxide Reprocessing Plant (THORP) at Sellafield in the United Kingdom uses the PUREX technology with a capacity of 1200 MTHM. Japan has a small reprocessing plant at Tokai-mura and is beginning operation of a reprocessing plant at Rokkasho-mura with a capacity of 800 MTHM per year. Russia has a 400 MTHM/year commercial reprocessing plant at Mayak. India has been

© Woodhead Publishing Limited, 2010

Spent nuclear fuel recycling

33

reprocessing spent nuclear fuel since the 1960s on smaller research scales. China has a reprocessing demonstration plant and is beginning to design a commercial reprocessing plant with a capacity of 800 MTHM/year in the near future. In France and the United Kingdom, improvement in the operations of the PUREX process continues to reduce the amount of waste generated. Improved sorting procedures and increased package concentrations have allowed the volume of waste generated to be reduced. However, volatile radionuclides from the processes are removed by caustic scrubbing and then released to the sea. This practice is not permitted in the US.

2.3

Advanced spent nuclear fuel recycling technologies

In the early years of reprocessing in the US, the goal was to separate pure plutonium for use in nuclear weapons. In recent years, as interests in commercial nuclear energy grew, the focus has shifted to an emphasis on non-proliferation, decreasing losses of fissile and fertile elements to waste, and optimizing waste production for final disposal. The main issues with the PUREX process are twofold. First, a pure plutonium stream is separated and increases the proliferation risk. Second, minor actinides and heat-generating radionuclides go into the waste stream, increasing the radiotoxicity and volume of waste, as well as subutilizing the emplacement of waste in a geological repository because of the thermal effects. Since 2002, the Advanced Fuel Cycle Initiative (AFCI) from the US DOE focuses on the R&D of fuel cycles that could have substantial environmental, non-proliferation, and economic advantages over the once-through fuel cycle. Specifically, the AFCI focuses on (1) separation technologies for spent nuclear fuel, (2) transmutation technologies for minor actinides and long-lived fission products from spent nuclear fuel, and (3) development of advanced proliferation resistant reactor fuels that will consume plutonium. The goals of the R&D are (1) to remove over 90% of the uranium, (2) to remove over 99% of cesium and strontium, and (3) to separate the transuranic elements (plutonium, neptunium, americium, and curium) for recycling. Although there are several emerging R&D technologies being investigated internationally (IAEA, 2008), there are two key technology areas, as discussed below.

© Woodhead Publishing Limited, 2010

34

Geological repository systems for safe disposal

2.3.1 Advanced aqueous technologies A new generation of fuel reprocessing technologies is now in development under the DOE programs. In particular, a family of liquid reprocessing technologies derived from PUREX is being investigated. This family is called UREX+ technologies. UREX consists of a suite of processes where specific groups of radionuclides are removed to tailor products and compositions of the desired products and waste streams. All of these processes involve an initial extraction of uranium and technitium from the spent fuel dissolver solution. Other extraction steps are then added to meet the various separation requirements. These UREX+ processes use four main extraction steps to separate U, Tc, Cs/Sr, FP, and Pu/MA into six separate fractions. Additional extractions could also be used to partition Pu/ Np, Am, and Cm. In the UREX+ processes, U is denitrated from the aqueous stream and stored as an oxide powder for reuse. Tc is incorporated in a metallic alloy, Sr/Cs can be separated but may contain dilute nitric acid or some organic chemicals, lanthanides are separated and stabilized, and Pu and minor actinides (Np, Am, and Cm) are separated as a group for reuse. Many of the UREX+ processes under investigation are still at a laboratory scale and have not been optimized with respect to minimizing the number of separation cycles or optimizing separation efficiencies. Potential product and waste/storage materials are summarized below (see Figure 2.1). 1. 2. 3. 4. 5. 6.

Uranium: storage as U3O8. Cs/Sr aluminosilicate for storage. Tc in metallic alloy for disposal. Other fission products in vitrified high-level waste for disposal. Cladding hulls are compressed and disposed as HLW. Tritium, collected as tritiated water, may be grouted or encapsulated for disposal. 7. Xenon and krypton gases may be immobilized in zeolite and disposed as HLW. 8. Carbon-14 is recovered as CO2, converted to carbonate, and disposed as HLW. 9. Transuranics may be recycled for fuel fabrication. 10. Iodine may be trapped in silver-coated zeolite, converted to potassium iodate, and disposed as HLW. Continuing work is still under way to optimize the detailed flowsheets and steps.

© Woodhead Publishing Limited, 2010

Spent nuclear fuel recycling

35

2.1 UREX+ process and waste/storage products.

2.3.2 Advanced pyroprocessing technology Pyroprocessing is a non-aqueous process conducted in molten chloride salts that make use of electrochemical dissolution, selective reduction, and adsorption to partition groups of elements. Pyroprocessing is currently not being used commercially, but has been the subject of much R&D. Pyroprocessing technology was originally proposed by the Argonne National Laboratory as a process with the potential to treat all DOE spent nuclear fuel. Later it was used as a potential treatment method for sodiumbonded fuel. As applied to SNF reprocessing, a demonstration project has been studied at the Idaho National Laboratory for reprocessing of Experimental Breeder Reactor II (EBR-II) since 1996. In this demonstration project, chopped fuel rod elements are placed in a steel anode basket in an electrorefiner that contains a KCl–LiCl molten salt eutectic system at upwards of 500 8C. Upon passage of a constant electrolysis current between the anode basket and the steel cathode, U, Pu, transuranic elements (TRU), the alkalis and alkaline earth metals, and rare earths are oxidized into the molten salt. The stainless steel from the cladding, most of the Zr, and the noble metals remain in the anode baskets. The oxidized uranium (U+3) is reduced to the metal and deposited on to the cathode. After a given period of time of electrolysis, the U cathode is removed and uranium is cast into

© Woodhead Publishing Limited, 2010

36

Geological repository systems for safe disposal

metal ingot for future reuse or storage. The salt containing Pu and TRU elements, alkalis and alkaline earths, and some fission products is removed from the electrorefiner, mixed with zeolite, and then mixed with glass and hot pressed into a glass-bonded ceramic waste form. The material remaining in the anode basket is volatilized from the molten salt and cast into metal ingots as the metal waste form. An NAS committee evaluated this process and found no technical barriers in the reprocessing of EBR-II SNF using the pyroprocessing process (NRC, 2000). Specifically, the committee concluded that: . .

The physical and mechanical behavior of the ceramic waste form under repository conditions should be comparable to that of borosilicate glass. The corrosion rate of the metal waste form appears to be low and similar to the stainless steel and C22 alloy to be used in Yucca Mountain.

Pyroprocessing is most suitable for metallic spent fuel, but can also process oxide fuel after a reduction step. In an electrorefiner containing molten electrolytic salt, the anode will contain metals such as Tc, zirconium, iron, and molybdenum. One cathode will contain most of the uranium as metal, while the other cathode contains rare earth fission products plus TRU elements. Fission products such as Sr and Cs will remain in the molten salt. The uranium metal is converted to an appropriate form for reuse, and TRU metal is also reused for fuel fabrication. The metal left at the anode is heated in a metal waste furnace to produce a solid metallic waste form for final disposal. The fission-product-laden salt is circulated through a zeolite ion exchange bed and the zeolite matrix is consolidated into a monolithic ceramic waste form for disposal (Fig. 2.2). R&D is underway at the Idaho National Laboratory to optimize the pyroprocessing technology. Key R&D areas include separation of Sr/Cs from the salt so a separate waste stream can be produced, process modeling, and better actinide recovery. The Korea Atomic Research Institute is also conducting R&D on pyroprocessing technology.

2.4

Impacts of spent nuclear fuel recycling on geological disposal

The Yucca Mountain repository is used in this paper as an example for analyzing the potential impacts of recycling, since Yucca Mountain is a wellstudied and fully analyzed site. Current inventory destined for disposal at Yucca Mountain has a legislative limit of 70 000 MTHM. This inventory consists of 63 000 MTHM of spent nuclear fuel from commercial sites, and the remaining 70 000 MTHM would consist of about 2333 MTHM of DOE

© Woodhead Publishing Limited, 2010

Spent nuclear fuel recycling

37

2.2 Pyroprocessing flow chart.

spent fuel and high-level waste. Between 11 000 and 17 000 waste packages will be emplaced in the repository. The DOE submitted the License Application (LA) to the Nuclear Regulatory Commission in June of 2008. In the LA, a Total Systems Performance Assessment (TSPA) was used to calculate potential releases of radionuclides from the repository under all plausible scenarios for up to a million years after repository closure. The TSPA assesses the capabilities of various barriers at the Yucca Mountain site. The performance of the repository is controlled by the natural and engineered features of the site, which act in concert to prevent or reduce the movement of radionuclides to the accessible environment. Three barriers are important to the waste isolation at Yucca Mountain: the upper natural barrier, the engineered barrier system, and the lower natural system. Details of calculations of infiltration of water into the mountain under various climatic and hydrological conditions are used to develop models for the seepage of groundwater into emplacement drifts underground. These models are incorporated into the TSPA model. Models of the engineered barrier system use these seepage rates along with the chemical and thermal conditions at the repository to calculate the degradation rates of the various engineered systems, i.e. drip shield, waste packages, and the

© Woodhead Publishing Limited, 2010

38

Geological repository systems for safe disposal

waste forms. The release rates of radionuclides from the waste forms and waste packages into the unsaturated zone below the repository are then estimated. The flow of water movement within the unsaturated zone is then used to calculate the transport of radionuclides to the saturated zone 300 meters below the repository. Radionuclide concentrations at wells 20 km from the repository are then used as potential doses to individuals. Recently, the EPA finalized the environmental protection standards for Yucca Mountain (EPA 40 CFR 197, 2008). In this standard, compliance will be judged against a standard of 15 mrem/year dose at times up to 1000 years after disposal and against a standard of 100 mrem/year dose at times after 1000 years and up to 1 million years after disposal. The results of the TSPA in the LA shows that the radionuclides that dominate the annual doses typically have a combination of unique characteristics such as (1) large initial inventory in the waste, (2) moderate to high solubility, (3) long half-life (e.g. ≥ 105 years), and (4) low sorption in the transport paths. The radionuclides that are important to dose also depend on the time frame considered (i.e. 10 000 or 1 000 000 years) because of the effect of radionuclide decay and the effect of retardation from sorption along the flow path. Ingrowth of radionuclide through the decay chain can also be an important process that determines the role and importance of radionuclide in the actinium, uranium, neptunium, and thorium decay chains. In the License Application, results of compliance calculations show that the mean annual dose to an individual is 0.24 mrem for the first 10 000 years after disposal and the medium annual dose to an individual is 0.9 mrem for a time period up to one million years after disposal (peak dose at 720 000 years). The LA also shows that the main contributors to the mean annual dose for 10 000 years after closure are (ranked from highest to lowest): Tc99, C-14, Pu-239, I-129, Cl-36, Pu-230, Se-79, and Np-237. The single largest contributor is Tc-99, which accounts for 51% of the maximum mean dose. For the time period between 10 000 years and one million years, LA shows the main contributors at the peak mean annual dose are (from highest to lowest): Pu-242, Np-237, Ra-226, and I-129. These four radionuclides account for about 77% of the total mean dose. Although Ra-226 has a relatively short half-life (1600 years), it is sustained through chain decay of the longer-lived radionuclides of Th-230 (half-life of 7.546104 years) and U-234 (half-life of 2.466105 years). An important strategy for the Yucca Mountain repository concept is to manage temperatures within and between emplacement drifts to allow water always to drain freely in the rock between the emplacement drifts. At the Yucca Mountain design, there are severe temperature constraints at various locations underground. For example, the rock temperature midway between the drifts must always remain below 96 8C. This constraint ensures that any

© Woodhead Publishing Limited, 2010

Spent nuclear fuel recycling

39

water flowing downward through the mountain will flow through the rocks, preventing the retention of large volumes of water above the repository that potentially could flood the repository as it cools. A temperature limit of 200 8C is imposed on rock surrounding the repository to prevent alteration of its crystalline structure. For the current inventory destined for Yucca Mountain, the temperature limits are met by setting the emplacement drifts 81 meters apart while specifying a linear heat load of 1.45 kW per meter and a maximum thermal load of 11.8 kW per waste package (DOE, 2008a). These constraints result in a repository footprint of 1150 acres for the disposal of 70 000 MTHM of waste. These thermal requirements are intimately linked to the characteristics of the waste inventory. The Yucca Mountain inventory consists of 90% commercial spent nuclear fuels and 10% government waste (vitrified HLW and DOE spent nuclear fuels). The decay heat characteristics are thus largely controlled by commercial spent nuclear fuels and the thermal requirements are mostly derived from commercial spent nuclear fuel characteristics. In general, in the first three hundred years after disposal, the heat is dominated by Sr-90 and Cs-137, while later in time it is dominated by long-lived isotopes of actinide elements plutonium and americium, specifically Pu-241 and its decay daughter Am-241. As spent nuclear fuel is recycled, there are three major changes to the inventory of residual wastes that may need disposal: 1. 2. 3.

Radionuclide composition and characteristics. Heat generation characteristics. Physical and chemical characteristics of waste forms.

These changes in inventory will result in profound changes to the performance of the repository in terms of both potential releases of radionuclides and the design of waste emplacement in the repository. Wigeland et al. (2006) performed scoping analyses on the potential impacts of separation and transmutation on the utilization (emplacement of wastes) of Yucca Mountain. They used thermal performance at Yucca Mountain to establish separation and transmutation criteria for commercial spent nuclear fuel. In this study, Wigeland used models that are simpler than those used by the Yucca Mountain project. Wigeland used thermal models that are decoupled from hydrological effects by assuming key hydrologic conditions (such as rock saturation states and surface water infiltration rates) as input. Results from the simplified models were compared with the more complicated coupled models and good agreements were found. It was found that if 99.9 % of the plutonium and americium are removed from spent PWR fuel, it is possible to increase the linear loading in the drift from the reference value of 1.1 MTHM/m to about 5.9 MTHM/m. This loading increase can be interpreted as an increase in area loading of the repository

© Woodhead Publishing Limited, 2010

40

Geological repository systems for safe disposal

by a factor of 5.4. If the separation efficiency is lowered to 99 and 90%, the drift loading can be increased to factors of 5.3 and 4.3 respectively. Once plutonium and americium are mostly removed, the decay heat is due almost entirely to Cs and Sr. It is shown by Wigeland that if Cs and Sr can be removed at 99.9% efficiency after 99.9% of the plutonium and americium are removed, the drift loading can be further increased to 47.0 MTHM/m, a factor 42.7 higher than the reference case. Wigeland and Morris (2006) also studied the impact of reprocessing of SNF on potential peak dose rates from Yucca Mountain. A DOE simplified model was used to calculate the transport of radionuclides as they are released from waste packages. An analysis was performed to see the impact on peak dose if the entire SNF inventory was put in a glass waste form with no change in radionuclide contents. It is shown that the waste form change has a large impact on dose, where the peak dose is reduced to 50% of the reference case due to the reduced degradation rate of the glass waste form. It is shown by Wigeland and Morris that if 99.9% of the uranium is removed from the inventory, the peak dose is reduced to about 85% of the reference case. On other hand, if both plutonium and uranium can be removed from SNF at 99.9% efficiency, the peak dose rate can be reduced by a factor of 5.0. Furthermore, if all actinides can be removed at an efficiency rate of 99.9%, the peak dose rate can be reduced by a factor of about 90. The peak dose rate is then dominated by Tc-99 and I-129. It is concluded by Wigeland that it is possible to dispose of reprocessed waste from 7 000 000 MTHM of PWR spent fuel and have essentially the same peak dose rate as for 70 000 MTHM of PWR spent fuel. For this reprocessed inventory, the repository temperature would be slightly lower than the reference case. Overall, the removal of actinides at a separation efficiency of between 99 and 99.9% would offer a significant range of benefit for the utilization of a geological repository such as Yucca Mountain. Ahn (2007) analyzed the impacts of removal of actinides from commercial spent nuclear fuel on the environmental impact of Yucca Mountain. In this study, the indicator for environmental impact is measured by the ‘radiotoxicity’ of a radionuclide, which is defined as the volume of water needed to dilute the radionuclide to the permissible concentration for drinking. Simplifying assumptions are used in this study, such as all waste packages are assumed to fail at the same time and radionuclides are assumed to be released either congruently or with solubility limited. With this, the study shows that with a reduction of the TRU nuclide inventory by a factor of 100, the environmental impact (i.e. radiotoxicity) would be reduced by the same factor. The analyses by Wigeland and Ahn are based on generic reprocessing and the waste form was assumed to be similar to vitrified glass. As the advanced reprocessing technologies are being developed, more information on waste

© Woodhead Publishing Limited, 2010

Spent nuclear fuel recycling

41

forms will be available for further analysis. For pyroprocessing technology, two waste forms are generated: the metal waste form and the ceramic waste form. The ceramic waste form is expected to be very similar to borosilicate glass; however, the characteristics of the metal waste form may be quite different from the glass waste form. One interesting aspect of the metal waste form is that Tc-99 will be incorporated into the metal waste form from pyroprocessing, and technetium will be in a reduced form. This reduced form of Tc-99 may have a significantly lower release rate than the reference case, potentially reducing the peak dose from the repository. The metal waste form has undergone extensive qualification examination and testing for acceptance to the Yucca Mountain repository (Frank et al., 2007). The metal waste form being tested is stainless steel and zirconium alloys. The empirical model that predicts the overall corrosion behavior of the metal waste form shows that uniform aqueous corrosion is the main degradation mechanism for radionuclide release. The study also shows that the metallic waste form can be very robust in the retention of Tc and other fission materials.

2.5

Future trends

The current Nuclear Waste Policy Act (NWPA) and the NRC regulations contain rules that govern the management and disposal of all classes of nuclear waste, both high-level and low-level waste. These rules were derived from waste generated from activities pertinent at the time, and are based on the ‘origin of generation of waste’. With new types of waste generated from potential reprocessing in the US, new approaches to managing nuclear waste may be considered. Gombert (2008) laid out an integrated waste management strategy for the Global Nuclear Energy Partnership program. In this strategy, logical risk-based considerations were used for potential disposition paths for the various waste generated by both UREX and pyroprocessing reprocessing technologies. Gombert found that all wastes generated by processes being developed under GNEP and AFCI programs can be disposed of under the current regulations; however, significant efficiencies can be realized if the separated wastes can be dispositioned based on their characteristics and risks rather than their origins. A recent decision was made by the incoming new US administration of Obama to halt or delay the development of the Yucca Mountain repository. This decision could have very significant impact on the whole nuclear energy policy. Current law does not provide an alternative repository site to Yucca Mountain and it does not authorize the DOE to open temporary centralized storage facilities without a permanent repository in operation. Without congressional action, therefore, the default alternative to Yucca Mountain would be indefinite on-site storage of spent nuclear fuel at reactor sites. The

© Woodhead Publishing Limited, 2010

42

Geological repository systems for safe disposal

new administration also indicates that it is in favor of a closed fuel cycle policy and intends to convene a study to address reprocessing and the alternatives of nuclear waste issues. A National Academy of Sciences study of reprocessing technologies found that none of the separation and transmutation technologies would eliminate the need of a geological repository (NRC, 1995). With further R&D, however, if uranium, plutonium, and other long-lived radionuclides can be recycled back into a reactor in the future, the residual waste could contain only shorter-lived fission products and would be quite benign in terms of radiotoxicity. If this were the case, performance requirements for the final disposal at a repository could be much easier to achieve as the longterm uncertainties would be much eliminated. At the same time, alternative disposal options – e.g. alternative disposal sites – may be considered.

2.6

References

Ahn J (2007), ‘Deterministic assessment of environmental impact of Yucca Mountain Repository measured by radiotoxicity’, Journal of Nuclear Science and Technology, 44, 431. DOE (Department of Energy) (2007), ‘Global Nuclear Energy Partnership Strategic Plan’, GNEP-167312. DOE (Department of Energy) (2008a), ‘Yucca Mountain Repository License Application’, DOE/RW-0573. DOE (Department of Energy) (2008b), ‘The Report to the President and the Congress by the Secretary of Energy on the Need for a Second Repository’. Frank S M, Kaiser D D and Marsden K C (2007), ‘Immobilization of technetium in a metallic waste form’, Global. Gombert D (2008), ‘Global Nuclear Energy Partnership Integrated Waste Management Strategy’, GNEP-WAST-WAST-AI-RT-2008-000214, Department of Energy. IAEA (International Atomic Energy Agency) (2005), ‘Status and trends in spent fuel reprocessing’, IAEA-TECDOC-1467. IAEA (International Atomic Energy Agency) (2008), ‘Spent fuel reprocessing options’, IAEA-TECDOC-1587. NRC (National Research Council) (1957), The Disposal of Radioactive Waste on Land, National Academy Press, Washington, DC. NRC (National Research Council) (1995), Nuclear Waste: Technologies for Separations and Transmutation, National Academy Press, Washington, DC. NRC (National Research Council) (2000), Electrometallurgical Techniques for DOE Spent Fuel Treatment, National Academy Press, Washington, DC. Wigeland R A and Morris E E (2006), ‘Processing requirements for PWR spent fuel to reduce estimated peak dose rate associated with potential releases from a geologic repository’, Argonne National Laboratory, ANL-AFCL-166. Wigeland R A, Bauer T H, Fanning T H and Morris E E (2006), ‘Separations and transmutation criteria to improve utilization of a geological repository’, Nuclear Technology, 154 (April), 95–106.

© Woodhead Publishing Limited, 2010

3 Near-surface, intermediate depth and borehole disposal of low-level and short-lived intermediate-level radioactive waste I . G . C R O S S L A N D , Crossland Consulting, UK

Abstract: This chapter is concerned with the permanent disposal of radioactive waste in near-surface, intermediate-depth and borehole disposal facilities. The main styles of disposal are described and their advantages and disadvantages are explained in terms of their ability to meet the requisite safety standards and, hence, their suitability for housing the various categories of radioactive waste. Key words: radioactive waste, near-surface disposal, borehole disposal, intermediate depth disposal, long-term safety.

3.1

Introduction

3.1.1 Historical background to near-surface disposal Decades before deep geological disposal was conceived as an idea, radioactive waste was being generated by activities such as the production of radium for therapeutic uses, research into weapons and atomic power, and, beginning in the 1950s, power production itself. Many of these wastes were dumped at sea but, where the level of radioactivity was relatively low and a suitable site existed, disposal at the surface was also used. The Drigg low-level waste disposal site in the UK, for instance, was opened in 1959. For the most part these disposal facilities consisted of a simple trench into which the waste was tumble tipped. The prohibition of sea dumping by the London Dumping Convention, first as a two-year moratorium in 1982 and later as an outright ban, increased the volume of waste requiring disposal on land and created a need for more and larger disposal facilities. In some cases this led to closer scrutiny of existing near-surface facilities and a move away from simple trench systems to more engineered structures. Following a 43 © Woodhead Publishing Limited, 2010

44

Geological repository systems for safe disposal

critical report from a parliamentary committee in 1986, for instance, disposal at the Drigg site changed from tumble tipping to a concept in which the wastes are encapsulated in concrete. This evolution in engineering practice has allowed near-surface disposal to continue to occupy an important niche in radioactive waste management. In Sweden, the Forsmark facility, opened in 1988, pioneered the disposal of short-lived low and intermediate waste at so-called ‘intermediate depths’: typically tens of metres below grade. A number of similar facilities, both existing and planned, have followed.

3.1.2 Current role of near-surface and borehole disposal in the overall context of radioactive waste management Longstanding practice in Europe classified radioactive waste according to the amount of radioactivity that it contained. Three basic categories were defined: low-, intermediate- and high-level wastes (LLW, ILW and HLW respectively). These categories were primarily designed to reflect the degree of hazard associated with each, especially when handling or storing them. It soon came to be recognised, however, that waste management is made more efficient if the waste categorisation anticipates the intended disposal route and, because the radionuclide half-life is an important determinant of the long-term safety of a disposal facility, this needs to be taken into account in the waste categorisation. The existing IAEA waste categorisation system (IAEA, 1994a) recognises this by subdividing ILW into long- and short-lived subcategories where ‘long half-life’ is taken to be greater than about 30 years; this allows strontium-90 and caesium-137 to be classified as short-lived. At the same time, the amount of long-lived radionuclides that are permitted in LLW is strictly limited. This leads to the possibility of disposing of LLW and short-lived ILW (collectively LILW) in an engineered near-surface facility. More recent developments (IAEA, 2009a) have recognised the usefulness of two new waste categories: the first is very low-level waste (VLLW), which should be sufficiently limited in both total and specific activity to allow disposal in a landfill (i.e. with minimal engineered barriers). The second is very short-lived waste, which is suitable for decay storage followed by release as non-radioactive waste. Low-level wastes have low specific activity and, normally, high volume: often the volume requiring disposal within a national programme will amount to hundreds of thousands or even millions of cubic metres. Highlevel wastes, on the other hand, are likely to amount to, at most, a few thousand cubic metres. While the low volumes of HLW and its very high radiation field provide strong motives for deep disposal, the large volumes of LLW demand more economic methods, and these can be justified on account

© Woodhead Publishing Limited, 2010

Disposal of intermediate radioactive waste

45

of its lower activity. Typically, these more economic methods employ various methods of volume reduction coupled with near-surface disposal. By providing disposal facilities with a range of depths and engineering sophistication, wastes may be allocated to disposal facilities in a costeffective way. As discussed later, wastes containing radionuclides with halflives of a few thousand years (typically, radium-226 and carbon-14) may be allocated for intermediate-depth disposal. Particular difficulties can arise with respect to disused sealed radioactive sources. The problem is that these do not fit well into the existing waste categories because, even when their activity is modest, their small volume ensures that the specific activity is high. Small sources may, nevertheless, be disposed to a near-surface facility by simply mixing the sealed source with LLW and then immobilising the whole inside a standard container. At higher activities, however, the source activity may exceed the limit for a single waste package. Furthermore, even when the source size is below the package limit, the presence of a concentrated source may create a ‘hot spot’ in a near-surface disposal that can create difficulties for the safety case*. Consequently, the permitted activity of sealed sources in a near-surface disposal will usually be limited by regulation; normally there will be some sort of sliding scale that allows higher activities for shorter-lived radionuclides. Borehole disposal offers a potential solution for the disposal of long-lived sealed sources by providing the necessary levels of safety without going to the great expense of deep geological disposal.

3.1.3 Defining the ‘near surface’: limits to human intrusion Radioactive waste disposal aims to make the disposed material permanently safe in a manner that ensures that future generations are not required to perform any active management or, once the period of institutional control is over, even to know that the disposal facility exists. An important issue for safety, therefore, is that of human intrusion, i.e. the possibility that humans could unwittingly damage the facility and cause it to become a source of radiation exposure to themselves and others. Typical human intrusion scenarios include the construction of a dwelling on top of the facility, road construction, archaeological investigations, exploratory drilling, etc. In addressing these issues, reasonable intrusion depths are 3 m for house construction and 10 m for construction of a major road (NEA, 1987). In general, 30 m may be taken to be the limit of what is feasible in terms of excavations from the surface for road or railway construction in a competent rock (Ouzounian and Ozanam, 2009). Deeper excavations may *

The safety case is the collection of arguments and evidence that is intended to demonstrate the safety of a facility. A safety case will usually include one or more safety assessments.

© Woodhead Publishing Limited, 2010

46

Geological repository systems for safe disposal

be associated with quarrying or open-cast mining, but one would expect to be able to eliminate these possibilities by siting a repository so as to avoid locations where there is potential for exploitation of minerals. Exploratory drilling can occur to any depth, of course, though the likelihood of such an intrusion diminishes significantly with the depth of disposal (Baker et al., 1997). A final consideration is that high-rise buildings could be compromised by the existence of underground voids (i.e. a repository) at depths of less than 50 m (Yamato, 2005). These figures suggest that, exploratory drilling apart, a reasonable value for the maximum depth of human intrusion is about 30 m. This value has been used by the BOSS (BOrehole disposal of Sealed Sources) borehole disposal system, which aims to avoid human intrusion by ensuring that the minimum depth of cover for buried disused sealed sources is 30 m. For an underground cavern this minimum depth may be increased to 50 m to allow for the possibility that high-rise buildings could be constructed immediately above it. Such figures are not to be regarded as absolute: where a facility is to be located in hilly terrain, for example, one may wish to admit the possibility of human intrusion by tunnelling (for a road, for example). Similarly, additional cover may be needed to offset the effect of erosion. While accepting these caveats, 30 m provides a useful discriminator between the near surface and intermediate depths, as we shall call them here.

3.2

Outline of the sections

Section 3.3 outlines the IAEA Safety Principles and requirements for radioactive waste disposal and then goes on to consider their main implications, while Section 3.4 investigates the safety of disposal facilities. Section 3.5 provides a description of the various styles of near-surface, intermediate-depth and borehole disposal facilities. Section 3.6 is concerned with the factors that influence the safety of such facilities. Section 3.7 briefly examines emerging and future trends in near-surface disposal. Finally, Section 3.8 provides a list of useful sources of information.

3.3

Safety requirements for near-surface disposal

3.3.1 IAEA safety principles and requirements With the exception of IAEA-sponsored international treaties, IAEA documents have no force of law. They do, nonetheless, form the basis for legislation and regulation for nuclear and radioactive waste issues in many countries. IAEA documents are arranged in a hierarchy. At the top are the fundamental safety principles (IAEA, 2006a) and the Basic Safety Standards (FAO et al., 1996).

© Woodhead Publishing Limited, 2010

Disposal of intermediate radioactive waste

47

The IAEA fundamental safety principles (shown in the box below) apply to every kind of nuclear facility and supersede the earlier, sector-specific documents such as the waste safety principles. In many respects the operation Principle 1. Responsibility for safety The prime responsibility for safety must rest with the person or organization responsible for facilities and activities that give rise to radiation risks. Principle 2. Role of government An effective legal and governmental framework for safety, including an independent regulatory body, must be established and sustained. Principle 3. Leadership and management for safety Effective leadership and management for safety must be established and sustained in organizations concerned with, and facilities and activities that give rise to, radiation risks. Principle 4. Justification of facilities and activities Facilities and activities that give rise to radiation risks must yield an overall benefit. Principle 5. Optimization of protection Protection must be optimized to provide the highest level of safety that can reasonably be achieved. Principle 6. Limitation of risks to individuals Measures for controlling radiation risks must ensure that no individual bears an unacceptable risk of harm. Principle 7. Protection of present and future generations People and the environment, present and future, must be protected against radiation risks. Principle 8. Prevention of accidents All practical efforts must be made to prevent and mitigate nuclear or radiation accidents. Principle 9. Emergency preparedness and response Arrangements must be made for emergency preparedness and response for nuclear or radiation incidents. Principle 10. Protective actions to reduce existing or unregulated radiation risks Protective actions to reduce existing or unregulated radiation risks must be justified and optimized.

© Woodhead Publishing Limited, 2010

48

Geological repository systems for safe disposal

(and the regulation of the operation) of a disposal facility is very similar to that of other nuclear facilities. The distinguishing feature of disposal is the recognition that at some stage the operator will relinquish or lose control over the facility. This may be a deliberate decision or it could be an involuntary act arising, for example, from a breakdown of organised society. For this reason, Principle 7 – protection of present and future generations – has special resonance for disposal and leads to a requirement to demonstrate, through post-closure safety assessment, that even in the absence of active control, the facility will remain adequately and permanently safe. The Basic Safety Standards (FAO et al., 1996) are concerned with the protection of human health from ionising radiation. These are expressed through the principles relating to justification, optimisation and risk limitation (Principles 4, 5 and 6). Coming below the fundamental safety principles and the Basic Safety Standards in the hierarchy of IAEA documents are the safety requirements. Requirements documents use the word ‘shall’ i.e. each requirement is obligatory. At the present time there are requirements documents that apply to near-surface disposal (IAEA, 1999) and to deep geological disposal (IAEA, 2009b). The document describing the requirements for near-surface disposal (IAEA, 1999) contains over 120 ‘shall’ statements but, in fact, all these can be more succinctly expressed through the 25 requirements that appear in the corresponding document for deep geological disposal (shown in Section 3.9) and, indeed, the IAEA has indicated its intention to develop a combined disposal requirements document. The implementation of these requirements is explained through three guidance documents that apply to near-surface disposal (IAEA, 2010a), deep geological disposal (IAEA, 2010b) and borehole disposal (IAEA, 2009b). Following standard regulatory practice (and in contrast to requirements documents), statements in guidance documents replace ‘shall’ with ‘should’. Without wishing to diminish the importance of the other requirements, it is probably true to say that the key safety issues for disposal are passive safety (Requirement 5), multiple safety functions (Requirement 7), containment of radionuclides (Requirement 8) and isolation of the waste so as to reduce the risk of human intrusion (Requirement 9).

3.4

Safety of disposal facilities

3.4.1 Operational safety It is convenient to separate safety issues into operational and post-closure phases where operational safety is concerned with the safety of workers and the public during the operational period. Here it is necessary to show that the facility is acceptably safe both in normal operation and in non-normal

© Woodhead Publishing Limited, 2010

Disposal of intermediate radioactive waste

49

operation, where the latter includes incidents, accidents and emergencies. This demonstration is usually done using an operational safety assessment. Operationally, a radioactive waste disposal facility is much like any other nuclear plant; it has a radiological protection plan (including an emergency plan) and there is careful control and monitoring of both worker doses and radioactive contamination around and within the facility. It is normal for repository construction and repository operation to be ongoing simultaneously and, if this is the case, it will need to be recognised in the operational safety assessment. The operational safety assessment and any other relevant information such as security arrangements, financial guarantees, record keeping, quality management, etc., will be contained in the operational safety case, whose purpose is to describe the operation of the facility and demonstrate its safety. Worker doses will depend on the nature of the operations and the type of waste that is received. If, for instance, the facility is designed to accept shortlived intermediate-level wastes that emit significant levels of gamma radiation, there will probably be a need for remote handling of waste packages so that worker doses can be kept within the regulatory constraints. From this example we see that the way in which the facility is designed to be operated will be a factor in determining the waste acceptance criteria, i.e. the formal definition of the type of waste that can, and cannot, be accepted for disposal. Other operational factors that are likely to figure in the waste acceptance criteria are the weight, geometry and lifting arrangements of waste packages and the levels of removable contamination permitted on their surfaces. The end of the operational period is often defined to coincide with the start of the institutional control period (discussed in Section 3.4.4); alternatively, it may be defined to include the institutional control period – it does not really matter so long as there is no ambiguity.

3.4.2 Post-closure safety Post-closure safety needs to be addressed in a separate post-closure safety assessment. This aims to show that, even though future generations may not know of its existence, the design of the facility is such that the facility is permanently safe. To achieve this, all disposal facilities deploy multiple barriers to provide separate safety functions (Requirement 7) whereby engineered and natural barriers work together to provide the requisite level of passive safety, containment and isolation (Requirements 5, 8 and 9 respectively). The design aims to ensure that no undue reliance is placed on any one barrier. That said, a near-surface disposal is likely to rely more on the engineering barriers than the natural ones. The acceptability of this practice rests on the radionuclides in the waste being short-lived.

© Woodhead Publishing Limited, 2010

50

Geological repository systems for safe disposal

The post-closure safety assessment will generally include a ‘design’ or ‘normal’ scenario, which estimates the expected radiological outcomes assuming that the facility performs as it is designed to do. This requires the post-closure safety assessment to anticipate the degradation of the repository structures (e.g. waste package corrosion, erosion of the repository cap) and to demonstrate that the repository will perform satisfactorily during and following events, such as storms or seismicity, that might reasonably be expected to occur during the assessment period, i.e. with a probability of, say, greater than 5%. To perform such a safety assessment it will be necessary to determine the storm and seismic hazards at the chosen site. For a near-surface disposal, key issues for post-closure safety are the integrity of the waste packages, the mechanical stability of the waste stack and the overlying cap, and the hydraulic properties of the surrounding environment. The main purpose of the repository cap, for instance, is to avoid or reduce infiltration of meteoric water into the waste. If the waste stack settles, say because of excess voidage, this may lead to cracking of the cap, increased water infiltration and leaching of radionuclides from the waste. Where a near-surface repository is located in the unsaturated zone, a key parameter is the distance between the base of the waste stack and the groundwater table. The greater this distance, the more radionuclides will be retarded as they migrate downwards towards the aquifer. Such effects need to be considered as part of the post-closure safety assessment. As another example, a below-grade near-surface repository sited in impermeable clay may be subject to ‘bathtubbing’ whereby the excavation fills up with meteoric water. Radionuclides will leach into this water and, as the water spills over on to the adjacent land, will contaminate the land and any crops that are growing there. Such examples illustrate the situations that might be analysed as part of the normal scenario. The post-closure safety assessment should also examine the consequences of less likely events such as severe storms and seismicity and demonstrate that, allowing for the lower probability of such events, the calculated risk is acceptably low. These ‘less likely scenarios’ usually include a series of ‘what if’ calculations that would probably include assessments of the consequences of premature failure of the engineered barriers or increased rates of erosion and flooding. Another standard type of scenario is that of human intrusion. This is usually considered to occur immediately after the end of the institutional control period and will normally include an assessment of the radiological consequences of someone building a house on top of the facility, eating food grown in contaminated soil and extracting drinking water (and, possibly, irrigation water) from a borehole that passes through or close to the facility into an underlying aquifer. In this context the main purpose of the

© Woodhead Publishing Limited, 2010

Disposal of intermediate radioactive waste

51

institutional control period is to prevent human intrusion until the radionuclides have decayed sufficiently that the radiological consequences of human intrusion are acceptable. The post-closure safety assessment is an important document for the derivation of the waste acceptance criteria. The assessment should provide, in particular, (1) a limit on the total radioactive content of the repository and (2) a limit on the specific activity of individual waste packages. These limits are often expressed in terms (separately) of beta/gamma emitting radionuclides and alpha emitting radionuclides. This has the advantage of simplicity and may also facilitate the link between waste categorisation and the disposal route. In this way, all waste classified as LLW (for example) would be eligible for disposal in the facility. Alternatively, limits on total activity and specific activity may be expressed as separate values for individual radionuclides. A problem with this second method is that the radionuclides that are of greatest interest for disposal (e.g. iodine-129) are generally difficult to measure, so that it is usually necessary to establish a correlation between each radionuclide of interest and a different ‘indicator’ radionuclide (usually a gamma emitter) that is easier to measure. Typical examples are correlations between iodine-129 and caesium-137 (both fission products) or between carbon-14 and cobalt-60 (both neutron activation products). Consideration of the plume of contaminated water that could flow from the repository once the engineered barriers have degraded leads to the establishment of limits on the total activity. Other calculations, such as (1) exposure of the waste following removal of the cap due to erosion and (2) use of contaminated soil in a human intrusion scenario, lead to limits on specific activity. A particularly useful exposition of the derivation of activity limits for near-surface disposal is provided by IAEA (2003a). The waste acceptance criteria may also include limits on other parameters such as container strength, composition of the immobilisation grout, free water content and voidage (e.g. IAEA, 2000a; Morales, 2005). In some cases, there may also be a limitation on cellulose content on the grounds that, under cementitious conditions, cellulose can degrade to an organic acid that forms a complex with actinide ions and so makes them soluble and mobile. In most cases it is the repository operator that puts the wastes into the disposal container and grouts them into place and, where this is the case, it will be the disposer, rather than the producer, that takes on some of the burden of ensuring that the packages comply with the waste acceptance criteria.

3.4.3 Safety of mining and milling wastes The mining, milling and processing of uranium, thorium and rare earth minerals produces a range of so-called NORM (naturally occurring

© Woodhead Publishing Limited, 2010

52

Geological repository systems for safe disposal

radioactive material) wastes that are high in volume. In general, mining wastes consist of excavated rock while tailings are the more finely divided residue that remains after the target element(s) is removed from the finely ground (i.e. milled) ore. In general, the proportion of the target element that is removed from the extracted ore is no more than a few tens of percents. Consequently, the concentration of (say) uranium in the waste rock and tailings materials is not so different from that of the original ore body. Ore processing may also produce wastes whose specific activity exceeds that of the original ore: an example is the radium-containing scales formed on the inside of pipes. For wastes like this, the term TENORM (technically enhanced NORM) may sometimes be used, especially in the USA. Invariably, NORM and TENORM wastes contain long-lived radioactive elements. Standard practice is to pile up the waste rock and the tailings on the surface. Sometimes a proportion of the waste rock may be returned to the mine, especially if it is needed for mechanically supporting the excavated cavities. However, the increase in volume that occurs when the rock is broken up ensures that the mine will never accommodate all the waste rock that it produces. Consequently, unless another use such as road construction can be found for the waste rock and tailings, it is inevitable that waste management facilities – rock piles for instance – will be left in situ around the mine. It is usual for the safety of these facilities to be governed by mining regulations rather than regulations for radiological protection; because these wastes are long-lived, there is usually an expectation that the facilities will be subject to perpetual care. The acceptability of this rests on the argument that the radionuclides are naturally occurring and that their specific activity and the associated hazard are low.

3.4.4 Significance of the institutional control period In terms of the IAEA requirements for disposal (Section 3.9), two requirements concern passive safety. Requirement 6 states that safety shall be ensured by passive means so as to minimise the need for action after the facility has been closed. Requirement 10 complements this, stating that in the post-closure period the site shall be appropriately supervised in order to protect and preserve the passive safety barriers. For the most part, the thing that this post-closure supervision is intended to protect against is human intrusion, and we may regard post-closure supervision as an engineered safety barrier whose main safety function is the prevention of human intrusion. Clearly, this safety function will have particular importance for near-surface disposal because of the greater risk of human intrusion there. The institutional control period is the time span over which post-closure supervision may be assumed to apply. A commonly used institutional

© Woodhead Publishing Limited, 2010

Disposal of intermediate radioactive waste

53

control period is 300 years. This is convenient because it represents ten halflives and, therefore, a factor of about 1000 reduction in the activity of two of the most commonly encountered fission products, strontium-90 and caesium-137. It also happens to represent a period of time over which, it may be credibly argued, institutional control could be continued. Evidence for this comes from many organisations such as faith bodies, legislative assemblies and long-established private institutions. Reduced to very simple terms, the institutional control period may be seen as the period of time over which radionuclides in a repository should decay to near-exemption levels. This has implications for the waste acceptance criteria, as explained in Section 3.6.2. For mining and milling wastes the situation is rather different. The long-lived (albeit naturally occurring) radionuclides demand that surface facilities should be subject to perpetual control to avoid human intrusion. In low population locations, regular inspections of fences and signs to warn off potential intruders may be sufficient given that the low specific activity of the wastes requires longterm exposure to accumulate a significant dose. In facilities that are located near urban areas, more innovative solutions may be sought. At Port Hope in Canada, for instance, large volumes of historical radium-containing wastes were dumped in various locations around urban areas. The community decided to deal with the potential hazard by moving the wastes to one site, covering them with a thick layer of earth and designating the flattened, covered piles as a recreational area for common use. Institutional control measures are sometimes described as either ‘active’ or ‘passive’ (IAEA, 2003b). Active measures include activities such as a security presence and environmental monitoring. Passive measures usually include archiving of repository documents, control of ownership and, perhaps, markers designed to indicate the presence of a hazard. The main advantage of categorising the measures in this way is that it recognises that some forms of institutional control are likely to be more durable than others. The approach may lead to the institutional control period being divided into active and passive periods.

3.5

Styles of near-surface disposal

3.5.1 General Radioactive waste disposal facilities are expensive to develop and this, together with the general issue of sustainability, provides a strong motivation for waste minimisation. This important subject falls outside the remit of this document apart from noting that simple housekeeping measures on nuclear plant can lead to very significant reductions in waste production. Especially useful in this respect are strict segregation of waste

© Woodhead Publishing Limited, 2010

54

Geological repository systems for safe disposal

streams and the avoidance of unnecessary waste creation by, for example, forbidding staff from taking disposable items (e.g. newspapers) into controlled working areas. Waste minimisation is concerned with waste prevention (indeed, this would be a better name for it) and should not be confused with volume reduction, which is a process that is applied after the waste has been created. Pre-disposal management of radioactive waste forms the basis for a suite of IAEA requirements and guidance documents (IAEA, 2000b, 2003c). While this subject largely falls beyond the scope of the present discussion, one aspect – volume reduction – has a strong influence on disposal costs and is so commonly used that it will be briefly described. Given that LLW may often include miscellaneous contaminated trash, volume reduction is often achieved by compaction or incineration. In waste compaction, waste is typically placed into 200 litre steel drums that are subsequently crushed in a high-force compactor to produce ‘pucks’. These are then placed inside a second (often concrete or concrete-lined) container and backfilled with a concrete grout (Fig. 3.1). In general, the larger this second container is, the greater will be the degree of volume reduction achieved. Where waste is capable of being incinerated, this may also be a useful volume reduction option, remembering that the resulting ash will have a considerably higher specific activity than the starting material and that it will itself need to be immobilised and packaged in some way. Incineration is also a useful option for incinerable liquids such as contaminated oils, which would otherwise be difficult to dispose. In the case of large-volume decommissioning wastes, worker doses will be minimised if these are placed in the vault intact and grouted into place. Because of the volume increase that occurs when fragmented components (e.g. heat exchangers) are placed into waste packages, it will often be found that emplacement of the intact component will also be most efficient in terms of volume utilisation.

3.5.2 Surface facilities: trenches and engineered vaults The needed level of sophistication of a repository design – for which one may read the number of engineered barriers – primarily depends on the nature of both the waste and the site. In desert areas with low rainfall, high evapotranspiration and a deep water table, safe disposal may be possible with a covered over mound or, more probably, simple trenches backfilled with excavated material. Nevertheless, safety assessments for such sites should still consider the full range of features, events and processes (FEPs), supplying justification for the exclusion of those that are not applicable. It will be necessary to go beyond the consideration of average rainfall and evapotranspiration values to consider the rainfall pattern, which may be

© Woodhead Publishing Limited, 2010

Disposal of intermediate radioactive waste

55

3.1 Sectioned waste package consisting of a series of ‘pucks’ (supercompacted 200 litre drums) cemented into a larger drum (courtesy Ondraf Niras).

associated with infrequent but intense rainfall events. Such facilities may be found in operation in the USA and Australia and are planned in Egypt, Iran and other countries. At sites in temperate and tropical regions the groundwater table will often be close to the surface and here engineering measures will usually be necessary (1) to reduce the rate of leaching of radionuclides from the waste (usually by encapsulating them in concrete) and (2) to divert meteoric water away from the wastes through the installation of a low-permeability cap on

© Woodhead Publishing Limited, 2010

56

Geological repository systems for safe disposal

top of the facility. Typically, wastes are cemented into steel or concrete containers and then placed inside concrete-lined vaults. The repository cap, which is put in place when the vaults are full, usually consists of at least a metre-thick layer of low-permeability clay, which is compacted after being put in place. This may be covered by an anti-bio-intrusion layer of loose rock of, maybe, half a metre, followed by soil that supports a layer of vegetation that provides some protection against erosion. Water management is an important issue when operating these facilities. During the operational phase when the wastes are still uncovered, meteoric water cannot be allowed to accumulate inside the waste cells and, if water does enter, measures must be taken to remove it. This may require the water to be decontaminated before it can be discharged to the environment. To reduce infiltration of meteoric water it is common to use a temporary ‘roof’ over a cell while it is in the process of being filled. This type of facility may be constructed on the ground surface or just below it. An above grade facility, i.e. one located directly on the ground surface, maximises the distance between the base of the facility and the groundwater table but needs some degree of lateral support. Placing the facility in a shallow excavation requires less lateral support and has some advantages with respect to access for waste placement and installation of a stable, erosion-resistant cap. Facilities of this type are found in France, Japan, Spain, the UK and Ukraine (Figs 3.2 and 3.3).

3.2 Aerial view of the closed near-surface repository at Centre de la Manche, France. The ‘corrugated’ surface is used to drain water off the cap (courtesy Andra).

© Woodhead Publishing Limited, 2010

Disposal of intermediate radioactive waste

57

3.3 Operations at the Drigg, UK, near-surface repository. Half-height ISO containers form the outer layer of the waste package (courtesy BNFL).

3.5.3 Subsurface facilities: silos, caverns and tunnels Intermediate depth disposal Specially created subsurface facilities for disposal of low- and short-lived intermediate-level wastes at intermediate depths are in operation in Finland and Sweden; others are planned in France, Japan and the Republic of Korea. In Sweden the Fo¨rsmark facility has been in operation since 1988. LLW is deposited in tunnels that are located about 60 m below the bottom of the Baltic Sea in crystalline bedrock, as shown schematically in Fig. 3.4. Each tunnel is subdivided into a series of concrete cells where the waste is placed without further cementation. Short-lived ILW is lowered into a 50-m deep underground concrete silo that is divided into individual shafts and surrounded by a clay buffer. Wastes are lowered into the shafts and then backfilled with a concrete grout (Skogsberg and Ingvarsson, 2006). Radioactivity was first discovered in France so the country has a long history of utilisation of radioactive materials and, consequently, many historic wastes, of which radium-containing wastes form a significant part. France also has large volumes of graphite waste following the decom-

© Woodhead Publishing Limited, 2010

58

Geological repository systems for safe disposal

3.4 Schematic view of the Fo¨rsmark LILW repository showing the silo and four disposal tunnels (copyright SKB, artist Jan M. Rojmar).

missioning of its first-generation, gas-cooled, graphite moderated reactors. These graphite wastes contain a range of radionuclides and, of those that occur in significant quantities, carbon-14 is the most long-lived. Both carbon-14 and radium-226 have half-lives of a few thousand years (to be precise, 5730 and 1600 years respectively). The radium wastes have much lower activity than the graphite wastes and a key objective of any disposal scheme will be to contain emitted radon. One option that is being considered is disposal in a specially created facility in a low-permeability clay or marl formation that is at least 50 m thick. The depth of the repository could be up to 200 m. Current plans (Ouzounian and Ozanam, 2009) envisage a simple tunnel with radium and graphite wastes housed in separate vaults. The waste package details have yet to be decided but the space-filling characteristics of rectilinear packages gives them a clear advantage in terms of efficiency of volume utilisation and avoidance of large voids that require backfilling. A similar facility, shown diagrammatically in Fig. 3.5, is proposed for use in Japan at Rokkasho (Yamato, 2005). In the Czech Republic a former limestone mine and munitions store has been converted to a repository for low- and intermediate-level waste. Under normal conditions the mine is dry though there is a possibility of flooding

© Woodhead Publishing Limited, 2010

Disposal of intermediate radioactive waste

59

3.5 Investigatory tunnel for the proposed intermediate depth repository at Rokkasho in Japan (after Yamato, 2005) (courtesy JNFL).

during very severe storm events. To accommodate such rare events, the design incorporates a novel hydraulic cage design that diverts water away from the wastes (Biurren et al., 2005). In the Republic of Korea a subsurface repository for LILW is being constructed close to the Wolsung Nuclear Power Plant (NPP) on the southeastern coast of the country. Current plans envisage a drift running from the surface to a depth of 80 m where, in the first phase, six underground silos, 27 m in diameter and 50 m deep, will provide disposal capacity for 100 000 drums, each of 200 litres. Within the drums, wastes are encapsulated in cement or polymer. Drums are loaded into the silo inside a rectilinear concrete overpack that holds 16 drums in a single layer 464 array. Deep disposal of long and intermediate-level waste (LILW) While it may be convenient to distinguish between ‘intermediate-depth’ and ‘deep geological’ disposal there is, in fact, little difference between them. The choice of depth primarily rests on one, or a combination, of the following: the regulatory requirements in the specific country, the chosen location and stakeholders’ perceptions of the potential hazard. Deep disposal of LILW may result when former mines are adapted for the

© Woodhead Publishing Limited, 2010

60

Geological repository systems for safe disposal

purpose. This strategy has been successfully pursued in Germany where three former mines, all of them unsaturated, serve as repositories for LILW. Two, those at Morsleben and Asse, are disused salt mines, where large cavities resulting from salt excavation have been partly filled with LLW. The third is the former iron mine at Konrad, where LILW is to be disposed in the old mine galleries, which remain unsaturated, at depths between 800 and 1300 m. More unusually, new deep facilities may be deployed for LILW disposal, as in Canada*, which envisages a disposal facility at Kincardine (near the Bruce power plant), where LILW from the power plants would be disposed in a limestone formation at a depth of 660 m. A second facility could be constructed at a depth of 500 m in granite underlying the Chalk River Laboratories site for LILW that is currently held there.

3.5.4 Mining and milling wastes The regulatory regime for mining and milling wastes may differ significantly from those applicable to radioactive waste from the nuclear industry. It is normal for these naturally occurring wastes to be excluded from the regulations governing radioactive material and for them to be controlled in a similar manner to other mildly toxic (but non-radioactive) wastes. Issues arising from their disposition may not be limited to the presence of toxic or radioactive materials. A common problem, for instance, is that oxidation of iron sulphide minerals in the wastes may lead to the formation of acidic compounds and acid run-off (IAEA, 2004a), just as it does in the similar phenomenon known as acid mine drainage. This can have a very negative effect on the surrounding environment, both from the acid itself and from the increased solubility of toxic metal ions in meteoric water. Surface accumulations of mining wastes can be ancient – rock piles are known that go back to the Bronze Age. These wastes frequently represent a minor hazard due to the presence of toxic and/or radioactive elements in low concentrations; there may also be a physical hazard from landslides when, as is very often the case, these wastes are deposited on sloping ground. A great deal of effort has been (and is still to be) expended on remediation of such facilities. Necessary actions usually include stabilising the rock piles and covering them with a metre or so of uncontaminated soil. Stabilisation may entail slope reduction, compaction, placing of supporting structures at the foot of the pile and water management. These measures reduce the physical hazard and prevent dispersion of the wastes. The cover has a number of purposes: it prevents wind-blown contamination, reduces radon *

Ken Nash, Chair, Nuclear Waste Management Organization, speech to Canadian Nuclear Association Seminar, 23 February 2006, http://www.nwmo.ca/.

© Woodhead Publishing Limited, 2010

Disposal of intermediate radioactive waste

61

exhalation (in the case of radium-containing wastes), mitigates external radiation levels on the surface of the pile and, by encouraging vegetation to grow, controls erosion and promotes stability. The properties of the mining and milling wastes will, clearly, depend on the nature of the operation, e.g. whether operations are limited to mining or whether they include ore concentration and smelting. Mining produces rock wastes, while ore processing produces powdered material that is mixed with water in a slurry form. For all these types of waste we find that rich ore bodies tend to produce wastes that are relatively small in volume but high in metal (e.g. uranium) concentration. The reverse is the case for low-grade ore bodies. Smelting produces molten wastes that usually solidify to a glassy slag; if radioactivity levels are low, this may be a useful pozzolanic (cementproducing) material when crushed to powder. Other wastes are likely to include contaminated items of plant, especially pipework that has become coated with scale and may be sufficiently radioactive to be classified as ILW. Wastes arising from milling and ore concentration are water-borne and need to be allowed to settle to allow the solid and liquid phases to separate. This is usually done through the creation of a tailings dam and lake. There are two main forms: valley dam impoundments and ring dyke impoundments (IAEA, 2004a). The mill tailings are pumped in the form of slurry on to the lake. Water agitation is kept to a minimum and the solids fall to the bottom so that the supernatant water can be removed to be reused. In some earlier designs (especially in desert locations) water removal may occur mostly by evaporation (IAEA, 2004b). Key safety issues during operation are the stability of the dam (keeping in mind seismicity, for instance) and management of contaminated water. It is possible, for instance, that contaminated water may seep through the deposited tailings into the underlying ground where potable water may be located. Measures to prevent this include specially created dewatering drifts below the tailings facility, use of wicks laid within the tailings as conduits for water removal and groundwater abstraction to lower the surrounding groundwater table. Water that is removed in this way may be used as process water; surplus water will need to be treated before discharge to the environment (IAEA, 2004b). When the operations cease, it is usual for tailings management facilities to be converted to disposal facilities. This will require a number of engineering measures that must be designed to meet the needs of the site. The tailings, for instance, will no longer be covered by water, which allows the possibility of wind-borne releases. There may also be some radon exhalation. This will necessitate the capping of the tailings pond with clean soil and a layer of vegetation. Other measures might include improvements with respect to mechanical stability and use of an anti-bio-intrusion layer and/or a low

© Woodhead Publishing Limited, 2010

62

Geological repository systems for safe disposal

permeability layer within the cap, which could include a geotextile membrane (IAEA, 2004a). Regardless of what engineering measures are implemented, the existence of an above-grade disposal facility containing long-lived or chemically toxic wastes will usually require some form of permanent management in the form of surveillance, environmental monitoring, repairs to the cap, etc. Where mining is an open-cast operation, it is more probable that the mine and tailings wastes will be used to backfill the excavated hole. It is even possible for this to be done during the course of operation. It is usual for the coarser mine wastes to be placed in the bottom of the hole so that the finer, more easily leachable tailing material may be placed above the level of the groundwater. If the hole is lined with clay to reduce interaction with the surrounding groundwater and if another use can be found for the surplus rock created by the increase in volume on excavation, such a design may provide a walk-away solution, i.e. it may require no long-term management.

3.5.5 Borehole facilities for large- and small-volume waste packages Surprisingly perhaps, the use of borehole facilities for radioactive waste storage or disposal is a relatively common practice throughout the world. Those that are most frequently found are the RADON borehole facilities developed in the former Soviet times and still in use, especially in Russia. These RADON boreholes may be conveniently subdivided into two types: small and large diameters. The small-diameter RADON boreholes consist of a roughly 10 cm diameter borehole leading to a larger stainless steel container placed at a depth of about 3 m (Sobolev et al., 2001). Disused sealed sources are dropped down the borehole so that they fall into the container. An unusual feature of these facilities is that the total activity of sources can be so great (PBq quantities) that it has been found useful to pour molten lead into the container to facilitate heat dissipation. While these facilities were originally intended for disposal, it seems that they are now regarded as storages. A similar type of facility has also been deployed (without the addition of lead) for disused sealed sources at Pelindaba in South Africa. The larger-diameter RADON boreholes (Fig. 3.6) are a more recent development (Prozorov et al., 2001). At the Moscow RADON site, boreholes are drilled to 40 to 50 m depth in clayey sediments above the saturated zone. Hole diameters can vary but the smallest is 1.9 m cased to a usable diameter of 1.5 m with the soil-casing annulus filled with bentonite– cement mortar. Wastes are packaged in 200-litre drums and emplaced in the borehole to within 2 m of the surface. Wastes are generally ungrouted in the

© Woodhead Publishing Limited, 2010

Disposal of intermediate radioactive waste

63

3.6 Large-diameter borehole at the MosRadon site near Moscow (source: IAEA TECDOC-1368).

borehole to aid retrieval should this be needed. At the present time the emplacement is classed as storage but this could, with regulatory approval, be simply converted to disposal by filling the voidage with a free-flowing bentonite–cement mortar. Even larger diameter boreholes have been used for radioactive waste

© Woodhead Publishing Limited, 2010

64

Geological repository systems for safe disposal

disposal in Western Australia at the Mount Walton Intractable Waste facility (Hartley et al., 1994) and in the USA where there are facilities at the Nevada test site (Cochrane and Hasan, 2005) and Savannah River (Fig. 3.7). At the Nevada test site a 3 m diameter hole was drilled to a depth of 37 m, this being about 200 m above the saturated zone. Drummed wastes are loaded into the hole to within 21 m of the surface and backfilled with excavated soil. PBq quantities of caesium-137 and transuranic waste have been disposed in this way. The safety case (Cochran et al., 2001a, 2001b) largely rests on the very low availability of water at the site.

3.7 Schematic diagram of the Greater Confinement Test facility at Savannah River (source: US NRC NUREG/CR-3774 V.5, Alternative methods for disposal of low-level radioactive waste. Task 2E: technical requirements for shaft disposal of low level radioactive waste).

© Woodhead Publishing Limited, 2010

Disposal of intermediate radioactive waste

65

3.5.6 The International Atomic Energy Agency (IAEA) borehole disposal of sealed sources (BOSS) concept for disused sealed sources A relatively new development in borehole disposal facilities is the BOSS (BOrehole disposal of Sealed Sources) concept that stems from IAEA’s efforts to retrieve, ‘condition’ and safely store disused radium sources through its African Regional Cooperative Agreement for Research, Development and Training related to Nuclear Science and Technology – the so-called ‘AFRA’ project. In this context ‘conditioning’ entails the sealing of radium sources into 3-mm thick stainless steel capsules. Later developments, drawing on concerns over radiological security, have widened the remit of the conditioning campaign to encompass sources other than radium. This has included the development of a mobile hot cell that allows high activity sources to be remotely handled and conditioned (Crossland et al., 2011). Recognising that storage is not a permanent solution, the AFRA project then turned its attention to devising a safe yet economic means of disposal. Much of the development work was performed for IAEA by the Nuclear Energy Corporation of South Africa (Necsa), which produced a borehole disposal concept that largely relies on the containment/corrosion resistance properties of stainless steel by placing the conditioned sources inside a concrete-filled 6-mm thick stainless steel container. The disposal containers are then placed at a suitable depth (but always greater than 30 m to avoid

3.8 Schematic diagram of the BOSS borehole disposal concept (copyright Crossland Consulting).

© Woodhead Publishing Limited, 2010

66

Geological repository systems for safe disposal

human intrusion) within a specially drilled borehole and concreted into place using a fluid cement grout (Fig. 3.8). In the absence of localised corrosion – something that can be assured quite simply under these cementitious conditions – corrosion rates are extremely low, so low, in fact, that most radioactive sources can be expected to decay to exemption levels before the disposal container is breached (IAEA, 2010c).

3.6

Designing for safety

3.6.1 Stakeholder views An early approach to site selection for near-surface disposal (IAEA, 1994b) adopted a strong technical focus, ranking the suitability of sites in terms of their physical characteristics, i.e. their geology, hydrology, geomorphology, etc. The danger with this approach is that it implies a search for the ‘best’ site when, in reality, experience has shown that, while the best is unattainable, virtually any site, with few exceptions, has the capability to host a disposal facility safely. This is not to say that all sites are equally good but, rather, that it is usually possible to overcome a site’s shortcomings through the use of more robust engineered barriers or, alternatively, by choosing a more appropriate geological horizon. A further difficulty with a site selection process that is focused on technical criteria is that it overlooks the issue of social acceptability. In many countries, the willingness (or not) of the local community to host a near-surface disposal facility is the single most important determinant in site selection. In most cases, the only communities willing to host such a facility are those that already have a nuclear facility in their locality. The argument that is usually deployed is that these communities have a higher level of understanding of nuclear issues and therefore appreciate the benefits of a new disposal facility in terms of jobs and inward investment. An argument that is less often heard is that such a ‘nuclear community’ may already hold significant quantities of waste in storage. If other communities are unwilling to receive this, the nuclear community has two choices: a continuation of the status quo or an agreement to host a new disposal facility. Where a community agrees to perform this role for the benefit of the country as a whole, it will, clearly, seek to accrue other benefits such as infrastructure improvements. For the owner of the wastes there may also be significant savings in transport costs, of course. Stakeholder involvement may not be limited to the issue of site selection; there are at least two cases where the local community has had a significant influence on the facility design. In the Port Hope area of Ontario, Canada, the local community wished to use relocated mining and milling wastes to create an area that could be used for local leisure activities, and this had

© Woodhead Publishing Limited, 2010

Disposal of intermediate radioactive waste

67

repercussions with respect to design. In Belgium, in 1999, three communities were initially competing to host a near-surface repository for LILW and, in July 2006, the government decided to locate the repository at Dessel where the STOLA local partnership (STudie- en Overleggroep Laagactief Afval– Dessel or, in English, the study and consultative group on low-level waste– Dessel) had been established as a ‘consultation platform’ to investigate the ‘technical feasibility and social acceptability of the final disposal of Belgian low-level and short-lived waste within municipal boundaries’ (STOLA, 2004). The proposed repository design has been strongly influenced by STOLA. In particular, the floor of the disposal modules is supported by pillars, creating a basement whose purpose is to make it possible to perform continuous visual checks and intervene in the case of cracks or leaks.

3.6.2 Waste acceptance criteria The operational and post-closure safety assessments for a near-surface disposal facility should be used to define the waste acceptance criteria that will determine the wastes that can, and cannot, be disposed to the facility. However, since we may also choose to categorise waste according to its disposal route, it follows that there should also be a strong link between waste categorisation and safety assessment. In practice, mostly for historical reasons, this link is not always evident. Broadly speaking, near-surface disposal is reserved for LLW and shortlived ILW (collectively, ‘LILW’). Typically, LLW consists of contaminated trash and other wastes and may include contaminated clothing, tissues and wipes, and disused items of plant such as NPP secondary circuit pipework and pumps. Short-lived ILW usually consists of NPP operational wastes, especially wastes generated by water cleanup plant. These may include spent ion exchange resins and sludges, both of which must be conditioned into a suitable waste form before disposal. Operationally, the main distinction between LLW and short-lived ILW is that the former has a sufficiently low specific activity to allow manual handling whereas the latter requires remote handling and, therefore, an investment in remote handling equipment, which may be considerable. The upper limit on specific activity for beta/gamma emitters in LLW is usually around 10 kBq/g, which, in theory, could produce a contact dose rate of up to 5 mSv/h. In practice, limits on the total activity that may be contained within a single package and the fact that most of the radionuclides are present as surface contaminants mean that packages will be self-shielding so that dose rates will usually be much lower than this. Where remote handling equipment is available, this will allow much higher activity wastes to be emplaced. In this case the limiting specific activity for disposable short-lived radionuclides is more likely to be

© Woodhead Publishing Limited, 2010

68

Geological repository systems for safe disposal

determined by post-closure safety rather than operational safety. The broad aim is that, after the 300-year institutional control period, the specific activity of the wastes should have fallen to an acceptable level. There are many ways of estimating this level and here we illustrate a simple method that might be applied. At the end of the institutional control period we may suppose that the repository cap is still in place, but an intrusion occurs by drilling through the cap into the waste and the drill cuttings are spread over the surface. This contaminated ground is then cultivated. The area of agricultural land needed to support a critical group of, say, five adults using modern, highinput agricultural practice is about 2500 m2 and a reasonable cultivation depth is 10 cm, which produces a total soil volume of 250 m3. If we assume that contaminated drill cuttings have a volume of about 0.5 m3 and are distributed through the soil volume, we derive a dilution factor of 500. Note that it is not necessary to assume uniform mixing because the higher doses and risks arising from more heavily contaminated areas will be offset by lower doses and risks from the less heavily contaminated areas. A reasonable expectation is that radiation exposure from consumption of contaminated food and water will be acceptable if the specific activity of the cultivatable soil falls below exemption levels. By way of example, the Basic Safety Standards document (FAO et al., 1996) has an exemption level for specific activity of caesium-137 of 10 Bq/g. For this radionuclide, a 300-year institutional control period represents around ten half-lives or about 1000 times reduction in activity. Combining this with the factor of 500 dilution in soil we may therefore expect the limiting specific activity for near-surface disposal of short-lived ILW to be about 5 MBq/g, which is about a factor of 500 greater than the beta/gamma limit for LLW (for this radionuclide). LILW will, of course, contain some long-lived radionuclides and it is necessary to set activity limits for these also. In this case we can largely ignore decay during the institutional control period so that, following the argument in the previous paragraph, the limiting values on specific activity will be around 100 (the dilution factor) times the exemption values. For uranium-238 and radium-226, this produces a limiting specific activity of 5 kBq/g. This estimate is close to those usually deployed in practice: France (Dutzer, 2002), for example, currently has a total alpha-specific activity limit of 4 kBq/g (calculated at 300 years) for the Centre de l’Aube near-surface disposal facility and the limit for the UK Drigg facility is similar. As a general rule, disposal at greater depths (as with the Fo¨rsmark facility) will permit higher specific activities to be disposed.

© Woodhead Publishing Limited, 2010

Disposal of intermediate radioactive waste

69

3.6.3 Disposal environment With few exceptions, the groundwater pathway is the most important pathway for radionuclide migration from a repository. Consequently, the hydrology and hydrogeology of a site will have a profound effect on the repository design. As a general rule, making a safety case will be simpler in environments that have little or no water present (e.g. desert environments) than in those that do. Deserts have the further advantages of low agricultural productivity and low population density, which reduce the risk of human intrusion. Consequently, for disposal in a very dry environment (as at the Nevada test site) it will usually be possible to achieve adequate levels of safety with relatively few engineered barriers. In contrast, where a repository is sited below or close to the water table, it will usually be necessary to deploy engineered barriers such as a repository cap to reduce infiltration of meteoric water and encapsulation of the wastes in concrete to reduce leaching of radionuclides. There are a number of characteristics that may reduce the technical suitability of a site to host a disposal facility. These include the presence of an active geological fault, periodic flooding, rapid erosion or landslip. Even in these cases, however, it would often be possible to engineer the site so that these effects were insignificant for safety. Waste repositories are usually fairly simple structures so that, in the case of a nearby active fault, the repository structures could be designed to have sufficient mechanical strength to resist the disruptive effect of fault movements. Similarly, periodic flooding could be prevented through flood defences or by raising the ground level at the facility. It may be possible to overcome erosion and landslip through quite simple civil engineering measures such as slope stabilisation. All these measures would, of course, increase costs and this itself may be a sufficient disincentive to cause the developer to look for an alternative site or to place the facility at greater depth. There are a few site characteristics that have the capability to eliminate a site from further consideration altogether, and these are generally nontechnical. Sites that have special religious, social or wildlife significance would fall into this category and might include, for example, cemeteries, battle sites and national parks.

3.6.4 Engineered barriers It is normal for developers to design repositories following the multi-barrier approach. This means that the design deploys a series of engineered and natural barriers that, so far as possible, act independently of each other so that, if one or two were to fail, safety would still be assured. The first engineered barrier usually consists of the waste form, which is

© Woodhead Publishing Limited, 2010

70

Geological repository systems for safe disposal

often created in two stages: conditioning and immobilisation. Conditioning is required if the waste has an unsuitable physical or chemical form; liquid and gaseous wastes, for instance, will not permitted to be disposed in this form and must be converted into solid substances. Similarly, chemically active wastes (e.g. alkali metals) must be conditioned by chemical reaction to form a more passive material. Immobilisation (sometimes called encapsulation) means that the conditioned waste is embedded in a solid matrix; this is usually concrete but could be a polymeric material or, more unusually these days, bitumen. When the waste is directly mixed with an immobilising material the resulting waste form is usually described as monolithic. The immobilisation matrix is likely to have a number of safety functions, such as radiation shielding during handling operations, reduction of the rate at which radionuclides may be leached from the waste by water, and consolidation of finely divided material. The conditioned and immobilised waste, i.e. the waste form, will invariably be held within a container and the whole is called a waste package. Containers are usually made from mild steel, cast iron, stainless steel or concrete. As already explained, the repository structure may take a number of forms. In some cases, the geometry of a facility may be simply determined by the footprint of the site. For near-surface disposal, semi-buried concrete vaults are commonly used. Where the vaults take the form of concrete-lined trenches, these may be subdivided into cells through the use of cross walls. Waste packages are emplaced into the cells or vaults and may be backfilled with concrete or previously excavated earth so as to fill up the gaps between the waste packages. The purpose of backfilling is to reduce voidage and, therefore, settlement of the waste so that subsequent shrinkage of the waste stack does not cause the repository cap to fail. The backfill may also act as a sorbent that reduces the rate of migration of radionuclides in groundwater. A repository cap may then be placed over the cells for the purpose of reducing the infiltration of meteoric water. Finally, institutional control, perhaps in the form of a security presence, fences, warning notices, etc., is implemented for the purpose of avoiding damage due to human intrusion. Depending primarily upon the nature of the site and the wastes, different kinds and combinations of engineered barriers will be deployed so as to achieve the requisite levels of operational and post-closure safety. For mining and milling wastes, for example, engineered barriers may consist of little more than an earth cover, measures to promote mechanical stability and regular monitoring of the condition of the mounded wastes. For LILW, on the other hand, we might expect to see an impermeable cap, monolithic waste forms, and an active security presence that is intended to prevent unauthorised access for hundreds of years.

© Woodhead Publishing Limited, 2010

Disposal of intermediate radioactive waste

71

3.6.5 Natural barriers Natural barriers are less important for near-surface disposal than for deep geological disposal primarily because, if an unwitting human intruder were to take up residence on top of the facility, there would be no natural barriers, only engineered barriers, between him and the waste. Nonetheless, the surrounding environment (and therefore the potential natural barriers) will be important in determining the repository geometry – especially its depth, which may be chosen to keep the base of the repository above the water table. Natural barriers are likely to have raised importance in fault scenarios (e.g. premature failure of the cap) or at very long timescales when, although radionuclides or toxic materials are present in only small quantities, all the engineered barriers will have become degraded. In a near-surface facility where infiltrating meteoric water leaches out radionuclides and carries them down to an underlying aquifer, the natural barriers are represented by (a) the unsaturated rock that lies between the base of the repository and the underlying aquifer and (b) the saturated rock of the underlying aquifer that lies between the base of the repository and the so-called discharge point, i.e. the location at which this contaminated water might be utilised for drinking and other uses. Often the discharge point is assumed to be a water abstraction well (e.g. IAEA, 2003b). In these circumstances, the effectiveness of the natural barriers is largely expressed through the following parameters: the net rate of infiltration of water through the wastes, the radionuclide retention coefficients for the saturated and unsaturated rocks, the distance between the base of the repository and the groundwater table, and the hydraulic conductivity, porosity and hydraulic gradient in the aquifer. The geochemical properties of the surrounding environment are also important, but usually in the sense that certain geochemical regimes are to be avoided, especially where natural groundwater comes into contact with repository materials. In near-surface disposal the most usual example is the avoidance of unwanted cement–groundwater reactions, such as sulphateinduced degradation of concrete. At greater depths – in borehole disposal, for example – high chloride levels in groundwater could lead to rapid steel corrosion.

3.6.6 Safety functions A very useful way of defining and describing the roles of the engineered and natural barriers in providing adequate levels of operational and post-closure safety is through an explanation of their safety functions. It has been found that this helps the project team to adopt a consistent approach to meeting the design aims; it may also be an important input to the regulatory review. In the case of a near-surface disposal, for example, we would expect to see

© Woodhead Publishing Limited, 2010

72

Geological repository systems for safe disposal

every major engineered component listed together with an explanation of its role in providing operational and/or post-closure. This may be conveniently presented as a table that should also show the timeframe over which each barrier is expected to operate.

3.7

Future trends

3.7.1 Remediation of historical near-surface disposal facilities In a few countries there are a number of disposals that were performed under (now) outdated regulation. Such disposals were legally made but they would not be permitted under modern-day regulation. The question then arises of whether these disposals should be remediated. This could be a relatively minor action such as improvements to the repository cap or, at the other extreme, it could include retrieval of the wastes followed by repackaging and a fresh disposal. ICRP (2000) has provided useful guidance on how to evaluate such a situation based on the principle that such a remediation should do more good than harm. In making such an evaluation, the risks to future generations are balanced against the risks to workers who would be required to perform the remediation. On the assumption that older regulations will not be so very different from those pertaining to the present day, there is a general expectation that if the disposals were compliant with the then-prevailing regulations, remediation will not extend so far as retrieval. There exist, however, some sites where such old disposals sit alongside ongoing disposals that are being made in compliance with current regulation. When assessing the post-closure safety of such sites, it will usually be necessary to consider the site as a whole, i.e. including all radionuclide releases for all facilities on the site. Given that new near-surface disposal sites can be extremely difficult to create, it is this aspect, rather than the approach recommended by ICRP (2000), that is likely to drive the need (or not) to remediate. Where such a situation is discovered, and pending a definitive resolution (which may require a government decision), the regulator may require future waste emplacements at the site to be regarded as storage, which entails certain requirements with respect to waste retrieval. Such situations have arisen at the Drigg site in the UK (Fig. 3.9) and at RADON sites in the Russian Federation. Given that the availability of radioactive waste disposal sites seems unlikely to keep pace with the increasing volume of waste requiring disposal, we may anticipate that this situation will be repeated elsewhere.

3.7.2 Intermediate depth disposal Intermediate depth disposal is an emerging trend for wastes that are unsuitable for near-surface disposal but not so radioactive or long-lived that

© Woodhead Publishing Limited, 2010

Disposal of intermediate radioactive waste

73

3.9 The Drigg LLW repository in north-west England showing the location of past, present and possible future disposals. Since the photograph was taken, Vault 8 has been nearly filled (courtesy Sellafield Ltd).

they require deep geological disposal. The archetypal examples of this are the intermediate depth facilities proposed in Japan and France (Section 3.5.3). In some ways, however, the term ‘intermediate depth’ is unfortunate because it implies that the needed depth of disposal directly correlates with the activity of the wastes – what might be simply (if crudely) paraphrased as ‘intermediate depth disposal for intermediate-level wastes’. In fact, depth of disposal is as much a function of the chosen site and stakeholder perceptions as the nature of the wastes. Compare, for instance, the different approaches towards LILW disposal in Canada and Belgium that were described in Sections 3.5.2 and 3.6.1. Nevertheless, proposals for intermediate depth disposal are indicative of a more optimised approach in which the economic benefits of disposal at an appropriate depth are brought to the fore.

3.7.3 Borehole disposal As described in Section 3.5, at the present time borehole disposal facilities may be primarily seen as a variant of near-surface disposal. For disposals at greater depths, borehole facilities offer two significant advantages. The first is flexibility in terms of disposal depth and choice of host rocks. The second is cost, which should be lower because of the reduced volume of rock that needs to be excavated per unit volume of waste. This is especially important for the many countries that, while they have no nuclear programme, do have significant numbers of long-lived disused radioactive sources requiring

© Woodhead Publishing Limited, 2010

74

Geological repository systems for safe disposal

disposal. For these countries the IAEA BOSS concept, or something similar, may offer a permanent solution at reasonable cost. The two great advantages of borehole disposal – flexibility and cost – are not limited to disused sealed sources; the use of very deep boreholes for disposal of spent fuel and HLW has been mooted by, for instance, SKB (1989). It seems to us, nonetheless, that development beyond the conceptual stage is unlikely for the foreseeable future given the investments already made in ‘conventional’ repository technology.

3.8

Sources of further information and advice

Almost without exception, national waste management agencies have policies promoting openness and transparency that require them to provide virtually free access to their reports and research findings. Many of these reports are available in English and most of them can be freely downloaded on the Internet. Hard copies can also be ordered over the Internet or requested by mail. Where hard copies are supplied, there may be a charge to cover administrative costs and postage. The British, Finnish, French, Swedish and US agencies (DOE) are particularly useful in this respect. The relevant organisations, addresses and websites are shown below. Organisation Full name and address Agence Nationale pour la Gestion des Déchets Andra Radioactifs 1/7, rue Jean Monnet, Parc de la Croix-Blanche, 92298 Châtenay-Malabry, Cedex, France International Atomic Energy Agency, Wagramer IAEA Strasse 5, PO Box 100, Vienna, Austria Nuclear Decommissioning Authority, Waste NDA Management Division (formerly UK Nirex Ltd) Curie Ave, Harwell, Didcot, Oxon, UK Nuclear Energy Agency of the Organisation for NEA-OECD Economic Development and Cooperation Issy-les-Moulineaux, Paris, France

Website www.andra.fr

www.ieae.org www.nda.gov.uk

www.nea.fr

Posiva

Posiva Oy, Olkiluoto, FI-27160 Eurajoki, Finland

SCK-CEN

Studiecentrum voor Kernenergie/Centre d'Etude www.sckcen.be de l'Energie Nucléaire Boeretang 200, BE-2400 MOL, Belgium

SKB

Svensk Kärnbränslehantering, Blekholmstorget www.skb.se 30, Box 250, SE-101 24 Stockholm

US DOE

US Department of Energy www.energy.gov 1000 Independence Ave, SW, Washington, DC 20585, USA

© Woodhead Publishing Limited, 2010

www.posiva.fi

Disposal of intermediate radioactive waste

75

The website of the Belgium research organisation SCK/CEN is also a useful source. Work performed within the European Community Research Framework programmes tends to be listed on project-specific websites and also on the websites of the organisations involved. Two other useful sources, both for international regulatory standards and technical information, are the IAEA and OECD-NEA. Virtually all IAEA reports (including all those listed in the references) and even draft safety standards are available in downloadable form.

3.9

IAEA requirements for geological disposal

IAEA requirements for geological disposal (IAEA, 2006b) are listed below. Requirement 1. Government responsibilities The government is required to provide an appropriate national legal and organisational framework within which disposal facilities for radioactive waste can be sited, designed, constructed, operated and closed. This shall include: confirmation at a national level of the need for different types of disposal facilities; the definition of the steps in the development and licensing for different types of facility; the clear allocation of responsibilities, the securing of financial and other resources, and the provision of independent regulatory functions related to each planned disposal facility. Requirement 2. Regulatory body responsibilities The regulatory body shall establish the regulatory requirements for the development of each type of disposal facility and shall set out the procedures for meeting the requirements for the various stages of the licensing process. It shall also set conditions for the development, operation and closure of each individual disposal facility and shall carry out such activities as are necessary to ensure that the conditions are met. Requirement 3. The responsibilities of the operator The operator of a disposal facility shall be responsible for its safety. The operator shall carry out safety assessments and develop a safety case, and shall carry out all the necessary activities for siting, design, construction, operation and closure, in compliance with the regulatory requirements and within the national legal infrastructure. Requirement 4. Importance of safety in the development process Throughout the development of a disposal facility, an appropriate understanding of the relevance and implications for safety of the available

© Woodhead Publishing Limited, 2010

76

Geological repository systems for safe disposal

options shall be developed by the operator, for achieving the ultimate goal of providing an optimised level of operational and post-closure safety. Requirement 5. Passive safety The operator shall site, design, construct, operate and close the disposal facility in such a way that safety is ensured by passive means to the extent possible and the need for actions to be taken after the closure of the facility is minimised. Requirement 6. Understanding and confidence in safety The operator of a disposal facility shall develop an adequate understanding of the facility and its host environment and the factors that influence its postclosure safety over suitably long time periods, so that a sufficient level of confidence in safety is achieved. Requirement 7. Multiple safety functions The host environment shall be selected and the engineered components of the facility designed so as to ensure that post-closure safety is provided by means of multiple safety functions; i.e. containment and isolation of the waste shall be provided by means of a number of barriers whose performance is achieved by diverse physical and chemical processes and whose individual adequacy shall be demonstrated. The overall performance of the disposal system shall not be unduly dependent on a single barrier or safety function. Requirement 8. Containment The engineered barriers, including the waste form and packaging, shall be so designed, and the host environment shall be so selected, as to provide containment of the waste during the period when radioactive decay has not yet significantly reduced the hazard posed by the waste and in the case of heat-generating waste when the waste produces heat energy in amounts that could adversely affect the containment. Requirement 9. Isolation The disposal facility shall be sited in a host environment depth that provides isolation of the waste from the biosphere and from humans over time periods of several hundreds of years for short-lived waste and at least several thousand years for high-level waste, with account taken of both the natural evolution of the disposal system and disturbing events. Requirement 10. Post-closure supervision An appropriate level of supervision shall be applied in order to protect and preserve the passive safety barriers to the extent that this is needed in order to fulfil the functions that they are assigned in the post-closure safety case. Requirement 11. Step-by-step development and evaluation

© Woodhead Publishing Limited, 2010

Disposal of intermediate radioactive waste

77

Disposal facilities shall be developed, operated and closed in a series of steps, each supported, as necessary, by iterative evaluations of the site, of the options for design, construction, operation and management, and of the performance and safety of the disposal system. Requirement 12. Preparation of the safety case and safety assessment A safety case and supporting safety assessment shall be prepared and updated by the operator, as necessary, at each step in the development, operation and closure of a disposal facility. The safety case and safety assessment shall be sufficiently detailed and comprehensive to provide the necessary technical input for informing the regulatory and other decisions needed at each step. Requirement 13. Scope of the safety case and safety assessment The safety case for a disposal facility shall describe all the safety relevant aspects of the site, the design of the facility, and the managerial and regulatory controls. The safety case and its supporting assessments shall illustrate the level of protection provided and shall provide assurance that safety requirements will be met. Requirement 14. Documentation of the safety case and safety assessments The safety case and its supporting safety assessments shall be documented to a level of detail and quality sufficient to inform and support the decision to be made at each step and to allow for their independent review. Requirement 15. Site characterisation The site for a disposal facility shall be characterised at a level of detail sufficient to support both a general understanding of the characteristics of the site, including its present condition, its probable natural evolution, possible natural events and also human plans and actions that may affect the facility or its vicinity over the period of interest with regard to safety, and a specific understanding of the impact on safety of features, events and processes associated with the site and the facility. Requirement 16. Design The disposal facility and its engineered barriers shall be designed to contain the waste with its associated hazard, to be physically and chemically compatible with the host geological and/or surface environment and to provide post-closure safety features that complement those afforded by the host environment. The facility and its engineered barriers shall also be designed to ensure safety during the operational period. Requirement 17. Construction A disposal facility shall be constructed in accordance with the design as described in the approved safety case and safety assessments. It shall be

© Woodhead Publishing Limited, 2010

78

Geological repository systems for safe disposal

constructed in such a way as to preserve the post-closure safety functions of the host environment that have been shown to be important by the safety case. The construction activities shall be carried out to ensure safety during the operational period. Requirement 18. Operation A disposal facility shall be operated in accordance with the conditions of the licence and the relevant regulatory requirements to maintain safety during the operational period, and in such a manner as to preserve the post-closure safety functions assumed in the safety case. Requirement 19. Closure A disposal facility shall be brought to closure in such a way that the safety functions shown by the safety case to be important for the post-closure period are provided for. Plans for closure, including the transition from active management of the facility, shall be well defined and practicable, so that closure can be carried out safely at an appropriate time. Requirement 20. Waste acceptance Waste packages and unpackaged waste accepted for emplacement in a disposal facility shall conform to criteria fully consistent with and derived from the safety case for the operational and post-closure safety of the disposal facility. Requirement 21. Monitoring programmes A programme of monitoring shall be carried out prior to and during the construction and operation of a disposal facility, and after closure if this is part of the safety case. This programme shall be designed to collect and update the information needed to confirm the conditions necessary for the safety of workers and members of the public and the protection of the environment during the operation of the facility and to confirm the absence of any conditions that could reduce the post-closure safety of the facility. Requirement 22. Post-closure and institutional controls Plans shall be prepared for the post-closure period to address institutional control and the arrangements for maintaining the availability of information on the disposal facility. These plans shall be consistent with passive safety and shall form part of the safety case on which authorisation to close the facility is granted. Requirement 23. Consideration of nuclear safeguards Nuclear safeguard requirements shall be considered in the design and operation of any disposal facility that may accept materials or wastes to which nuclear safeguards apply. Nuclear safeguards shall be implemented in such a way as not to compromise the safety of the disposal facility.

© Woodhead Publishing Limited, 2010

Disposal of intermediate radioactive waste

79

Requirement 24. Management systems Management systems* to provide for assurance of quality shall be applied to all safety-related activities, systems and components throughout all the steps of the development and operation of a disposal facility. The level of assurance for each element shall be commensurate with its importance to safety. Requirement 25. Existing disposal facilities The safety of existing disposal facilities shall be assessed periodically, when a modification is planned, when changes with respect to the authorised conditions happen and at least every ten years. In the event that any safety requirements are not met, measures shall be put in place to upgrade the safety of the facility.

3.10

References

Baker A J, Chambers A V, Jackson C P, Porter J D, Sinclair J E, Sumner P J, Thorne M C and Watson S P (1997), ‘Nirex 97: An assessment of the post-closure performance of a deep waste repository at Sellafield; Volume 3: The groundwater pathway’, Nirex Report S/97/012-3. Biurren E, Haverkamp B and Kuc´erka M (2005), ‘Closure of a chamber in the Richard Underground Repository’, in WM’05 Conference, 27 February–3 March 2005, Tucson, Arizona. Cochran J R and Hasan A (2005), ‘Greater confinement disposal of radioactive waste in borehole facilities’, in Disposal of Low Activity Radioactive Waste: Proceedings of an International Symposium on Disposal of Low Activity Radioactive Waste, 13–17 December 2004, Cordoba, Spain, Session 5, p. 393, IAEA, Vienna. Cochran J R, Crowe B and Colarusso A (2001a), ‘Results of the performance assessment for the classified transuranic waste disposed at the Nevada Test Site (1)’, in WM’01 Conference, 25 February–1 March 2001, Tucson, Arizona. Cochran, J R et al. (2001b), ‘Compliance assessment document for the transuranic wastes in the greater confinement disposal boreholes at the Nevada Test Site, Volume 2: Performance assessment’, SAND2001-2977, Sandia National Laboratories, Albuquerque, New Mexico. Crossland I G, Jova-Sed L A and Nachmilner L (2011), ‘A comprehensive system for the disposal of disused sealed sources’, in Radioactive Waste Management: Science and Technology, A special edition of International Journal of Environmental Engineering Science (IJEES), 2011, 3(1), in press. Dutzer M (2002), ‘From waste packages acceptance criteria to waste packages acceptance process at the Centre de l’Aube Disposal Facility’, in Radioactive Waste Products (RADWAP 2002) Proceedings, edited by R D Odoj, P Baier, P * The term ‘management system’ reflects and includes all the initial concepts of ‘quality control’ (controlling the quality of products) and its evolution through ‘quality assurance’ (the system to ensure the quality of products) and ‘quality management’ (the system to manage quality).

© Woodhead Publishing Limited, 2010

80

Geological repository systems for safe disposal

Brennecke and K Kuehn, 23–26 September 2002, Forschungzentrum Juelich GmbH, Germany. FAO et al., Food and Agriculture Organization of the United Nations, International Atomic Energy Agency, International Labour Organisation, Nuclear Energy Agency of the Organisation for Economic Co-Operation and Development, Pan American Health Organization and World Health Organization (1996) International Basic Safety Standards for Protection against Ionizing Radiation and for the Safety of Radiation Sources, Safety Series 115, IAEA, Vienna. Hartley B M, Wall B, Munslow-Davies L, Toussaint L F, Hirschberg K-J, Terry K W and Shepherd M (1994) ‘The establishment of a radioactive waste disposal facility in Western Australia for low level waste’, in Proceedings of the 9th Pacific Basin Nuclear Conference, Nuclear Energy, Science and Technology Pacific Partnership, 1–6 May 1994, Sydney, Australia. IAEA (1994a), ‘Classification of radioactive waste’, Safety Series, Safety Guide G1.1, International Atomic Energy Agency, Vienna. IAEA (1994b), ‘Siting of near surface disposal facilities’, Safety Series, Safety Guide No. 111-G-3.1, International Atomic Energy Agency, Vienna. IAEA (1999), ‘Near surface disposal of radioactive waste’, Safety Standards Series No. WS-R-1, International Atomic Energy Agency, Vienna. IAEA (2000a), ‘Inspection and verification of waste packages for near surface disposal’, TECDOC-1129, International Atomic Energy Agency, Vienna. IAEA (2000b), ‘Predisposal management of radioactive waste, including decommissioning’, Safety Standards Series WS-R-2, International Atomic Energy Agency, Vienna. IAEA (2003a), ‘Derivation of activity limits for the disposal of radioactive waste in near surface disposal facilities’, TECDOC-1380, International Atomic Energy Agency, Vienna. IAEA (2003b), ‘Radioactive waste management glossary’, International Atomic Energy Agency, Vienna. IAEA (2003c), ‘Predisposal management of low and intermediate level radioactive waste’, Safety Standards Series WS-G-2.5, International Atomic Energy Agency, Vienna. IAEA (2004a), ‘The long term stabilization of uranium mill tailings. Final Report of a Co-ordinated Research Project 2000–2004’, TECDOC-1403, International Atomic Energy Agency, Vienna. IAEA (2004b), ‘Treatment of liquid effluent from uranium mines and mills. Report of a Co-ordinated Research Project 1996–2000’, TECDOC-1419, International Atomic Energy Agency, Vienna. IAEA (2006a), ‘Fundamental safety principles’, Safety Series SF1, International Atomic Energy Agency, Vienna. IAEA (2006b), ‘Geological disposal of radioactive waste safety requirements’, Safety Standards Series WS-R-4, International Atomic Energy Agency, Vienna. IAEA (2009a), ‘Classification of radioactive waste, general safety guide,’ Safety Standards Series GSG-1, International Atomic Energy Agency, Vienna. IAEA (2009b), ‘Borehole disposal facilities for radioactive waste’, Safety Standards Series SSG-1, International Atomic Energy Agency, Vienna. IAEA (2010a), ‘Near surface disposal of radioactive waste’, Draft Safety Guidance DS356, International Atomic Energy Agency, Vienna, in preparation.

© Woodhead Publishing Limited, 2010

Disposal of intermediate radioactive waste

81

IAEA (2010b), ‘Geological disposal of radioactive waste’, Draft Safety Guidance DS334, International Atomic Energy Agency, Vienna, in preparation. IAEA (2010c), ‘BOSS technical manual’, Waste Technology Section, International Atomic Energy Agency, Vienna, in preparation. ICRP (2000), ‘Protection of the public in situations of prolonged radiation exposure. The application of the Commission’s system of radiological protection to controllable radiation exposure due to natural sources and long-lived radioactive residues’, ICRP Publication 82, Pergamon. Morales A (2005), ‘El Cabril low and intermediate level waste disposal facility (Spain): new acceptance criteria’, in WM’05 Conference, 27 February–3 March 2005, Tucson, Arizona. NEA (1987), ‘Shallow land disposal of radioactive waste: reference levels for the acceptance of long lived radionuclides. A Report by an NEA Expert Group’, Nuclear Energy Agency of the Organisation for Economic Co-Operation and Development, Paris. Ouzounian G and Ozanam O (2009), ‘Disposal project for low-level long-lived radioactive waste in France’, in WM2009 Conference, 1–5 March 2009, Phoenix, Arizona. Prozorov L, Tkatchenko A, Titkov V and Korneva S (2001), ‘Prospects of large diameter well construction at ‘‘RADON’’ sites’, in WM’01 Conference, 25 February–1 March 2001, Tucson, Arizona. SKB (1989), ‘Storage of nuclear waste in very deep boreholes: feasibility study and assessment of economic potential’, SKB Technical Report 89-39, SKB, Stockholm. Skogsberg M and Ingvarsson R (2006), ‘Operational experience from SFR – Final repository for low- and intermediate-level waste in Sweden’, in Topseal Conference Session III, September 2006, http://www.euronuclear.org/events/ topseal/transactions/ . Sobolev I A, Ojovan M I and Karlina O K (2001), ‘Management of spent sealed radioactive sources at regional facilities ‘‘RADON’’ in Russia’, in ICEM’01 8th International Conference on Environmental Management, 30 September–4 October 2001, Bruges, Belgium. STOLA, 2004, ‘Belgian low-level and short-lived waste: does it belong in Dessel? Choosing a sustainable solution’, STOLA, Dessel, November 2004. Note that STOLA ceased to exist in July 2005 and was immediately succeeded by STORA, which is tasked with monitoring all nuclear affairs in Dessel. STOLA documents are downloadable from http://www.stora.be/. Yamato A (2005), ‘The current status of the JNFL subsurface disposal plan for relatively high low level radioactive waste’, in Proceedings of an International Conference: Safety of Radioactive Waste Disposal, Tokyo, 3–7 October 2005, p. 303, IAEA, Vienna.

© Woodhead Publishing Limited, 2010

4 Underground research facilities and rock laboratories for the development of geological disposal concepts and repository systems I. BLECHSCHMIDT and S. VOMVORIS, National Cooperative for the Disposal of Radioactive Waste–NAGRA, Switzerland

Abstract: Underground research laboratories (URLs) play an important and multi-faceted role in the development of deep geological repository systems for the disposal of radioactive waste, both from a scientific and technological point of view as well as for building public confidence. For more than three decades several countries have been conducting extended experimental and demonstration programmes in such facilities. The results produced from these investigations in URLs have proven to be valuable both in generic terms, i.e. developing and assessing the disposal concept and improving general acceptance, and in specific terms, as an essential means for detailed characterisation, design and assessment of potential repository systems. Key words: underground research laboratories, role in radioactive waste disposal programmes, case studies, Grimsel Test Site, Mont Terri Project, in situ experiments, large-scale experiments, bentonite, FEBEX, CRR, LTD, GMT, ESDRED, TEM, HG-A.

4.1

Introduction

4.1.1 Definition and roles of underground rock laboratories (URLs) The acronym ‘URL’ stands for underground research laboratory or underground rock laboratories. It has become the accepted generic term for underground facilities in which activities are carried out in support of radioactive waste repository development programmes (NEA, 2001a). A URL is any underground facility in which characterisation, testing, 82 © Woodhead Publishing Limited, 2010

Underground research facilities and rock laboratories

83

Table 4.1 Relationship between laboratory studies, in situ experiments in URLs and natural analogues (modified after Kickmaier, 2002) Method

Characteristics

Laboratory experiments (‘classic’ lab studies)

Well-defined boundary Weeks to years conditions, unrealistic environment Defined but complex Several years to decades boundary conditions, realistic environment (‘repository relevant’) Boundary conditions less Up to millions of years well defined, realistic environment

Underground research laboratory (in situ experiments) Natural analogues

Duration of the ‘experiments’ (observation period)

technology development and/or demonstration activities are carried out in support of the development of deep geological repositories for radioactive waste. It may be an elaborate, purpose-built facility in which large research programmes can be carried out over many years or a quite simple facility, for example one attached to existing underground excavations, in which quite specific investigations may be made. URLs are located in rocks that are considered to be suitable for the construction of radioactive waste repositories, such as granite, salt, clay/shale, volcanic tuff, etc. Experiments in underground rock laboratories represent one of the complementary approaches that are needed for building the scientific understanding and for gathering the diversity of data to analyse the potential behaviour of repository systems over long time spans. URLs provide the link between investigations in ‘classic’ laboratories, where spatial scale is quite small, and natural analogues, where spatial and temporal scale are much larger (Table 4.1). The work in URLs allows an in-depth investigation of the selected geological environment and the engineered components of the repository, providing the opportunity to test models at more appropriate (repositoryrelevant) scales and boundary conditions than can be achieved from the surface only. Consequently, the RD&D programmes in the URLs form an important part of the overall radioactive waste management strategies (Fig. 4.1). In national radioactive waste programmes the general objectives of underground research facilities can be summarised as follows (modified after McCombie and Kickmaier, 2000): . .

Developing the technology and methodology needed for underground experimentation (especially important for countries with little tradition of tunnelling, mining or resource exploitation) Providing data that improve the basic understanding of the behaviour of

© Woodhead Publishing Limited, 2010

84

Geological repository systems for safe disposal

4.1 URLs as part of the overall waste management strategy (modified after Kickmaier, 2002).

. .

repository system components – and of their interactions – as input to qualitative and quantitative assessments Demonstration and, at an advanced stage, optimisation of engineering components and system processes Building confidence within the scientific and technical community as well as within the public that the understanding of the important processes governing repository performance is adequate

The value of URLs in communicating with the public and with ‘decision makers’ has been recognised as a valuable non-technical asset. Although the nature of the facilities and their often remote locations preclude site visits by vast numbers of the public, many thousands of members of the general public are received each year in the operating URLs. In addition, illustrative material from experiments in these facilities is frequently used in public documents on disposal produced from within and outside the waste management community. A more recent trend has been towards the use of URLs to demonstrate the feasibility of the technology proposed for implementation in underground repositories. Currently many experimental programmes in URLs are focusing on engineering practicability and demonstrating the behaviour of large multi-component systems. The high costs and long timescales that must be considered for such experiments make them an appropriate focus for international cooperation.

© Woodhead Publishing Limited, 2010

Underground research facilities and rock laboratories

85

4.1.2 Chapter set-up Section 4.1 gives an introduction to underground research laboratories (URLs). The different types of URLs and their role in the development of deep geologic repositories are described in Section 4.2. Section 4.3 summarises the basics for planning and designing an underground research facility taking into account most recent examples. Public outreach and the role of URLs as training platforms are discussed in Section 4.4. A few case studies of in situ characterisation and testing, engineered barriers and demonstration experiments are highlighted in Section 4.5. The conclusions and future needs are summarised in Section 4.6.

4.2

Types of URLs and their role in the staged development of geological repositories

4.2.1 Different types of URLs Many different definitions have been used in the past thirty years to describe the work which is performed in URLs such as ‘first-generation URLs’, ‘rock characterisation facilities – RCFs’, ‘off-site URLs’, ‘second-generation URLs’, ‘site-specific URLs’, ‘on-site URLs’ or ‘performance confirmation facilities’, are some of the most frequently used names. As the objectives between the various types of URLs can differ a lot, leading to a certain degree of confusion, for the remainder of this chapter we are using the most accepted definitions, as is shown below. We will distinguish URLs in two major categories: . .

Generic URLs Site-specific URLs

Generic URLs (equivalent to off-site URLs or first-generation URLs) are independent of final disposal sites and comprise facilities that are developed for research and testing purposes at a site that will not be used for waste disposal; they provide, however, information that is expected to support the disposal of radioactive waste elsewhere. Both non-destructive as well as destructive experiments are performed (experiments often culminate with a post-mortem phase) and a flexible approach, in which modifications are possible, is warranted. The major roles of generic URLs are: . .

Development of methodology (preparations for in situ testing – surface/ underground) and testing of the transferability of data obtained in the laboratory to in situ tests (e.g. sorption processes, rock mechanics) Collection of information (host rock – barrier properties, engineering feasibility, interactions between host rock and engineered barriers, work

© Woodhead Publishing Limited, 2010

86

. .

Geological repository systems for safe disposal process, empirical observations, data sets/observations for model testing, independent evidence, data for direct use, e.g. by use of transfer factors, modelling, etc.) Concept testing and demonstration (e.g. engineered barriers – emplacement, evolution, retrieval) Platform for interaction (professional community – different organisations, different disciplines; broad public)

Site-specific URLs (on-site URLs, second-generation URLs or rock characterisation facilities) as the name suggests, are located in the host rock in an area that is considered as a potential location for a repository. They include facilities that are developed for specific investigations at the given site and may, indeed, be a forerunner to the development of a repository at the site. Because they may become part of a future repository, activities performed as well as the URL itself should not unduly affect the potential future repository; for example, they should not have a negative impact on the host rock performance and their planning and execution is part of the development of the specific repository project with more formal requirements. The major differences between these two URL categories are summarised in Fig. 4.2. A special case within the site-specific URLs is the ‘performance confirmation facility’, which is continuously gaining more relevance as the

4.2 Major differences between ‘generic URLs’ and ‘site-specific URLs’ (based on Kickmaier, 2002).

© Woodhead Publishing Limited, 2010

Underground research facilities and rock laboratories

87

Table 4.2 Overview of generic Underground Research Laboratories (modified after IAEA, 2001; NEA, 2001a; NEA 2001b) Generic URL Operation

Host rock, depth

Comments

Whiteshell Canada Underground AECL Research Laboratory (URL) 1984–2003

Granite 240–420 m

Purpose built; generic Shaft sealing on-going

Olkiluoto Research Tunnel 1992–

Finland Posiva

Granite (tonalite) 60–100 m

Purpose built; parallel to repository facilities

Amelie 1986–1992

France ANDRA

Bedded salt

Pre-existing tunnels

Fanay-Auge`res 1980–1990

IRSN

Granite Uranium mine

Pre-existing tunnels

Sediments (shale) 250 m

Pre-existing tunnels

Permian rock salt anticline Mining levels 490– 800 m, cavern 950 m

Pre-existing tunnels

Tournemire (Tournemire Research Tunnel) 1990– Asse Mine 1965–1997

Country Organisation

IRSN Germany GSF

Tono 1986–2006

Japan JNC/JAEA

Sediments, uranium Pre-existing tunnels mine 130 m

Kamaishi 1988–1998

Granite, Fe mine

Pre-existing tunnels

Mizunami Underground Research Laboratory (MIU) 2004–(shaft sinking initiation)

Granite 1000 m (shaft)

Purpose built; generic Surface investigations since 1996

Horonobe Underground Research Laboratory 2005–(shaft sinking initiation)

Sedimentary rock 500 m (shaft)

Purpose built; generic Surface investigations since 2000

KURT– Korean Underground Research Tunnel 2006–

South Korea KAERI

Granite 90 m

Purpose built; generic

Stripa Mine 1976–1992

Sweden SKB

Granite, Fe mine 360–410 m

Pre-existing tunnels

© Woodhead Publishing Limited, 2010

88

Geological repository systems for safe disposal

Table 4.2 (cont.) Generic URL Operation

Country Organisation

¨ spo¨ Hard Rock A Laboratory (HRL) 1995–

Host rock, depth

Comments

Granite 200–460 m (ramp/ spiral)

Purpose built; generic

Grimsel Test Site (GTS) 1984–

Switzerland Nagra

Granite 450 m

Purpose built; parallel to existing tunnels

Mt Terri (FMT) 1995–

SNHGS

Opalinus Clay 400 m

Purpose built; parallel to existing tunnels

Climax, Nevada 1978–1983

USA US-DOE

Granite, mine 420 m

Pre-existing tunnels

G-Tunnel, Nevada 1979–1990

Tuff 300 m

Pre-existing tunnels

Busted Butte, Yucca Mountain, Nevada 1998–

Bedded tuff 100 m

Purpose built; generic

development of deep disposal of radioactive waste moves forward in several countries. It is expected that in the near future they will warrant the designation of ‘third-generation URLs’ and they will comprise facilities designed for confirming key phenomena of importance during and after waste emplacement in an existing repository, e.g. provide input to the decision-making for the final closure of the repository.*

4.2.2 Past and present URLs Tables 4.2 and 4.3 summarise past and present URLs in a succinct form. Consistent with the definitions above, the underground facilities are broadly categorised as in generic and site-specific URLs. The first category is further divided into the ones using pre-existing underground facilities, i.e. tunnels originally developed for different purposes, and the purpose-built laboratories. Examples of generic URLs are presented in Figs 4.3 to 4.7. During the last 40 years more than 20 facilities have been utilised for underground research and testing (Table 4.2); about one-third of these facilities are currently in active operation or under construction. The number of site-specific facilities, as expected, is increasing, e.g. the ONKALO facility in Finland (Fig. 4.11), the Bure URL (Meuse/Haute *

Excerpt from P. Zuidema’s presentation at the celebration ceremony for the 25th anniversary of the Grimsel Test Site (GTS) in Switzerland (September 2009).

© Woodhead Publishing Limited, 2010

Underground research facilities and rock laboratories

89

Table 4.3 Overview of site-specific Underground Research Laboratories (modified after IAEA 2001; NEA 2001a; NEA 2001b) Site-specific URL Operation

Country Organisation

Host rock, depth

Comments

High-Activity Disposal Experiment Site URL, Mol (HADES-URF) 1984–

Belgium Boom clay (plastic Purpose built; sitespecific GIE EURDICE clay) 230 m

ONKALO, Olkiluoto 2003–

Finland Posiva

Granite (tonalite) 500 m (ramp)

Purpose built; sitespecific

Meuse/Haute-Marne (Bure URL) 2000–

France ANDRA

Shale (indurated clays) 450–500 m

Purpose built; sitespecific

Gorleben 1985–1990

Germany BfS, DBE

Salt dome below 900 m

Purpose built; sitespecific

Konrad 1980–

Limestone, Fe mine Facility in former 800–1300 m iron mine

Morsleben (ERAM) 1981–1998

Salt dome, K/salt mine below 525 m

Facility in decommissioning stage

Salt (bedded) 655 m

Operating repository since 1999*

Welded tuff 300 m (ramp)

Purpose built; sitespecific

Waste Isolation Pilot Plant (WIPP) 1982–(1999) Exploratory Studies Facility (ESF), Yucca Mountain, Nevada 1996–

USA US-DOE

* A part of the underground facility continues as a performance confirmation facility or third-generation URL.

Marne) in France (Fig. 4.10) and the Exploratory Studies Facility (ESF) at the proposed Yucca Mountain site in the USA. The ESF (Fig. 4.9) has already completed a broad in situ experimentation programme at the rock formation (tuff) of the proposed high-level waste repository. At the time of writing this text, the Bure URL has completed the first phase of tests at the formation of interest (Callovo-Oxfordian) and is developing the next phase of underground testing. In Spring 2009, ONKALO passed the mid-point of the target depth and had already characterised important geological features crossed by the access ramp. Both in Finland and in the USA the intention has been to integrate part of the URL in the infrastructure of the repository facility. For ANDRA’s programme in the Meuse/Haute Marne region this is still an open question and depends on the final location of the repository area, which is to be proposed around 2015. Examples of site-specific URLs are presented in Figs 4.8 to 4.11.

© Woodhead Publishing Limited, 2010

90

Geological repository systems for safe disposal

4.3 Grimsel Test Site (GTS), generic URL in granite, purpose-built parallel to existing tunnels, 400 m, horizontal ‘ramp type’ access (1983–1984: construction/1984–1986: basic site characterisation/1986–: scientific programme organised in GTS project phases).

4.4 A¨spo¨ Hard Rock Laboratory, purpose-built generic URL in granite, 450 m, ramp and shaft access (1986: investigations/1988: selection of site/1990–1995: construction/1995– : scientific programme) (# SKB 2009). © Woodhead Publishing Limited, 2010

Underground research facilities and rock laboratories

91

4.5 Mont Terri rock laboratory (FMT), generic URL in clay formation, purpose-built parallel to existing motorway tunnels, horizontal ‘ramp type’ access, operated by swisstopo (modified after Bossart and Thury, 2008).

The A¨spo¨ underground facility in Sweden deserves a special mention here, since it is a purpose-built laboratory but will not be part of the repository facility (Fig. 4.4). This addresses an important issue, namely whether or not an underground research laboratory should always be constructed as the first step of a repository facility. In the Swedish case, it is argued that the answer depends on many different aspects, but most importantly on the characteristics of the rock formation at hand; for example, in widespread rocks with properties varying homogeneously in space, fundamental research could be carried out in a representative generic URL. The two key assumptions for this statement to be valid are homogeneity and representativeness. Homogeneity is used here in a statistical sense; for example, the spacing between two large-scale fracture zones or the thickness of fracture zones follows a statistical distribution that has characteristics that are homogeneous in space (a constant mean value and standard deviation for a second order approximation). Representativeness, the second key assumption, is associated with scale; the location and geometric characteristics of the URL should be such that they can capture the smallscale variability of the key properties of the formation and can allow upscaling to make an inference about the parameters at large scales. Under these two conditions it can be confidently argued that a generic purpose-built URL at depths similar to those of the planned repository can suffice for performing most of the engineering and scientific tests, and

© Woodhead Publishing Limited, 2010

92

Geological repository systems for safe disposal

4.6 Mizunami URL (status of excavation in December 2009: main shaft: 413.0 m; ventilation shaft: 436.3 m), purpose-built generic URL in granite with sedimentary cover, shaft access (1996: project initiated/2003: construction started) (# JAEA 2009).

definitely for carrying out a lot of the illustration or demonstration tests. It should be emphasised that this statement does not imply that during construction of the repository there will be no test facility or in situ testing. On the contrary, it is expected that a number of test niches and galleries will be constructed to gain specific data for the rock formations of interest and feed into the finalisation of the designs and safety assessments. These data would be (a) of a confirmatory nature and (b) a prerequisite for the engineering design. The need, however, to perform fundamental tests for

© Woodhead Publishing Limited, 2010

Underground research facilities and rock laboratories

93

4.7 KURT–KAERI underground research tunnel, purpose-built generic URL in granite, ramp access (2003: project initiation and site characterisation/2004: basic and detailed design and achieved construction license/2005: start construction/2006: completion of the tunnel/2007: start of operation) (# KAERI 2009).

4.8 HADES (Mol) URL, site-specific URL in plastic (boom) clay, shaft access (1980: construction start/1985: URL operational/1987: first extension / 1997–2000: second shaft /2001–2002: connecting gallery/ 2006-2007: PRACLAY gallery) (# SCK-CEN 2009).

© Woodhead Publishing Limited, 2010

94

Geological repository systems for safe disposal

4.9 YUCCA Mountain Exploratory Studies Facility (ESF), site-specific URL in welded tuff, 300 m, ramp access (1987: selection of site/1994–: construction/1997–: drift scale thermal tests started/1999–2008: demonstration experiments/2009: programme stopped/re-evaluation of options) (courtesy of the US Department of Energy).

4.10 Meuse/Haute-Marne URL at Bure, site-specific URL in CallovoianOxfordian Argillites, shaft access (1999: authorisation for installation/ 2002: construction suspended/2003: construction restarted) (# Andra 2009).

© Woodhead Publishing Limited, 2010

Underground research facilities and rock laboratories

95

4.11 ONKALO Underground Rock Characterisation Facility, sitespecific URL (URL in granite), ramp access (2004–2011: construction/ 2004–: investigation/2010: characterisation depth /2011: excavation complete/2015–: construction of repository (# POSIVA 2009).

concept development or model confirmation, e.g. a long-term migration test to verify the performance assessment models, could have already been satisfied with the generic URL.

4.2.3 The two bounding approaches for developing a URL programme As the title suggests the two approaches described here are the two extreme cases, presented to more clearly illustrate the points. In reality, it is expected that a hybrid approach will be followed that lies somewhere in the middle. The top-down approach starts with a very systematic evaluation of the needs for the URL in the context of the repository programme and derives the specific experiments from these needs. The bottom-up approach starts from an evaluation of the possible experiments that can be performed and builds up experiments or activities that would serve the repository programme needs. From a technical and scientific point of view the first approach would seem the preferable one. However, as mentioned in Section 4.1, the roles of a URL extend beyond technical ones, e.g. focal points for training, know-how build-up, etc. It is in this context, and for programmes that are at an early stage, that the bottom-up approach is usually followed. Thus, the welljustified literature survey of what has been or is being done in other URLs occasionally tends to become the proposed URL programme rather than providing input for the programme of the new URL. As the repository

© Woodhead Publishing Limited, 2010

96

Geological repository systems for safe disposal

programme in a country develops, e.g. as the first set of provisional safety analyses is completed or as engineering designs are developed, the planning can focus on new experiments that should serve the specific needs of and be derived from the high-level requirements of the respective repository programme.

4.2.4 Evolution of URL programmes over time The research, development and demonstration (RD&D) programme to be followed in a URL depends on the stage of the national programme, the knowledge and expertise available internationally for similar geologic formations or repository designs and the long-term plans for the repository development. The RD&D programme and the resulting activities in the URL evolve dynamically over time and should be targeted to fulfilling the evolving needs of its various users. ‘Users’ here is used broadly to define the various technical disciplines involved in a geological repository realisation, e.g. those represented in the site and field characterisation group, in the repository design and engineering group or in the performance assessment group, as well as the non-technical disciplines, e.g. the group responsible for communication with the various stakeholders, policy- and decision-makers and the public. It suffices to mention that a URL, both an existing one as well as a newly built one, can benefit from experience accumulated to date elsewhere. The strategy adopted most of the time in selecting medium- to long-term experimental programmes is a trade-off between the following two aspects: (a) a set of experiments that have enough overlap with existing programmes to strengthen the confidence, through verification or benchmarking, that the developments and results are confirmed by international comparison and (b) a set of experiments or projects that focus either on activities specific to the repository system selected for the particular programme or on novel topics for which a significant contribution to the state-of-knowledge can be accomplished. The initial experimental programmes are usually determined by shortterm needs, partly because during the short term one has to perform a certain type of fundamental characterisation tests to develop the basic understanding and databases for the design of more complex experiments. The basic evolution of user needs and respective programmes in a URL is shown in Fig. 4.12 for the case of the Grimsel Test Site (GTS) in Switzerland (www.grimsel.com); the GTS has been in operation since 1984 and for each of the three past decades of operation the main focus is as summarised in this figure. At the beginning – in the early 1980s – emphasis was placed on developing the appropriate tools and techniques that would be needed for carrying out more complicated experiments. This was a very important

© Woodhead Publishing Limited, 2010

Underground research facilities and rock laboratories

4.12

97

Evolution of needs → example of the Grimsel Test Site (GTS).

objective at that time because existing technologies from other industries (e.g. oil and gas industry, mining, tunnel engineering) had to be adapted to the requirements of radioactive waste disposal: the low-permeability host rock formations to be evaluated, the longevity of equipment and instrumentation and the desire for non-intrusive or non-destructive testing techniques are just a few examples of the additional requirements posed in the context of radioactive waste disposal. A fundamental toolkit was obtained during that period, which allowed designing and performance of more complex experiments but also had a significant contribution to the site exploration techniques employed by Nagra at that time. Three examples of such techniques are (a) the multi-packer hydraulic systems developed, which were employed in the crystalline rock characterisation from a tunnel at the Piz Pian Grand potential repository site (Nagra, 1988), (b) the evaluation of techniques and materials for sealing of boreholes drilled from underground exploration tunnels, as preparation for the underground characterisation of the Wellenberg potential repository site (Coons et al., 1987; see also Blu¨mling and Adams, 2008) and (c) the cross-hole tomography studies, as

© Woodhead Publishing Limited, 2010

98

Geological repository systems for safe disposal

preparation for the second phase of the crystalline characterisation programme in Switzerland (Niva et al., 1988). Similar examples can be found in other first-generation URLs, e.g. at the A¨spo¨ Hard Rock laboratory in Sweden (SKB, 1996). New technologies are of course being tested and evaluated continuously, but compared to the first decade this is done rather on an as-needed or ad hoc basis; such an example is wireless data transmission techniques that have been tested at GTS since 2006. The second decade shows an additional emphasis on performance assessment questions, in particular testing and verification of models for radionuclide migration, study of the effects of special materials to be used in repository construction (e.g. concrete) on the natural system or other issues such as migration of repository-generated gas. A unique characteristic of the GTS among rock laboratories worldwide is the existence of a radiation controlled zone (IAEA Level B/C) in one of the investigation tunnels, which allows experiments to be carried out with radioactive tracers in the geosphere under realistic conditions. With this set-up, it has been possible to check the results of small-scale laboratory experiments in the field and to test directly model calculations of the migration of radioactive substances. During this decade, the emphasis expanded to include large-scale engineering demonstration experiments, such as FEBEX (full-scale engineered barriers experiment) and GMT (gas migration test). These types of experiments, associated with a certain maturity of the URL and associated repository development programmes, also take place elsewhere, as shown in Table 4.3. Since the end of the 1990s, further steps have been taken towards more integrated (and more complex) projects with: (a) field experiments under repository-relevant boundary conditions, as far as possible (large-scale, long-term, realistic hydrogeological conditions, etc.), and (b) projects addressing the implementation of a geological repository (engineering feasibility, operational aspects, closure, monitoring and possible effects of repository construction on the surrounding rock). Example case studies of detailed programmes are presented in Section 4.5.

4.3

Planning and designing an underground research facility: basic considerations

This section is meant as a short primer mainly for programmes that want to develop their own URL facility. Our starting point is that national strategies for URL development strongly depend on the following: 1.

What is the timescale for the radioactive waste disposal and how do activities feature in this timescale?

© Woodhead Publishing Limited, 2010

Underground research facilities and rock laboratories 2.

3.

4. 5. 6.

99

Has the potential host rock been selected and what requirements does this decision (or lack thereof, if the host rock is still open) have on the URL programme? Is the implementation of an URL needed to develop and test a disposal concept and is going underground the most efficient way to satisfy research and testing needs? Can desired information be obtained by cooperating in work performed in the URL of another country? Can an existing underground facility be adapted for generic URL work in a cost-effective manner? Is the overall waste-disposal programme sufficiently advanced to provide continuity when the URL work under consideration is completed?

In the following two sections the requirements, timing and resources needed for planning a URL are discussed in the context of the six points listed above.

4.3.1 General requirements for implementation of a sitespecific URL Key issues to consider in this context are discussed using the ONKALO Underground Rock Characterisation Facility in Finland, which represents an example of a site-specific URL currently under construction. A fundamental requirement is that a site-specific URL must be constructed without jeopardising the long-term safety of the selected repository site. Fulfilling this requirement also renders it possible later to link the facility to the repository as a part of it (Posiva, 2003a). The URL buildings and structures should comply with the land use plan of the region, ensuring the needs of present operations and future projects in the area (Posiva, 2003a). The URL shall be based on proven technology that should help to minimise technical risks and to improve working conditions. It shall enable the operation of the final disposal facility according to the defined time schedule and the collection of sufficient information and knowledge of the repository host rock and other underground conditions to ensure the long-term safety of the final disposal and to meet the requirements of the construction licence for the repository. Consequently, the investigation programme for the URL shall be drawn up to allow the collection of sufficient information and knowledge required for the application of the construction licence. The occupational and operational safety level must correspond to the safety level of the planned repository. Furthermore, visits to the URL shall be possible during and after its construction. The URL shall comply with the plans of the final

© Woodhead Publishing Limited, 2010

100

Geological repository systems for safe disposal

disposal facility, e.g. compliance of URL underground structures with repository design to allow disposal tunnel characterisation, use of the horizontal and vertical deposition concepts, use of one-level and two-level repository layouts and the possibility of extension of the repository. The URL construction shall keep the natural conditions of the bedrock, and conditions otherwise favourable with respect to long-term safety, sufficiently unchanged in the repository area. It shall keep the environmental impacts small in order not to affect the natural surroundings considerably or change the local habitants’ living conditions. Different stages of investigation phases can be defined (Posiva, 2003b): .

.

.

.

First stage. Surface-based investigations before the construction of the access ramp/shaft starts. The main results are an improved description of the target rock volume and the access locations of the URL and the establishment of baseline conditions. Second stage. Construction of an access ramp and/or a shaft down to the planned repository depth accompanied by parallel investigations on the surface. The main products are monitoring responses of the geosphere to the construction activities, the improved characterisation of the target rock volumes and the completion of the detailed design of the URL. Third stage. Completion of the URL construction at the target depth, together with underground investigations, including site-specific tests of repository technology and experiments related to the long-term performance of the multi-barrier system. Fourth stage. The repository excavation stage. The underground characterisation of the site will continue to be an important activity during the operational stage, especially when designing and constructing the deposition tunnels.

4.3.2 Timing of URL development and the resources required Having a URL offers a lot of advantages, as has been discussed above, but it also carries a substantial commitment in terms of resources, both human and financial. Thus, the decision to develop one’s own URL has to be evaluated within the whole repository development programme. Construction costs for a URL may easily be on the order of 100 million euros and, once a URL is started, a significant portion of a disposal programme’s budget may be used to support it. The entire life cycle costs of the URL, including decommissioning, are significantly higher than these initial costs. Andersson (1999) reports that four European URLs spend between 5 and 11 million euros annually on research and development.

© Woodhead Publishing Limited, 2010

Underground research facilities and rock laboratories Table 4.4

101

Participation and financing schemes

The entire URL – a joint programme

This has been the case with the Stripa project, where each of the partners had a ‘share’ of the whole project. The financial responsibility was according to the shares held and access was guaranteed to all results.

Joint programme This is a variation of the above scheme, whereby each partner – no fixed sharing has the same ‘voting’ rights but there is no fixed key of contribution. URL run by one organisation – participation fee

Under this scheme each partner organisation contributes directly to the operational costs of the URL and then additionally participates with personnel or resources in the experiments of interest.

URL run by one organisation – participation in a project

This is a variation of the participation scheme mentioned above. Here a partner organisation participates in or conducts its own project. The projects are the constituent elements of the URL and it is each project that contributes to the URL costs.

The timing of site-specific URL development strongly depends on the following points: . . . .

Are specific data needed that can only be obtained in a site-specific URL? Have all necessary data been collected before the system is disturbed? Have all technical, logistical and regulatory preconditions been met? Is the programme ready to demonstrate full capability to build a repository?

Table 4.4 shows the different financing models that have been followed so far for setting up cooperation projects in URLs. Note that a cooperative programme would, in most cases, be associated with generic URLs.

4.4

Public outreach and the role of URLs as training platforms

4.4.1 Public outreach In parallel with their main role as research facilities, URLs have also become established as important platforms for active interaction with the interested public and thus contribute significantly to the acceptance of the scientific and engineering work being performed in the area of geological disposal. By supplying direct information to the public from (and in) a ‘realistic’ environment, the feasibility of safe disposal can be illustrated and conveyed in a convincing manner. As part of the public outreach programmes, thousands of visitors are guided through the underground facilities every

© Woodhead Publishing Limited, 2010

102

Geological repository systems for safe disposal

year and can form their own impression of the studies being carried out. They can look over the shoulders of the researchers and experience the engineering and scientific work at close hand.

4.4.2 URLs as training platforms URLs also play an important role in the dissemination of accumulated knowledge and experience to a broad spectrum of young scientists worldwide. The GTS, for example, serves as an internationally recognised training platform for future generations of experts in the management of radioactive waste. Most importantly, it offers its infrastructure and databases for educational purposes. The educational activities take into account the needs of implementers, regulators and research organisations. As members of IAEA Network Centres of Excellence, ‘Training in and Demonstration of Waste Disposal Technologies in Underground Research Facilities’, many URLs provide their facilities and experts with knowledge for theoretical and practical training in all aspects of underground research, engineering and related issues.

4.5

Case studies

4.5.1 In situ characterisation and testing (near-field and farfield processes) Tests with radionuclides colloid and radionuclide retardation (CRR) experiment Studies of natural colloids in deep groundwaters from widely ranging environments (e.g. Yucca Mountain, Wellenberg, Schwarzwald, Oklo, Cigar Lake, etc.) have been ongoing for several decades. In addition, laboratory programmes on colloid generation, stability and classical batch experiments studying radionuclide uptake on colloids (and the possibilities for subsequent release) have been reported and a significant database already exists. It is generally accepted that five requirements must be fulfilled to prove that colloid-facilitated transport of radionuclides in a potential repository host rock may be of significance to the long-term performance of a waste repository: . . . . .

Colloids must be present. Colloids must be mobile. Colloids must be stable under the given groundwater conditions (geochemical and hydrogeological environment). Radionuclide association with the colloids must take place. The association must be irreversible.

© Woodhead Publishing Limited, 2010

Underground research facilities and rock laboratories

103

4.13 Three-dimensional view of the CRR test site. The view direction is perpendicular to the experimental shear zone.

The colloid and radionuclide retardation (CRR, www.grimsel.com) experiment was dedicated to improving the understanding of the in situ retardation of safety-relevant actinides and fission products associated with bentonite colloids in the vicinity of the engineered barrier system (EBS)/host rock interface. In addition to a series of in situ dipole experiments that were carried out at the Grimsel Test Site (GTS), the project partners, namely ANDRA (F), ENRESA (E), FZK-INE (D), JNC (J), USDOE/Sandia (USA) and Nagra (CH), funded an extensive programme of laboratory and modelling investigations. The aims of CRR were: examination of the in situ migration of bentonite colloids in fractured rocks, investigation of the interactions between safety-relevant radionuclides and bentonite colloids in the laboratory and in situ and, in addition, the testing of the applicability of numerical codes for representing colloidmediated radionuclide transport. The starting point of the CRR experiment was at the interface between the bentonite buffer (EBS) and the geosphere. A radionuclide and bentonite colloid cocktail was injected into a dipole flow field of 2.2 m length and the breakthrough of radionuclides and bentonite colloids was monitored and compared (Fig. 4.13). The results of the CRR experiment can be summarised as follows: . .

Bentonite colloids were generated at the bentonite–host rock interface under laboratory conditions. High bentonite colloid concentrations (up to 1000 mg/l) remain stable

© Woodhead Publishing Limited, 2010

104

. . .

Geological repository systems for safe disposal even over long timescales (months), showing very slow aggregation rates in the Grimsel groundwater. A high recovery of up to 90% of the injected bentonite colloids in the dipole flow field indicated high colloid mobility under the applied (short-term and fast-flow) experimental conditions. Batch experiments revealed a strong decrease in sorption to fracture infill and granite in the presence of colloids. The recovery of tri- and tetravalent actinides in the in situ experiment was increased from about 30% in the absence of to 60% in the presence of bentonite colloids. Colloidally transported radionuclides arrived slightly earlier at the extraction borehole than did conservative tracers.

The following conclusions can be drawn from the CRR project: .

.

.

The bentonite buffer is a potential source of artificial colloids within the EBS of a potential repository. To date, there is no in situ evidence for the formation and release of such colloids into the geosphere. Ongoing in situ experiments might give a first indication but, as they were not specifically designed for this purpose, these data might be biased by artefacts of the experimental layout. Bentonite colloids were found to be mobile within the tested shear zone. However, the migration of particles strongly depends on the internal structure of the shear zone and on the (applied) flow field. To date, there are no in situ experiments that were performed under realistic (i.e. repository-relevant) groundwater flow conditions. The radionuclide uptake of bentonite colloids has been proven for at least the tri- and tetravalent actinides, both with classical batch experiments in the laboratory and with in situ experiments. Within the experimental timescales, the colloidal association of these radionuclides appears to be irreversible.

Tests with radionuclides: long-term diffusion (LTD) It is well known that matrix diffusion is an important process in retarding radionuclides in fractured rocks. Rock matrix diffusion is particularly important when determining dose and risk calculations for weakly and nonsorbing radionuclides such as 129I and 14C. However, there have been few long-term field scale experiments to evaluate matrix diffusion of radionuclides in fractured rock with a minimal disturbance to the in situ condition. The long-term diffusion (LTD, www.grimsel.com) project is a series of experiments that aims to obtain quantitative information on matrix diffusion under in situ conditions. Partners and contributors are JAEA and AIST (2006–2008) of Japan, NRI and RAWRA of the Czech Republic,

© Woodhead Publishing Limited, 2010

Underground research facilities and rock laboratories

105

HYRL of Finland and Nagra and PSI of Switzerland. The project is divided into several work packages in which are: .

.

. .

An in situ monopole diffusion experiment where radionuclide tracers diffuse into undisturbed rock matrix with subsequent geochemical analysis of matrix samples combined with predictive and post-mortem modelling exercises to increase confidence in the modelling of long-term and large-scale diffusion processes. Characterisation of the pore space geometry (including determination of in situ porosity for comparison with laboratory-derived data) using 14Cdoped PMMA resin injection and NHC-9 chemical porosimetry techniques. A study of natural tracers in the rock matrix to elucidate evidence for long-term diffusion processes. Detailed characterisation of the flow paths in a water-conducting fracture and investigation of the in situ matrix diffusion paths in core material from earlier GTS experiments.

Circulation of a cocktail of sorbing, weakly sorbing and non-sorbing radionuclides (3H, 22Na, 131I and 134Cs) in the monopole was started in June 2007 and continued until August 2009 (Fig. 4.14). Water samples retrieved on a regular basis have been analysed at the Paul Scherrer Institute (PSI).

4.14 In situ monopole diffusion experiment where radionuclide tracers diffuse into the undisturbed rock matrix (borehole 8 m long, diameter 56 mm).

© Woodhead Publishing Limited, 2010

106

Geological repository systems for safe disposal

Compared with predictive models carried out by four modelling teams, a much higher decrease in Cs was seen in the analysis of water samples than was predicted in the models. This was due to sorption on the borehole wall, and was confirmed after the borehole was overcored in November 2009. Analysis of overcored subsamples showed that the diffusion profiles of 22Na, stable I and 3H were more or less as predicted by the models. Overall the main conclusions so far from Phase 1 of the LTD project can be summarised as follows: . . . .

A stronger than expected Cs sorption was found on the borehole wall in the in situ diffusion experiment. A 10–20% decrease in porosity was found in samples measured in situ compared with samples measured in the laboratory. Evidence of unlimited connected porosity was obtained from natural tracer studies. Evidence of matrix diffusion retarding 237Np and 235U was found after only 60 days circulation in an advective fracture.

In the next phase of the LTD project, the monopole will be overcored in order to investigate the Cs sorption sites and concentrations of the radionuclides that have diffused into the rock matrix from the borehole. Future in situ experiments include a novel di-monopole experiment that allows in situ monitoring of matrix diffusion to be carried out and one to study diffusion from a shear zone using soluble, non-sorbing radionucide tracers is planned.

4.5.2 Engineered barriers Large-scale process testing full-scale engineered barrier system experiment (FEBEX) The FEBEX experiment at the Grimsel Test Site (www.grimsel.com) consists of an in situ full-scale engineered barrier system (EBS) test performed under natural conditions (Fig. 4.15). A ‘mock-up’ test, at almost full scale, runs in parallel in the laboratories at Ciemat in Madrid (a full description of the ‘in situ’ and the ‘mock-up’ test, covering the period 1994– 2004 is included in ENRESA, 2006, and references therein). The FEBEX experiment has been the research subject in three subsequent, European research projects: FEBEX, FEBEX-II and NF-PRO, under the leadership of ENRESA. Starting in 2008, a consortium of four partners (SKB, Ciemat, Posiva and Nagra) was brought together, which continues running the in situ experiment and the mock-up as part of Grimsel Phase VI under the name FEBEXe (e = extension) until 2012, when excavation of the second heater is planned.

© Woodhead Publishing Limited, 2010

107

4.15

The current layout of the FEBEXe experiment at the GTS.

Underground research facilities and rock laboratories

© Woodhead Publishing Limited, 2010

108

Geological repository systems for safe disposal

The experiment is based on the Spanish reference concept for the disposal of radioactive waste in crystalline rock in which the canisters enclosing the conditioned waste are placed horizontally in drifts and surrounded by a clay barrier constructed of highly compacted bentonite blocks. With the start of the heating in 1997, the FEBEX experiment (heater 2 and EBS; see Fig. 4.1) is the longest running full-scale experiment. A constant temperature of 100 8C has been maintained at the heaters/bentonite contact during this time, while the bentonite buffer has been slowly hydrating with the water naturally coming from the rock. A total of 632 sensors of diverse types were installed in the clay barrier, the rock mass, the heaters and the service zone to measure the following variables: temperature, humidity, total pressure, displacement, water pressure, etc. Partial dismantling of the in situ test was carried out during 2002. After five years of heating, heater 1 was removed and the materials recovered (bentonite, metals, instruments, etc.) have been analysed to investigate the different types of processes undergone, while the second heater continued. The aim of the FEBEX experiment is to study the behaviour of the nearfield components (EBS, host rock) for a high-level radioactive waste repository in crystalline rock. The three main objectives are: . . .

To demonstrate the feasibility of handling and constructing an engineered barriers system. To study the thermohydromechanical (THM) processes in the near field. To study the thermohydrogeochemical (THG) processes in the near field.

As the experiment has now been running over more than 11 years, a unique dataset including crucial long-term observations in terms of sealing, saturation and corrosion has been generated. The current status is that the outer buffer is practically saturated and hydration is in progress at the inner buffer. This was predicted by THM models, but occurs at a slower pace than expected. Especially, one can observe that: . . .

Temperatures in the bottom part continue being higher than those in the sides and upper part for each section. Temperature trends are increasing very slowly in the buffer and in the rock close to the heater. The total pressure in general continues increasing into the buffer and also in the contact bentonite/rock and bentonite/plug. The water content in the buffer continues increasing at all points. The biggest differences can be observed in the inner part but it is clearly rising elsewhere.

Modelling efforts also continue within FEBEXe with the objective to:

© Woodhead Publishing Limited, 2010

Underground research facilities and rock laboratories . . .

109

further elucidate the observed behaviour and the processes involved, assess how the results can be extrapolated to the near field in the long term and assess the transferability of the results to the national disposal concepts of the FEBEXe partners.

Large-scale process testing: gas migration test (GMT) The gas migration test (GMT, www.grimsel.com) (under the leadership of RWMC/Japan) has been a world premiere project, focusing on the migration of waste-generated gas – in the GMT case simulated by nitrogen – through the engineered barrier system (vent and bentonite/sand barrier) in a large-scale, in situ, realistic set-up (Fig. 4.16). The results of the field test are reported elsewhere (see, for example, Vomvoris et al., 2003, Shimura et al., 2006). Important conclusions, e.g. on the robustness of the engineered barriers to gas migration, have been reached and, more importantly, the experience gained with GMT has been used to design similar large-scale experiments elsewhere (e.g. the HG-A test at the FMT).

4.16 Gas migration test (GMT) at the GTS.

© Woodhead Publishing Limited, 2010

110

Geological repository systems for safe disposal

4.5.3 Demonstration experiments Engineering studies and demonstration of repository design (ESDRED) The European research project ESDRED (www.esdred.info) comprises various experiments aiming at the demonstration of technological concepts for handling and disposal of radioactive waste. The long low-pH shotcrete plug in GTS is part of ESDRED Module 4 whose object is the construction and test of the low-pH shotcrete plug at full scale (Fig. 4.17). The aim of the test is to demonstrate the support capacity of such a plug under realistic conditions (Spillmann et al., 2009). This involves the swelling pressure of a bentonite buffer applied to one side of the plug. In order to demonstrate the applicability of the experimental concept a full-scale and fully monitored test of a low-pH shotcrete plug was carried out at the Grimsel Test Site (www.grimsel.com). To simulate real conditions of a filled repository, the plug is loaded with the swelling pressure of a bentonite buffer. The bentonite was provided with an artificial hydration system (vertical mats) to accelerate the saturation process and to impose a pore water pressure in the bentonite buffer. The cementitious materials used in repositories plugs must be designed to support either the rock of the underground works or mechanically a swelling material seal. The concrete used for these two applications must comply with two main requirements:

4.17 Layout of the ESDRED shotcrete plug including the monitoring components of the TEM project at the GTS.

© Woodhead Publishing Limited, 2010

Underground research facilities and rock laboratories . .

111

The pH has to be as low as possible (below 11) in order to minimise chemical interaction with the bentonite or with the surrounding geological medium. The concrete has to be emplaced by the shotcreting technique.

Suitable concrete formulations have been developed and a short low-pH shotcrete plug was tested at the A¨spo¨ URL (Sweden). The full-sized bentonite and shotcrete plug in the Grimsel Test Site was completed in March 2007. The plug is monitored by a network of hard-wired sensors that measure total pressure, water pressure and water content. Test and evaluation of monitoring (TEM) techniques The ESDRED plug experiment also provides the opportunity to test and evaluate different monitoring methods (TEM project, www.grimsel.com) in a situation that would have similarities with a repository vault-end seal. Six additional sensors constitute the wireless monitoring network. The data from the wireless monitoring sensors (two pore pressure, two total pressure and two water content sensors) are temporarily stored on a battery powered data logger. A magneto-inductive transmitter at the tunnel face and a receiver in the access tunnel provide the wireless data transmission. To protect the standard data logger and magneto-inductive system from the expected high swelling pressures of the bentonite (up to 5 MPa), they were installed in small boreholes. The data are then transmitted by MI antennas to the data acquisition unit located in the tunnel. Non-intrusive monitoring techniques have the advantage that the technical barrier and surrounding excavation damage zone are not disturbed by the monitoring system. A regular fan of six monitoring boreholes was drilled to evaluate the seismic tomography as a non-intrusive monitoring technique. A seismic source and receivers explore the area between the boreholes. Repeat measurements allow investigating changes in the water content or pressure during the saturation process.

4.5.4 Hydromechanical evolution of backfill and host rock Gas path through host rock and along seal sections: HG-A experiment The excavation damaged zones (EDZs) around cylindrical excavations, such as sealed sections in tunnels or shafts, have been the subject of extensive experimental and theoretical investigations in the field of underground waste disposal. Conceptual and numerical models have been developed to describe the hydromechanical processes associated with the creation and evolution of damaged zones during the operational phase and after backfilling of the underground structures. However, only limited data

© Woodhead Publishing Limited, 2010

112

Geological repository systems for safe disposal

from tunnel-scale experiments are available for a comprehensive validation of the EDZ models. The HG-A experiment is a large-scale experiment focusing on hydromechanical processes as part of a long-term geoscientific research programme at the Mont Terri Underground Rock Laboratory located in the Opalinus Clay formation in the Jura Mountains of Switzerland and operated by swisstopo/Swiss Geological Survey (www. mont-terri.ch). A horizontal microtunnel with a diameter of 1 m and a length of 13 m was drilled in a claystone formation. Monitoring was carried out before, during and after tunnel excavation. After installing monitoring instruments in the open tunnel, the end of the tunnel was backfilled with sand (test interval) and a large hydraulic packer was emplaced in the seal section. The packer was inflated and subsequently the test interval was saturated with synthetic pore water (a key reference for this case study is Marschall et al., 2008). Figure 4.5 shows the layout of the Mont Terri URL with the HG-A experimental site located in a niche of Gallery 04. The specific aims of the HG-A experiment are to: . . .

provide evidence for barrier function of the Opalinus Clay on the tunnel scale, investigate mechanical self-sealing of EDZ after tunnel closure (packer inflation) and provide evidence for a gas transport capacity of Opalinus Clay for both the intact host rock and EDZ.

The microtunnel was excavated during February 2005 using a steel augur from a niche in Gallery 04. The microtunnel was excavated parallel to a bedding strike, so that bedding runs along the tunnel. This was done to imitate the expected relationship between bedding and emplacement tunnel orientation in a deep repository in Northern Switzerland, where bedding is expected to be almost horizontal and emplacement tunnels to be subhorizontal (Nagra, 2002). The experimental layout is shown in Fig. 4.18. Excavation was monitored using piezometers and deformation gauges installed in boreholes. This was subsequently supplemented with additional piezometer boreholes and borehole stressmeters (Fig. 4.19). The first 6 m of the microtunnel was lined with a steel casing immediately after excavation to stabilize the opening. The gap behind the liner was then cement-grouted, but not sealed. The next 3 m from 6–9 m along the tunnel forms the seal section where a purpose-built hydraulic megapacker (940 mm in diameter and 3000 mm in length) was installed. The final 4 m of the microtunnel is the test section. This section was instrumented with piezometers, extensometers, strain gauges and time domain reflectometers (TDRs) to measure pressure, deformation and water content. After instrumentation the test section was filled with sand behind a retaining wall. The seal section was instrumented with piezometers, total pressure cells

© Woodhead Publishing Limited, 2010

Underground research facilities and rock laboratories

113

4.18 Gas path through the host rock and along seal sections: the HG-A experiment layout.

4.19 Schematic drawing of the microtunnel and the site instrumentation. Grey zone coding refers to the steel liner (black), the seal section (grey in the middle part) and the backfilled test section (grey part at the end).

and TDRs prior to the installation of the megapacker. Following the installation of the megapacker the volume between the retaining wall and the megapacker was filled with a cement grout. The post-excavation experimental sequence is summarised in Fig. 4.20. The test and sealing sections were instrumented during 2006 and the megapacker was emplaced and first inflated in November 2006 (Phase 2).

© Woodhead Publishing Limited, 2010

114

Geological repository systems for safe disposal

4.20

Post-excavation experimental sequence.

During 2007 a series of water injection tests were performed to test the performance of the flow control units and the long-term monitoring system. A year-long multi-rate water injection was started in January 2008 and is still ongoing. At this point, the experiment is approaching the end of Phase 3. It is anticipated that Phase 4, gas injection, will be started during 2010 and Phase 5, the post-gas hydraulic testing, will start in 2012. Following Phase 5 it is planned to excavate and perform post-mortem characterisation of the test and sealing sections. The following preliminary conclusions can be drawn from the HG-A project (Marschall et al., 2008): 1.

Hydromechanical processes. The responses to injection and changes in megapacker pressure indicate a tightly coupled system where imposed test section pressure changes cause deformation of the test and sealing section, while stress changes in the seal section lead to deformation of the test section and consequent pressure changes. In addition to these short-term responses the measurements indicate ongoing deformation of the test section, presumably resulting from the response to far-field stresses and possible softening of the rock due to resaturation. The coupled nature of the responses together with the ongoing transient

© Woodhead Publishing Limited, 2010

Underground research facilities and rock laboratories

2.

3.

4.6

115

responses to excavation and subsequent resaturation make interpretation complex. Flowpaths and self-sealing. The hydraulic response suggests that flow paths are dominated by the EDZ around the tunnel and that the intact host rock represents a significant low permeability barrier. Initial interpretations and models suggest that the total conductance of the EDZ is reducing with time. It is assumed that this reduction relates to reconsolidation and swelling of the rock, gel creation and sealing of fractures within the EDZ. The influence of the megapacker pressure on EDZ properties is harder to determine and will require more detailed consideration of the coupled responses seen during the low and high effective stress periods. Barrier function of the undisturbed rock. The low diffusivity responses to saturation and pressurisation of the test section that have been observed in the monitoring boreholes support the current estimates of the diffusivity of the intact Opalinus Clay. To date, the uncertainty in the storativity prevents any better estimate of metre-scale hydraulic conductivity.

Future trends

Experience from more than 30 years has shown that the investigation and testing programmes in Underground Research Laboratories form an important part of the overall waste management strategies of the various national programmes. URL projects contribute in many different ways to the geological disposal programme and their role evolves with time, stage of the programme, state of know-how and technology. Broad international experience exists to design and implement a URL programme in a fast and economic way. In the last few decades, many of the established URLs emerged as internationally well-known research laboratories in the area of safe disposal of radioactive waste in deep geological repositories. They are characterised by a broad spectrum of scientific and engineering projects, close international collaboration and know-how exchange and active, transparent dialogue with the public and their elected representatives.

4.6.1 What has been obtained from URLs . .

Development of methods and equipment for underground characterisation and testing of the reliability of the different methods. Determination of reliability of surface-based methods of site characterisation.

© Woodhead Publishing Limited, 2010

116 . . . . . . .

Geological repository systems for safe disposal Application of site-exploration strategies and strategies to adapt underground systems as more information is acquired. Testing and development of conceptual and numerical models of processes potentially relevant to radionuclide transport through rock. Quantification of impacts of excavation on the local system. Further development and assessment of excavation techniques. Simulation of effects caused by emplacement of radioactive waste. Demonstration of engineered-barrier systems (feasibility). Demonstration of repository engineering on a large scale.

4.6.2 Future needs Within a waste disposal programme, URLs have a direct technical function and also a confidence-building role. In advanced programmes, there is a trend towards large-scale, realistic, integrated experiments in which a number of interacting components and/or processes are simultaneously studied. As fundamental scientific problems move towards solution and the implementation of repositories draws closer, the confidence building role grows in importance. Large-scale, long-term, integrated experiments play a key role in raising technical and public confidence. Full- or large-scale experiments performed to date have highlighted potential optimisation areas (feedback to Nagra’s programme, for example, leading to selection of a combination of bentonite blocks and granular bentonite as buffer). The timing of these experiments should be determined considering the overall geological disposal programme and the steps required leading up to the submission of the licence application. In emerging programmes, the need for know-how development in some of the fundamental issues will exist, but whether this need is satisfied with a new URL or being involved in cooperative projects in existing URLs is an issue that has to be addressed by each programme individually. Visits to URLs represent not only an occasion for gathering visual impressions but also an opportunity for in-depth discussions between technical experts and decision makers. URLs also provide an excellent focus for mutually beneficial international cooperative efforts. In any cost–benefit analysis of the value of a URL in a national programme, the purely scientific/technical contributions of the facility must not be considered in isolation; the less direct or intangible benefits associated with raising of technical and public confidence in the competence of the implementer must also be weighed. Finally, URLs provide a training platform for the transfer of knowledge and experience, a role gaining increasingly in importance.

© Woodhead Publishing Limited, 2010

Underground research facilities and rock laboratories

4.7

117

References

Andersson J (1999), ‘A study of the co-operation of research in underground facilities within the EU on aspects of disposal of radioactive waste’, European Commission, Brussels. Blu¨mling P and Adams J (2008), ‘Grimsel Test Site – Investigation Phase IV: borehole sealing, Nagra Technical Report NTB 07-01, Nagra, Wettingen, Switzerland. Bossart P and Thury M (2008), ‘Mont Terri Rock Laboratory. Project, Programme 1996 to 2007 and Results’, Report Swiss Geological Survey 3, swisstopo, 3084 Wabern, Switzerland. Coons A, Bergstro¨m P, Gnirk et al. (1987), ‘State-of-the-art report on potentially useful materials for sealing nuclear waste repositories’, Nagra Technical Report NTB 87-33, Nagra, Wettingen, Switzerland. ENRESA (2006), ‘Full-scale engineered barriers experiment: updated final report (1994–2004)’, Transedit, ISSN1134-380X. IAEA (2001), ‘The use of scientific and technical results obtained from underground research laboratory investigations for the geological disposal of radioactive waste’, IAEATECDOC-1243. Kickmaier W (2002), ‘The role of rock laboratories’, Bulletin 34, pp. 4–9, Nagra, Wettingen, Switzerland. McCombie C and Kickmaier W (2000), ‘Underground research laboratories: their roles in demonstrating repository concepts and communicating with the public’, in Euradwaste 1999: Radioactive Waste Management Strategies and Issues: Fifth European Commission Conference on Radioactive Waste Management and Disposal and Decommissioning, edited by C Davies, Luxembourg, 15–18 November 1999, EUR 19143 EN, pp. 274–281. Marschall P, Trick T, Lanyon G W, Delay J and Shao H (2008), ‘Hydro-mechanical evolution of damaged zones around a microtunnel in a claystone formation of the Swiss Jura Mountains’, ARMA, San Francisco, California. Nagra (1988), ‘Untersuchungen zur Standorteignung im Hinblick auf die Endlagerung schwach- und mittelaktiver Abfa¨lle: Berichterstattung u¨ber die Untersuchungen der Phase 1 am potentiellen Standort Piz Pian Grand (Gemeinden Mesocco und Rossa, GR)’, Nagra Technical Report NTB 88-19. Nagra, Wettingen, Switzerland. Nagra, (2002), ‘Projekt Opalinuston – Synthese der geo-wissenschaftlichen Untersuchungsergebnisse. Entsor-gungsnachweis fu¨r abgebrannte Brennelemente, verglaste Hochaktive sowie Langlebige mittelaktive Abfa¨lle’, Nagra Technical Report NTB 02-03, Nagra, Wettingen, Switzerland. NEA (2001a), ‘The role of underground laboratories in nuclear waste disposal programme’, OECD/Nuclear Energy Agency, Paris. NEA (2001b), ‘Radioactive Waste Management Committee: going underground for testing, characterisation and demonstration (a technical position paper)’, 6/ REV, OECD/Nuclear Energy Agency (OECD/NEA), Paris. Niva B, Olsson O and Blu¨mling P (1988), ‘Grimsel Test Site – radar crosshole tomography with application to migration of saline tracer through fracture zones’, Nagra Technical Report NTB 88-31, Nagra, Wettingen, Switzerland.

© Woodhead Publishing Limited, 2010

118

Geological repository systems for safe disposal

Posiva Oy (2003a), ‘ONKALO: underground rock characterisation facility – main drawing stage’, Posiva Working Report 2003-26, Posiva Oy, Helsinki. Posiva Oy (2003b), ‘ONKALO: underground characterisation and research programme (UCRP)’, Posiva Report 2003-03, Posiva Oy, Helsinki. Shimura T, Fujiwara A, Vomvoris S, Marschall P, Lanyon B, Ando K and Yamamoto S ‘Large scale gas migration test at Grimsel Test Site’, in Proceedings of the 11th International High-Level Radioactive Waste Management Conference, 30 April–4 May 2006, Las Vegas, Nevada, ANS/ HLRWM, p. 4. SKB (1996) A¨spo¨ Hard Rock Laboratory, 10 Years of Research, SKB, Stockholm, Sweden.. Spillmann T, Fries T, Wetzig V, Koch AS, Garcia-Sin˜eriz J-L, Ba´rcena I (2009), Beton mit tiefer Alkalita¨t: Demonstrationsversuche im Felslabor Grimsel: Stollenverschluss aus niedrig-pH Spritzbeton, Tec21 SIA/21, pp. 17–19. Vomvoris S, Lanyon B, Marschall P, Ando K, Adachi T, Fujiwara A and Yamamoto S (2003), ‘Sand/bentonite barriers and gas migration: the GMT large-scale in-situ test in the Grimsel Test Site’, in Materials Research Society Proceedings on Scientific Basis for Nuclear Waste Management XXVI, Vol. 757, Materials Research Society, Warrendale, Pennsylvania. Websites of the various URL projects and http://www.ha.nea.fr/html/rwm/docs/ 2001/index.html.

© Woodhead Publishing Limited, 2010

5 Crystalline geological repository systems: characterisation, site surveying and construction technologies and techniques A . J . H O O P E R , Alan Hooper Consulting Ltd, UK

Abstract: This chapter describes the technologies and techniques that are available or are being developed for the characterisation of crystalline sites, where the repository host rock is a strong igneous or metamorphic rock that is water saturated. Individual sections are presented on each of the categories of geoscientific information, namely lithologies, geological structure, rock mechanics and geotechnical properties, hydrogeology, geochemistry and radionuclide transport properties, which typically are sought and then must be integrated in a site characterisation programme. Further sections consider respectively: disturbance by excavation and waste emplacement, the stability of crystalline rocks in the context of waste disposal and the feasibility of construction in such rocks. Key words: deterministic and stochastic treatment as a function of length scale, spatial variability of flow, geochemical and transport properties, fractures and fracture networks, borehole drilling and testing, development and testing of conceptual models.

5.1

Introduction

This chapter describes the technologies and techniques that are available or are being developed for the characterisation of crystalline sites. In the context of geological disposal of radioactive wastes, the term crystalline is used to describe strong igneous or metamorphic rocks such as granites or consolidated tuffaceous rocks. Such rocks typically have low porosities and groundwater is present predominantly in discontinuities, or fractures, within the rock matrix. Much attention is devoted to the characterisation of the fracture systems in crystalline rocks since it is these that will provide the

121 © Woodhead Publishing Limited, 2010

122

Geological repository systems for safe disposal

principal pathway for any groundwater flow and radionuclide transport away from the repository. Crystalline rocks have many properties that make them suitable host rocks for an underground disposal system located hundreds of metres below ground. In some countries, such as Sweden and Finland, the crystalline rocks that have been investigated persist from the prospective repository depth all the way to the ground surface, whereas in other countries they are typically overlain by sedimentary rocks whose characteristics must also be understood in order to develop the necessary designs and safety assessments, as discussed in Chapter 6. This provides a clear-cut example of the necessity to ensure that the site characterisation techniques and technologies that are selected for use are appropriate to the geological and hydrogeological conditions at a site. Sweden and Finland have two of the most advanced programmes in the world dealing with the geological disposal of spent nuclear fuel and both have identified prospective sites for locating a repository in the crystalline bedrock (SKB, 2008; Andersson et al., 2007). Significant contributions have been, and continue to be, made to the development of technologies and techniques for characterising crystalline sites in these countries. The spent fuel and/or high-level waste disposal programmes of other countries, including Canada (AECL, 1994), France (Andra, 2005) and Switzerland (Nagra, 1994), have made notable contributions also, even when at the stage of testing the feasibility of disposal in crystalline rock, as did investigations in the UK of a potential crystalline host rock, overlain by sedimentary rocks, for the disposal of predominantly long-lived intermediate-level wastes arising from the reprocessing of spent fuel (Nirex, 1997). The bulk of the chapter is taken up with a description of the technologies and techniques that can be used to characterise key aspects of crystalline sites to provide the sort of information required to develop repository designs, safety assessments and the necessary geological understanding. This analysis is strongly informed by the published experiences of national radioactive waste management programmes such as those listed above, which in turn have used the experience of other industries such as mineral extraction, geotechnical engineering, geothermal energy, water resource exploration and, in some cases, hydrocarbon exploration. Individual sections are presented on each of the categories of geoscientific information, namely lithologies, geological structure, rock mechanics and geotechnical properties, hydrogeology, geochemistry and radionuclide transport properties, which typically are sought in a site characterisation programme. However, an important point to note is that each category of information must be related to others so that a consistent picture of the site emerges rather than a collection of separately obtained findings that will not give the necessary overall understanding of a site. These sections on techniques and

© Woodhead Publishing Limited, 2010

Crystalline geological repository systems

123

technologies are followed by shorter sections that consider, respectively, disturbance by excavation and waste emplacement, the stability of crystalline rocks in the context of waste disposal and the feasibility of construction in such rocks. This is followed by a short discussion of possible future trends in the field of characterising prospective repository sites in crystalline rocks.

5.2

Lithologies

The classical starting point for the description of a site is the distribution of different rock types, involving the identification of rock units on the basis of properties that are potentially relevant to aspects of the repository design and/or safety assessment.

5.2.1 Geophysical surveys The location of the site itself may be defined as a result of regional geophysical and mapping surveys that enable the identification of a suitably sized volume of the potential host rock. The relevant geophysical surveys at this stage are often conducted as low-level aerial surveys and include the following: . . .

Magnetic surveys that determine the distinctive magnetic properties of different rock types. Radiometric surveys that determine the distinctive radionuclide content of different rock types. Electromagnetic surveys that determine the distinctive conductivities of different rock types.

These surveys are very useful in providing information on the larger-scale structural elements, such as lineaments, within the rock mass. Such large structural elements are often used to define the boundaries within which the prospective repository site will be located. Gravity surveys are also useful at the early, reconnaissance, stage of site characterisation to identify the different rock types on the basis of distinctive gravitational properties. However, this geophysical technique is not well suited to aerial surveying and is most often used in manual surveys conducted on the ground surface.

5.2.2 Surface mapping Geophysical surveys are especially valuable when there is no outcrop of the crystalline rock because it is overlain by sedimentary cover, or perhaps relatively thick Quaternary deposits (soils and gravels deposited over the last 1.5 million years or so), or when the area of interest is below the sea or some

© Woodhead Publishing Limited, 2010

124

Geological repository systems for safe disposal

other large body of water. However, geological mapping of outcrops at the surface or of exposures in trenches, when available, provides much useful information, particularly when combined with the current state of development of electron microscopy and analytical techniques. The identification of the mineralogical and chemical compositions and the microstructural grain size in outcropping rocks provides an important input to the subdivision of the crystalline rock mass into different rock units.

5.2.3 Borehole drilling Boreholes are drilled in order to obtain similarly detailed information on the rocks occurring at depth. Two distinct types of drilling are used in repository site investigation: core drilling and percussion drilling. In core drilling, intact drill cores are taken out of the borehole. In suitable geological conditions this technique can be used throughout the length of a borehole and there are many examples of a continuous core record being obtained over depth intervals of more than one thousand metres. The drill cores are examined in detail, much as in the case of outcrop mapping, and typically samples are also taken for other sorts of analyses. In percussion drilling the rock is crushed in the borehole and the investigations to identify the lithological properties are done on the crushed material when it is removed from the borehole. A number of geophysical tools have been developed for use in boreholes and some of these can complement the sampling afforded by abstraction of drill cores or crushed rock. In relatively simple rocks, the density/photoelectric factor (PEF) and neutron tools that are used to measure rock porosity will also resolve the dominant rock type. Other lithology-sensitive measurement techniques, such as the elemental capture spectroscopy (primarily measuring silicon, calcium, iron and sulphur) and natural gamma-ray tool (measuring thorium, uranium and potassium), can be used if the rock is more complex.

5.2.4 Role of rock units The identification of different rock units as subdivisions in the crystalline host rock is valuable in establishing the spatial distribution of key physical and chemical properties within the rock volume. For example, in the work carried out by SKB, the Swedish Nuclear Fuel and Waste Management Company, to characterise the candidate repository sites at Forsmark and Laxemar, rock units are identified to be associated with certain fracture network characteristics, such as fracture density and permeability, or thermal conductivities, resulting from the precise mineral content (Stephens et al., 2007; Wahlgren et al., 2008). Such properties are important in

© Woodhead Publishing Limited, 2010

Crystalline geological repository systems

125

evaluating the mechanical and/or hydrogeological properties of the rock mass and in developing repository designs.

5.3

Geological structure

There are strong relationships between the rock types and the aspects of geological structure that are important to repository development, so many of the techniques and technologies discussed above are valuable in characterising the structure. The structure of a prospective repository site, placed within a regional context, can be seen as the basis for developing ideas on the location of the repository and for influencing other elements of the site characterisation programme. The focus of attention in respect of a crystalline site is a description of the structure in terms of the discontinuities in the rock mass, since these will determine the mechanical and hydrogeological properties that are important in determining the safety of the repository and should control decisions on design in order to achieve safety.

5.3.1 Length scales A fairly standard approach has developed whereby discontinuities in the rock mass are classified in terms of their different length scales and are then subject to different means of characterisation according to that classification. The term deformation zone is used to describe extensive discontinuities that are likely to include large numbers of individual fractures. These deformation zones are subdivided in terms of the likely implication that their size will have for repository safety and design. At the top end of the length scale regional deformation zones typically extend over tens of kilometres, and, if present, are likely to represent an important boundary for the site investigations. Major deformation zones more local to the prospective repository site may extend over a length of the order of one kilometre. At a potentially suitable repository site there will be relatively few major deformation zones such that these are typically characterised deterministically in site characterisation programmes conducted to date. The classification of minor deformation zones refers to a length scale of a few hundred metres where the lengths of individual fractures also can be of this magnitude. In most crystalline rocks that have been investigated to date, the number of deformation zones in this category within the prospective repository volume has been such that they cannot be investigated individually and a geostatistical method is required to characterise their properties. Finally, there are individual fractures that may have length scales on the order of ten metres.

© Woodhead Publishing Limited, 2010

126

Geological repository systems for safe disposal

5.3.2 Characterisation of deformation zones As already noted, surface mapping and aeromagnetic surveying provide good information on the regional deformation zones and these zones may also have clear signatures in the topography of the region. Given their likely significance for civil engineering projects and resource exploitation, e.g. mineral extraction, there will often be good pre-existing geoscientific information. Major deformation zones more local to the prospective repository site are also readily identified by the same methods, but greater precision is required in defining their location at depth so borehole drilling is often used in conjunction with surface-based information to locate these deformation zones, at depth. Particularly for steeply dipping deformation zones, the use of inclined or deviated boreholes can prove invaluable. More detailed investigations are required in order to characterise the frequency and spatial distribution of minor deformation zones. Recent work in Sweden has shown the value of magnetic surveys conducted manually from the ground surface, which are more sensitive than the previously conducted aerial surveys (Triumf, 2007). The identification of magnetic anomalies is used in conjunction with detailed surface mapping of fracture sets to identify the distribution of the minor deformation zones within the surveyed area. Boreholes can then be drilled to intersect the deformation zone indicated from the survey, as in the case for the major deformation zones. For gently dipping deformation zones, advances in the technology of surface-based seismic reflection surveys have proved successful in identifying minor deformation zones in recent years. This represents a potentially powerful technique when a vertical seismic profiling (VSP) tool can be used to pick up the seismic signature in a borehole drilled subsequently to intersect the deformation zone (Juhlin, 2007).

5.3.3 Characterisation of fractures and fracture networks Information on frequency, orientation and dimensions is required to understand the rock mechanical and hydrogeological significance of fractures and fracture networks. This information is obtained by examination of fractures retrieved in drillcore and by obtaining images of the borehole wall. Special techniques have been developed to correctly orientate the drill core in three-dimensional space and hence to correlate the fracture being examined in the retrieved core with the image of the fracture at depth in the borehole wall. Geophysical and hydrogeological data obtained from the borehole are also used to characterise the fractures and valuable evidence on the fracture type can be obtained from the mineralogical signature of any fracture filling that is observed. This information on the fractures and fracture networks can then be used to identify domains

© Woodhead Publishing Limited, 2010

Crystalline geological repository systems

127

occurring between the deterministically defined deformation zones within the rock mass on the basis of the predominating fracture characteristics.

5.3.4 Discrete fracture network models The geostatistical information on fractures and fracture networks required by other disciplines is typically presented as a discrete fracture network (DFN) model (e.g. Munier, 2004). Typically the DFN will comprise, at a minimum, the following components for each fracture set identified on the basis of its orientation: . . .

5.4

The orientation of the set. The size distribution of fractures in the set (often expressed in terms of areal dimensions). The amount, or intensity, of fracturing (e.g. areal density per unit volume of rock).

Rock mechanics and geotechnical properties

Information is required on the mechanical properties of the rock mass in which the waste deposition will occur and through which access tunnels and/ or shafts will be constructed. In particular, it is required to understand the strength of the rock, its deformation properties under loading and the in situ stress regime. Given the occurrence of deformation zones and fractures in many crystalline rocks, this information is required on each of the main structural elements within the rock mass and has to be in a form that allows analysis of the response of the overall rock mass at a range of different length scales according to the element of repository design that is of interest.

5.4.1 Mechanical properties The strength properties, such as uniaxial compressive strength and tensile strength, of the intact rock not intersected by fractures can be determined by a suite of well-known laboratory tests on drillcore samples. Similarly, the mechanical properties of fractures in the rock can be determined by carrying out tilt tests and direct shear tests on drillcore samples containing a fracture. The deformation properties of the intact rock, classically expressed in terms of Young’s modulus and Poisson’s ratio, can be obtained from the results of uniaxial and triaxial compression tests on drillcore samples. The results from the laboratory tests are typically used in empirically based rock mass classification systems to determine the rock mass deformation and strength properties that are relevant to repository design and to determining the response to mechanical or thermal loading. The

© Woodhead Publishing Limited, 2010

128

Geological repository systems for safe disposal

empirically based rock mass classification systems draw on a large database generated from geotechnical engineering projects and are widely used in underground tunnelling and civil engineering works in other industries. There is an increasing trend to use theoretically based numerical modelling to determine the relevant rock mass properties and such an approach has been used alongside the empirical approach in SKB’s site characterisation programme, for example (Glamheden et al., 2007).

5.4.2 In situ stress The in situ stress can be determined by using various techniques. It is resolved into maximum horizontal stress, minimum horizontal stress and vertical stress, so that both the magnitudes of these and the orientation of the maximum and minimum stress direction are required. Commonly applied methods include overcoring, where a small diameter borehole is drilled at the base of a cored borehole. Strain gauges are fixed to the wall of this borehole and the instrumented rock is then retrieved to the surface by overcoring, allowing the direct measurement of the stress. Techniques involving hydraulic pressure are also commonly used to measure in situ stress where the hydraulic pressure on an identified fracture in a borehole wall is increased and decreased to determine the pressure at which it just opens and closes. Hydraulic fracturing is also used where the hydraulic pressure in a section of a borehole is increased until fracturing of the rock is induced.

5.5

Hydrogeology

Information on the hydrogeology of a potential repository site is required to compile a hydrogeological description on a regional and local scale that is sufficiently detailed to support judgements on the suitability of the site and to meet the information needs of repository design, safety assessment and environmental impact assessment. The hydrogeological description is universally presented as a conceptual model and the hydrogeological information obtained in site characterisation is aimed at supporting the definitions of boundary conditions for the model and of the initial conditions and justifying the hydraulic parameter values that are assigned to various components of the model. Figure 5.1 provides a typical schematic illustration of the major features of a groundwater flow system, in this example showing the SW–NE cross-section of the regional groundwater flow system affecting the Borrowdale Volcanic Group rocks at Sellafield as determined by Nirex (1997).

© Woodhead Publishing Limited, 2010

Crystalline geological repository systems

129

5.1 A schematic illustration of the major features of the groundwater flow system in the region around the former potential repository zone (PRZ) at Sellafield (copyright United Kingdom Nuclear Decommissioning Authority; reproduced by kind permission).

5.5.1 Hydraulic parameters Much of the information on the hydrogeology of a site comes from the programme of testing that is carried out in the various boreholes drilled at the site. Tests can be carried out in single boreholes or can involve cross-hole or interference tests between two or more boreholes. The characteristics of interest primarily concern the derivation of hydraulic parameters and include measures of environmental pressure, environmental and freshwater heads, hydraulic conductivity, permeability, transmissivity, storativity, specific storage, hydraulic diffusivity and measures of connectivity. Various methods are used for the interpretation of each identified parameter, e.g. by direct measurement (associated with the respective testing or data acquisition technique), through preliminary interpretation or through modelling. A range of hydrogeological testing methods are available which themselves yield different types of information. The methods identified in a recently conducted review (Golder Associates, 2006) are: . . . . . . . .

pumping tests conducted in open boreholes without packers; interference measurements conducted in open observation boreholes; single-hole packer tests; interference tests conducted in observation boreholes equipped with multi-packer systems; fluid logging; flow logging; differential flow logging; and long-term monitoring of pressure in boreholes equipped with multi-level piezometers or multi-packer systems.

© Woodhead Publishing Limited, 2010

130

Geological repository systems for safe disposal

5.5.2 Hydrogeological conceptual model It is recognised as of great importance to relate the targets and design of the hydrogeological testing programme to the development of the hydrogeological conceptual model of the site. Particularly at the early stages of a site characterisation programme there will be the potential for viable alternative conceptual models. An important role of the testing programme is to enable discrimination, where possible, between such alternatives and to identify the conceptual and parametric uncertainties associated with the model that emerges from this process as having the best justification. The structural building blocks of the hydrogeological conceptual model(s) are the hydrogeological units, which are recognised on the basis of their hydraulic properties and initial condition. An important element in the design of a hydrogeological testing programme is a consideration of how hydrogeological units will be defined and which part of the system will be treated deterministically and which stochastically. As noted in the discussion on structure, a fractured crystalline rock formation will probably be assumed to have stochastically distributed properties whereas major deformation zones are likely to be represented deterministically in terms of location and size. However, within such deterministically represented features, the hydraulic parameters may be represented as constant or stochastically distributed. In many cases, it may be uncertain at the outset how the various hydrogeological units in the system are to be treated for modelling purposes and the initial programme of investigations and

5.2 Discrimination between deterministic and stochastic treatment of hydrogeological features of different length scales as applied by United Kingdom Nirex Limited in its investigation of the Borrowdale Volcanic Group basement rocks (copyright United Kingdom Nuclear Decommissioning Authority; reproduced by kind permission).

© Woodhead Publishing Limited, 2010

Crystalline geological repository systems

131

hydrogeological testing will be designed to resolve this. Figure 5.2 shows how discrimination between the deterministic and stochastic treatment of hydrogeological features was applied by United Kingdom Nirex Limited in its investigation of the Borrowdale Volcanic Group basement rocks.

5.5.3 Hydrogeological measurements Single-hole tests are mainly conducted to derive the hydraulic parameter transmissivity. In rock formations where the flow geometry is less complex the derived transmissivity could be used to derive the hydraulic conductivity of the tested formation. In fractured rock formations the derived transmissivity is representative of only a small proportion of the test interval and in most cases it would not be clear how many fractures and what totally hydraulically effective aperture could be attributed to the transmissivity. In fractured rock masses, it is essential to recognise the importance of scale in relating the results of hydrogeological testing to the large-scale behaviour of hydrogeological units. The further inputs required to construct a hydrogeological model in fractured crystalline rock can be established from a combination of tests, typically as follows: . .

.

Interference tests give information on the connectivity of the system and its relation to scale and direction. Fluid and flow logging measurements use fluid conductivity, temperature or flow in open holes under static or pumping conditions to seek to identify discrete features (fractures) that are hydrogeologically active and thus enable derivation of the flowing fracture frequency. Long-term discrete pressure measurements in boreholes are used for deriving the static pressure and equivalent freshwater head of the isolated section and their natural (or induced) fluctuations. The monitoring installations are also used as observation points during interference tests.

Characterising fractures and fracture networks The upscaled hydrogeological properties for the units that are modelled stochastically are usually determined through the use of hydrogeological discrete fracture network (DFN) models, where similar principles apply as described in the discussion on the use of DFN models in the characterisation of geological structure. Figure 5.3 shows a realisation of the fracture network within a 2-km-sided cube of the Borrowdale Volcanic Group produced from a discrete fracture network model. In this figure flowing features are represented as planar fractures and each fracture is tessellated

© Woodhead Publishing Limited, 2010

132

Geological repository systems for safe disposal

5.3 A realisation of the flowing features within a 2-km-sided cube of the Borrowdale Volcanic Group rocks produced from a discrete fracture network model (copyright United Kingdom Nuclear Decommissioning Authority; reproduced by kind permission).

into domains with sides of approximately 40 m, within which the transmissivity is taken to be constant. It is particularly important in constructing a hydrogeological DFN to be able to characterise the fracture population that is capable of carrying groundwater flow. The borehole flow logging tool developed by Posiva in Finland has proved very effective in characterising individual fractures that are conducting flow at the point where they are intersected in a monitored borehole (O¨hberg and Rouhiainen, 2000). However, other fractures may also be capable of conducting flow; the concept of potentially flowing fractures has commonly been used whereby fractures that are identified from drillcore logging and/or borehole imaging and geophysical measurements as having the characteristics of a flowing fracture (e.g. open aperture, mineralogical evidence) are also taken into consideration. Figure 5.4 shows how a model domain can be constructed to simulate groundwater flow in the fracture network of the crystalline host rock of a repository. The figure is taken from a simulation of the volume of the Borrowdale Volcanic Group of rocks bounded by major

© Woodhead Publishing Limited, 2010

Crystalline geological repository systems

133

5.4 The model domain used for the calculations of groundwater flow through the fracture network in the Borrowdale Volcanic Group rocks at Sellafield (copyright United Kingdom Nuclear Decommissioning Authority; reproduced by kind permission).

deformation zones having approximate dimensions of 3 km62 km6750 m (Nirex, 1997).

5.5.4 Testing the conceptual model Good scientific practice has been developed over a number of years involving the testing of the hydrogeological model that results from the process outlined here against other information obtained from site investigations that has not been used in the development of the model. Various types of information have been used depending on the nature of the prospective repository site, and in particular whether the crystalline rock is overlain by sedimentary formations. In general the types of information used have included the following: . . . .

5.6

conservative tracer tests; groundwater level responses in boreholes to large-scale drawdowns, induced in interference tests, for example; surface and near-surface groundwater levels and discharges; and hydrochemistry of water sampled from fractures and from the rock matrix.

Geochemistry

Geochemistry is taken here to mean the chemical characteristics of: .

the groundwater in the fracture network in the crystalline rock;

© Woodhead Publishing Limited, 2010

134 . .

Geological repository systems for safe disposal the pore water in the crystalline rock matrix; and the minerals formed on the surface of fractures (including secondary minerals such as calcite).

It includes the associated media such as colloids, microbes and gases that, with the groundwater and rocks, make up the present subsurface environment at the prospective repository site. It is important to understand the geochemistry for a number of reasons, and these dictate the sampling and analysis that is undertaken. Obtaining good-quality groundwater samples that are representative of the undisturbed conditions at depth is notoriously difficult, given the necessity to use borehole technology to obtain the samples combined with the paucity of groundwater and low hydraulic conductivities that characterise a potentially suitable rock mass for the siting of a repository. Similarly, it has not been possible until recently to characterise reliably the hydrochemistry of the pore water in a crystalline rock matrix for the purposes of a repository project. Improvements in the techniques used to obtain and analyse samples have been made on the basis of international experience in the field such that the difficulties can be overcome sufficiently to obtain enough good quality samples to characterise the hydrochemistry of a prospective repository site. An important contribution has been made by the development of protocols to qualify representative samples that are not significantly contaminated by materials introduced by drilling or by the sampling method (e.g. Laaksoharju et al., 2008a). Some geochemical measurements, such as pH or redox potential, Eh, can be made by specially designed down-hole instruments.

5.6.1 Multi-disciplinary approach To various degrees, the sampled groundwater compositions reflect processes relating to ancient through to modern water/rock interactions and mixing of groundwaters of different origins. Such complex data require a multidisciplinary approach to their interpretation in order to produce a hydrogeochemical model that can be used, for example, in hydrogeological, radionuclide transport and safety assessment modelling. The hydrogeochemical model should represent a site-scale hydrogeochemical interpretation that is fully integrated with the corresponding geological and hydrogeological models. Laaksoharju et al. (2008b) propose that the objective for such a model in the Swedish geological repository programme is that it should clearly show the following: . .

the major lithological and structural units comprising the site; knowledge of the major groundwater flow directions from hydrogeological modelling;

© Woodhead Publishing Limited, 2010

Crystalline geological repository systems . . . .

135

the relationship of chemistry (i.e. mixing of end members, chemical reactions, etc.) to these major hydraulically conducting pathways; a clear indication of the groundwater types representative of the main hydraulic rock mass; units characterised by fractures (fracture zones) of lower transmissivities; and an indication of the chemistry of the rock matrix pore space fluid/ groundwater.

5.6.2 Application of geochemical information The specific uses to which such a model can be put are highly dependent upon the location of the site and the repository concept under consideration. However, a general scheme can be defined for the application of geochemical information and hence the targets for geochemical characterisation, as, for example, by Bath (2008): 1. 2.

3.

4.

5.

Performance of the engineered barrier system: groundwater compositions at repository depth and up-gradient of repository location. Groundwater flow directions and rates: locations of groundwater volumes with distinct compositions; hydrochemistry of transmissive deformation zones and fractures; hydrochemical properties at groundwater flow model boundaries; and depth of active modern groundwater circulation. Pathways for radionuclide transport and retention in the geosphere: groundwater compositions at repository depth; hydrochemistry of transmissive deformation zones and fractures down-gradient of repository location; and geochemical description of radionuclide transport pathways including fracture minerals and adjacent rock matrix. Palaeohydrogeology (i.e. groundwater flows and compositions in the past): locations of groundwater volumes with distinct compositions and hydrochemical properties at groundwater flow model boundaries. Biosphere pathways and processes, including the geosphere–biosphere interface zone: hydrochemistry of shallow groundwaters, soil waters and surface waters; geochemical composition of soils; and baseline geochemical conditions of the surface environment.

Identification of objectives The approach that has been developed for determining the samples and analyses that are required involves the identification of the specific objectives that characterisation of the geochemistry has to achieve in

© Woodhead Publishing Limited, 2010

136

Geological repository systems for safe disposal

support of such a scheme. Bath (2008) identifies the following objectives that are generally recognisable in the crystalline rock programmes that have been documented to date. Develop a redox model This can be used in assessing the performance of engineered barriers and inform the speciation of redox-sensitive radionuclides. The following information would be used: . . . .

Use geochemical modelling to interpret Eh measurements and redoxsensitive solute data with particular attention to sulphate reduction, FeII/FeIII equilibria and methane/carbon dioxide equilibria. Support the interpretation of the sulphate/sulphide system with data on stable sulphur-34/sulphur-32 isotopic ratios. Use microbial population data to develop an understanding of microbial mediation of chemical conditions in general and redox in particular. Use data on redox parameters and microbial populations as a function of depth from the surface to understand the development of a redox front with infiltration.

Understand controls on groundwater composition . .

Use graphical and statistical numerical analyses of data for non-reactive solutes (e.g. chloride) and stable isotopes (e.g. oxygen-18/oxygen-16 ratios) to determine solute sources and groundwater mixing ratios. Determine the capacity of rocks and minerals in the prospective repository location to control or buffer the key hydrochemical conditions such as Eh, pH and total inorganic carbon.

Evaluate groundwater ages and solute travel times .

. . .

Use non-reactive solutes and stable isotopes to: support and test the interpretation of groundwater flow paths in the hydrogeological model; constrain groundwater sources and mixing (as above); and calibrate a palaeohydrogeological model. Use chloride and oxygen-18/oxygen-16 data from extracted rock matrix waters to characterise matrix diffusion and the palaeohydrochemical equilibrium between matrix porewater and mobile groundwater. Use tritium data to identify where modern groundwater recharge occurs. Interpret carbon-14 and carbon-13/carbon-12 data to obtain groundwater ages.

© Woodhead Publishing Limited, 2010

Crystalline geological repository systems . .

137

Interpret chlorine-36 data in terms of chlorine residence times and groundwater ages. Interpret helium-4 data in terms of groundwater mixing and residence times, using estimates of its in situ production from the uranium and thorium present in the rocks.

Support development of the surface compartments of the hydrogeological model . .

Use tritium data to identify the depth of penetration of recently recharged water. Use chloride and other geochemical and isotope data to identify areas of groundwater discharge.

Evaluate palaeohydrogeological stability . . .

Use observations of calcite (secondary mineral) morphology to interpret past variations of groundwater salinity. Use iron, manganese and rare earth element contents of calcite and other secondary minerals to interpret past stability or variability of redox conditions. Use oxygen-18/oxygen-16 and carbon-13/carbon-12 data to identify calcite sources and qualitative ages, and also to interpret carbon biogeochemistry.

Quantify concentrations and fluxes of analogue solutes and mobile species .

.

Interpret dissolved concentrations of naturally occurring uranium, thorium, radium and other solutes that are analogues for repository radionuclides to test consistency with the outputs of radionuclide transport models. Interpret variations of concentrations of colloids, microbes and gases in terms of understanding their sources, compositions, reactions and transport/retention mechanisms.

Establish baseline hydrochemical conditions . .

Determine the chemical compositions of undisturbed groundwaters and surface waters in the siting area. Interpret dissolved gas compositions in shallow groundwaters and in soil gases.

© Woodhead Publishing Limited, 2010

138

Geological repository systems for safe disposal

Identify any hazardous or anomalous geochemical conditions or scenarios . .

5.7

Interpret the production, concentrations and outgassing of radon, methane and any other hazardous dissolved gases at the prospective repository depth. Identify any other geochemical conditions that pose potential environmetal hazards (e.g. the presence of arsenic or fluoride) that could affect the performance of the engineered barrier system (e.g. sulphide mineral oxidation to give sulphuric acid, penetration of low ionic strength waters that would degrade swelling clays) or that could influence radionuclide transport (e.g. naturally occurring organic complexants).

Radionuclide transport

It is considered that on the macroscopic scale radionuclide transport in crystalline rock will occur along advective flow channels that are within deformation zones and fractures. The retardation processes that are most often considered to act to limit the rate of radionuclide transport in these flow channels are rock-matrix diffusion and sorption. The characterisation of the relevant processes uses a combination of in situ and laboratory techniques and technologies. It relies heavily on information on lithologies, geological structure, hydrogeology and geochemistry, both to provide the framework for the transport model and to provide data on the compositions and mineralogies of rocks and fracture infills, and of groundwater and matrix porewater.

5.7.1 Transport data As exemplified by Crawford (2008), the transport data required include effective diffusivities for radionuclide transport in the rock matrix, rockmatrix porosities, the surface area of the porosity, cation exchange capacities of the minerals in the rocks and fracture infills, and the sorption properties for radionuclides on to rocks and fracture infills in contact with groundwater of various compositions. All of this information can be obtained from what are now relatively standard laboratory measurements, albeit that they are demanding to perform under the carefully controlled conditions required to ensure that they provide representative data. Other disciplines in the overall site characterisation are important in ensuring that the rocks and fracture infills used in making the measurements are representative of those that will be contacted along the advective flow path and that appropriate synthetic groundwaters are used in the radionuclide sorption measurements.

© Woodhead Publishing Limited, 2010

Crystalline geological repository systems

139

The effective diffusivity is obtained through the formation factor that is defined as the ratio of effective diffusivity of a solute to the free diffusivity of that solute at infinite dilution in water. The formation factor is a geometrical parameter dependent upon the porosity, tortuosity and constrictivity of the pore space available for diffusion into the rock matrix. Formation factors can be measured both in the laboratory, using through-diffusion experiments involving tritiated water and by electrical resistivity measurements, and in situ, using a high-resolution, downhole electrical resistivity tool. The in situ measurements are important in ensuring that the formation factor is representative of conditions at depth where the rock is subjected to stress and to identify where artificially high formation factors may have resulted from opening of pores in the laboratory samples in the absence of the stress field.

5.7.2 Flow data The extent of the interaction of radionuclides dissolved in groundwater in the advective flow channels with the fracture surface and the rock matrix will be controlled by the surface area in contact with the flowing water for a given water flow rate. The ratio of the flow wetted surface to flow rate is termed the hydrodynamic transport resistance or, more conveniently, the Ffactor (Andersson et al., 1998). The necessary information on flow and its distribution in the crystalline rock mass comes from the hydrogeological model and will follow the approach to characterising the hydrogeological properties of deterministically defined deformation zones and stochastically modelled fracture networks as described above. Typically a two-dimensional model will be constructed of the deterministic deformation zones such that it provides the lateral and depth-dependent variation of the transmissivity, calibrated against borehole measurements in the deformation zone. A hydrogeological discrete fracture network (DFN) model is used to provide F-factors for representative flow paths in the stochastically modelled fracture network and the frequency of the occurrence of the modelled flow paths. Information on representative flow path lengths and hydraulic gradients is required to support this technique. It is recognised that there is the potential to underestimate the effect of flow channelling in individual borehole measurements either because the flow channel is not intersected or, in the case that its properties are inferred, the flow rate that the channel can support is underestimated because of the bulk properties of the structure in which it occurs. Flow-channelling effects in the connected fracture network should be captured in the hydrogeological DFN modelling, but other, more-localised, channelling is usually accounted for by scoping calculations until the stage of undertaking underground

© Woodhead Publishing Limited, 2010

140

Geological repository systems for safe disposal

investigations. Once underground, mapping of open fractures and determination of flowing fracture frequencies, coupled with information on the relationship between rock mechanical properties and fracture transmissivities, would lead to an improved understanding of the extent and nature of flow-channelling effects.

5.7.3 Description of transport characteristics The information on radionuclide retardation by rock-matrix diffusion and sorption can be combined with F-factors for the various sections of flow paths from the repository depth to the surface in safety assessment models, which in this way capture the relevant site characterisation information in evaluating the transport of radionuclides along these flow paths. The transport properties derived by the approach described here can be tested by various designs of tracer tests where the movement of tracer solutes in flow paths between sections of injection boreholes and receptor boreholes is measured under controlled conditions. Single well injection and withdrawal tests have also been used in some characterisation programmes.

5.8

Disturbance by excavation or waste emplacement

The excavation of the repository underground works, comprising the deposition areas, access tunnels, drifts and shafts will perturb the natural system in a number of ways. The rock surrounding the excavations can be subject to mechanical, hydrogeological and chemical disturbance, as will be outlined below. The main impact of waste emplacement considered in respect of geological repositories is the thermal perturbation caused by the disposal of heat-generating waste. There are also chemical perturbations to be considered but these are more concept-specific. These processes of disturbance will not occur separately and independently of one another. The coupling between the different processes requires to be understood sufficiently to assess the implications for repository design and safety. This typically involves the construction of thermal–hydraulic– mechanical–chemical (THMC) coupled process models and considerable international effort has been devoted to develop capabilities in this area, using carefully controlled experiments, often conducted in underground laboratories in validation testing of the outputs of such models (NEA, 2001).

© Woodhead Publishing Limited, 2010

Crystalline geological repository systems

141

5.8.1 Mechanical disturbance Deformation of the rock Deformation of the rock mass surrounding the excavated opening can result from the response of the rock to the in situ stress field. This is most significant when there is a high, anisotropic stress field such as found at the underground research laboratory at Lac do Bonnet (Canada) that was operated in the past in support of the Canadian spent fuel waste management programme (Martin, 1990). Using information obtained from field experiments on such deformation phenomena, the orientation of tunnels and their cross-sectional geometry and dimensions can be optimised to minimise the extent of such deformations. Disturbance to fracture system Opening of new and pre-existing fractures in the rock wall surrounding the excavations and movement along fracture planes can occur in response to stress relief. This type of disturbance is often split into subcategories. ‘Excavation damage’ is used to describe perturbations that are not reversible, in extreme cases involving the movement of rock blocks or mechanical spalling of smaller rock fragments. ‘Excavation disturbance’ describes perturbations that may be transient in nature where the relevant aspects of the natural system are expected to return towards their unperturbed state following the closure of the repository, e.g. changes in aperture of fractures in response to stress relief. Characterisation of the rock mechanics and geotechnical properties as described above is designed to understand the quality of the rock mass and hence its response to excavation, among other things. Thus the extent and nature of any excavation damage will be assessed in advance and appropriate measures taken in terms of design of the underground openings. Limited damage of the rock surface resulting from the excavation process itself is unavoidable, but a range of excavation methods have been developed for use in crystalline rock and the extent of surface damage can be controlled by the choice of method if necessary. In recent years large-scale underground experiments have been conducted to measure the nature and extent of excavation disturbance in the rock wall surrounding an excavated tunnel in terms of fracture opening, movement on fractures and any resulting changes in hydraulic properties, e.g. the ZEDEX experiment conducted in the Aspo Hard Rock Laboratory in Sweden (Emsley et al., 1997). Such experiments use similar techniques to those applied in surface-based site characterisation, although it is worth mentioning the value of acoustic emission tests to detect the extent of

© Woodhead Publishing Limited, 2010

142

Geological repository systems for safe disposal

movement of fractures. The information obtained on the characteristics of the damaged/disturbed zones can be used to construct appropriate structural and hydrogeological models of these features for use in conjunction with the equivalent model for the undisturbed system.

5.8.2 Hydrogeochemical disturbance The excavation of the underground works will create a void in the rock mass at essentially atmospheric pressure and result in a hydraulic pressure gradient towards the excavations. This means that the pre-existing, natural groundwater system will be perturbed by flows towards the excavations. The amount of such perturbing flow will depend on the pressure gradient and flow path characteristics, and also on the groundwater control measures that are implemented in constructing the underground works. This so-called drawdown has the potential to be used to confirm the understanding of the hydrogeology. Once the repository has been filled with wastes and sealed and closed, it would be expected that the pre-existing pressure field would be restored. However, one of the potentially important long-term impacts of the drawdown would be to disturb the groundwater chemistry at the repository depth and in the surrounding rocks. This requires careful consideration, particularly when, as is often found, there are marked variations in groundwater salinity as a function of depth, and water of different salinity would then be introduced into the repository system from the overlying rocks. The hydrogeological and geochemical models described earlier can be used to evaluate the extent and spatial distribution of the resulting hydrochemical changes. The capability to carry out such evaluations can be tested beforehand by comparing the predictions of equivalent models in relation to the sites of underground research laboratories against information on the hydrochemical perturbation that occurs as a consequence of their excavation (e.g. Vieno et al., 2003). The introduction of air will also perturb the hydrochemistry and the mineralogical chemistry, particularly when one bears in mind that chemically reducing conditions are usually found at depth. However, the fact that the groundwater flow is towards the repository excavations when they contain air means that the depth of such disturbance into the rock wall will be limited. Methods exist from other industries such as mining and civil engineering to assess the extent of phenomena, such as acid mine drainage (where sulphide minerals such as pyrite are oxidised in air to produce sulphuric acid), that must be taken into account in describing the perturbed condition of the repository.

© Woodhead Publishing Limited, 2010

Crystalline geological repository systems

143

5.8.3 Thermal disturbance In repository designs for the disposal of heat-generating wastes in crystalline rocks it is typically found that the temperature of the rock surrounding the emplaced waste will be increased by a few tens of degrees Centigrade for a few hundred years and by just a few degrees over timescales on the order of one thousand years or more. As noted above, this will cause effects that interact with the effects of other forms of disturbance, e.g. localised changes in rock stresses and hydraulic pore pressures in the immediate vicinity of the waste. Combined modelling and experimental testing, particularly using heater tests in underground research laboratories, is used to gain an adequate understanding of the coupling of the thermal processes with others. Thermal rock spalling, where the thermally induced stresses could cause detachment of fragments of rock, currently receives a great deal of attention since it has the potential to degrade the performance of some engineered barrier systems that would adjoin the rock surface. This provides a good example of the interaction between site characteristics and repository design since the engineered barrier design or its method of emplacement would probably need to be adapted to overcome this difficulty.

5.8.4 Chemical disturbance by the wastes or engineered barrier materials The emplacement of wastes and engineered barriers will introduce a number of materials that are foreign in respect of the pre-existing natural system. The disturbance created will depend strongly upon the chemical nature of the materials in relation to the natural hydrochemistry and mineral chemistry. Many engineered barrier materials are selected on the basis that they will not react significantly with the groundwater and consequentially their impact on the chemistry of the natural system will be minimal. The materials that have been identified as having the greatest potential impact in this respect are cement-based materials, variously used in grouts within certain waste packages and as repository construction materials, and metallic wastes or waste containers that will corrode to generate gas under the prevailing hydrochemical conditions. Groundwater that contacts cement-based materials will be conditioned to high alkalinity and will leach the chemical constituents of the cement, typically leading to increased concentrations of sodium and potassium (in the short term) and of calcium. Extensive integrated modelling and experimental work, including the use of natural analogue systems, has been conducted to develop models for the evaluation of the nature and extent of the change to groundwater chemistry and mineral composition along flow paths leading from the repository (Nirex, 2002a).

© Woodhead Publishing Limited, 2010

144

Geological repository systems for safe disposal

The generation of gas has the potential to create an overpressure and, if it exceeds the solubility limit, a separate gas phase that will interact with the groundwater flow system. Again, extensive integrated modelling and experimental work has been conducted to develop models for the evaluation of any physical disruption of the surrounding rock mass and of two-phase gas-water flow in fractures (Nirex, 2002b).

5.9

Stability

The scientific and technical factors taken into account in selecting a candidate repository site for characterisation include, in particular, its longterm stability and its buffering capacity with respect to external and internal perturbations, along with its ability to accommodate the waste disposal facility and its ability to prevent or attenuate the potential release of radionuclides. In building a safety case, it is therefore important to assess: . . .

the features, events and processes (FEPs) that could influence the evolution of the geosphere; the long-term stability of the favourable conditions displayed by the host formation; and the buffering capacity of the formation with respect to perturbations.

The key issue is to evaluate the resilience of the main safety functions of the geosphere (including its flow and transport properties) to natural perturbations. The relevance of various naturally occurring processes and events will depend upon the timeframe to be considered, but timeframes on the order of about one million years are typically considered. Stability, in this sense, does not imply that steady-state conditions exist; the geosphere is constantly evolving, although in many cases rather slowly, and such evolution is perfectly acceptable for safe geological disposal. What is important is that this evolution is properly understood.

5.9.1 Geoscientific understanding In a general sense, crystalline rock formations are regarded as very stable: many (in Scandinavia and Canada in particular) have ages on the order of billions of years and are associated with very low surface erosion rates. This has been an important reason in many countries for considering such formations for radioactive waste repositories. Therefore many crystalline rocks provide intrinsically stable environments, particularly from a mechanical standpoint, and provide good buffering against external events and processes. Considerably younger crystalline rocks have also been considered, e.g. in Japan, but their relative youth does not imply that they are, necessarily, any

© Woodhead Publishing Limited, 2010

Crystalline geological repository systems

145

less stable, or less suitable for disposal purposes. There may, however, be intrinsic differences between very old and relatively recent rocks, especially with regard to the types of deformation zones present and the variability of fracture orientations, which could have implications with regard to stability. The natural phenomena (processes and events) identified to have the potential to perturb the stability of a crystalline rock have remained constant over a number of years and are as follows: . . . .

seismicity (in respect of the reactivation of existing fractures, faults and deformation zones); climate change (leading to either colder or warmer surface temperatures, and hence, variously, to ice sheets, permafrost or sea level change); uplift and erosion; and volcanism.

The geoscientific understanding of these phenomena was evaluated at a workshop of the OECD Nuclear Energy Agency Geosphere Stability Project (NEA, 2009). Regarding the understanding and handling of perturbations that affect the stability, the findings of the workshop, following a review of the results of site characterisation and research programmes, can be broadly summarised as follows: . . . . . .

There is, in general, good confidence in the understanding of the magnitude, cause, characteristics and frequency of perturbing phenomena. There is more limited confidence as to where and when a perturbation will occur. Confidence is also more limited regarding the volume of rock affected by a perturbation, but there are methods available to study such issues. The extent to which a repository is affected by a perturbation can often be conservatively addressed by using bounding and/or pessimistic approaches. Conclusions regarding many of these phenomena are supported by the results of natural analogue studies. Such evidence is useful in supporting a safety case, but often relevant site-specific observations provide stronger evidence than those from a general natural analogue study.

5.9.2 Geosphere stability in safety cases There are assessment tools (deterministic, probabilistic and bounding) for taking account of uncertainties in relation to geological stability in a safety case and also a wealth of examples of how these tools have been applied.

© Woodhead Publishing Limited, 2010

146

Geological repository systems for safe disposal

The bounding analyses often carried out as part of a safety case are likely to provide pessimistic bounds, which are also likely to be broader than the bounds produced by the use of so-called proxy data, obtained from investigations of analogous geological systems. Ultimately the evolution of the geosphere and the associated stability issues must be evaluated in a repository and site-specific context. For aspects of stability to which the repository design or safety assessment is particularly sensitive, there may be a need to constrain the assessment with more realistic site-specific information.

5.9.3 Implications for repository design As an example, analyses of the effects of seismic perturbations on the stability of the geosphere have shown that they have important implications for repository design, especially in relation to selecting the location of the repository within the rock mass to provide a suitable separation from active or potentially active deformation zones or faults. At the repository scale, the effects of stress magnitude and stress orientation, combined with consideration of the potential reactivation of deformation zones and the larger fractures in future seismic events, provide important constraints on the location and orientation of disposal tunnels and on the positioning of waste containers. In some geological environments (perhaps, for example, in Japan) the orientation of the maximum horizontal stress may be expected to remain approximately constant for a considerable time in the future, so that it is the effects of future seismicity that are of greatest interest. This appears also to be the case in other, seismically quieter countries, such as Sweden and Finland; here also it is considered that the greatest perturbation to the geological stability would be caused by a future seismic event during rapid glacial retreat as a result of a future transition from a cold climate state to a warmer one. An understanding of the long-term stability of the hydrogeochemistry may be very important in cases where the intrusion of altered groundwaters at repository depth could degrade the performance of engineered barriers, for example. The construction of a palaeohydrogeochemical model to understand how the system has responded to equivalent perturbations in the past is noted above as an important objective of geochemical characterisation of a site.

5.10

Feasibility of construction

The wealth of experience obtained from tunnelling and civil engineering projects in crystalline rocks has enabled the construction of empirical rock mass classification systems to inform the design and choice of construction

© Woodhead Publishing Limited, 2010

Crystalline geological repository systems

147

and support methods in such environments. Provided that the rock mechanics and geotechnical properties have been characterised appropriately, the orientation, geometry and dimensions of underground openings can be optimised in terms of creating a suitably stable environment in which the necessary repository operations of emplacing waste containers and the associated engineered barriers can be conducted safely. Following this approach, tunnels or vaults having diameters or spans on the order of 20 metres or more can be excavated at prospective repository depths on the order of 500 metres in good-quality crystalline rock and with minimal rockbolting will give openings that will remain stable for about 100 years or more.

5.10.1 Construction methods Well-established construction methods are available to excavate such openings, involving the use of drilling and blasting techniques or of tunnel-boring machines. The comparative attributes of these two general methods for excavating tunnels at depth in crystalline rock have been evaluated at the Aspo Hard Rock Laboratory (HRL) in Sweden (Ba¨ckblom et al., 2004). Also, in connection with evaluations of the potential mechanical disturbance caused by excavation, various controlled blasting methods have been developed to enable the damage to the rock surface to be limited, if required. If the excavation has to pass through difficult features such as fracture zones carrying high water flows, a suite of methods has been developed to deal with these situations both from experience in underground research laboratories and in other industries. In the example of high water flows in fracture zones, grouting is likely to be used, as exemplified by the experience gained at the Aspo HRL.

5.10.2 Stabilising the excavations As already noted, the excavated opening is likely to be stabilised by rock bolting with the frequency and length of the rock bolts being determined by the local rock quality. Consideration would have to be given to the choice of materials used in connection with this control method to avoid introducing materials that may be detrimental to repository performance, e.g. organic resins. The spalling of small fragments of rock can be prevented by the use of shotcrete, a surface coating of a specially formulated concrete. Consideration would have to be given to the chemical perturbation that this would cause in the long term, which might necessitate its removal prior to backfilling and sealing the relevant section of the repository. Similar considerations will apply to any use of concrete in structural components of the underground workings, such as floor slabs below waste handling

© Woodhead Publishing Limited, 2010

148

Geological repository systems for safe disposal

machines, and some programmes are studying the potential development of specially formulated concretes that will minimise the chemical perturbation caused by their use (e.g. Vuorinen et al., 2004). Groundwater inflows into the excavations can be controlled to some extent by grouting the flowing fractures or more simply by piping the water into a sump if the flow rate does not warrant grouting. The construction of sealing systems to isolate various compartments of the underground system is covered as part of the discussion of engineered barriers in another chapter.

5.10.3 Design considerations As noted in the discussion on stability, the location and layout of the underground works will be optimised relative to key site characteristics and their evolution in response to potential natural perturbations that may occur in the future. The example of avoiding deformation zones and fractures that may be reactivated has been discussed already and other features, such as fractures carrying excessive flow of groundwater, are also typically identified as requiring to be avoided. A formal process of identifying requirements that should be met to ensure the safety functions of the engineered barriers and geosphere, as now used in the most advanced programmes, provides a framework for making these design decisions (NEA, 2004).

5.11

Future trends

Existing techniques and technologies already provide a high capability to characterise crystalline sites, as evidenced by the interpreted results of site characterisation programmes conducted in Finland and Sweden and the associated identification of remaining issues. In addition, good experience continues to be gained from underground research laboratories such as the Aspo HRL and from the ONKALO rock characterisation facility that is being developed at the selected repository site at Olkiluoto in Finland. Nevertheless, possible improvements have been identified, not least because waste management programmes have an obligation to take account of all relevant scientific and technological advances. An obvious development will be in the use of information obtained from investigations undertaken underground to test the understanding gained from surfacebased investigations, in particular the various models used to represent that understanding. Although research facilities such as the Aspo HRL and the Canadian URL have provided a dress rehearsal of these techniques, this way of testing and calibration of the data and modelling that are to be used in repository design and safety assessment will provide new insights. Aspects that are considered important in these advanced programmes are as follows:

© Woodhead Publishing Limited, 2010

Crystalline geological repository systems . . .

149

calibrating the size distribution and size intensity models for fractures and fracture systems at repository depth from underground investigations; obtaining information on in-plane flow channelling from investigations of coupling of rock mechanics properties and fracture transmissivities in tunnels; and characterising the nature and extent of any changes in hydraulic conductivity that are induced by excavation disturbance as a function of local conditions.

Technological improvements that may be made in connection with underground investigations include: . . .

methods for stabilising investigation boreholes, particularly when intersecting features carrying high water flows; laser scanning to produce three-dimensional surface mapping of rock walls; and methods for measuring water inflows that reflect the previously unperturbed natural system more reliably.

Various technological improvements have been identified for possible future use in surface-based investigations, although in some cases they may also have application underground. A number of these relate to geophysical tools for use in boreholes, as follows: . . . .

a nuclear magnetic resonance tool to relate measured petrophysical properties to fluid flow; an ultrasonic tool (acoustic televiewer) to determine fracture orientation; a downhole tool for in situ measurement of rock thermal conductivity as a function of localised rock characteristics; and ‘walk-around’ borehole seismic techniques to image the key features in the rock mass between boreholes.

In addition, there are recognised benefits in conducting logging while drilling, rather than after the borehole has been drilled to target depth, and this technique has been trialled successfully. It has been recognised that there is value in understanding the occurrence and spatial distribution of different fracture minerals on the site scale, so attempts will be made to develop such models in relation to the role that fracture minerals play both in determining radionuclide transport properties and in buffering the hydrochemistry (e.g. with respect to pH and Eh). In the field of modelling in relation to site characterisation, there are likely to be improved techniques for treating spatial variability in relation to geological structure and hydrogeology and for modelling coupled THMC

© Woodhead Publishing Limited, 2010

150

Geological repository systems for safe disposal

processes over length scales of interest in repository design and safety assessment. Several initiatives are underway in the area of evaluating the stability of crystalline sites (NEA, 2009). An example is the International Tectonics Meeting (ITM) methodology of the Nuclear Waste Management Organisation (NUMO) of Japan, which considers the likelihood of tectonic activity affecting the stability of a potential repository in Japan. It is aimed at assessing the likelihood of a site being significantly affected by volcanic events and rock deformation processes within 104 years and the potential scale and nature of any such impacts as a function of their type and geographic distribution.

5.12

Sources of further information

The site descriptive modelling reports reports published by the Swedish Nuclear Fuel and Waste Management Company (SKB, 2008) and the Finnish waste management company, Posiva Oy (Andersson et al., 2007), respectively on the Forsmark and Olkiluoto sites, and the references cited therein, provide a comprehensive, state-of-the-art view of the characterisation of crystalline sites. These overall site description reports are supported by tiers of reports that provide successively more details on the acquisition and use of information from the relevant geoscientific disciplines that have been discussed in this chapter. All such reports are readily accessible (in May 2009) through the respective websites of these two organisations, respectively www.skb.se and www.posiva.fi. An assessment of geosphere characterisation techniques conducted in the context of the United Kingdom radioactive waste management programme (Nirex, 2007, and references cited therein) provides a good summary of the capabilities of a comprehensive range of techniques used in waste management programmes and other relevant industrial applications.

5.13

References

AECL (1994), ‘Environmental impact statement on the concept for disposal of Canada’s nuclear fuel waste’, Atomic Energy of Canada Limited Report AECL-10711, COG-93-1, Atomic Energy of Canada Ltd, Ottawa. Andersson J, Ahokas H, Hudson J, Koskinen L, Luukonen A, Pitka¨nen P, Mattila J, Ikonen A and Yla¨-Mella M (2007), ‘Olkiluoto Site Description 2006’, Posiva Oy Report POSIVA 2007-03, Posiva Oy, Olkiluouto, Finland. Andersson J, Hermanson J, Elert M, Gylling B, Moreno L and Selroos J-O (1998), ‘Derivation and treatment of the flow-wetted surface and other geosphere parameters in the transport models FARF31 and COMP23 for use in safety assessment’, SKB Report R-98-60, Svensk Ka¨rnbra¨nslehantering AB, Stockholm, Sweden.

© Woodhead Publishing Limited, 2010

Crystalline geological repository systems

151

Andra (2005), Dossier 2005 Granite, Agence Nationale pour la Gestion des De´chets Radioactifs, Paris, France. Ba¨ckblom G, Christiansson R and Lagerstedt L (2004), ‘Choice of rock excavation methods for the Swedish deep repository for spent nuclear fuel’, SKB Report R-04-62, Svensk Ka¨rnbra¨nslehantering AB, Stockholm, Sweden. Bath A (2008), ‘Geochemistry strategy for GeoCORE’, Intellisci Ltd, Technical Note 0607.19 prepared for Nirex, Nuclear Decommissioning Authority, Harwell. Crawford J (ed.) (2008), ‘Bedrock transport properties Forsmark, Site descriptive modelling SDM-Site Forsmark’, SKB Report R-08-48, Svensk Ka¨rnbra¨nslehantering AB, Stockholm, Sweden. Emsley S, Olsson O, Stenberg L, Alheid H-J and Falls S (1997), ‘ZEDEX – a study of damage and disturbance from tunnel excavation by blasting and tunnel boring’, SKB Report TR-97-30, Svensk Ka¨rnbra¨nslehantering AB, Stockholm, Sweden. Glamheden R, Fredriksson A, Ro¨shoff K, Karlsson J, Hakami H and Christiansson R (2007), ‘Rock mechanics Forsmark, site descriptive modelling Forsmark stage 2.2’, SKB Report R-07-31, Svensk Ka¨rnbra¨nslehantering AB, Stockholm, Sweden. Golder Associates (2006), ‘Geosphere Characterisation Project – state of the art in hydrogeological testing and interpretation’, Report 05527634.500/A.0 prepared for Nirex, United Kingdom Nirex Limited, Harwell. Juhlin C (2007), ‘Integrated interpretation of surface and borehole (VSP) seismic data along profiles 2 and 5, Forsmark, Sweden’, in Geology – Background Complementary Studies, Forsmark Modelling Stage 2.2, edited by M Stephens and K Skagius, SKB Report R-07-56, Svensk Ka¨rnbra¨nslehantering AB, Stockholm, Sweden. Laaksoharju M, Smellie J, Tullborg E-L, Gimeno M, Hallbeck L, Molinero J and Waber N (2008a), ‘Bedrock hydrogeochemistry Forsmark, site descriptive modelling SDM-site Forsmark’, SKB Report R-08-47, Svensk Ka¨rnbra¨nslehantering AB, Stockholm, Sweden. Laaksoharju M, Smellie J, Tullborg E-L, Gimeno M, Go´mez J, Auque´ L, Molinero J, Gurban I, Hallbeck L, Buckau G, Gascoyne M and Wallin B (2008b), ‘The hydrogeochemical modelling approach used within the Swedish site investigation programme’, in Abstracts, 33rd International Geological Congress, Oslo, compiled by K Ahlbom and M Stephens, SKB Report R-0897, Svensk Ka¨rnbra¨nslehantering AB, Stockholm, Sweden. Martin C (1990), ‘Characterising in-situ stress domains at the AECL underground research laboratory’, Canadian Geotechnical Journal, 27(5), 631–646. Munier R (200), ‘Statistical analysis of fracture data, adapted for modelling discrete fracture networks – version 2’, SKB Report R-04-66, Svensk Ka¨rnbra¨nslehantering AB, Stockholm, Sweden. Nagra (1994), ‘Kristallin-1 safety assessment report’, Nagra Technical Report 93-22, Nationale Genossenschaft fu¨r die Lagerung radioaktiver Abfa¨lle, Baden, Switzerland. NEA (2001), ‘Going underground for testing, characterisation and demonstration (a technical position paper)’, Nuclear Energy Agency Report NEA/RWM(2001) 6/ REV, OECD Nuclear Energy Agency, Paris.

© Woodhead Publishing Limited, 2010

152

Geological repository systems for safe disposal

NEA (2004), ‘Engineered barrier systems (EBS) design requirements and constraints’, Nuclear Energy Agency Report 4548, OECD, Paris. NEA (2009), ‘Stability and buffering capacity of the geosphere for long-term isolation of radioactive waste – application to crystalline rock’, Nuclear Energy Agency Report 6362, OECD, Paris. Nirex (1997), ‘Nirex 97 – an assessment of the post-closure performance of a deep waste repository at Sellafield’, Nirex Science Report S/97/012, United Kingdom Nirex Limited, Harwell. Nirex (2002a), ‘Research on the alkaline disturbed zone resulting from cement– water–rock reactions around a cementitious repository’, Nirex Report N/054, United Kingdom Nirex Limited, Harwell. Nirex (2002b), ‘Review of the work undertaken on gas migration in the geosphere’, Nirex Report N/023, United Kingdom Nirex Limited, Harwell. Nirex (2007), ‘Geosphere characterisation project-status report October 2006’, Nirex Report N/136, United Kingdom Nirex Limited, Harwell. O¨hberg A and Rouhiainen P (2000), ‘Posiva groundwater flow measuring techniques’, Posiva Oy Report POSIVA 2000-12, Posiva Oy, Olkiluoto, Finland. SKB (2008), ‘Site description of Forsmark at completion of the site investigation phase-SDM-site Forsmark’, SKB Report TR-08-05, Svensk Ka¨rnbra¨nslehantering AB, Stockholm, Sweden. Stephens M, Fox A, La Pointe P, Simeonov A, Isaksson H, Hermanson J and O¨hman J (2007), ‘Geology Forsmark-site descriptive modelling Forsmark stage 2.2’, SKB Report R-07-45, Svensk Ka¨rnbra¨nslehantering AB, Stockholm, Sweden. Triumf C-A (2007), ‘Oskarshamn site investigation – assessment of probable and possible dolerite dykes in the Laxemar sub-area from magnetic total field data and digital elevation models’, SKB Report P-07-223, Svensk Ka¨rnbra¨nslehantering AB, Stockholm, Sweden. Vieno T, Lehikoinen J, Lo¨fman J, Nordman H and Me´sza´ros F (2003), ‘Assessment of disturbances caused by construction and operation of ONKALO’, Posiva Oy Report POSIVA 2003-06, Posiva Oy, Olkiluoto, Finland. Vuorinen U, Lehikoinen J, Imoto H, Yamamoto T and Alonso M (2004), ‘Injection grout for deep repositories, subproject 1 – low-pH cementitious grout for larger fractures, leach testing of grout mixes and evaluation of the long-term safety’, Posiva Oy Working Report 2004-46, Posiva Oy, Olkiluoto, Finland. Wahlgren C-H, Curtis P, Hermanson J, Forssberg O, O¨hman J, Fox A, La Pointe P, Drake H, Triumf C-A, Mattsson H, Thunehed H and Juhlin C (2008), ‘Geology Laxemar-site descriptive modelling SDM-site Laxemar’, SKB Report R-08-54, Svensk Ka¨rnbra¨nslehantering AB, Stockholm, Sweden.

© Woodhead Publishing Limited, 2010

6 Clay geological repository systems: characterisation and site surveying technologies and techniques J . D E L A Y , National Radioactive Waste Management Agency, France

Abstract: The concept of a disposal facility in a clay site relies on the idea that long-term safety will be assured by the argillaceous rock and the repository layout. Therefore, the main objective of the studies carried out on the host formation is to determine its confinement abilities and its stability over a long period of time. Since the physical properties of the argillaceous rocks constrain the layout of the underground structures, the natural properties of the host formation are of concern in the survey strategy. This chapter sets out the technologies and techniques that will be implemented in the framework of such a geological survey. Key words: clay site, argillaceous rock, survey strategy, geological survey, physical properties of the rocks.

6.1

Foreword

The argillaceous sedimentary formations capable of hosting a disposal facility must (Andra, 2005): . .

.

Protect the waste packages from superficial phenomena, mainly climatic (erosion, glaciation) and human activities. Ensure the confinement of radionuclides, i.e. possess intrinsic properties restricting water migration and promoting the retardation of radionuclides. These properties must be ensured over long periods of time (from thousands to several hundred thousand years). Maintain these favourable properties in spite of the perturbations induced on the rock by the construction of the facility (excavation damaged zone, i.e. fractures caused by stress redistribution around the underground structures after excavation). In addition, the barrier properties of the rock must remain intact despite the effects from heat153 © Woodhead Publishing Limited, 2010

154

Geological repository systems for safe disposal generating waste packages and the degradation of the packages and the engineered barriers.

In Europe, three argillaceous sedimentary formations are the subjects of intensive research carried out through both surface surveys and underground laboratory experiments. They are: (1) the Boom Clay at Mol in Belgium (ONDRAF: Organisme National des De´chets Radioactifs et des Matie`res Fissiles, 2001), (2) the Callovo-Oxfordian at Bure, in France (Andra: Agence Nationale pour la Gestion des De´chets Radioactifs, 2005) and (3) the Opalinus Clay at Benken (Nagra: Nationale Genossenschaft fu¨r die Lagerung radioaktiver Abfa¨lle, 2002a) and Mont Terri (Bossart and Thury, 2008) in Switzerland. These rock formations consist mostly of clay minerals, carbonates, quartz and other silicates. In a natural environment, these components may exist in greatly varying proportions and all intermediary facies may be found, whether pure clay, limestone or siltstone. In addition to these major components, organic matter and accessory minerals, such as pyrite, can also be found (Boisson, 2005; Yven et al., 2007). Siliceous shale (porcelanite) and diatomaceous shale studied for nuclear waste disposal in Japan (Horonobe site) may also be considered as equivalents of argillaceous rocks (Hama et al., 2007). However, due to poor compaction and fractures, their hydraulic conductivity is high and a disposal concept cannot rely only on the confining properties of this rock. In Canada a Deep Geologic Repository for low- and intermediate-level radioactive waste has been proposed in a massive Ordovician argillaceous limestone overlain by 200 m of Ordovician shale (Jensen et al., 2009; Mazurek, 2004). In this particular case, the hydraulic conductivity of the argillaceous limestone and the shale is very low and the rock formation is able to ensure the confinement of radionuclides. Techniques for the survey of the host formation are identical to those implemented in the case of more argillaceous host formations. The major aims of a survey carried out on an argillaceous sedimentary site are (1) to identify the geometry of the host formation, i.e. its thickness and sedimentary or tectonic boundaries, (2) to reconstruct the depositional environment and (3) to assess the variability in space of the physical characteristics of the rock. The size of the studied site varies according to the geographic and geological context and the concept and waste inventory of the disposal facility. This site covers the footprint of the facility and a zone extending from several hundred metres to several kilometres around the footprint. This chapter illustrates the site surveys carried out in the framework of deep geological disposal projects in clay formation. It is essentially focused on surface techniques and technologies. It mainly relies on the French, Swiss

© Woodhead Publishing Limited, 2010

Clay geological repository systems

155

and Belgium experiences. The chapter shows examples of lessons learned in following the practical applications of safety requirements (IAEA, 2007) or safety guides (IAEA, 1994) internationally acknowledged.

6.2

Specific features of a clay site survey

6.2.1 Characteristics of the lithology and consequences for the confining properties The confining properties of argillaceous rock formations come from their texture, the nature and relative proportion of their minerals, their low permeability and their deformation properties under stress. Regarding migration and retention properties, the differences encountered between various rock formations of this type depend on their porosities, i.e. on the degree of compaction and diagenetic cementation, and on the relative proportions of the minerals (Samper et al., 2008; Wersin et al., 2008). Clay minerals consist of alumina-silicate crystals in the form of platelets, which trap water molecules. These platelets are electrically charged at their surface with a negative charge. Thus, they are able to retain positively charged ions (sorption process). The negatively charged ions are pushed back from the platelets, reducing thereby the space between the clusters of platelets in which they move (Sammartino et al., 2003) and thus delaying the migration. The pore waters of such rocks are generally reducing and have circumneutral pH values. These environmental conditions are efficiently buffered by the mineral assemblage (Gaucher et al., 2009). Carbonates are in equilibrium with the carbon dioxide dissolved in water. These minerals regulate the pH of the medium, ensuring thereby the chemical stability of the pore water (Vinsot et al., 2008a). The pore water composition also controls the solubility limits of the radionuclides. One of the difficulties met in the characterisation of argillaceous sites is to collect in situ interstitial water unperturbed by sampling conditions. In the survey phases carried out in boreholes from the surface, this characterisation is made on interstitial water extracted from solid samples. The analysis of this water, associated with modelling taking into account the fluid/rock equilibria, provides a good estimation of the pore water composition (Gaucher et al., 2009). Recent results obtained through direct sampling in dedicated boreholes have significantly enhanced the knowledge of the phenomena constraining the chemical composition of pore water (Vinsot et al., 2008b). The porosity of argillaceous rocks greatly varies accordingly, namely to the degree of compaction and diagenesis of the sediments. As an example, porosities of the Opalinus Clay at Mont Terri and Benken, in Switzerland,

© Woodhead Publishing Limited, 2010

156

Geological repository systems for safe disposal

and of the Callovo-Oxfordian clays at Bure, in France, are 12 to 18% in volume (Mazurek et al., 2008). In more compacted clay rocks, it can be much less. In the Callovo-Oxfordian formation the pore size is about 10 to 100 nanometres. This particular geometry accounts for the very low hydraulic conductivity values of these rocks (1012–1014 m/s; Delay, 2007). Consequently, in clay rocks, the chemical elements dissolved in water migrate mainly through diffusion, i.e. through the effect of their own movement, rather than advection, i.e. carried by the flowing water. In a natural environment located at a depth of a few hundred metres, this porosity is usually water saturated. However, a gaseous phase may exist and the size of the pores act as a capillary barrier. In the case of a nuclear waste disposal facility, gas migration mechanisms are also studied since the excavation of the drifts will lead to partial dewatering of the rock, and some components of the disposal facility will release gases. These gases, mostly hydrogen produced by the corrosion of metals, may increase the pore pressure and may also lead to dilatant micro-fracturing in the vicinity of the repository (Marschall et al., 2005; Boulin et al., 2008).

6.2.2 Mechanical and thermomechanical properties of clay rocks Owing mainly to their high content in clay minerals, argillaceous rocks display a complex rheological behaviour. Indeed, as a rule, argillaceous rocks present a low mechanical resistance and a poor ability to conduct heat. However, their creep and swelling properties lead to the self-sealing of fissures or fractures (Bastiaens et al., 2007; Bock et al., 2010). This selfsealing may lead to nearly complete sealing of the excavation-damaged zone of the underground structures after repository closure. The proportions of carbonates or silts, the water content or other specific features may significantly modify each of these mechanical properties. Thus, the Opalinus Clay studied at the Benken site and the Callovo-Oxfordian clays studied at Bure behave differently according to their mechanical strength (uniaxial strength). The strength of Opalinus Clay decreases with an increasing carbonate content, whereas it increases in the case of the Callovo-Oxfordian clays. These different behaviours can be accounted for by micro-structural variations. Results of carbonate micro-fabric investigations confirm that Opalinus Clay contains mainly coarse-grained shell fragments, while the Callovo-Oxfordian clays show a more homogeneous distribution of fine-grained fragments, mainly non-biogenic carbonates. In conclusion, not only their carbonate content but also their grain size distribution, their shape and spatial distribution control the mechanical

© Woodhead Publishing Limited, 2010

Clay geological repository systems

157

behaviour of clays (Klinkenberg et al., 2009). Besides, an increase in quartz content increases the thermal conductivity of the rock. However, the mechanical behaviour of an underground structure depends not only on the intrinsic properties of the rock but also on specific factors such as the depth of the structure and its orientation with respect to the natural stress field. Excavation conditions, support and lining also play a role (Blu¨mling et al., 2007).

6.2.3 Characteristics required from the geological environment Besides the intrinsic characteristics of the argillaceous rock, the host formation and overlying and underlying formations must display certain favourable characteristics in order to be considered as a possible site for hosting a nuclear waste disposal facility. A conceptual geological model has therefore to be developed to establish the favourable nature of the geological environment. This conceptual geological model describes the evolution of the host formation from the phase of sedimentary deposit up to its present state. This model must also be able to describe its future evolution over the time required for the demonstration of repository safety. Firstly, this model is geometrical since it determines the dimensional characteristics of the argillaceous host formation and overlying and underlying formations. This model is also genetic since it helps to reconstruct the burial history of the rock and its micro-structural and geochemical evolution. Finally, this model is dynamic since it attempts to describe the evolution of the pore water migrating through diffusion in the host formation and through advection in more permeable formations. Over the area covering the disposal footprint, the argillaceous host formation must be: .

Thick enough to ensure the long-term confinement of radionuclides. Large-scale hydraulic conductivity must therefore be assessed through tests or through a set of pressure measurements carried out over a long period of time. Since the concentration profiles of natural solutes in the clay pore water (ions, isotopes, gases, i.e. so-called natural tracers) provide information on the vertical transport (Fig. 6.1) (Nagra, 2002b; Bossart and Thury, 2008), the transport mechanisms can be assessed by analysing their concentrations in pore waters extracted from cores. The CLAYTRAC project (Mazurek et al., 2009) on natural tracer profiles across argillaceous formations evaluates the relevance of natural tracer data in the understanding of past geological evolution. Data from nine sites have been analysed to support the scientific studies of the geological barrier and the feasibility of a safe geological repository.

© Woodhead Publishing Limited, 2010

158

Geological repository systems for safe disposal

6.1 Chloride concentration profile in the Opalinus Clay layer at Mont Terri (Switzerland). This profile is accounted for by the diffusion mechanisms of ions in the layer and the difference in concentrations with the Trias and Dogger pore waters at the boundaries of this layer (Mont Terri Project – swisstopo Reports, from Bossart and Thury, 2008; reproduced by permission of swisstopo).

6.2 Burial history of the sedimentary formations in the Zu¨rcher Weinland (Switzerland) (from Nagra, 2002a, 2002b; reproduced by permission of swisstopo).

.

.

Deep enough to ensure long-term protection against erosion and climatic surface phenomena. The recent and past evolution of a potential siting area must therefore be analysed with respect to glacial and fluvial erosion, sedimentation and seismicity arising from crustal movement seismicity (Nagra, 2002a). Homogeneous enough with limited tectonic discontinuities and litholo-

© Woodhead Publishing Limited, 2010

Clay geological repository systems

159

gical variability. This analysis relies on the paleogeographic reconstruction of the depositional environment and the geological history over millions of years (Fig. 6.2). The tectonic setting of the site must also be analysed in order to identify more specifically the past tectonic evolution and the different tectonic stress fields that may have existed or may still exist in the formation. The analysis of the past evolution helps us to understand the sequence of occurrence of faults or folds, and their amplitude. This analysis leads to an assessment of the future tectonic evolution of the site and the hydrogeological role of the tectonic structures and the consequences on future radionuclide release into the geosphere. The simpler the structure of the geological formations, i.e. plane, homogeneous and distant from active tectonic zones, the easier it is to apprehend their characteristics as a whole. The tectonic intra-plate zones of this type of formation are considered as the most favourable for a disposal facility. This comprehensive characterisation of a site survey must ensure that nothing will interfere with the integrity of the repository. The surface area of the studied sector may cover scales of up to several hundred square kilometres (Fig. 6.3). More broadly, all types of impact, not only radiological, are of interest. For instance, the construction and operation of a disposal facility must have no impact on existing natural resources (ground water, geothermal resources, iron ore, oil and gas exploitation). This involves identifying natural resources and assessing their potential.

6.3

Survey tools

Geological characterisation methods fall under three categories: (1) surface methods, (2) survey and observation boreholes, and (3) survey drifts. Specific techniques and technologies are associated with each method to describe or measure the characteristics of the underground. These methods have been developed over time by oil (Downey, 2004) or mining exploration companies (Petit-Dominguez et al., 2008), by civil engineering companies in the framework of the construction of tunnels and by organisations in charge of the management and protection of hydrogeological or geothermal resources. More recently, drift excavation technologies have also made headway, thanks to the refinement of techniques applied in the framework of large infrastructural works such as hydroelectrical installations (Sapigni et al., 2003).

© Woodhead Publishing Limited, 2010

160

Geological repository systems for safe disposal

6.3 Potential implantation of a disposal site in the Benken region (Switzerland) and boundaries of the sector surveyed through threedimensional seismic and investigation boreholes (from Nagra, 2002a, 2002b; reproduced by permission of swisstopo (BA091588)).

6.3.1 Surface methods Surface-based geological methods, such as outcrop mapping and microtectonic analyses, are used to characterise the sedimentary geologic setting with particular attention given to interpreting stratigraphic relationships, geometry and structure (i.e. faults and fracture networks). These methods can be complemented by near-surface ‘light’ geophysics acquisition (e.g. electrical recording, radar, etc.) in order to determine specific layer geometry

© Woodhead Publishing Limited, 2010

Clay geological repository systems

161

and localise discontinuities. The geologic models supported and derived from these data are used, in part, for the static correction of two- and threedimensional seismic data necessary to compensate for velocity variations in overburden. Two-dimensional and three-dimensional seismic methods are used primarily to reconstruct the geometric framework of the geological media (bedrock stratigraphy, presence of faults or fractures) (Lavergne, 1986). Results are available in the form of line data (two-dimensional) or blocs of data (three-dimensional). At depths commonly considered for a geological waste disposal facility (a few hundred metres) these methods can reveal the presence of faults a few metres in vertical throw. Three-dimensional technical resolution is better than two-dimensional as it relies on the spatial coherence of data sets and emphasises the identification of low-amplitude discontinuities (Birkha¨user et al., 2001; Andra, 2005). Stratigraphic inversion or analysis of seismic data has been used to infer petrophysical attributes such as images of porosity variation in carbonates and siltstones. However, in argillaceous formations, only a qualitative estimate of variability can currently be made on the basis of such data. A seismic survey requires the completion of a well calibration to identify the depth of seismic reflectors and the related geological layers, i.e. to convert the time information into depth information by a reliable velocity model (Boyer and Mari, 1997; Birkha¨user et al., 2001). Finally, in order to complete this inventory of surface methods, the installation of a seismological measurement network should be included in this category. This network provides recordings of the natural seismicity and contributes to the general survey of the site.

6.3.2 Drillings and borehole measurements The design of a drilling programme starts with the ranking of the desired objectives and information. Ideally, one should be able to collect simultaneously in the same borehole the maximum information on hydrogeology, rock mechanics, geochemistry, geophysics, etc. However, the technical requirements specific to each of these domains are sometimes incompatible and this set of information often cannot be obtained in a single borehole. Thus, the inability to carry out all the measurements in the same borehole involves drilling several dedicated boreholes with specific technical construction characteristics (cored boreholes, destructive boreholes), as well as designing a measurement programme and a completion adapted to the known or assumed geological and hydrogeological conditions. In practical terms, one of the issues in clay drilling is borehole stability. This stability depends, in part, on the techniques used for drilling and the associated mud.

© Woodhead Publishing Limited, 2010

162

Geological repository systems for safe disposal

Polymer muds are commonly used in sedimentary rocks, but clay drilling is often more efficient with oil-based mud or silicate polymer mud.

6.3.3 Underground structures The construction of a survey shaft or drift is a common means to survey a geological site considered for the construction project of a tunnel in difficult conditions. In the case of studies dedicated to the construction of a radioactive waste disposal facility, pilot laboratories are also built with the view to acquire complementary data and clearing uncertainties regarding the behaviour of the formation and the stability of the underground structures. This type of survey has been conducted in methodological laboratories at Mol, in Belgium (Mertens et al., 2004), and Mont Terri, in Switzerland (Bossart and Thury, 2008), as well as in the Andra laboratory at Bure (Delay et al., 2007a). The construction follow-up of shafts, drifts and ramps greatly enhances the knowledge of the geological environment by providing a change of scale in the sedimentological and structural observations (Langer, 1991). In addition, a good assessment of the ground water productivity and analysis of the producing horizons (porous or fractured horizons) can be made on the basis of water inflows.

6.4

Survey strategy

6.4.1 Analysis of available data Preliminary studies for setting up a geological survey programme include the analysis of all available data (Stewart, 2002). Geophysical, climatic, hydrological, hydrogeological, seismic and other data are collected in addition to geological maps established with more or less accuracy according to the region. This may include information derived from historic oil drilling or seismic campaigns conducted within the sedimentary basins.

6.4.2 Definition of the survey zone boundaries The boundaries of the areas and depths to be surveyed are determined by the predictable extension of radionuclide migration over very long times and the occurrence and location of potential discharge zones; i.e. the potential future changes of the hydrogeological system have also to be taken into account. These potential discharge zones may be controlled by large extension (or regional scale) hydrogeological systems (McEwen, 2007). These constraints require study areas of several thousand square kilometres.

© Woodhead Publishing Limited, 2010

Clay geological repository systems

163

6.4 Extent of the Boom Clay and locations of the investigation boreholes (Belgium) (from Wemaere et al., 2008; reproduced by permission of CEN SCK).

The data necessary for carrying out such large-scale analyses can be found in scientific or technical papers published by national geological surveys. In practical terms, on-site surveys are restricted to a sector limited by faults, outcrops of layers and hydrological limits (river or coastal boundary), which constitute the local limits of regional hydrogeological models. Thus, in France (Delay et al., 2007b), regional hydrogeological studies covered the Paris Basin as a whole but surveys carried out by Andra in 2003 in a preliminary survey phase only covered a sector of about 700 km2. During this campaign, a total of eight boreholes were drilled from five platforms distributed throughout this sector. Figure 6.4 shows, for the Mol site (Belgium), the zone covered by the survey and locations of the boreholes used to build the hydrogeological model (Wemaere et al., 2008). If the sector is large enough, the repository location can be optimised within a less extended zone considered as favourable. In the case of the French project, this zone is called the ‘transposition zone’, i.e. a zone where the results obtained at the underground research laboratory may be considered transferable, both with respect to the formation-specific confining properties and to the nature

© Woodhead Publishing Limited, 2010

164

Geological repository systems for safe disposal

6.5 Map covering the sector studied in France. Location of the Bure underground research laboratory (URL) and boundaries of the transposition zone. Survey grid (seismic lines) used at Bure to determine the level of knowledge on the transposition zone and define the locations of new boreholes. E and F drilling pads are outside the boundaries of the transposition zone (source: Andra).

and occurrence of perturbations induced by the construction and operation of the repository (Fig. 6.5).

6.4.3 Selection of the survey grid and key parameters of the tools adapted to this grid The amount of information to be acquired on the defined perimeter is determined according to several criteria. Most important is the knowledge

© Woodhead Publishing Limited, 2010

Clay geological repository systems

165

about the variability scales of the geological environment. This variability is sedimentary (characteristic length of the sedimentary bodies, extension of the various facies, etc.), but it also comprises structural features (density, length of the structures, etc.). Based on this information, and taking into account available data, it is possible to determine the means and different associated survey grids necessary to fulfil the characterisation objectives. The survey campaign conducted in France in 2007 and 2008 on the transposition zone illustrates this approach. It relies on a 3 km63 km grid of the zone. The length of 3 km was retained not only because it corresponds to a characteristic frequency of the variability of the Callovo-Oxfordian layer (second-order sedimentary variations) but also of the underlying and overlying formations. This survey area is also consistent with the area in which surface and subsurface investigations were conducted during the development and operation of the underground laboratory at Bure. The approach was implemented based on an uncertainty analysis following previous campaigns. Each square was assigned an uncertainty with respect to the others according to the quantity and quality of available data. The objective was to develop a survey approach that would yield data as consistent as possible throughout the transposition zone. The differences between the squares of the grid were minimised by adding complementary ‘survey points’ corresponding to the boreholes. The result converged towards the implantation of four drilling platforms (A to D, Fig. 6.5) within the transposition zone. Two other platforms (E and F, Fig. 6.5) were established according to specific geological criteria related to the fractures of the underlying and overlying formations outside the transposition zone. Since vertical steep structures are under-represented in vertical boreholes, a two-dimensional seismic grid was designed to investigate such a structure with the knowledge of typical fault characteristics and geometry. In order to calibrate the seismic records according to borehole data, these seismic lines pass through the locations of the drilling platforms.

6.4.4 Specific features Detailed studies may be conducted on specific objects if they can play a role in the construction, operation or long-term safety of the disposal facility. They may be structural objects, such as faults or fractures located within, or at a distance from, the footprint of the disposal facility, or they may be current phenomena such as seismic activity and topographic modifications linked, for instance, to glacial rebound (Boulton et al., 2004). The scope of the characterisation studies will be adapted according to the specific features of each disposal project. However, certain recurrent themes of study must be taken into account in the design programmes from the beginning. For instance, specific to clays, the anisotropy of the miner-

© Woodhead Publishing Limited, 2010

166

Geological repository systems for safe disposal

alogical structure linked to the orientation of the minerals and the conditions of their sedimentation may lead to anisotropies of the confining properties, such as the hydraulic conductivity and diffusion coefficient (Wenk et al., 2008). The issue of ‘abnormal’ pressures must also be addressed at the design stage of the measurement and analysis programmes in settings with clay rocks. ‘Abnormal’ pressures are pressures measured in the ground that are not induced by hydraulic head or salinity gradients. Indeed, it is quite common to find ‘abnormal’ pressures (over- or underpressures) in clays. This phenomenon, identified and well documented, thanks to deep oil exploration boreholes (Hunt, 1996 pp. 290–320), is often assigned to compaction or gas production phenomena encountered at great depths. Other in situ and laboratory studies conducted on these subjects (Neuzil, 2000; Rousseau-Gueutin et al., 2009) have led to consider osmotic components, at least partially, for overpressures or underpressures. However, it is likely that in most cases these ‘abnormal’ pressures are caused by coupled phenomena. These phenomena may be hydrodynamic with biphasic components (capillarity), hydromechanical, thermohydromechanical or thermo-osmotic. Modifications of the hydrodynamic limits induced by climatic (glaciation/deglaciation) or topographical changes (erosion) could partially account for the occurrence and/or persistence of these situations. Whatever the reason, the existence of ‘abnormal’ pressure heads reveals the occurrence of low formation scale permeabilities (Fig. 6.6).

6.5

Technologies

Geoscientific technologies and measurement tools are under continuous development and numerous specialised journals and professional associations provide information on the latest enhancements. The sections below describe the main technologies used in this field but only elaborate on those that have undergone a specific development relevant to the survey of a radioactive waste disposal site.

6.6

Geological mapping

Establishing or, more broadly, updating geological maps is an essential stage for any disposal project. If observation conditions are favourable, tectonic and sedimentary surveys using light surface geophysics (electrical, VLF, light seismic, radar) are commonly undertaken. A combination of these measurements determines the stratigraphy of the layers and may reveal structures such as buried valleys infilled to depths of tens to hundreds of metres by drift (Jørgensen et al., 2003). When the surveyed sector is too vast,

© Woodhead Publishing Limited, 2010

Clay geological repository systems

167

6.6 Pressure and permeability profiles obtained within the Paleozoic sedimentary sequence beneath the Bruce site, Ontario, Canada. Environmental head profiles have been corrected for salinity. Anomalous under- and overpressure conditions, indicative of low formation scale bedrock permeabilities, are observed in the Silurian, Ordovician and Cambrian sediments (reproduced by permission of NWMO).

the aim of the geological mapping is limited to identifying the zones of fractures or sedimentary features.

6.7

Geophysical seismic surveys

These technologies are most commonly used by oil companies for the survey of deep levels (over 1000 metres). They must be adapted to the characterisation of argillaceous sites at depths of several hundred metres. Attempts have been made at preserving the high frequencies of the signal in order to increase the resolution power. In the same way, the corrections of surface velocity anomalies (static corrections) must be the object of particular care since, when they are correctly applied, the quality of the seismic imagery in the first few hundred metres can be significantly improved. Usually, acquisition parameters are defined according to the objective characteristics (nature, depth, continuity, desired resolution, etc.). Most often, they rely on the feedback of previous seismic campaigns conducted within the proposed area. One or two days should therefore be dedicated to

© Woodhead Publishing Limited, 2010

168

Geological repository systems for safe disposal

on-site tests, prior to acquisition, in order to verify and optimise these parameters. As regards acquisition parameters, the data processing sequence and interpretation must be adapted to the specific features of the site. Besides the static corrections mentioned above, the velocity analyses must be the object of a strict follow-up.

6.8

Drilling

6.8.1 Drilling programme Once the geometry of the studied formations and the survey grid are defined, drilling is a rapid means for collecting information on the physical and chemical nature of the rock and interstitial fluids. Depending on the characterisation objectives, the boreholes can be vertical, slanted or deviated; vertical boreholes are suited to the study of tabular structures, deviated or inclined boreholes to the study of subvertical fractures and subhorizontal or directional boreholes to the survey of future drifts axes. Cored boreholes Cored boreholes provide a detailed knowledge of the geology and direct observation of the petrophysical, sedimentological and structural features. In addition, thanks to an adapted sampling, a detailed mineralogical characterisation of each horizon can be achieved in the laboratory as well as various direct measurements such as density, porosity and permeability. In clay formations, coring is carried out with a drilling mud that ensures the stability of the borehole walls, the lifting of the cuttings and the cooling of the tool. However, drilling mud creates a ‘cake’ along the walls, which induces a skin effect that conceals the hydraulic properties of the geological formation. The fluid used for hydrogeological measurements is, if possible, a polymer-based mud. For stress measurements, non-hydrated rock core samples and a very high-quality borehole wall surface can be obtained in most argillaceous levels using oil-based mud. Percussion-air drilling In general, percussion drilling is implemented in aquifers overlying and underlying the host formation, and the boreholes are often dedicated to hydrogeochemical characterisation. Thus, fluids are collected in these formations to provide data for the regional hydrogeological and hydrogeochemical models.

© Woodhead Publishing Limited, 2010

Clay geological repository systems

169

The main constraints of boreholes dedicated to hydrogeochemical sampling are as follows: .

.

The drilling technique must be as little contaminating as possible with respect to the interstitial fluids extracted through pumping. Overpressures must be avoided to prevent in-depth invasion of the formation by a possibly contaminating fluid. All fluids and their additives used at the drilling stage (grease, oil, water, etc.) must be sampled eventually to trace the source of chemical pollutants. All the equipment used for drilling or for hydrogeological tests must be clean (physically and chemically). This equipment must release no product or substance liable to alter the chemical quality of the water.

Percussion-air drilling is recommended for short boreholes (less than 200 metres deep) in usually productive formations. This type of drilling is rapid and only requires light equipment and tools commonly used by drillers in the exploitation of water resources for human and agricultural needs. For deep boreholes, i.e. depths ranging between 200 and 800 metres, the most effective technique is reverse-air circulation percussion drilling. With reverse-air circulation the cuttings are lifted inside the rods instead of the hole/tube annular. Thus, since they have not been in contact with the overlaying formations liable to contaminate them, the rock cuttings and formation fluid cuttings reaching the surface are more representative of the formation (Fig. 6.7). Geological programme The geological programme comprises the analysis of the cuttings or cores and the follow-up of the drilling parameters. The monitoring of mechanical and hydraulic parameters while drilling provides information on the characteristics of the drilled layers. When drilling mud, it is standard practice to monitor the levels of the mud tanks (total losses, inflows) and the presence of gas. There are also devices for monitoring the chemical composition of the mud. In cored boreholes the tagging of the mud with chemical elements or fluorescent tracers allows a precise management of the mud flows during the drilling and pumping phases. It enables quality control of the samples dedicated to geochemical analysis. These parameters are recorded and displayed in real time using data acquisition and processing systems provided by ‘mud-logging’ companies. This real-time analysis is necessary in the case of an exploration borehole as monitored parameters can vary rapidly. A constant rate of drilling and maintaining other borehole parameters as steady as possible will ensure the best quality measurements.

© Woodhead Publishing Limited, 2010

170

Geological repository systems for safe disposal

6.7 Principle of the reverse-air percussion drilling technique (source: Andra).

6.8.2 Geophysical wireline logging Well logging allows a detailed record of the geologic formations penetrated by a borehole. The log is based on physical measurements made by geophysical tools lowered into the hole. Historically, the first logging measurements consisted of basic electrical logs (resistivity) and spontaneous potential (SP) logs. They were used to identify oil bearing formations. In the case of an investigation survey they are also used to evaluate the zone around the borehole invaded with the drilling mud.

© Woodhead Publishing Limited, 2010

Clay geological repository systems

171

Electric, sonic and porosity tools, as well as tools for measuring natural gamma radiations and gamma–gamma density radiations, are used to estimate the physical properties of the rock formations. These indirect measurements are also made accurately to determine the depth at which data are obtained through coring. Thus, sonic and acoustic tools combined with density measurements are used to calibrate the reflection seismics. High-resolution electrical or acoustical imaging logs are used to visualise the formation, compute the formation dip and analyse thinly bedded and fractured geological layers. With a high-quality resolution, they can display sedimentary features such as porous horizon distribution or clay layer orientation and organisation. Two main types of tools are available to measure the porosity of the drilled formations: neutron source tools and nuclear magnetic resonance (NMR) tools. NMR tools must be retained for borehole porosity measurements since neutron source tools alter the natural carbon-14 content of the formation fluids. Besides a total bulk porosity value, the NMR tool also provides free fluid and bound water porosity estimates, as well as a qualitative estimation of the permeability (Coates et al., 1999). The analysis of these measurements made in different types of borehole (Fig. 6.8) brings an understanding of the formation lateral and horizontal variability (Lefranc et al., 2008).

6.8 Interpretation of natural gamma-ray logs. Example of a lithostratigraphic correlation in the Callovo-Oxfordian clay formation at Bure (Andra, 2005).

© Woodhead Publishing Limited, 2010

172

Geological repository systems for safe disposal

6.9 Example of an interpretation of hydrogeological logs during a pumping test in a Dogger borehole at Bure. It is to be noted that an important inlet is identified through thermal logging and spinner and two secondary inlets are identified through fluid logging (source: Andra).

6.8.3 Hydrogeological logging tools Two types of wireline tools are used in the hydrogeological logging programmes: (1) flow logging tools, i.e. mechanical, thermal and chemical tools for locating and quantifying the flow zones and (2) fluid sampling probes. Flow logging tools can be used without pumping to highlight the differences in heads and flows between permeable levels. Generally, flow loggings are associated with a pumping phase to characterise water inflows with transmissivities ranging between 109 and 105 m2/s. Measurement of the water level in the borehole during the pumping phase also provides data to estimate a global transmissivity value. This value is compared with the individual measurements obtained through wireline flow loggings. Figure 6.9 shows loggings recorded in the pumping phase associating various techniques in the same borehole. Flow logging Flow logging is used in boreholes when vertical water fluxes generate velocities higher than ca. 5 mm/s. Each of the inlets can then be identified and, assuming a constant head throughout the tested aquifer, the

© Woodhead Publishing Limited, 2010

Clay geological repository systems

173

transmissivity value of each zone can be deduced. This method is considered well suited to boreholes with transmissivitiy values ranging between 107 and 105 m2/s. Different types of borehole flow measuring tools are available on the market. The most widely used are hydrogeological micro-spinners. However, hydrogeological tools are more accurate than oil tools because the spinner is replaced by a turbine with a much better sensitivity and resolution. Electric conductivity fluid logging Fluid conductivity logging (Tsang et al., 1990) has been developed in Switzerland in the framework of regional siting programmes for high-level waste. After the drilling phase, the mud in the borehole is replaced with water with a contrasting electrical conductivity with respect to the pore water. During a pumping phase, electrical conductivity logs show spikes at the inflow points. These peaks broaden and move along the borehole according to the inlets flow rates. In the analysis a numerical inversion simulation is used to obtain the locations and rates of the inlets. Fluid conductivity flow logging has been successfully applied in the Bure area boreholes with transmissivity values ranging between 109 and 105 m2/s (Delay and Distinguin, 2004). Heat-pulse flowmeters Heat-pulse flowmeters are used to measure flows using either the travel time of a thermal pulse or the dilution of heat released by a continuous heat source. A Finnish adaptation of the method has been implemented to characterise the detailed heterogeneity of flow rates ranging between about 0.2 ml/min and over 1 l/min (O¨hberg and Rouhiainen, 2000). Thus, for 25cm diameter boreholes (suitable for a 4-inch pumping device) the transmissivity covered by this type of tool ranges between 108 and 106 m2/s. Samplers When sampling cannot be made from the surface at the outflow of the pumping device, samplers are handled with a cable. The most sophisticated samplers can bring out fluids at the bottom-hole pressure without modifying the fluid/gas ratio. Sampling with a cable consists in lowering into the borehole, at a given depth, a fluid sampling device. The sampling is triggered off either by remote

© Woodhead Publishing Limited, 2010

174

Geological repository systems for safe disposal

control from the surface or by an integrated time switch. There are numerous types of samplers: standard, high pressure (which keeps the sample under pressure), large size and inert (with a special lining of the sampling chamber, e.g. Teflon).

6.8.4 Packer tests For borehole measurements of low hydraulic conductivities in clay rocks, owing to low flows, the borehole sections to be characterised must be isolated with inflatable packers. In these measuring chambers, different types of equipment are used to apply a hydraulic pulse in underpressure or overpressure conditions. Isolation of the test interval is achieved by inflating the packers. Following the isolation of the interval from the lower and upper parts of the borehole, a variation in pressure is applied to the interval (slug or pulse test). The recovery of the pressure or water level to its initial state is recorded. The interpretation of the pressure variations during the recovery of the initial state, according to time, helps to determine the permeability and storativity of the rock formation (Hsieh et al., 1981). The tubing which supports the down-hole equipment constitutes the ‘test string’. The packer consists of a set of sensors (pressure, temperature) and devices (shutting tools, etc.). A surface control and acquisition system is used for recording and displaying in real time all the parameters collected at the bottom of the borehole (pressure, temperature, flow rate, conductivity) and for controlling the tools (opening and shutting of the valves, inflating and deflating the packers). Usually, the test string is lowered into the borehole to carry out a set of tests, starting with the deepest one. When a test is completed the packers are deflated and the system is moved up to the next test interval, and so on. Borehole slug tests A slug test is carried out to assess the permeability of the formation in the immediate vicinity of the borehole for values ranging between 107 and 1010 m/s. A slug test consists in monitoring, in the tubing supporting the down-hole equipment, the return to equilibrium of a water column initially in equilibrium with the formation. In the case of an overpressure choc, the test fluid in the borehole penetrates into the formation, whereas in the case of a depression choc the flow is reversed and the formation fluid enters the borehole. The latter option must be retained whenever borehole conditions are favourable so that injecting the test formation with a foreign fluid can be avoided.

© Woodhead Publishing Limited, 2010

Clay geological repository systems

175

Borehole pulse tests A pulse test is carried out to assess the permeability of the formation in the immediate vicinity of the borehole for values ranging between 109 and 1014 m/s. Pulse tests require a device that can apply a choc and equipment to measure the pressure in the test interval when the flow is insignificant. The reliability of a pulse test requires assessing as accurately as possible the volume of the fluid involved in the test. This volume depends on the compressibility of the down-hole equipment and the physical properties of the fluid. The equivalent compressibility of the system therefore depends on the compressibility of the fluid (test fluid) and equipment (metal parts) in the test chamber, on the packers (rubber + inflating fluid), as well as on the elasticity of the borehole walls in the test zone.

6.8.5 Geomechanical tests: acquisition of a stress profile Several technologies are available to determine the stress state throughout a sedimentary sequence. The variations in horizontal and vertical natural stresses are the consequences of the rheological properties and geometrical characteristics of the stratigraphic layers as well as the loading history of the rock mass. Technologies typically used in deep boreholes in the framework of characterisation surveys are described in the following paragraphs (Wileveau et al., 2007) (Fig. 6.10). Other methods, such as overcoring, used in mines or underground laboratories are not well adapted to deep boreholes. Traditional hydraulic fracturing test and the hydraulic test on pre-existing fracture (HTPF) method The hydraulic fracturing technique is used to estimate the value of the minimum principal stress (Haimson, 1993). It relies on the fact that a hydraulic fracture will be created and will propagate normal to the minimum principal stress. The pressure at which the fracture closes, i.e. closure pressure, is an estimate of the minimum principal stress. When the borehole is parallel to a principal stress, the normal to the fracture plane is the orientation of the minimum principal stress. Acoustic and electrical wireline logging tools are used to detect the fracture dip and orientation. Borehole images are acquired before the tests and compared with images acquired after the test. Instead of creating a new fracture, the same methodology may be used to determine the stress component acting normal to a pre-existing plane of weakness, either a bedding plane or a pre-existing fracture. This is the so-

© Woodhead Publishing Limited, 2010

176

Geological repository systems for safe disposal

6.10 In situ stress profile established at Bure (France). It is to be noted that when the content in clay is high (Callovo-Oxfordian), the minor and vertical constraint values are very similar (source: Andra, from Wileveau et al., 2007).

called HTPF (hydraulic test on pre-existing fracture) method (Cornet, 1986). Combined with hydraulic fracturing information, such tests help to determine additional components of the stress tensor. Sleeve fracturing test The sleeve fracturing test consists in the strictly mechanical reopening of a hydraulic fracture. This test evaluates the stress concentration at the borehole wall in a given orientation. Desroches and Kurkjian (1999) propose to interpret this test with the same assumptions as those proposed by Haimson (1993) to interpret reopening fractures in low-permeability rocks. Systematic analysis of breakouts in boreholes Another method consists in analysing the position and orientation of breakouts in deviated boreholes of different orientations (Brudy and

© Woodhead Publishing Limited, 2010

Clay geological repository systems

177

Zoback, 1999). In deviated boreholes, the orientation of breakouts along the hole provides information on the shape of the stress ellipsoid (Peska and Zoback, 1995; Haimson, 2007). This information provides a ranking of the stress components and an estimate of the maximum horizontal principal stress, when knowledge about the minimum horizontal stress σh (sigmah) and the vertical stress σv (sigmav) is available.

6.8.6 Borehole long-term monitoring equipments Once the borehole and associated in situ measurements are completed borehole instrumentation can be installed. Generally, such completions are installed to observe the stabilisation of formation pressures, which in lowpermeability sediments may occur over periods of 1 to 2 years. It takes, for instance, one or two years for a hydraulic head to stabilise in a deep borehole in a formation with a 1013 m/s average permeability. Hydrogeological equipment commonly installed in a deep borehole falls under three main types: .

.

.

Pressure sensors (sometimes associated with temperature and conductivity sensors) installed below the water level, or in depth in front of the inlets. This type of equipment is used in the most permeable geological formations. Pressure and temperature sensors installed in a concrete backfilled measuring chamber at the bottom of the borehole. These types of sensor are put in place permanently without the intent to retrieve them (Delay et al., 2007c). Their advantage is that they are perfectly isolated from the rest of the borehole, but their drawback is that they cannot be replaced, or recalibrated, in case there seems to be a problem with their functioning. Wireless or wireline data transmission has been used successfully. Multi-packer completions are used to follow up simultaneously up to twenty static levels in the same borehole. This type of device is relatively costly, but it is very effective even with low permeability environments.

Sometimes, other types of completion are installed with aims other than strictly hydrogeological ones. In France, a sophisticated completion was installed in a borehole from the surface at a depth of 537 metres to follow up the diffusion of radioactive tracers (Delay et al., 2007c). This equipment, installed in 2004, is still functioning in 2009. Finally, in certain geothermal contexts, innovative optic fibre temperature measuring devices have been installed in boreholes to detect transmissive formations or zones over the long term (Yamano and Goto, 2005).

© Woodhead Publishing Limited, 2010

178

Geological repository systems for safe disposal

6.8.7 Workover operations When it is technically possible, it may be sensible to perform additional wireline logs or surveys. Such measurements can be particularly useful in quantifying the evolution of the geometry and temperature in the borehole, or changes in formation permeabilities. Workover operations are commonly scheduled in survey programmes to avoid costs associated with drill rig stand-by.

6.9

Underground structures surveys

Underground structures, whether built for a laboratory or a disposal facility, provide a unique opportunity to observe geological layers in large size and to carry out in-depth geological analysis. In addition, the geomechanical behaviour of the structures can be assessed in situ following the excavation of the front face. In the case of the construction of a deep disposal facility, a scientific follow up will be conducted to confirm the previous observations provided by surface investigations and will include the development of experimental niches, drifts or experimental cells or vaults. Any operation linked to the excavation or safety of the excavation worksite can be adopted by the investigation programme. For example, boreholes advanced beyond the excavation face to detect pressurised water or gas zones can be used to collect samples of uncontaminated groundwater in the more permeable formations.

6.10

Core lab analysis

When a borehole is cored, the samples are packaged in a specific way according to the type of analysis. Four main types of sampling and analysis are carried out in clays: .

.

Sampling for geological studies: mostly mineralogical and microstructural analysis. The identification of fossils and microfossils is also carried out to interpret the depositional environment (Thierry et al., 2006). Sampling for geomechanical studies. In clays, it is important that a sample will not undergo structural modifications due to decompression to maintain it as close as possible to its in situ state. Thus, following its extraction from the core barrel, the sample is stored in a special cell where it is cemented under load. The main elastic, poroelastic and thermal properties can then be determined in the laboratory (Homand et al., 2006).

© Woodhead Publishing Limited, 2010

Clay geological repository systems .

.

179

Sampling for permeability measurements (permeability to water or gas). The samples, also packaged in a confining cell, are the subject of measurements carried out parallel and perpendicular to the stratigraphy (Delay et al., 2006). Sampling for geochemical studies. The samples are packaged in specific cells or gas-tight bags as soon as they are extracted from the core barrel. In these geochemical cells air is replaced either by vacuum or by a gas, which restricts the reactions with the rock minerals. Sophisticated leaching or squeezing extraction techniques are implemented in specialised laboratories to determine the geochemistry of the pore water (Gaucher et al., 2006). To extract the gases from the cores, the samples are usually packaged directly on site in air-tight cells and gas extraction is carried out on the spot.

6.11

Integration of results

The common goal of the characterisation studies, sample analysis and modelling is to provide documents in support of the licensing for the construction and operation of a nuclear waste disposal facility. It usually consists of three types of technical documents using data acquired on the sites: (1) a geosynthesis, (2) a disposal facility design and (3) a safety analysis. The current practice is to gather the analysis and field measurements in a geosynthesis that provides a global understanding of past and current functioning of the geological system. In parallel, the scientific results are used in the design of the disposal facility, i.e. in the description of its general architecture and the geometry of the modules and disposal cells. In this respect, the rock mechanics results are of particular significance for clay rocks. The geosynthesis is used to build the conceptual model integrating the confinement properties of the geological environment. The conceptual model serves as a frame to describe the evolution of the disposal facility through scenarios. These scenarios take into account the sensitivity of the geological environment to perturbations induced by the construction and operation of the disposal facility, including its closure and long-term monitoring stages, and finally the long-term geological evolution of the whole system during the post-closure phase. In addition, the safety functions, i.e. the objectives to be achieved by a disposal facility, are analysed in a safety analysis according to various scenarios.

© Woodhead Publishing Limited, 2010

180

Geological repository systems for safe disposal

6.12

Future trends

In recent years, significant breakthroughs have been made by the various national and international agencies (NEA, IAEA, etc.) thanks to the surveys and research studies carried out in the framework of international collaborations. They are taken into account when updating safety requirements (IAEA, 2006, 2007). Numerous research programmes and partnerships carried out in the framework of European projects have also provided opportunities to share human and technical means as well as experimentation sites. In this respect, projects carried out at the Mol and Mont Terri sites have played a pioneering role in international cooperation by making available to the scientific community the first underground structures specifically dedicated to this type of research. Currently, the surveys of sedimentary basins and argillaceous environments are making rapid headway, thanks, in part, to the efforts of oil companies interested in CO2 sequestration in deep geological layers. In some respects, this subject is very close to activities carried out in the past ten years in the framework of nuclear waste disposal. As regards subjects more specifically related to nuclear waste disposal, in spite of successful attempts made in boreholes from the surface, the fields of experimentation dealing with the interactions between the geological environment (clays and shales) and disposal materials (glass, iron, cement), as well as subjects dealing with the sealing of shafts and drifts, will continue to be studied, for the greater part, in URLs and on samples in surface laboratories. The same applies to studies dedicated to transport and microbiological phenomena. Two other major scientific and technical axes must be developed from boreholes. The first one deals with the geomechanical characterisation of argillaceous formations. At the present, it is difficult to predict reliably over the long term, the behaviour of a shaft or a drift, and the extent of the damaged zone around these structures. The presence of hydromechanical couplings also raises numerous theoretical and experimental questions, regarding in particular the dewatering and rehydratation phenomena of clays. The second axis is the analysis of gas migration in argillaceous rocks. Whether in the frame of the analysis of natural gas migration or radionuclide migration induced by the generation of gas in the disposal facility, experiments carried out from the surface should be considered to acquire a better understanding of the elementary phenomena and hydraulic and geomechanical couplings.

© Woodhead Publishing Limited, 2010

Clay geological repository systems

6.13

181

Sources of further information

An exhaustive review of websites on existing or potential radioactive disposal facilities has been published by Rempe (2007). The reader will find there all the sources of information on all ongoing projects worldwide. The status of these projects is regularly updated by the DOE in a document called Geological Challenges in Radioactive Waste Isolation, the fourth edition of which came out in April 2006 (Witherspoon and Bodvarsson, 2006). The reader will find in this issue a synthesis made by each country involved in a disposal facility project. The agencies in charge of radioactive waste management bring out, according to the legislations or administrative processes, synthesis reports presenting the progress of their projects to their safety authority and their government. These reports are detailed syntheses of sites and projects. They are the objects of international peer reviews and are made available to the public. The most recent ones are: . . .

For Belgium, the SAFIR 2 (ONDRAF, 2001) report. For Switzerland, reports on the ‘Opalinus Clay’ project (Nagra, 2002a, 2002b, 2002c). For France, Dossier 2005 (Andra, 2005).

Two files are scheduled in France in the coming years, one in 2013 for Public Debate and the other in 2015 for the Licensing of the Bure site, and one in Canada in 2012 for Public Hearing regarding the Bruce site project. Information regarding the Canadian project on the proposed Bruce site Deep Geologic Repository may be found at http://www.nwmo.ca/. Numerous congresses, symposiums and journals regularly present stateof-the-art progress in the various fields of earth sciences. Scientific journals compile the approaches of the academic community, research agencies, waste producers and reviewers of research programmes. Specialised publications also take stock of the progress of scientific works. Some of the most recent are as follows: . .

The document published by the Mont Terri consortium: Mont Terri Rock Laboratory Project, Programme 1996 to 2007 and Results (Bossart and Thury, 2008). Papers published following the congress ‘Clays in natural and engineered barriers for radioactive waste confinement – Lille 2007’, organised by Andra (France), SKB (Sweden), Nagra (Switzerland) and ONDRAF (Belgium) (Special Issue 2008, Physics and Chemistry of the Earth; see the papers mentioned in the reference).

© Woodhead Publishing Limited, 2010

182 .

Geological repository systems for safe disposal Reviews in Engineering Geology XIX – Deep Geologic Repositories (Rempe, 2008) published by the Geological Society of America.

Regarding geophysics, and more specifically the two- and three-dimensional seismic reflection, the websites of the main contractors present their activities and major research works that are under way. Most of them offer integrated services, from acquisition to interpretation. Among others are: http://www.cgg.com/ http://www.westerngeco.com/ http://www.dmt.de/en/home.html http://www.beicip.com/

6.14

Acknowledgements

The bulk of this review is based on information provided by Andra, SCKCEN, Nagra and the Mont Terri Project. I warmly thank Andreas Gautschi (Nagra) and Mark Jensen (NWMO) for reviewing this chapter. I would also thank Isabelle Wermaere (SCKCEN), Paul Bossart (Mont Terri Project – Swisstopo), Ce´line Righini (Andra), Agne`s Vinsot (Andra), Lise Feuillaˆtre (Andra), Georges Vigneron (Andra) and Anne de Henning for their contribution to this work.

6.15

References

Andra (2005), Dossier 2005: Ce´de´rom Argile – plaquettes, synthe`se, tomes, re´fe´rentiels et glossaire, Chaˆtenay-Malabry, Available from www.andra.fr (Accessed 14 August 2009). Bastiaens W, Bernier F and Li X (2007), ‘SELFRAC: experiments and conclusions on fracturing, self-healing and self-sealing processes in clays’, Physics and Chemistry of the Earth, 32, 600–615. Birkha¨user P, Roth P, Meier B and Naef H (2001), ‘3D-Seismik: Ra¨umliche Erkundung der mesozoischen Sedimentschichten im Zu¨rcher Weinland’, Nagra Technical Bericht NTB 00-03. Blu¨mling P, Bernier F, Lebon P and Martin D C (2007), ‘The excavation damaged zone in clay formations time dependent behavior and influence on performance assessment’, Physics and Chemistry of the Earth, 32, 588–599. Bock H, Dehandschutter, Martin D C, Mazurek M, de Haller A, Skoczylas F and Davy C (2010), ‘Self-sealing of fractures in Argillaceous Formations in the context of geological disposal of radioactive waste – Review and synthesis’, OECD NEA Report, NEA, Paris, in press. Boisson J Y (ed.) (2005), ‘Catalogue of characteristics of argillaceous rocks studied with respect to radioactive waste disposal issues’, in OECD/NEA IGSC Working Group on the Characterisation, the Understanding and the Performance of Argillaceous Rocks as Repository formations (the ‘‘Clay Club’’), OECD/ NEA, Paris.

© Woodhead Publishing Limited, 2010

Clay geological repository systems

183

Bossart P and Thury M (2008), ‘Mont Terri Rock Laboratory Project, Programme 1996 to 2007 and results’, Reports of the Swiss Geological Survey 3, Swiss Geological Survey, Wabern, Switzerland. Boulin P F, Angulo-Jaramillo R, Daian J-F, Talandier J and Berne P (2008), ‘Pore gas connectivity analysis in Callovo-Oxfordian argillite’, Applied Clay Science, 42, 276–283. Boulton G, Chan T, Christiansson R, Ericsson L O, Hartikainen J, Jensen M R, Stanchell F W and Wallroth T (2004), ‘Thermo-hydro-mechanical (T-H-M) impacts of glaciation and implications for deep geologic disposal of nuclear waste’, Elsevier Geo-Engineering Book Series, 2, 299–304. Boyer S and Mari J-L (1997), Seismic Surveying and Well Logging, Editions Technip, Paris. Brudy M and Zoback M D (1999), ‘Drilling-induced tensile wall-fractures: implications for determination of in-situ stress orientation and magnitude’, International Journal of Rock Mechanics and Mining Sciences, 36, 191–215. Coates G R, Xiao L and Prammer M G (1999), NMR Logging – Principles and Applications, Halliburton Energy Services Publication H02308, Houston, Texas. Cornet F H (1986), ‘Stress determination from hydraulic test on pre-existing fractures – the H.T.P.F. method’, in Proceedings of the International Symposium on Rock Stress and Rock Measurements, Stockholm, Lulea, Centek, pp. 301–312. Delay J and Distinguin M (2004), ‘Hydrogeological investigation in deep wells at the Meuse/Haute Marne Underground Research Laboratory, Northeastern France’, in Engineering Geology for Infrastructure Planning in Europe, LNES 104, Springer, Berlin, pp. 219–225. Delay J, Trouiller A and Lavanchy J M (2006), ‘Proprie´te´s hydrodynamiques du Callovo-Oxfordien dans l’est du basin de Paris: comparaison des re´sultats obtenus selon diffe´rentes approches’, CR Geosciences, 338, (12–13), 892–907. Delay J (2007), ‘Permeability measurements in Argillaceous Rocks at the Meuse/ Haute/Marne Underground Research Laboratory, France’, in IAH Selected Paper 10, Chapter 7, Taylor & Francis, London, pp. 87–110. Delay J, Vinsot A, Krieguer J M, Rebours H and Armand G (2007a), ‘Making of underground experimental programme at the Meuse/Haute-Marne Underground Research Laboratory, Northeastern France’, Physics and Chemistry of the Earth, 32, 2–18. Delay J, Rebours H, Vinsot A and Robin P (2007b), ‘Scientific investigation in deep wells at the Meuse/Haute-Marne Underground Research Laboratory, Northeastern France’, Physics and Chemistry of the Earth, 32, 42–57. Delay J, Dewonck S and Distinguin M (2007c), ‘Characterisation of a clay rich rock through development and installation of specific hydrogeological and hydrogeochemical equipment at the Meuse/Haute-Marne Underground Research Laboratory’, Physics and Chemistry of the Earth, 32, 393–407. Desroches J and Kurkjian A (1999), ‘Applications of wireline stress measurements’, SPE 48960, SPE Reservoir Evaluation and Engineering, 2, 451–461. Downey M W (2004), ‘Oil and natural gas exploration’, Encyclopedia of Energy, 2004, 549–558. Gaucher E C, Blanc P, Bardot P, Braibant G, Buschaert S, Crouzet C, Gautier A,

© Woodhead Publishing Limited, 2010

184

Geological repository systems for safe disposal

Girard J P, Jacquot E, Lassin A, Negrel G, Tournassat C, Vinsot A and Altmann S (2006), ‘Modelling the porewater chemistry of the CallovianOxfordian formation at a regional scale’, CR Geosciences, 338, (12–13), 917– 930. Gaucher E C, Tournassat C, Pearson F J, Blanc P, Crouzet C, Lerouge C and Altmann S (2009), ‘A robust model for pore water chemistry of clayrock’, Geochimica et Cosmochimica Acta, in press, accepted manuscript, (Available online 22 July 2009). Haimson B C (1993), ‘The hydraulic fracturing method of stress measurement: theory and practice’, Comprehensive Rock Engineering, 3, 395–412. Haimson B C (2007), ‘Micromechanisms of borehole instability leading to breakouts in rocks’, International Journal of Rock Mechanics and Mining Sciences, 44, 157–173. Hama K, Kunimaru T, Metcalfe R and Martin A J (2007), ‘The hydrogeochemistry of argillaceous rock formations at the Horonobe URL site, Japan’, Physics and Chemistry of the Earth, 32, 170–180. Homand F, Shao J F, Giraud A, Auvray C and Hoxha D (2006), ‘Pe´trofabrique et propriete´s me´caniques des argilites’, CR Geosciences, 338, (12–13), 882–891. Hsieh P A, Tracy J V, Neuzil C E, Bredehoeft J D and Silliman S E (1981), ‘A transient method for determining the hydraulic properties of ‘tight’ rocks – 1. Theory’, International Journal of Rock Mechanics and Mining Sciences, 18, 245–252. Hunt J M (1996), Petroleum Geochemistry and Geology, W. H. Freeman and Company, New York. IAEA (1994), ‘The siting of geological disposal facilities. A safety guide, Safety series 111-G-4.1, IAEA, Vienna, Available from www-ns.iaea.org/standards/ documents/ (accessed 6 October 2009). IAEA (2006), ‘Geological disposal of radioactive waste’, Safety Requirements WSR-4, IAEA, Vienna, Available from www-ns.iaea.org/standards/documents/ (accessed 6 October 2009). IAEA (2007), ‘Disposal of radioactive waste, Draft Safety Requirements DS 354, IAEA, Vienna, Available from www-ns.iaea.org/standards/documents/ (accessed 6 October 2009). Jensen M, Lam T, Luhowy D, McLay J, Semec B and Frizzell R (2009), ‘Ontario power generation’s proposed L&ILW deep geologic repository: an overview of geoscientific studies’, GeoHalifax 2009, CGS-IAH(CNC), Halifax, Nova Scotia, Canada. Jørgensen F, Lykke-Andersen H, Sandersen P B E, Auken E and Nørmark E (2003), ‘Geophysical investigations of buried Quaternary valleys in Denmark: an integrated application of transient electromagnetic soundings, reflection seismic surveys and exploratory drillings’, Journal of Applied Geophysics, 53, 215–228. Klinkenberg M, Kaufhold S, Dohrmann R and Siegesmund S (2009), ‘Influence of carbonate microfabrics on the failure strength of claystones’, Engineering Geology, 107, 42–54. Langer M (1991), ‘Engineering geological investigations for planning and construction of an underground repository for low-level radioactive wastes’, Engineering Geology, 30, 115–126.

© Woodhead Publishing Limited, 2010

Clay geological repository systems

185

Lavergne M (1986), Me´thodes Sismiques, Editions Technip, Paris. Lefranc M, Beaudoin B, Chile`s J P, Guillemot D, Ravenne C and Trouiller A (2008), ‘Geostatistical characterization of Callovo-Oxfordian clay variability from high-resolution log data’, Physics and Chemistry of the Earth, 33, S2–S13. McEwenT (2007), ‘Site selection and characterisation, deep geological disposal of radioactive waste’, Radioactivity in the Environment, 9, 77–111. Marschall, P, Horseman, S and Gimmi, T (2005), ‘Characterisation of gas transport properties of the Opalinus Clay, a potential host rock formation for radioactive waste disposal’, Oil and Gas Science and Technology – Revue de l’Institut Franc¸ais du Pe´trole, 60/1&2. Mazurek M (2004), ‘Long-term used nuclear fuel waste management – geoscientific review of the sedimentary sequence in Southern Ontario’, Institute of Geological Sciences, Technical Report TR 04-01, University of Bern, Bern. Mazurek M, Gautschi A, Marschall P and Vigneron G, Lebon P and Delay J (2008), ‘Transferability of geoscientific information from various sources (study sites, underground rock laboratories, natural analogues) to support safety cases for radioactive waste repositories in Argillaceous Formations’, Physics and Chemistry of the Earth, 33, S95–S105. Mazurek M, Alt-Epping P, Bath A, Gimmi T and Waber H N (2009), ‘Natural tracer profiles across Argillaceous Formations: the CLAYTRAC project’, OECD/ NEA Report 6253, OECD Nuclear Energy Agency, Paris, France. Mertens J, Bastiaens W and Dehandschutter B (2004), ‘Characterisation of induced discontinuities in the Boom Clay around the underground excavations (URF, Mol, Belgium)’, Applied Clay Science, 26, 413–428. Nagra (2002a), ‘Projekt Opalinuston – Synthese der geowissenschaftlichen Untersuchungsergebnisse (Project Opalinus Clay – synthesis of geological investigation results)’, Nagra Technical Report NTB 02-03, Nagra, Wettingen. Nagra (2002b), ‘Project Opalinus Clay – Safety Report’, Nagra Technical Report NTB 02-05, Nagra, Wettingen. Nagra (2002c), ‘Projekt Opalinuston – Konzept fu¨r die Anlage und den Betrieb eines geologischen Tiefenlagers (Project Opalinus Clay – concept for the facilities and the operation of a deep repository)’, Nagra Technical Report NTB 02-02, Wettingen, Nagra. Neuzil C E (2000), ‘Osmotic generation of ‘‘anomalous’’ fluid pressure in geological environment’, Nature, 403, 182–184. O¨hberg A and Rouhiainen P (2000), ‘Posiva groundwater flow measuring techniques’, Posiva Oy Report 2000-12, Posiva, Olkiluoto. ONDRAF (2001), CD-rom SAFIR 2: (NIROND 2001-06 E – December 2001) (Nirond 2001-05 E – December 2001), Brussels, Available from www.ondraf.be (accessed 14 August 2009). Petit-Dominguez M D, Rucandio M I, Galan-Saulnier A and Garcia-Gimenez R (2008), ‘Usefulness of geological, mineralogical, chemical and chemometric analytical techniques in exploitation and profitability studies of iron mines and their associated elements’, Journal of Geochemical Exploration, 98, 116–128. Peska P and Zoback M D (1995), ‘Compressive and tensile failure of inclined well bores and determination of in situ stress and rock strength’, Journal of Geophysical Research, 100, 12792–12812.

© Woodhead Publishing Limited, 2010

186

Geological repository systems for safe disposal

Rempe N T (2007), ‘Permanent underground repositories for radioactive waste’, Progress in Nuclear Energy, 49, 365–374. Rempe N T (2008), Deep Geologic Repositories, Reviews in Engineering Geology XIX, The Geological Society of America, Boulder, Colorado. Rousseau-Gueutin P, de Greef V, Gonc¸alve`s J, Violette S and Chanchole S (2009), ‘Experimental device for chemical osmosis measurement on natural clay-rock samples maintained at in situ conditions: implications for formation pressure interpretations’, Journal of Colloid and Interface Science, 337, 106–116. Sammartino S, Bouchet A, Preˆt D, Parneix J-C and Tevissen E (2003), ‘Spatial distribution of porosity and minerals in clay rocks from the Callovo-Oxfordian formation (Meuse/Haute-Marne, Eastern France) – implications on ionic species diffusion and rock sorption capability’, Applied Clay Science, 23, 157– 166. Samper J, Dewonck S, Zheng L, Yang Q and Naves A. (2008), ‘Normalized sensitivities and parameter identifiability of in situ diffusion experiments on Callovo–Oxfordian clay at Bure site’, Physics and Chemistry of the Earth, 33, 1000–1008. Sapigni M, La Barbera G and Ghirotti M (2003), ‘Engineering geological characterization and comparison of predicted and measured deformations of a cavern in the Italian Alps’, Engineering Geology, 69, 47–62. Stewart S (2002), ‘Exploring the continental shelf for low geological risk nuclear waste repository sites using petroleum industry databases: a UK case study’, Engineering Geology, 67, 139–168. Thierry J, Marchand D, Fortwengler D, Bonnot A and Jardat R (2006), ‘Les ammonites du Callovien-Oxfordien des sondages Andra dans l’Est du basin de Paris: synthe`se biochronostratigraphique, inte´reˆtes pale´oe´cologique et pale´obioge´ographique’, CR Geosciences 338, (12–13), 834–853. Tsang C F, Hufschmied H and Hale F V (1990), ‘Determination of fracture inflow parameters with a borehole fluid conductivity logging method’, Water Resources Research, 26, 561–578. Vinsot A, Appelo C A J, Cailteau C, Wechner S, Pironon J, De Donato P, De Cannie`re P, Mettler S, Wersin P and Ga¨bler H-E (2008a), ‘CO2 data on gas and pore water sampled in situ in the Opalinus Clay at the Mont Terri rock laboratory’, Physics and Chemistry of the Earth, 33, S54–S60. Vinsot A, Mettler S. and Wechner S (2008b), ‘In situ characterization of the CallovoOxfordian pore water composition’, Physics and Chemistry of the Earth, 33, S75–S86. Wemaere I, Marivoet J and Labat S (2008), ‘Hydraulic conductivity variability of the Boom Clay in north-east Belgium based on four core drilled boreholes’, Physics and Chemistry of the Earth, 33, 2008, S24–S36. Wenk H R, Voltolini M, Mazurek M, Van Loon L and Vinsot A (2008), ‘Preferred orientations and anisotropy in shales: Callovo-Oxfordian shale (France) and Opalinus Clay (Switzerland)’, Clays and Clay Mining, 56, 285–306. Wersin P, Soler J M, Van Loon L, Eikenberg J, Baeyens B, Grolimund D, Gimmi T and Dewonck S (2008), ‘Diffusion of HTO, Br, I, Cs+, 85Sr2+ and 60Co2+ in a clay formation: results and modelling from an in situ experiment in Opalinus Clay’, Applied Geochemistry, 23, 678–691. Wileveau Y, Cornet F H, Desroches J and Blumling P (2007), ‘Complete in situ stress

© Woodhead Publishing Limited, 2010

Clay geological repository systems

187

determination in an argillite sedimentary formation’, Physics and Chemistry of the Earth, 32, 866–878. Witherspoon P A and Bodvarsson G S (2006), ‘Geological challenges in radioactive waste isolation, Fourth Worldwide Review’, Report LBNL-59808, Ernest Orlando Lawrence Berkeley National Laboratory – University of California, Berkeley. Yamano M and Goto S (2005), ‘Long-term monitoring of the temperature profile in a deep borehole: temperature variations associated with water injection experiments and natural groundwater discharge’, Physics of the Earth and Planetary Interiors, 152, 326–334. Yven B, Sammartino S, Ge´raud Y, Homand F and Villie´ras F (2007), ’Mineralogy, texture and porosity of Callovo-Oxfordian argillites of the Meuse/HauteMarne region (Eastern Paris Basin)’, Me´m. Soc. Geol. France, 178, 73–90.

© Woodhead Publishing Limited, 2010

7 Assessing the long-term stability of geological environments for safe disposal of radioactive waste K.J. WILSON and K.R. BERRYMAN, GNS Science, New Zealand

Abstract: The future stability of the earth’s crust is a critical factor in the site selection process for geological repositories of hazardous radioactive waste. The chapter compares the timeframes of geological repository operation with the timeframes of geological processes. The potential effects of tectonic instability and climate change are discussed and common techniques of assessing and modelling crustal stability are outlined. Key words: volcano-tectonic stability, geological repository, climate change and nuclear waste, long-term tectonic deformation, modelling tectonic stability.

7.1

Introduction

The stability of the earth’s crust over the lifetime of a nuclear waste repository is a critical factor in the site selection process for geological repositories. Events such as earthquakes and volcanic eruptions can have varying effects on the safe operation of the geological repository, depending on the size and proximity of the event. The effects may range from mild shaking to catastrophic rupture or exposure of the waste repository. Fortunately the principles and processes driving geologic instability, namely plate tectonics and climate change, are well understood. There are a number of techniques used by geoscientists to characterise the level of risk and to predict the long-term behaviour of the earth’s crust. In this chapter we review the timeframes relevant to nuclear waste repositories and compare these with the timeframes over which various geological processes operate. In general, geological repositories are expected to be effective (i.e. keep the hazardous waste isolated from the environment) 188 © Woodhead Publishing Limited, 2010

Assessing the long-term stability of geological environments

189

for 100 000 to 1 000 000 years. Many methods have been developed for geoscientists to assess the present and short-term geological stability (e.g. 100 years) of a site, but assessing the long-term stability requires an understanding of how volcano-tectonic processes change and migrate over time. Processes to consider include the migration of existing fault zones into new areas, the cumulative impact of slow uplift and erosion, and the possibility of new volcanoes or geothermal zones developing. At timescales greater than several thousands of years the potential effects of climate change, such as sea level change and ice sheet loading, also need to be considered. We describe the geological and geophysical techniques that can be used to quantify past and present deformation, and also describe how models of long-term crustal stability are developed.

7.2

Long-term volcano-tectonic stability issues for safe disposal of radioactive waste

There are various timeframes over which the stability of geological environment needs to be assessed when considering potential geological repository sites and assessing their future performance. As geological repositories are constructed, filled and interned, and as the radioactive waste decays, there are changing levels of acceptable risk from geological instability. We will discuss the timescales over which the geological repositories are operational and contrast this with the timescales over which geological processes change or evolve. Construction and filling of the repositories is expected to take decades. During this initial period the waste is at its highest level of radiation and it is not fully sealed from the biosphere; therefore the hazard from the waste is at its maximum. Some geological repository projects require that the repository remain open for a certain period after filling for monitoring or retrieval of the waste. For example, the US Nuclear Regulatory Commission requires nuclear waste be fully retrievable for 50 years after emplacement and it is planned to keep the proposed Yucca Mountain repository open for 100 years after it is filled. Once a repository is sealed the time period over which it must provide protection depends upon the rate of radioactive decay and release of radionuclides. A common benchmark when assessing the hazard of the waste, if it were to be released into the environment, is the comparison of the waste radioactivity to the radioactivity of natural uranium ore, the reasoning being that once the waste has decayed to levels equal to that found in nature then it is no longer a hazard to the biosphere. The time over which nuclear waste decays to ‘natural levels’ depends on the nature of the waste; processed high-level waste (HLW) will decay to natural levels within a few

© Woodhead Publishing Limited, 2010

190

Geological repository systems for safe disposal

thousand years, while spent fuel will take one to several hundreds of thousands of years (Chapman, 2006). Ten thousand years is generally considered a minimum time period that a geological repository will operate for and most organisations assess the stability of geological repositories over timescales of > 100 000 years. The USA has radiation standards for geological repositories based on millirem levels; a millirem is a measure of radiation dose absorbed by humans. New standards introduced in 2008 require repositories to have a radiation limit of 15 millirem per year for the first 10 000 years after disposal and 100 millirem per year between 10 000 to 1 million years (EPA, 2008). The Nuclear Waste Management Organisation of Japan (NUMO) focuses on a time period of 100 000 years for assessing tectonic stability (Apted et al., 2004). Geological processes affecting the stability of the earth’s crust are primarily related to earthquakes, volcanoes and climate change. The relevance of each geological hazard in relation to deep repositories varies over different timescales. For example, the probability of an ice sheet covering large parts of the earth in the next few thousand years is extremely low but the probability of earthquakes along plate boundaries is very high. Geological processes likely to operate in the short term (<104 years) involve earthquakes along known fault zones at plate boundaries, seismicity within stable intraplate regions, tsunamis and volcanic eruptions at known volcanic zones (typically at plate boundaries but also at intraplate hotspots; Fig. 7.1). Some climate change effects will potentially operate on short timescales, such as changes in groundwater hydrology and increased coastal erosion due to sea level rise and intensified storminess. Over the medium term (104 – several 106 years) the plate boundaries (Fig. 7.1), at which most of the earth’s deformation occurs, will stay in approximately the same location, but the associated zones of seismicity and volcanism will probably migrate within the wider plate boundary zone. An example of this is the migration of an area of upper plate uplift along the Chilean plate boundary related to the subduction of the Nazca Rise. Because subduction occurs obliquely along this margin, the Nazca Rise, a prominent ridge on the downgoing oceanic plate, is moving southward along the margin at approximately 7 km per 100 000 years. The associated zone of upper plate uplift (recorded by marine terraces) also moves southward along the margin, causing significant changes in the pattern of upper plate uplift detectable at timescales of 100 000 to 1 000 000 years (Hsu, 1992; Hampel et al., 2004). Similarly, zones of volcanism can migrate around the plate boundary zone. For example, in the Taupo volcanic zone (TVZ) within the backarc of the Hikurangi subduction zone, New Zealand, the locations and types of volcanism have changed in the past 2 million years (Ma). Andesitic volcanism started in the TVZ at 2 Ma, ryholitic volcanism began at ~1.6 Ma and the main area of rifting is migrating eastward over

© Woodhead Publishing Limited, 2010

Assessing the long-term stability of geological environments

191

7.1 Tectonic plate boundaries of the earth – these are the zones where most of the earth’s seismicity and volcanism occurs. Also shown are the regions of distributed deformation and volcanic hotspots. Key: 1. Divergent plate boundaries; 2. Transform plate boundaries; 3. Convergent plate boundaries; 4. Plate boundary zones, broad regions of diffuse deformation; 5. Major hotspots (source: adapted from USGS).

time (Wilson et al., 1995). Hotspot volcanism also migrates on timescales of > 1 000 000 years. For example, the Snake River Plain/Yellowstone hotspot has migrated to the northeast at approximately 40 km/million years over the past 15 Ma (Nash et al., 2006). At timescales > 10 000 years climate changes in the form of natural glacial–interglacial cycles. These cycles are generally related to orbital cycles that operate on periodicities of 23 000, 41 000 and 100 000 years (Hays et al., 1976). Due to strains associated with the loading and unloading of ice sheets at high latitudes and global sea level changes on the order of 150 m, climate cycles can have significant effects on the stability of the earth’s crust. On timescales greater than several million years the location and nature of plate boundaries can change. For example, the Himalayas have formed since the Indian plate collided with the Eurasian plate approximately 45 million years ago (Fig. 7.2). A new divergent plate boundary, manifest as the East African Rift, has developed through Ethiopia in the past 20 Ma (Pik et al., 2008). Long-term climate fluctuations also occur on timescales of tens of millions of years. For example, the early Eocene period (55–40 Ma) and late Cretaceous period (100–65 Ma) are recognised as times of warmer temperatures, or ‘thermal maximums’ (Poulsen et al., 2003; Zachos et al.,

© Woodhead Publishing Limited, 2010

192

Geological repository systems for safe disposal

7.2 Schematic diagram showing the northward migration of the Indian continental plate and its eventual collision with the Eurasian plate. This first-order plate boundary change occurred over timescales greater than several millions of years (source: adapted from USGS).

2008). Causes of these long-term climate variations may be related to factors such as plate tectonics (e.g. continent positions changing with respect to the poles and influencing oceanic circulation), mega-scale volcanism and methane release from continental shelves.

© Woodhead Publishing Limited, 2010

Assessing the long-term stability of geological environments

7.3

193

Geochemical stability issues for safe disposal of radioactive waste

Knowledge of the geochemistry and flow patterns of groundwater are of critical importance to the effective operation of geological repositories. Both of these factors can influence the corrosion rate of the engineered barriers encasing the waste, plus the entrainment rate and flow path of mobile radionuclides that are eventually released from the repository. While highly impermeable host rocks are typically selected for geological repositories, the host rock will still contain some groundwater (the exception being evaporite formations, which are impermeable). Ideally, the groundwater will be anoxic (reducing) with very low rates and volumes of flow (Chapman, 2006). Geological processes that can affect groundwater geochemistry over time are primarily concerned with volcanic and geothermal activity, climate change causing salinity and oxygenation variations, and faulting affecting the rock permeability. In general nuclear waste repositories will not be sited in or near active volcanic zones. However, there can be the possibility of future volcanic intrusions into previously non-volcanic areas (Mahoney et al., 2009). A magmatic intrusion close to the surface of the earth’s crust (e.g. within 8 km) will cause significant changes to groundwater. The flow patterns will be affected by the convective circulation of heated water and geochemistry will be altered, particularly by mixing of near-neutral pH water and steam to produce oxidising ClSO4 waters (Henley and Ellis, 1983). Reducing conditions are favourable in geological repositories because many radioelements such as uranium, neptunium, technetium and selenium are more soluble in fresh oxidising groundwater. Many radionuclides vary in solubility according to pH; for example, 239Pu and 238U are less soluble under alkaline conditions (Nirex, 2001). Magmatic intrusions and geothermal heating in the vicinity of nuclear waste repositories therefore have the potential to alter the flow, oxidation and pH of the groundwater and this can affect the solubility and distribution of radionuclides. Climatic variations can affect the geochemistry of groundwater in several ways. One example is the changing salinity of groundwater caused by sea level changes. With changing salinity the performance of bentonite buffers is altered; compacted bentonite is a commonly proposed engineered barrier for nuclear waste (Arcos et al., 2003; Tanaka et al., 2007). On a short timescale anthropogenic-induced climate warming in the next several thousand years will probably cause a rise in sea level with associated salinisation of groundwater in coastal areas. Over longer timescales continental areas at high latitudes, such as on the Canadian and Fennoscandian shields, show evidence of hypersaline brines forming during glacial periods of the Pleistocene. The brines migrate through the crystalline basement in response

© Woodhead Publishing Limited, 2010

194

Geological repository systems for safe disposal

to ice sheet loading and unloading (Starinsky and Katz, 2003). Strains associated with ice sheet loading have also been proposed as a mechanism for injecting fresh, oxygenated water into deep groundwater (Nirex, 2005a; Talbot, 1999). Fault rupture has also been suggested as a mechanism by which permeability pathways may be opened into deep repositories, potentially leading to a ‘freshening’ of the groundwater. Fault rupture will not necessarily create a permeable pathway through rock because the fault plane can be filled with a ‘fault gouge’ material of pulverised, very fine grained rock, which may not be permeable. However, there have been reports of water well levels changing in response to earthquakes at distances of > 1000 km from the earthquake epicentre, showing that groundwater circulation can be affected by earthquakes (Montgomery and Manga, 2003).

7.4

Potential climate change issues for safe disposal of radioactive waste

Climate and the geological environment are interrelated in many complex ways and therefore climate change can potentially affect the stability of the geological environment. In recent times, the consideration of future climate change is mainly concerned with global warming. There is evidence that the global average temperature is rising due to anthropogenic increases in carbon dioxide in the atmosphere (IPCC, 2007). There are also natural cycles of climate change related to orbital variations of the earth (also called Milankovich cycles); at present the earth is in an interglacial period but less than 20 000 years before the present it was in a glacial period. The timescales and effects of both anthropogenic-induced global warming and natural Milankovich cycles on geological repositories will be discussed. Evidence of global warming comes from a variety of datasets including rising average surface temperatures, rising sea levels and decreasing snow and ice cover (IPCC, 2007). These trends have been documented over the past several decades (with some records extending back to 1850 AD). Therefore the timescale of global warming is on the order of tens to hundreds of years, although predicting future trends remains controversial and typically has a large degree of uncertainty. There are many known and potential impacts of global warming but the phenomena most likely to impact geological repositories are those related to the coastal zone, namely rising sea levels and increased storminess. Predictions of sea level rise for the period up to 2099 AD range from 0.18 to 0.59 m (IPCC, 2007). Beyond the next century some scenarios incorporating ice sheet melting predict up to 7–10 m of sea level rise in the next millennium (Gregory et al., 2004; Kerr, 2006; Overpeck et al.,

© Woodhead Publishing Limited, 2010

Assessing the long-term stability of geological environments

195

2006). Geological repositories located near the coastline and near sea level will probably see an increased frequency of seawater inundation due to sea level rise, e.g. the inundation of repository access tunnels or ventilation shafts. Accompanying sea level rise will be an increase in the inland extent of storm surge inundation. On relatively flat coastlines a small rise in sea level (<0.5 m) could see the inland extent of storm surge inundation increase by kilometres, so that even repositories at seemingly large distances inland may become vulnerable. With higher sea levels and closer proximity of the coastline to geological repositories it is possible that the groundwater will become more saline. Increases in groundwater salinity may in turn have geochemical effects such as increasing corrosion or reducing the efficiency of the repository protective barriers (McKinley and Alexander, 2009). There is some evidence of increased storm activity related to global warming, e.g. an increase in the intensity of hurricanes (Knutson and Teleya, 2004; Webster et al., 2005). The greater intensity and/or frequency of storms could increase rates of coastal erosion and the inundation of storm surges. The Dounreay nuclear power station in Scotland provides an example of the effects of coastal erosion. A shaft that was used for nuclear waste disposal from 1959 to 1977 extends from the Dounreay station. The upper part of the shaft is located within 15 m of 5-m high seacliffs and is at risk of being breached due to coastal erosion processes within the next 160– 240 years (Nirex, 2005b). A seacliff stabilisation project is being carried out at Dounreay to protect the shaft while the plant decommissioning work is carried out. Sites will vary in vulnerability to coastal erosion according to local conditions such as geology (e.g. rock type and bedding plane geometry), coastal exposure and the presence of natural barriers such as sand dunes, reefs and estuaries. Over a longer timescale (1000’s of 100 000’s years) the stability of the geologic environment in most areas is likely to be affected by natural climate cycles, specifically the return to a glacial period. Predictions of the future return to a glacial period range from 55 000 years (based on orbital variations) to never (taking into account anthropogenic CO2; Berger and Loutre, 2002). The effects of a glacial climate on the geological environment will vary markedly with latitudinal position from generally less impact at the equator to severe impact at high latitudes. A case study in Britain found the greatest impacts of glaciation on geological repositories there would be erosion and changes in hydrology (Nirex, 2005a). Glaciers and ice sheets are known from the geological record to erode vast areas and there is a risk that after ice retreat waste repositories could be exposed or at least become closer to the surface. However, areas of past glacial erosion are relatively simple to identify from geomorphology such as U-shaped valleys, striations and moraines; therefore it is likely that repositories can be sited to avoid such features or possibly be buried deep

© Woodhead Publishing Limited, 2010

196

Geological repository systems for safe disposal

enough to mitigate the effects. It is also likely that glaciation would cause significant changes in groundwater and surface water hydrology. The Nirex report (2005a) suggests that the most significant potential hydrologic effect is large ice sheets driving oxygenated water deep into the underlying geology, particularly along pre-existing faults. The change in surface topography caused by ice sheet erosion and deposition may also change the drainage patterns of a region, possibly resulting in significant changes to fluvial and groundwater systems. During the emplacement, occupation and retreat of large glaciers and ice sheets there is adjustment of the lithosphere to the weight (and de-weighting) of the ice mass, called glacio-isostatic adjustment. For waste repositories located near continental margins at high latitudes the future elevation of the site with respect to sea level over the timescale of a glacial cycle may need to be considered. Also related to ice sheet unloading are the relationships between glacio-isostasy and seismicity. Observations of postglacial fault activity and historic seismicity in cratonic shield areas formerly covered by large ice sheets (mainly eastern Canada and Fennoscandia) suggests that there is a pulse of seismicity due to crustal stress changes following ice sheet retreat (Muir-Wood, 2000; Stewart et al., 2000; Ma et al., 2008; Olesen et al., 2008). The implications for nuclear waste repositories is that pre-existing bedrock faults, which show no sign of recent activity, may be reactivated at certain times within glacial–interglacial cycles.

7.5

Using geological, geophysical and geochemical techniques for quantifying stability for safe disposal of radioactive waste

The future stability of the earth’s crust can be assessed by studying past and present crustal deformation. Data from such studies can then be used to model the future behaviour of the crust (modelling techniques will be discussed in Section 7.6). Of particular importance is the ability to recognise and understand the patterns of tectonic instability. For example, a potential repository site not undergoing any measurable deformation at present may be located along the path of a volcanic chain; knowing the rate of volcanic chain migration is critical to assessing future stability at the site. The commonly used techniques of quantifying the stability of the geosphere are discussed below.

7.5.1 Geological mapping Basic geologic field mapping will identify rock types that are suitable or unsuitable for hosting geological repositories. Bedrock properties preferable

© Woodhead Publishing Limited, 2010

Assessing the long-term stability of geological environments

197

for repositories are low permeability and strong coherency. Preferable bedrock types generally fall into three categories: hard crystalline rock (e.g. granite, gneiss, volcanic tuff), argillaceous rock (e.g. claystone, mudrock, shale) and evaporites (bedded or domed salt deposits). Geological mapping will identify the location, and likely thickness, of these types of rocks, as well as some of the geotechnical properties. For example, highly fractured, thin shale will not be appropriate for a geological repository, but thick, relatively non-fractured shale may be suitable. Importantly, geologic mapping will also rule out many areas; e.g. regions of relatively young, non-compacted sandstone and limestone would be discounted due to the high permeability of such rocks. Field mapping and airphoto analysis can identify indicators of mass movement. Of most importance to geological repositories is the hazard of deep-seated bedrock landslides; deformation from such failures has been recorded at depths of up to 250 m (Petley and Allison, 1997). It is likely that most repositories will be located at > 250 m depth, but such surficial disturbance could greatly alter the rock weathering processes, groundwater levels and strain patterns in the bedrock, while also affecting the surface infrastructure and access tunnels. Shallow landslides are more frequent and widespread than deep-seated landslides but will generally not greatly affect a geological repository unless such an event occurs during the initial construction phases. Landslide hazard can be assessed by mapping the active landslides and the landforms of previous landslides and studying the correlations between landslides and other landscape characteristics such as rock type, bedding direction, slope angle and aspect and rainfall (e.g. Yamagishi et al., 2002; Van Den Eeckhaut et al., 2009). Other important geological features that will typically be identified by geological mapping are fault lines and volcanoes. Fault lines are usually recognised by the juxtaposition of two rock types that are not in chronological or stratigraphic order. For example, when Quaternary sediments (past 1.8 Ma) are overlain by Cretaceous (145–65 Ma) units; they cannot have been deposited in such an order and therefore must be separated by a fault line. Further details on how active fault lines are mapped and analysed will be discussed in Section 7.5.3. Active volcanoes are very obvious features in the landscape and simple to map at a coarse scale, but most volcanoes represent many eruptive events and at a detailed scale the ‘volcano’ may be represented as several edifices, vents and deposits. The classification of volcanoes and determination of frequency and magnitude to underpin probabilistic hazard can be a complex exercise (see, for example, Mahoney et al., 2009; Wetmore et al., 2009). Dormant or inactive volcanoes can be less obvious in the landscape but are usually identified by morphology (e.g. a cone shape or caldera lake) and geology (i.e. basaltic, andesitic or other such volcanic rocks). There are various dating methods

© Woodhead Publishing Limited, 2010

198

Geological repository systems for safe disposal

used, such as radiocarbon, luminescence, 40K–40Ar and 39Ar –40Ar, to attain the age of volcanic features so that the history and hazard of the volcano can be assessed.

7.5.2 Measuring current crustal deformation using global positioning system (GPS) When subjected to tectonic stresses, the earth’s crust will deform. Global positioning system (GPS) techniques measure mm-scale movement of the earth’s surface related to active tectonic deformation. GPS measurements are taken at survey points permanently attached to the ground either by intermittent (survey-style) or continuous (daily, automated) collection of phase and pseudorange data from the constellation of GPS satellites that orbit the earth. GPS measurements enable the quantification of tectonic strain at different timescales (e.g. interseismic tectonic deformation over periods of decades, to coseismic movement over seconds during an earthquake) and at different spatial scales (e.g. regional to continental scale deformation). GPS techniques are useful in assessing tectonic stability of potential nuclear waste repositories as they enable the determination of regional tectonic strain rates, the location of possible new blind faults with no obvious surface expression (e.g. ‘hidden faults’) and the detection of aseismic slip. Elevated strains around a geological repository could disrupt the repository in numerous ways, including through the reactivation or formation of new faults, creep deformation of the engineered barriers and flexural folding of the host rock. An example of GPS techniques being used to measure tectonic strain is at Yucca Mountain, where a dense GPS network was installed to monitor the potential geological repository site. Prior to GPS installation it was thought that the Yucca Mountain area was under relatively little strain because nearby Quaternary faults have extremely low slip rates (<0.02 mm/year). However, analyses of the GPS data indicates a right-lateral strain across the region with > 0.7 mm/year of deformation currently accumulating on unidentified structures in the Yucca Mountain region (Wernicke et al., 1998; Hill and Blewitt, 2006). At a larger scale GPS has been used to map the tectonic strain across most plate boundaries around the world; it can also help identify blocks within tectonically active areas that are undergoing smaller amounts of deformation, relative to the surrounding region. Faults with no previously identified surface expression (hidden or blind faults) can be detected in some cases with the aid of GPS measurements (e.g. Donnellan et al., 1993; Stevens et al., 2002), and zones of very high strain where faults may develop in the future can also be identified. For example,

© Woodhead Publishing Limited, 2010

Assessing the long-term stability of geological environments

199

in southern Kyushu, Japan, a zone of high left-lateral shear not identified by active faulting studies was defined from GPS studies (Wallace et al., 2009). The shear zone is interpreted to be associated with the subduction of a ridge, but because the ridge subduction point is continually migrating along the margin, the zone of high shear also migrates and does not localize in one place long enough to allow a fault zone to develop that is identifiable at the ground surface (Wallace et al., 2009). Based on the migration rate of the subducting ridge, the future location of a high left lateral shear strain can be predicted. GPS techniques can also be used to detect aseismic slip, which is when a fault slips without an accompanying earthquake. Aseismic fault creep has been detected on a few crustal faults (e.g. Azzaro et al., 2001; Lyons et al., 2002), but this typically occurs along faults with high slip rates that are identifiable at the surface. GPS techniques are most useful when used to identify aseismic slip at subduction zones. These are called slow slip events in which decimetres of movement occur on the interface between two plates over timescales of days to months (e.g. Dragert et al., 2001; Obara et al., 2004; Douglas et al., 2005). For example, the Manawatu slow slip event on the Hikurangi subduction interface, New Zealand, produced 35 cm of slip over 18 months. If this movement had been released suddenly it would likely have generated a MW 7.0 earthquake (Wallace and Beavan, 2006). Slow slip events along subduction interfaces are unlikely to be a direct hazard to nuclear repositories as there is no ground shaking associated with the events and the events occur at depths over several kilometres. However, slow slip events may alter the tectonic strain pattern in the upper plate of subduction margins and the identification of such events allows a better understanding of the seismic behaviour of subduction zones.

7.5.3 Active fault mapping and paleoseismology The presence of active faults is one of the soundest pieces evidence that an area is seismically active and will continue to be so in the future. The common definition of an active fault is one that has ruptured in the Holocene period (past 11 700 years). However, when considering the location of nuclear waste repositories it is necessary to use a longer timeframe and consider all faults that have been active within the Quaternary period (past 1.8 Ma). Active faults can pose a direct hazard to a repository by fault rupture through the repository (hence disruption of the engineered barrier and alteration of groundwater flow paths). Active faults near to the repository can also pose a hazard of severe ground shaking, landsliding, changing groundwater levels, uplift and possible oxidation, as well as altering the local tectonic strain field. A common approach to locating active faults is first to use aerial photos

© Woodhead Publishing Limited, 2010

200

Geological repository systems for safe disposal

to identify geomorphic scarps and lineaments, followed by mapping and trenching the fault to constrain the frequency, timing and size of past events. This field of research is called paleoseismology. Active faults can often be recognised in the landscape by a fault scarp; on a flat surface this will be recognisable as a linear step cross-cutting all features of the surface or, at the foot of mountain ranges, fault scarps can be identified by a distinct change in the surface slope (Fig. 7.3). The bedrock around active faults can be sheared and weakened by fault movements, so often fault lines produce alignments of linear valleys as streams take advantage of the weakened bedrock. Active faults can also be recognised by offset landscape features; e.g. streams, ridgelines and terraces can be displaced by faults (Fig. 7.3).

7.3 Photograph of the Alpine Fault, a major strike-slip fault in South Island, New Zealand. Prominent geomorphic markers that help to identify the active fault are shown (source: K. Wilson).

© Woodhead Publishing Limited, 2010

Assessing the long-term stability of geological environments

201

Once a fault line has been identified in the landscape it is important to understand several parameters. When did it last rupture? How frequently does it rupture? What size are the earthquakes? Paleoseismic investigations typically involve mapping the fault, observing local geology, geomorphology and rock outcrops, and trenching the fault. When mapping fault lines, offset features of the landscape will be recorded. Some of the offset features such as river terraces will have a known age or be able to be dated; therefore a fault slip rate can be calculated. For example, if a fault offsets a last glacial age river terrace (c. 20 000 years old) by 10 m the fault will have a slip rate of 0.5 m/1000 years (0.5 mm/year). The timing of the last rupture can be determined by dating the youngest offset feature. The frequency of fault rupture can be determined from dating successively offset landscape features (e.g. a sequence of abandoned channels) or by trenching. Fault trenching involves digging a large pit across the fault scarp, typically > 20 m long and > 2 m deep. Within fault trenches successive sedimentary units offset by the fault can be dated to determine the timing of past earthquakes (e.g. Lienkaemper et al., 2002; Kelson et al., 2006; Villamor et al., 2007). For some well-studied faults the size of the past earthquakes can be estimated using scaling relationships developed from historical earthquakes. These commonly relate earthquake magnitude and measurable geomorphic features such as the surface rupture length, the maximum surface displacement and the average surface displacement (Wells and Coppersmith, 1994; Hanks and Bakun, 2002). For example, Field et al. (2008) assembled active fault data for California and converted fault data to equivalent earthquake magnitude and frequency so that the historical record of earthquakes could be supplemented by earthquakes identified in the geological record. The paleoseismology methods discussed above can be applied to active faults with an identifiable scarp. However, many active faults, particularly those with very long recurrence intervals, do not display an obvious surface expression. Alternative methods of identifying active fault locations are required. For example, in Australia, active faults with very low slip rates (~0.02 mm/year) bounding the Flinders Range, south-central Australia, have been identified (Quigley et al., 2006; Hillis et al., 2008). The scarps of the faults were covered by Holocene sediments but exposures of Neoproterozoic schist (500–1000 Ma) thrust over Pleistocene (< 1.8 Ma) river gravels within stream exposures allowed identification of fault activity within the Quaternary period. Other techniques to recognise zones of active faulting include the identification of drainage pattern asymmetries (where active faults have altered drainage patterns, e.g. Cox, 1994, Meghraoui et al., 2001), identification of liquefaction features (where ground shaking causes liquefaction of soft sediments such as alluvial sands, e.g. Guccione, 2005; Singh and Jain, 2007), identification of seismites and seismically

© Woodhead Publishing Limited, 2010

202

Geological repository systems for safe disposal

triggered clastic dikes (sedimentary layers shaken by earthquakes, typically recognised by the reorganisation of magnetic signals in the sediment, e.g. Marco and Agnon, 2005; Levi et al., 2006) and identification of earthquaketriggered speleothem breakage events in caves (e.g. Lemeille et al., 1999; Pons-Branchu et al., 2004). Active fault mapping identifies areas in which nuclear repositories should not be sited, while the results of paleoseismology studies provide important information for seismic hazard analysis. Seismic hazard analyses use earthquake catalogues compiled from instrumental and historic data. However, in areas with short historic records or with seismogenic sources with long recurrence intervals, paleoseismology is needed to characterise low-frequency but large-magnitude events.

7.5.4 Historical seismological record The historical seismological record is derived from two sources: the instrumentally recorded earthquake catalogue and written historical records of earthquakes in the pre-instrumental period. The timescales over which the two sources record earthquakes varies. For example, Italy and Japan have written historical records of earthquakes dating back > 1000 years whereas written historical records for New Zealand date back to only 1840 AD due to its short history of European occupation. Historic records of earthquakes come from a variety of sources, usually written records but also oral histories. For example, the 1700 MW 9 Cascadia subduction earthquake is recorded in Native American oral histories and there is also a written account of the resulting tsunami in Japanese literature (Satake et al., 1996). The instrumentally recorded earthquake catalogue varies by region and magnitude. For example, there is considered to be a complete record of global shallow earthquakes ≥ M 7 since the early 1900s and a complete record of global shallow ≥ M 6 events only since the early 1960s (Pe´rez and Scholz, 1984). The instrumental record for smaller events depends upon the seismic network of each country. For example, New Zealand has had a seismic network since the 1930s, whereas the Ecuador national seismic network was only installed in 1990. Mapping the distribution of historical seismicity provides a clear indication of tectonically active and inactive regions (Fig. 7.4). When considering the seismological issues for location of a nuclear waste repository a probabilistic seismic hazard analysis (PSHA) will usually be undertaken. The aim of this is to understand the hazards of earthquake shaking and abrupt fault displacement at any site. The first step in a PSHA is to locate all known earthquake sources in the area (usually within 100–200 km of the site). One input for the earthquake sources is the mapped active faults, which indicate the sources for large earthquakes (described in the

© Woodhead Publishing Limited, 2010

Assessing the long-term stability of geological environments

203

7.4 An extract from the New Zealand earthquake catalogue showing the size and depth of earthquakes in New Zealand of ≥ M 4.0 from 1964 to 1997. This shows large areas of New Zealand that are seismically quiescent. The earthquake catalogue data are used in probabilistic seismic hazard analysis (see Stirling et al., 2001).

previous section). The other input is the historic seismicity record, which records the moderate-to-large background earthquakes. Background earthquakes are those events less than or equal in size to the ground-rupturing fault events, and can occur both on and away from the faults. They are wellillustrated by maps of seismicity (Fig. 7.4). The background earthquakes that do not occur on mapped faults may represent tectonic strain accumulation in between major faults or events on unmapped or blind faults (Stirling et al., 2009). Because earthquakes of <M 6.5 generally do not rupture the ground surface they will not be recorded by paleoseismic methods; therefore the frequency of such events needs to be derived from the historic siesmicity record.

© Woodhead Publishing Limited, 2010

204

Geological repository systems for safe disposal

An important use of historical seismicity catalogues is in the understanding of earthquake magnitude–frequency relationships. Large earthquakes occur less frequently, while small earthquakes occur with great frequency (specifically, the Gutenberg–Richter relationship shows that the logarithm of the frequency of the earthquake occurrence is a decreasing function of magnitude; Gutenburg and Richter, 1944). These relationships mean that the historic seismicity record can use the frequency of the smallto-moderate sized earthquakes to predict the frequency of large earthquakes. The magnitude–frequency relationships are key components of probabilistic seismic hazard analyses.

7.5.5 Indicators or tectonic uplift or subsidence Uplift and subsidence of the earth’s surface occurs in response to a variety of processes, e.g. glacial loading, volcanic intrusion or tectonic forcing. Vertical deformation of the earth’s surface should be considered in the process of siting nuclear waste repositories for two main reasons. Firstly, the initial burial depth of a repository could increase or decrease over time if the earth’s surface is raised or lowered. For example, if a repository is buried at a depth of 300 m at a site undergoing uplift at 3 mm/year, it will only take 100 000 years for the repository to reach the surface (if surface erosion keeps pace with the rock uplift). Secondly, vertical deformation at the earth’s surface may reflect an active ‘hidden’ structure at depth. Uplift and subsidence can be measured over a range of time and spatial scales (e.g. Litchfield et al., 2009). For short-term (minutes to decades) measurements of vertical deformation, common methods include: . . . . .

Continuous GPS (continuous measurement of a point to cm-scale accuracy for vertical motion) Geodetic surveying (levelling to mm-scale precision between widely spaced locations and comparing survey results over years to decades) Tide gauges (these record relative sea-level changes and can record land movement over years to decades) InSAR (using radar satellite images from different times to detect changes in the surface elevation to centimetres of precision) Biologic or geomorphic markers such as shorelines and coral atolls can record land-level changes over years to decades in rapidly deforming areas

Over the medium term (1000–100 000 years) geomorphic surfaces such as river terraces and marine terraces can record uplift. For example, river terraces that formed by aggradation during the last glacial have been uplifted above modern river levels along the parts of the Hikurangi margin, New Zealand. The terraces record differential uplift related to tectonic

© Woodhead Publishing Limited, 2010

Assessing the long-term stability of geological environments

205

processes along the subduction margin (Litchfield and Berryman, 2006). Marine terraces record former shorelines; if the sea level history is known and the terrace can be dated, then the comparison between the terrace elevation and modern sea level provides a measure of tectonic uplift (e.g. Berryman, 1993; Hsieh et al., 2004; Ota and Yamaguchi, 2004). Marine terraces are typically used to record uplift in the Holocene period or since the late Pleistocene (last 125 000 years). Buried shorelines or marginal marine sediments can also record tectonic subsidence (e.g. Cochran et al., 2006). Techniques to measure tectonic upift and subsidence over the long term (100 000’s to millions of years) usually apply to large areas and reflect plate tectonic processes such as continental collision or rifting. Marine sedimentary rocks high in the landscape can record long-term uplift, as can mudstone-porosity measurements and low-temperature thermochronology. Peneplains, large areas of formerly flat relief that are uplifted and incised, also record long-term tectonic uplift. Similarly, marine planation surfaces, formed by erosion when the sea level was at a low stand, can record tectonic subsidence when identified in marine seismic lines. Tectonic processes causing uplift and subsidence can vary from local faulting on a small scale to plate boundary collisional and extensional processes at the large scale. Marine and fluvial terraces can be particularly important features in nuclear waste repository investigations as they can represent the surface expression of uplift on blind faults. Blind faults are those that do not rupture the ground surface; hence no surface scarps are produced. Where fluvial or marine terraces, which are assumed to have initially formed flat (or slightly downstream-sloping for fluvial terraces), are warped, it may indicate tectonic deformation at depth. For example, blind thrust faults have been mapped using the deformation patterns recorded by mid-late Pleistocene fluvial and marine terraces on the eastern edge of the Apennine mountains in central Italy; these blind faults are thought to be the source of several historical M 5.2 earthquakes (Vannoli et al., 2004). Marine terraces can often also record uplift on unmapped offshore faults; e.g. at the Pakarae River mouth area, New Zealand, a suite of seven Holocene terraces records uplift that is associated with an offshore thrust fault (Ota et al., 1991; Wilson et al., 2006). On a larger scale, the distribution of uplift and subsidence at collisional plate boundaries can indicate areas of greater or lesser tectonic strain. For example, along the northern coast of California, in the vicinity of the Mendecino Triple Junction, there are many marine terraces. The uplift rates are generally very high (up to 4 mm/year) and variable due to complex interactions between the San Andreas fault, the southern end of the Cascadia subduction zone and the Mendecino fracture zone (Merritts, 1996). Large-scale subsidence can occur due to crustal thinning or by

© Woodhead Publishing Limited, 2010

206

Geological repository systems for safe disposal

tectonic loading. Crustal thinning occurs at divergent plate boundaries where continental plates are rifted apart, or within transtensional basins along transform plate boundaries. Tectonic loading occurs adjacent to large mountain ranges or volcanoes where the weight of the massif depresses the crust. Subsidence can also occur at subduction margins characterised by tectonic erosion; this is when the subducting plate scrapes material off the underside of the upper plate thus thinning the upper plate. However, this process generally occurs within 100 km of the subduction trench.

7.5.6 Geophysical techniques for detecting crustal structure and volcanic intrusions There are several geophysical methods that can be applied when seeking to understand the structure and stability of the earth’s crust. Techniques such as gravity surveys and seismic tomography can be used for investigating crustal structure, while aeromagnetics, magnetotellurics and heat flow can be used for detecting volcanic intrusions, such as magma chambers, as well as crustal structure. These techniques and their application to assessing geological stability are outlined below. Aeromagnetic measurements characterise the earth’s magnetic characteristics as measured from the air. The data are collected using a magnetometer aboard an aircraft; the magnetometer records variations in the intensity of the ambient magnetic field. After corrections have been made for atmospheric and regional affects a map is compiled that generally shows the distribution and relative abundance of magnetic minerals in the upper parts of the earth’s crust. As different rocks will contain different amounts of magnetic minerals this technique can be used relatively quickly, but coarsely, to map the major structures of the upper crust. Features such as major fault lines, folds and basins can be identified (but no estimate can be made of the age of the structures). Of relevance to geological repository investigations is that aeromagnetics is a useful tool for identifying magnetic anomalies that may be associated with subsurface volcanic rocks or magma bodies at shallow depth. In currently active volcanic regions the surface geology is often dominated by recent volcanic sediments such as pyroclastic flows; similarly, ancient volcanic features can be covered by non-volcanic Quaternary sediments. For example, a high-resolution aeromagnetic survey of the Amargosa Valley identified 20 anomalies that may represent basaltic volcanoes buried beneath 100–400 m of alluvium (O’Leary et al., 2002). This site is located only 25 km from Yucca Mountain and it has been estimated that the 20 newly identified volcanic centres would raise the probability of Yucca Mountain site disruption by 40% (Perry et al., 2004; Smith and Keenan, 2005).

© Woodhead Publishing Limited, 2010

Assessing the long-term stability of geological environments

207

Just as different rock units vary in magnetism, they also vary in density. Gravity surveys are a method of detecting lateral changes in gravity related to subsurface rock density. The surveys can be carried out at a variety of scales, from airborne surveys of large areas to higher resolution ground surveys over several kilometres. Raw gravity measurements are corrected for elevation, latitude and topography and the regional gravity field is subtracted, which yields a map of gravity anomalies. Positive gravity anomalies are often associated with the presence of anomalous high-density bodies in the shallow crust compared with a ‘standard’ earth model, whereas gravity lows are associated with low-density bodies, which may be located anywhere in the crust or the upper mantle. If there is some knowledge of the density of rocks in the area, crustal structure models can be produced. The models try to fit the gravity data as closely as possible but there is often more than one solution; therefore gravity surveys are best used in conjunction with other techniques such as seismic tomography (see below) and aeromagnetics. Gravity techniques have often been used successfully in volcano-tectonic regions where basement faults bounding the volcanoes are typically buried beneath young volcanic deposits. For example, basement faults beneath the Tongariro Volcanic Centre, New Zealand, are covered by up to 800 m of volcanic deposits but vertical displacements of 250–300 m on basement faults have been modelled using gravity data; this allows minimum rates of volcanic rift subsidence to be resolved (Cassidy et al., 2009). Gravity models of the Summa–Vesuvius volcano, Italy, allow identification of lineations related to normal faults in the carbonate basement up to 2 km beneath the surface (Cella et al., 2007). Beneath the volcanic island of Pico, Azores, gravity anomalies corresponding to lowdensity bodies are thought to represent shallow, small intrusions of magma at the intersections of tectonic lineaments, highlighting how gravity can help understand both crustal structure and volcanic hazards (Nunes et al., 2006). Seismic tomography is another geophysical method that is used for understanding deep crustal structure. Tomography uses the velocity of seismic waves to build three-dimensional models of the earth’s crust. Areas through which waves move quickly tend to be cooler or consist of denser rock. Conversely, areas where waves move slowly indicate warmer or less dense rock. Seismic tomography can be carried out at a variety of scales – from using global networks of seismometers to detect regions of upwelling and downwelling deep in the earth’s mantle, to using regional seismic networks to study tomography at a plate boundary or around an individual volcano. It can be particularly useful for detecting areas of partial melt and fluids in the crust. For example, at the scale of plate boundary, seismic tomography of the central North Island, New Zealand, shows the depth of the subducted Pacific Plate at up to 65 km depth and significant volumes of partial melt beneath the backarc volcanic zone (Reyners et al., 2006). At a

© Woodhead Publishing Limited, 2010

208

Geological repository systems for safe disposal

smaller scale, seismic tomographic studies of Mount St Helens, USA, have shown velocity anomalies associated with a linear zone of high seismicity, probably a fault, and with a magma chamber at 2–3.5 km depth (Waite and Moran, 2009). Another geophysical method for understanding crustal structure is magnetotelluric measurements. Magnetotelluric data maps the spatial variation of the earth’s electrical resistivity by measuring naturally occurring electric and magnetic fields at the earth’s surface. In general, rocks containing fluids, such as water, and rocks at high temperature will have a low resistivity. Conversely, dry and cold rocks will have high resistivities. Magnetotelluric measurements can produce images of the earth’s electrical resistivity structure from a few hundred metres to several hundred kilometres depth depending on the frequency of the electromagnetic signals. This method is particularly sensitive to fluids in the crust because fluids dramatically lower the resistivity of the host rocks. For example, magnetotelluric surveys across the San Andreas Fault show that the fault zone is characterised by a vertical zone of low electrical resistivity; this low resistivity is due to the presence of fluids on the highly fractured fault zone to a depth of approximately 4 km (Unsworth et al., 1997). Magnetotellurics is frequently used in volcanic and geothermal regions to image magma chambers, volcanic vents and regions of hydrothermal alteration. Beneath the Aso volcano, Japan, magnetotelluric surveys could differentiate which craters were underlain by active hydrothermal systems and which were underlain by consolidated magma (Kanda et al., 2008). Heat flow measurements are a further geophysical method used primarily in monitoring volcanoes, as well as for detecting melt intrusions and hydrothermal heating at some distances from volcanic centres. This is important for nuclear waste repository tectonic stability as anomalously high rates of crustal heat flow may be an early precursor to the onset of volcanism. The heat flow within the continental crust depends on the natural radioactivity in the crust, the tectonic setting and heat flux from the mantle below (Stein, 1995). An example of the effects of high heat flow due to magma intrusion is at Yellowstone, USA, where the average heat flow is 30 to 40 times greater than that of an average continental crust. Crustal heat flow is usually measured using data from deep boreholes and wells. In volcanically active zones infrared remote sensing is used to monitor surface heat flow. Groundwater temperature and chemistry is also monitored as this is affected by shallow magma intrusions. Precursor increases in heat flow have been detected using satellite data on several active volcanoes. For example, a temperature anomaly of several hundred degrees above average was detected in the months leading up to the Lascar volcano eruption in Chile in 1986 (Rothery et al., 1988).

© Woodhead Publishing Limited, 2010

Assessing the long-term stability of geological environments

7.6

209

Modelling long-term stability for safe disposal of radioactive waste

Extrapolating current levels of knowledge and confidence estimates far into the future has been recognised by nuclear safety commissions throughout the world as a challenging problem. The recent regulatory guide from the Canada Nuclear Safety Commission (CNSC, 2006) typifies how the longterm safety issue (and by inference the contributing hazard and risk assessment) is being addressed. CNSC note that ‘the demonstration of safety will rely less on quantitative predictions and more on qualitative arguments as the timescale increases’, and that ‘long term quantitative predictions should therefore not be considered as guaranteed impacts, but rather as safety indicators’. These statements represent a pragmatic view in the absence of mature procedures. Inevitably, quantitative assessment, including uncertainty analysis, will become the expectation in future, even if a qualitative approach is acceptable for the moment. At present the extent of hazard data and quantitative modelling of tectonic processes is limited, but with increasing computing capacity, new dating techniques and determination, a longer time series of information and modelling is achievable. The challenge before the scientific and risk community is to develop robust quantitative approaches to bound uncertainties and to entertain the appropriate alternative tectonic and volcanic evolution patterns in an area of interest over extended timeframes. Instrumental records exist for perhaps one or two centuries at best, historical documentation of events exists for a few millennia at most and the geological record loses its resolution quite quickly as one looks back in time. In many environments a reasonable level of preservation of events and processes in the geological archive may exist for 10–100 thousand years, but over longer periods event preservation will be more patchy, and the ability to date events accurately and calculate rates of process requires considerable resource. Additionally, it becomes increasingly important to evaluate whether large-scale boundary conditions, such as motions of the tectonic plates, can be assured to be constant. For example, the assumption that the volcanic front in a subduction margin setting is essentially fixed in space and time because it coincides with melting at the c. 100 km depth contour on the subduction interface may not be valid over long periods of interest such as 1 million years, because subduction processes can evolve on this timeframe. Thus, extrapolation of hazard estimates, based on well-constrained datasets for perhaps the past 100 thousand years at most, to 1 million years, brings great uncertainty. Traditional statistical approaches to uncertainty treatment at long return periods mean that the extreme tails of probability density functions (PDFs) of hazard parameters govern the hazard estimates (e.g. Stepp et al., 2001;

© Woodhead Publishing Limited, 2010

210

Geological repository systems for safe disposal

Hanks et al., 2006). Until recently, few data or models have existed to constrain the upper reasonable bound on statistical relations, and thus there is a tendency for hazard estimates to be extremely conservative. Studies that specifically address upper bound constraints on hazard parameters are important in the context of long-term geological stability assessment, but thus far have been largely restricted to the earthquake ground motion field. Approaches include modelling (e.g. Andrews et al., 2007) and statistical studies (e.g. Rhoades et al., 2008; Strasser et al., 2009). Similar approaches to bounding other key parameters of the hazard model or to limit the range of evolutionary tectonic paths that are reasonable over certain time periods in a region offer hope for more certain long-term assessments of site stability.

7.7

Future trends

A range of innovative and alternate approaches to site suitability assessment over and above traditional deterministic and latterly probabilistic methodologies are being developed (Fig. 7.5). The deterministic and probabilistic approaches are well developed and indeed expected for nuclear power plants

7.5 Steps in the process of evaluating the long-term stability of the geological environment for the purposes of radioactive and spent nuclear fuel repositories.

© Woodhead Publishing Limited, 2010

Assessing the long-term stability of geological environments

211

and increasing to geological repository siting and safety assessment (e.g. Cornell, 1968; Reiter, 1990; Stepp et al., 2001; IAEA 2002; Gu¨rpinar, 2005). To a large extent the innovations extend existing procedures or bring techniques from other disciplines to the hazard area. Many approaches are designed to capture a reasonable representation of uncertainty in hazard and risk calculations and to identify and separate epistemic (uncertainty because of lack of knowledge) from aleatory (variability that inherently exists in natural earth systems) components (SSHAC, 1997; Bommer and Abrahamson, 2006). Because quantitative data are often lacking for all aspects of tectonic hazard assessment at critical facilities, judgement is required to assess many aspects of the hazard and risk assessment. An experienced group of experts may be used to parameterise the quantitative models or assess the appropriate range of alternate geologic models that may characterise the site, or, in terms of very long-term assessment, how a region may evolve over periods of 100 000– 1 000 000 years. Considerable effort has been placed on developing a structured environment in which expert judgement can be used (e.g. SSHAC, 1997; Coppersmith et al., 2009). Coppersmith et al. (2009) list the following key requirements in achieving successful expert elicitation: . . . . . . .

Experts should be trained in probability theory, uncertainty quantification and ways to avoid common cognitive biases. Comprehensive and user-friendly databases should be provided to the expert. Experts should be required to evaluate all potentially credible hypotheses. Workshops and other interactions among the experts and proponents of alternative viewpoints should be encouraged. Assessment interviews should start with conceptual models and develop towards parameter assessments. Feedback should be provided to the experts to give them insight into the significance of their assessments to the hazard results. Complete documentation should include the technical basis for all assessments.

There remains considerable debate about the relative merits of probabilistic and deterministic hazard procedures (e.g. Bommer, 2002; Klu¨gel, 2005). In our view future developments will see a narrowing of these conflicts and in fact a recognition that deterministic scenarios should often be accepted into probabilistic methodologies as alternate viable models or parameters. These are readily structured in decision trees (often called logic trees), which are increasingly being formulated logically to structure probability estimates of components of hazard and risk assessment (e.g. SSHAC, 1997; Kessler and McGuire, 1999; Coppersmith et al., 2009). Decision trees often become very

© Woodhead Publishing Limited, 2010

212

Geological repository systems for safe disposal

large for assessment of complex systems, and increasingly a Monte Carlo sampling approach, paying heed to the weighting of alternate parameters or branches at each level of the decision tree, is used to generate the hazard curves (e.g. Jaquet et al., 2000; Perry et al., 2001). Particularly within the carbon sequestration field there has been an emphasis on deterministic scenario modelling using ‘features, events and processes’ or FEPs (e.g. Maul et al., 2004). Such an approach has also been developed occasionally in the nuclear waste industry (e.g. Swift et al., 1999; NEA/OECD, 2000) and particularly in the Swedish waste repository programme (e.g. Last et al., 2004; Kozak et al., 2006). These individual FEPs may be considered viable alternate geologic models, processes or event chains, which can readily be absorbed into probabilistic hazard methodology. Fuzzy logic theory has also been applied to probabilistic hazard estimation as a means of capturing uncertainty in data inputs (e.g. Elshayeb, 2004; Goldsworthy et al., 2004) Another modelling technique that may be of substantial value in forecasting long-term evolution of a geological repository is the use of the finite element approach. This technique is now used extensively to model the evolution of geological structure, groundwater flow, performance of radioactive waste containment and nuclear facility engineering (e.g. Rautman et al., 1995; Chen et al., 2004; Vespa et al., 2005). As computational capacity increases it appears feasible that sufficiently large or detailed meshes may be able to be defined for scales ranging from repository site area geology to a regional context such that process-driven evolutionary models of long-term site stability could be developed. Such models would be a valuable approach to compare alongside statistical methods embedded in deterministic probabilistic calculations.

7.8

Summary

Assessment of the stability of the geologic environment for the purposes of nuclear waste repositories requires a long-term (100 000–1 000 000 years) understanding of tectonic and climatic processes. This is significantly longer than comparable stability assessments for other critical facilities, such as nuclear power plants and large dams, which typically have a lifetime of <100 years. When considering the tectonic stability of a potential geological repository site, scientists and engineers need to understand the present stability (e.g. frequency and size of earthquakes, proximity and effects of volcanic activity and impacts of anthropogenic climate change), cumulative effects over long timescales (e.g. gradual effects of slow uplift and erosion), and how the tectonic and climatic processes may change over time (e.g. formation of new fault zones and volcanoes and impact of glacial periods). Modelling is used to extrapolate current levels of knowledge and confidence

© Woodhead Publishing Limited, 2010

Assessing the long-term stability of geological environments

213

estimates far into the future. Deterministic and probabilistic modelling approaches are well developed for tectonic hazard assessment but more work is required to refine upper bounds of hazard parameters as the extreme events become increasing important at long timescales. Future trends will see the development of existing procedures and the integration of techniques from other disciplines to the hazard area.

7.9

Sources of further information and advice

Chester D K, Degg M, Duncan A M and Guest J E (2000), ‘The increasing exposure of cities to the effects of volcanic eruptions: a global survey’, Global Environmental Change, Part B: Environmental Hazards, 2(3), 89–103. Houghton B, Rymer H, Stix J, McNutt S and Sigurdsson H (1999), Encyclopaedia of Volcanoes, Elsevier, Amsterdam, 1417 pp. Lantuejoul C (2002), Geostatistical Simulation: Models and Algorithms, SpringerVerlag, 256 pp. Yeats R S, Sieh K E and Allen C R (1996), Geology of Earthquakes. Oxford University Press, 576 pp.

7.10

Acknowledgements

We would like to thank our colleagues, Steve Sparks, Mick Apted, Charles Connor and Laura Wallace for many useful discussions. We would also like to thank Nicola Litchfield and Matt Gerstenberger for reviewing this manuscript.

7.11

References

Andrews D J, Hanks T C and Whitney J W (2007), ‘Physical limits on ground motion at Yucca Mountain’, Bulletin of the Seismological Society of America, 97, 1771–1792. Apted M, Berryman K, Chapman N, Cloos M, Connor C, Kitayama K, Sparks S and Tsuchi H (2004), ‘Locating a radioactive waste repository in the Ring of Fire’, Eos, Transactions, American Geophysical Union, 85, 465–467. Arcos D, Bruno J and Karnland O (2003), ‘Geochemical model of the granite– bentinite–groundwater interaction at Aspo HRL (LOT experiment)’, Applied Clay Science, 23, 219–228. Azzaro R, Mattia M and Puglisi G (2001), ‘Fault creep and kinematics of the eastern segment of the Pernicana Fault (Mt Etna, Italy) derived from geodetic observations and their tectonic significance’, Tectonophysics, 333, 401–415. Berger A and Loutre M F (2002), ‘An exceptionally long interglacial ahead?’, Science, 297, 1287–1288. Berryman K R (1993), ‘Age, height, and deformation of Holocene terraces at Mahia Peninsula, Hikurangi subduction margin, New Zealand’, Tectonics, 12, 1347– 1364. Bommer J J (2002), ‘Deterministic vs. probabilistic seismic hazard assessment: an

© Woodhead Publishing Limited, 2010

214

Geological repository systems for safe disposal

exaggerated and obstructive dichotomy’, Journal of Earthquake Engineering, 6, 43–73. Bommer J J and Abrahamson N A (2006), ‘Why do modern probabilistic seismichazard analyses often lead to increased hazard estimates? Bulletin of the Seismological Society of America, 96, 1967–1977, doi: 10.1785/0120060043. Canada Nuclear Safety Commission (CNSC) (2006), ‘Assessing the long term safety of radioactive waste management’, Regulatory Guide G–320. Cassidy J, Ingham M, Locke C and Bibby H (2009), ‘Subsurface structure across the axis of the Tongariro Volcanic Centre, New Zealand’, Journal of Volcanology and Geothermal Research, 179, 233–240. Cella F, Fedi M, Florio G, Grimaldi M and Rapolla A (2007), ‘Shallow structure of the Somma-Vesuvius volcano from 3D inversion of gravity data’, Journal of Volcanology and Geothermal Research, 161, 303–317. Chapman N A (2006), ‘Geological disposal of radioactive wastes – concept, status and trends’, Journal of Iberian Geology, 32, 7–14. Chen H, Chen L, Ma H, Li Y, Zhang J, He Q and Wang J (2004), ‘3-D finite element modeling for evolution of stress field and interaction among strong earthquakes in Sichuan-Yuannan region’, Acta Seismologica Sinica, 17, 625– 743. Cochran U, Berryman K, Mildenhall D, Hayward B, Southall K, Hollis C, Barker P, Wallace L, Alloway B and Wilson K (2006), ‘Paleoecological insights into subduction zone earthquake occurrence, eastern North Island, New Zealand’, Geological Society of America Bulletin, 118, 1051–1074. Coppersmith K J, Jenni K E, Perman R C and Youngs R R (2009), ‘Formal expert assessment in probabilistic seismic and volcanic hazard analysis’, in Volcanic and Tectonic Hazard Assessment for Nuclear Facilities, edited by Charles B Connor, Neil A Chapman and Laura J Connor, Cambridge University Press. Cornell C A (1968), ‘Engineering seismic risk analysis’, Bulletin of the Seismological Society of America, 58, 15831606. Cox T (1994), ‘Analysis of drainage-basin symmetry as a rapid technique to identify areas of possible Quaternary tilt-block tectonics: an example from the Mississippi embayment’, Geological Society of America Bulletin, 106, 571–581. Donnellan A, Hager B H, King R W and Herring T A (1993), ‘Geodetic measurement of deformation in the Ventura basin region, southern California’, Journal of Geophysical Research, 98, 21,727–21,739. Douglas A, Beavan J, Wallace L and Townend J (2005), ‘Slow slip on the northern Hikurangi subduction interface, New Zealand’, Geophysical Research Letters, 32, L16305, doi:10.1029/2005GL023607. Dragert H, Wang K and James T (2001), ‘A silent slip event on the deeper Cascadia subduction interface’, Science, 292, 1525–1528. Elshayeb Y (2004), ‘Overcoming uncertainties in risk analysis: trade-offs among methods of uncertainty analysis’, in Comparative Risk Assessment and Environmental Decision Making, edited by Igor Linkov and Abou Bakr Ramadan, Springer, pp. 287–295, ISBN 1402018959, 9781402018954. EPA (2008), ‘Public health and environmental radiation protection standards for Yucca Mountain, Nevada’, Federal Register, 73(200), 61256–61289. Field E H, Dawson T E, Felzer K R, Frankel A D, Gupta V, Jordan T H, Parsons T, Petersen M D, Stein R S, Weldon R J and Wills C J (2008), ‘The Uniform

© Woodhead Publishing Limited, 2010

Assessing the long-term stability of geological environments

215

California Earthquake Rupture Forecast, Version 2 (UCERF 2)’, USGS Open File Report 2007-1437. Goldsworthy M, Danko´ G and Fodor J (2004), ‘Probabilistic and fuzzy approach to safety assessment for the Ba´taapa´ti (U¨veghuta) Site’, in Annual Report of the Geological Institute of Hungary, 2003. Gregory J M, Huybrechts P and Raper S C (2004), ‘Threatened loss of the Greenland ice-sheet’, Nature, 428, 616. Guccione M J (2005), ‘Late Pleistocene and Holocene paleoseismology of an intraplate seismic zone in a large alluvial valley, the New Madrid seismic zone, central USA’, Tectonophysics, 408, 237–264. Gu¨rpinar A (2005), ‘The importance of paleoseismology in seismic hazard studies for critical facilities’, Tectonophysics, 408, 23–28. Gutenberg B and Richter C (1944), ‘Frequency of earthquakes in California’, Bulletin of the Seismological Society of America, 34, 185188. Hampel A, Adam J and Kukowski N (2004), ‘Response of the tectonically erosive south Peruvian forearc to subduction of the Nazca Ridge: analysis of threedimensional analogue experiments’, Tectonics, 23, TC5003, doi: 10.1029/ 2003TC001585. Hanks T C and Bakun W H (2002), ‘A bilinear source-scaling model for M-logA observations of continental earthquakes’, Bulletin of the Seismological Society of America, 92, 1841–1846. Hanks T C, Abrahamson N, Board M, Boore D, Brune J and Cornell A (2006), ‘Workshop on extreme ground motion at Yucca Mountain’, US Geological Survey Open-File Report OF2006-1277. Hays J D, Imbrie J and Shackleton N J (1976), ‘Variations in the Earth’s orbit: pacemaker of the ice ages’, Science, 194, 1121–1132. Henley R W and Ellis A J (1983), ‘Geothermal systems ancient and modern: a geochemical review’, Earth-Science Review, 19, 1–50. Hill E M and Blewitt G (2006), ‘Testing for fault activity at Yucca Mountain, Nevada, using independent GPS results from the BARGEN network’, Geophysical Research Letters, 33, doi:10.1029/2006GL026140. Hillis R R, Sandiford M, Reynolds S D and Quigley M C (2008), ‘Present-day stresses, seismicity, and Neogene-to-Recent tectonics of Australia’s ‘passive’ margins: intraplate deformation controlled by plate boundary forces’, The Nature and Origin of Compression in Passive Margins, Geological Society, London, Special Publication, 306, 71–90. Hsieh M-L, Liew P-M and Hsu M-Y (2004), ‘Holocene tectonic uplift on the Huatung coast, eastern Taiwan’, Quaternary International, 115–116, 47–70. Hsu J T (1992), ‘Quaternary uplift of the Peruvian coast related to the subduction of the Nazca Ridge: 13.5 to 15.6 degrees latitude’, Quaternary International, 15/ 16, 87–97. IAEA (2002), ‘Safety guide on evaluation of seismic hazards for nuclear power plants’, No. NS-G-3.3. IPCC (2007), ‘Climate Change 2007: Synthesis Report, Contribution of Working Groups I, II and III to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change’, Core Writing Team, R K Pachauri and A Reisinger (eds), IPCC, Geneva, Switzerland, 104 pp. Jaquet O, Low S, Martinelli B, Dietrich V and Gilby D (2000), ‘Estimation of

© Woodhead Publishing Limited, 2010

216

Geological repository systems for safe disposal

volcanic hazards based on Cox stochastic processes’, Physics and Chemistry of the Earth, Part A: Solid Earth and Geodesy, 25, 571–579. Kanda W, Tanaka Y, Utsugi M, Takakura S, Hashimoto T and Inoue H (2008), ‘A preparation zone for volcanic explosions beneath Naka-dake crater, Aso volcano, as inferred from magnetotelluric surveys’, Journal of Volcanology and Geothermal Research, 178, 32–45. Kelson K I, Streig A R, Koehler R D and Kang K-H (2006), ‘Timing of Late Holocene Paleoearthquakes on the Northern San Andreas Fault at the Fort Ross Orchard Site, Sonoma County, California’, Bulletin of the Seismological Society of America, 96, 1012–1028, doi: 10.1785/0120050123. Kessler J H and McGuire R K (1999), ‘Total system performance assessment for waste disposal using a logic tree approach’, Risk Analysis, 19, 915–931. Kerr R A (2006), ‘A worrying trend of less ice, higher seas’, Science, 311, 1698–1701. Klu¨gel J-U (2005), ‘Problems in the application of the SSHAC probability method for assessing earthquake hazards at Swiss nuclear power plants’, Engineering Geology, 78 (3/4), 285–307. Knutson T R and Tuleya R E (2004), ‘Impact if CO2-induced warming on simulated hurricane intensity and precipitation: sensitivity to the choice of climate model and convective paramterization’, Journal of Climate, 17, 3477–3495. Kozak M W, Beyleveld C and Carolissen A (2006), ‘Safety assessment of low- and intermediate-level waste disposal at Vaalputs, South Africa’, American Geophysical Union, Fall Meeting 2006, Abstract H24D-03. Last G V, Rohay V J, Schelling F J, Bunn A L, Delamare M A, Dirkes R L, Hildebr R D, Morse J G, Napier B A, Riley R G, Soler L and Thorne P D (2004), ‘A comprehensive and systematic approach to developing and documenting conceptual models of contaminant release and migration at the Hanford Site’, Stochastic Environmental Research and Risk Assessment, 18, 109–116. Lemeille F, Cushing M, Carbon D, Grellet B, Bitterli T, Flehoc C and Innocent C (1999), ‘Co-seismic ruptures and deformations recorded by speleothems in the epicentral zone of the Basel earthquake’, Geodinamica Acta, 12, 179–191. Levi T, Weinberger R, Aı¨ fa T, Eyal Y and Marco S (2006), ‘Earthquake-induced clastic dikes detected by anisotropy of magnetic susceptibility’, Geology, 34, 69–72. Lienkaemper J L, Dawson T E, Personius S F, Seitz G G, Reidy L M and Schwartz D P (2002), ‘A record of large earthquakes on the Southern Hayward Fault for the past 500 years’, Bulletin of the Seismological Society of America, 92, 2637– 2658. Litchfield N J and Berryman K R (2006), ‘Relations between postglacial fluvial incision rates and uplift rates in the North Island, New Zealand’, Journal of Geophysical Research, 111, doi:10.1029/2005JF000374. Litchfield N J, Ota Y and Merrits D (2009), ‘Tectonic uplift and subsidence’, Chapter 7, in Volcanic and Tectonic Hazard Assessment for Nuclear Facilities, edited by Charles B Connor, Neil A Chapman and Laura J Connor, Cambridge University Press. Lyons S N, Bock Y and Sandwell D T (2002), ‘Creep along the Imperial Fault, southern California, from GPS measurements’, Journal of Geophysical Research, 107, B10, 2249, doi:10.1029/2001JB000763. Ma S, Eaton D W and Adams J (2008), ‘Intraplate seismicity of a recently

© Woodhead Publishing Limited, 2010

Assessing the long-term stability of geological environments

217

deglaciated shield terrane: a case study from Northern Ontario, Canada’, Bulletin of the Seismological Society of America, 98, 2828–2848. McKinley I G and Alexander W R (2009), ‘Special issues for coastal sites’, Chapter 28, in Volcanic and Tectonic Hazard Assessment for Nuclear Facilities, edited by Charles B Connor, Neil A Chapman and Laura J Connor, Cambridge University Press. Mahoney S H, Sparks R S J, Connor L J and Connor C B (2009), ‘Exploring longterm hazards using a Quaternary volcano database’, Chapter 13, in Volcanic and Tectonic Hazard Assessment for Nuclear Facilities, edited by Charles B Connor, Neil A Chapman and Laura J Connor, Cambridge University Press. Marco S and Agnon A (2005), ‘High-resolution stratigraphy reveals repeated earthquake faulting in the Masada Fault Zone, Dead Sea Transform’, Tectonophysics, 408, 101–112. Maul P, Savage D, Benbow S, Walke R and Bruin R (2004), ‘Development of a FEP Database for the geological storage of carbon dioxide’, in Proceedings of the 7th International Conference on Greenhouse Gas Control Technologies (GHGT7), Vancouver, Canada, 5–9 September 2004. Meghraoui M, Delouis B, Ferry M, Giardini D, Huggenberger P, Spottke I and Granet M (2001), ‘Active normal faulting in the Upper Rhine graben and paleoseismic identification of the 1356 Basel earthquake’, Science, 293, 2070– 2073. Merritts D J (1996), ‘The Mendecino triple junction: active faults, episodic coastal emergence, and rapid uplift’, Journal of Geophysical Research, 101, 6051–6070. Montogomery D R and Manga M (2003), ‘Streamflow and water well responses to earthquakes’, Nature, 300, 2047–2049. Muir-Wood R (2000), ‘Deglaciation tectonics: a principal influence on intraplate seismogenesis at high latitudes’, Quaternary Science Reviews, 19, 1399–1411. Nash B P, Perkins M E, Christensen J N, Lee D-C and Halliday A N (2006), ‘The Yellowstone hotspot in space and time: Nd and Hf isotopes in silicic magmas’, Earth and Planetary Science Letters, 247, 143–156. NEA/OECD (2000), ‘Features, events and processes (FEPs) for geologic disposal of radioactive waste’, Nuclear Energy Agency/Organisation for Economic Development, Paris, France. Nirex (2001), ‘The scientific foundations of deep geological disposal’, Nirex Report N/016, Available from http://www.nda.gov.uk/documents/loader.cfm?url=/ commonspot/security/getfile.cfm&pageid=12461 (Accessed 20 April 2009). Nirex (2005a), ‘Glaciation and deep disposal of radioactive waste’, Nirex Report N/ 103, Available from http://www.nda.gov.uk/documents/upload/Glaciationand-deep-disposal-of-radioactive-waste-An-Overview-Nirex-Report-N-1032005.pdf (Accessed 20 April 2009). Nirex (2005b), ‘Summary note for CoWRM on the impact of rising sea level on coastal sites with radioactive waste stores’, Nirex Technical Note 484385, Available from http://www.nda.gov.uk/documents/upload/Summary-note-forCoRWM-on-impact-of-rising-sea-levels-on-coastal-sites-with-radioactivewaste-stores-A-Technical-Note-2005.pdf (Accessed 20 April 2009). Nunes J, Camacho A, Frabca Z, Montesinos F G, Alves M, Viera R, Velex E and Ortiz E (2006), ‘Gravity anomalies and crustal signature of volcano-tectonic

© Woodhead Publishing Limited, 2010

218

Geological repository systems for safe disposal

structures of Pico Island (Azores)’, Journal of Volcanology and Geothermal Research, 156, 55–70. Obara K, Hirose H, Yamamizu F and Kasahara K (2004), ‘Episodic slow slip events accompanied by non-volcanic tremors in southwest Japan subduction zone’, Geophysical Research Letters, 31, L23602, doi:10.1029/2004GL020848. O’Leary D W, Mankinen E A, Blakely R J, Langenheim V E and Ponce D A (2002), ‘Aeromagnetic expression of buried basaltic volcanoes near Yucca Mountain, Nevada’, US Geological Survey Open File Report, 02–020. Olesen O, Bungum H, Dehls J, Lindholm C, Pascal C and Roberts D (2008), ‘Netotectonics in Norway; mechanisms and implications’, 33rd International Geological Congress, Oslo, Norway, Abstract 1398408. Ota Y and Yamaguchi M (2004), ‘Holocene coastal uplift in the western Pacific Rim in the context of Late Quaternary uplift’, Quaternary International, 120, 105– 117. Ota Y, Hull A G and Berryman K R (1991), ‘Coseismic uplift of Holocene marine terraces in the Pakarae River area, eastern North Island, New Zealand’, Quaternary Research, 35, 331–346. Overpeck J T, Otto-Bliesner B L, Miller G H, Muhs D R, Alley R B and Kiehl J T (2006), ‘Paleoclimate evidence for future ice-sheet instability and rapid sealevel rise’, Science, 311, 1747–1750. Pe´rez O J and Scholz C H (1984), ‘Heterogeneities of the instrumental seismicity catalog (1904–1980) for strong shallow earthquakes’, Bulletin of the Seismological Society of America, 74, 669–686. Perry F V, Valentine G A, Desmarais E K and WoldeGabriel G (2001), ‘Probabilistic assessment of volcanic hazard to radioactive waste repositories in Japan; intersection by a dike from a nearby composite volcano’, Geology, 29(3), 255– 258. Perry F V, Cogbill A H and Oliver R D (2004), ‘Reducing uncertainty in the hazard of volcanic disruption of the proposed Yucca Mountain radioactive waste repository using a high-resolution aeromagnetic survey to detect buried basaltic volcanoes in alluvial-filled basins’, Geological Society of America Abstracts and Programs, 36(5), 33. Petley D N and Allison R J (1997), ‘The mechanisms of deep-seated landslides’, Earth Surface Processes and Landforms, 22, 747–758. Pik R, Marty B, Carignan J, Yirgu G and Ayalew T (2008), ‘Timing of East African Rift development in southern Ethiopia: implication for mantle plume activity and evolution of topography’, Geology, 36, 167–170. Pons-Branchu E, Hamelin B, Brulhet J and Bruxelles L (2004) Speleothem rupture in karst; tectonic or climatic origin? U–Th dating of rupture events in Salamandre Cave (Gard, southeastern France)’, Bulletin de la Socie´te´ Ge´ologique de France, 175, 473–479. Poulsen C J, Gendaszek A S and Lacob R L (2003), ‘Did the rifting of the Atlantic Ocean cause the Cretaceous thermal maximum?’, Geology, 31, 115–118. Quigley M C, Cupper M L and Sandiford M (2006), ‘Quaternary faults of southcentral Australia: palaeoseismicity, slip rates and origin’, Australian Journal of Earth Sciences, 53, 285–301. Rautman C A, Longenbaugh R S and Ryder E E (1995), ‘Use of geostatistical modeling to capture complex geology in finite-element analysis’, in Proceedings

© Woodhead Publishing Limited, 2010

Assessing the long-term stability of geological environments

219

of the Sixth Annual International Conference on High Level Radioactive Waste Management, Las Vegas, Nevada, 30 April–5 May 1995, pp. 352–354, ASCE 07844-0082-2, 810 pp. Reiter L (1990), Earthquake Hazard Analysis: Issues and Insight, Columbia University Press, New York. Reyners M, Eberhart-Phillips D, Stuart G and Nishimura Y (2006), ‘Imaging subduction from the trench to 300 km depth beneath the central North Island, New Zealand, with Vp and Vp/Vs’, Geophysical Journal International, 165, 565–583. Rhoades D A, Zhao J X and McVerry G H (2008), ‘A simple test for inhibition of very strong shaking in ground-motion models’, Bulletin of the Seismological Society of America, 98, 448–453. Rothery D A, Francis P W and Wood C A (1988), ‘Volcano monitoring using short wavelength infrared data from satellites’, Journal of Geophysical Research, 93 (B7), 7993–8008. Satake K, Shimazaki K, Tsuji Y and Ueda K (1996) ‘Time and size of a giant earthquake in Cascadia inferred from Japanese tsunami records of January 1700’, Nature, 379, 246–249. Singh S and Jain A (2007), ‘Liquefaction and fluidization of lacustrine deposits from Lahaul-Spiti and Ladakh Himalaya: geological evidences of paleoseismicity along active fault zone’, Sedimentary Geology, 196, 47–57. Smith E I and Keenan D L (2005), ‘Yucca Mountain could face greater volcanic threat’, Eos, Transactions, American Geophysical Union, 86, 317, 321. SSHAC, Senior Seismic Hazard Analysis Committee (1997), ‘Recommendations for Probabilistic Seismic Hazard Analysis: Guidance on Uncertainty and Use of Experts’, US Nuclear Regulatory Commission, US Department of Energy, Electric Power Research Institute, NUREG/CR-6372, UCRL-ID-122160, Volumes 1 and 2. Starinsky A and Katz A (2003), ‘The formation of natural cryogenic brines’, Geochimica et Cosmochimica Acta, 67, 1475–1484. Stein C A (1995), ‘Heat flow of the Earth’, in Global Earth Physics, A Handbook of Physical Constants, AGU Reference Shelf 1, edited by T Ahrens, pp. 144–158. Stepp J C, Wong I, Whitney J, Quittmeyer R, Abrahamson N, Toro G, Youngs R, Coppersmith K, Savy J, Sullivan T and Yucca Mountain PSHA Project Members (2001), ‘Probabilistic seismic hazard analyses for ground motions and fault displacement at Yucca Mountain, Nevada’, Earthquake Spectra, 17, 113–151. Stevens C W, McCaffrey R, Bock Y, Genrich J F, Pubellier M and Subarya C (2002), ‘Evidence for block rotations and basal shear in the world’s fastest slipping continental shear zone in NW New Guinea’, in Plate Boundary Zones, edited by S Stein and J Freymueller, AGU Geodynamics Series 30, pp. 87–99. Stewart I S, Sauber J and Rose J (2000), ‘Glacio-seismotectonics: ice sheets, crustal deformation, and seismicity’, Quaternary Science Reviews, 19, 1367–1389. Stirling M W, McVerry G and Berryman K R (2001), ‘A new seismic hazard model for New Zealand’, Bulletin of the Seismological Society of America, 92(5), 1878–1903. Stirling M, Berryman K, Wallace L, Litchfield N, Beavan J and Smith W (2009), ‘Multidisciplinary probabilistic tectonic hazard analysis’, Chapter 19, in

© Woodhead Publishing Limited, 2010

220

Geological repository systems for safe disposal

Volcanic and Tectonic Hazard Assessment for Nuclear Facilities, edited by Charles B Connor, Neil A Chapman, Laura J Connor, Cambridge University Press. Strasser F O, Abrahamson N A and Bommer J J (2009), ‘Sigma: issues, insights, and challenges’, Seismological Research Letters, 80, 40–56. Swift P, Barr G, Barnard R, Rechard R, Schenker A, Freeze G and Burck P (1999), ‘Feature, event, and process screening and scenario development for the Yucca Mountain total system performance assessment’, in Proceedings of the NEA Workshop on Scenario Development, Madrid, Spain, 10–12 May 1999, Nuclear Energy Agency of the Organisation for Economic Co-operation and Development. Talbot C J (1999), ‘Ice ages and nuclear waste isolation’, Engineering Geology, 52, 177–192. Tanaka Y, Hasegawa T and Nakamura K (2007), ‘Modeling swelling characteristics and permeability of several compacted bentonite affected by saline water’, in Clays in Natural and Engineered Barriers for Radioactive Waste Confinement International Meeting, Lille, France, pp. 553–554. Unsworth M J, Malin P E, Egbert G D and Booker J R (1997), ‘Internal structure of the San Andreas fault at Parkfield, California’, Geology, 25, 359–362. Van Den Eeckhaut M, Moeyersons J, Nyssen J, Abraha A, Poesen J, Haile M and Deckers J (2009), ‘Spatial patterns of old, deep-seated landslides: a case study in the northern Ethiopian highlands’, Geomorphology, 105, 239–252. Vannoli P, Basili R and Vanlensise G (2004), ‘New geomorphic evidence for anticlinal growth driven by blind-thrust faulting along the northern Marche coastal belt (central Italy)’, Journal of Seismology, 8, 297–312. Vespa A, Haapaniemi H, Luukkanen P, Nurkkala P and Saarenheimo A (2005), Application of finite element model updating to a feed water pipeline of a nuclear power plant. Nuclear Engineering and Design, 235, 1849–1865. Villamor P, Van Dissen R J, Alloway B V, Palmer A S and Litchfield N J (2007), ‘The Rangipo Fault, Taupo Rift, New Zealand: an example of temporal sliprate and single-event displacement variability in a volcanic environment’, Geological Society of America Bulletin, 119, 529–547; doi: 10.1130/B26000.1. Waite G P and Moran S C (2009), ‘VP structure of Mount St. Helens, Washington, USA, imaged with local earthquake tomography’, Journal of Volcanology and Geothermal Research, doi: 10.1016/j.volgeores.2009.02.009. Wallace L M and Beavan R J (2006), ‘A large slow slip event on the central Hikurangi subduction interface beneath the Manawatu region, North Island, New Zealand’, Geophysical Research Letters, 33, doi:10.1029/2006GL026009. Wallace L M, Ellis S M, Miyao K, Miura S, Beavan R J and Goto J (2009), ‘Enigmatic, highly active left-lateral shear zone in southwest Japan explained by aseismic ridge collision’, Geology, 37(2), 143–146, doi: 10.1130/G25221A.1. Webster P J, Holland G J, Curry J A and Chang H-R (2005), ‘Changes in tropical cyclone number, duration and intensity in a warming environment’, Science, 309, 1844–1846. Wells D L and Coppersmith K J (1994), ‘New empirical relationships among magnitude, rupture length, rupture width, rupture area, and surface displacement’, Bulletin of the Seismological Society of America, 84, 974–1002. Wernicke B P, Davis J L, Bennett R A, Elosegui P, Abolins M J, Brady R J, House

© Woodhead Publishing Limited, 2010

Assessing the long-term stability of geological environments

221

M A, Niemi N A and Snow J K (1998), ‘Anomalous strain accumulation in the Yucca Mountain area, Nevada’, Science, 279, 2096–2100. Wetmore P H, Hughes S S, Connor L J and Caplinger M L (2009), ‘Spatial distribution of eruptive centers about the Idaho National Laboratory’, Chapter 16, in Volcanic and Tectonic Hazard Assessment for Nuclear Facilities, edited by Charles B Connor, Neil A Chapman and Laura J Connor, Cambridge University Press. Wilson C, Houghton B F, McWilliams M O, Lanphere M A, Weaver S D and Briggs R (1995), ‘Volcanic and structural evolution of Taupo Volcanic Zone, New Zealand: a review’, Journal of Volcanology and Geothermal Research, 68, 1–28. Wilson K J, Berryman K R, Litchfield N J and Little T A (2006), ‘A revision of Mid to Late Holocene marine terrace distribution and chronology at New Zealand’s most tectonically active coastal location, Pakarae River, North Island, New Zealand’, New Zealand Journal of Geology and Geophysics, 49, 477–489. Yamagishi H, Ito Y and Kawamura M (2002), ‘Characteristics of deep-seated landslides of Hokkaido: analyses of a database of landslides of Hokkaido, Japan’, Environmental and Engineering Geoscience, 8, 35–46. Zachos J C, Dickens G R and Zeebe R E (2008), ‘An early Cenozoic perspective on greenhouse warming and carbon-cycle dynamics’, Nature, 451, 279–283.

© Woodhead Publishing Limited, 2010

8 Far-field process analysis and radionuclide transport modelling in geological repository systems M . M A Z U R E K , University of Bern, Switzerland

Abstract: The differing roles of the geosphere in safety cases considering crystalline-basement and sedimentary-rock (clays, shales, marls) host formations for deep disposal of radioactive waste are discussed. These roles are then illustrated using the Swiss and Finnish programmes as examples. The analysis of relevant FEPs (features, events and processes) and scenarios provides the basis for the quantification of radionuclide transport. The key geosphere-related FEPs for Opalinus Clay in Switzerland and for crystalline rocks at Olkiluoto in Finland are compared and their effects on calculated radionuclide fluxes are evaluated. Finally, recent developments in host-rock characterisation are discussed, including in situ characterisation of matrix porosity and of the microstructure of the pore space and chemical characterisation of pore fluids. Key words: sedimentary and crystalline environments, safety functions, FEPs (features, events and processes) and scenarios, modelling of geosphere transport and retardation, matrix porosity and diffusion coefficient.

8.1.

Framework

The far field is the region of the geosphere where the thermal, hydraulic, chemical and mechanical effects caused by the presence of the repository are small or negligible; i.e. it relates to the system under natural conditions. The term host rock can be identical to geosphere (e.g. in crystalline-basement environments without sedimentary cover), but frequently it relates only to a specific unit within the geosphere in which the repository is located. In most but not all cases, the host rock represents the part of the geosphere that contributes most strongly to the confinement of the waste, and it may be

222 © Woodhead Publishing Limited, 2010

Far-field process analysis and radionuclide transport modelling

223

surrounded by units with less favourable and/or less well predictable properties, such as lithologically heterogeneous sequences or aquifers. High-level radioactive waste, the main focus of this chapter, contains most of the activity that results from the operation of nuclear power plants and includes spent fuel and vitrified waste originating from reprocessing. Intermediate- and low-level wastes contain much less activity but account for larger waste volumes.

8.1.1 Role of the geosphere in a repository system There is international consensus that a repository system has the following broad safety functions (e.g. NEA, 2004): . .

Isolation of the waste from the human environment. Containment of radionuclides within the repository system and barrier to radionuclide release and migration. This function can be further subdivided (e.g. Nagra, 2002b) into: ○ long-term confinement of the waste within the engineered barrier system and the immediately surrounding rock; ○ attenuation of releases to the environment by various processes occurring in the host rock of the repository.

The host rock of a deep geological repository is one of several components of the multi-barrier disposal system, including engineered and natural barriers (e.g. waste matrix, canister, bentonite buffer, host rock, other units of the geosphere). Key functions of the host rock in support of the main safety functions include (e.g. NEA, 2004): . . .

to provide a stable, reducing geochemical environment; to provide a mechanically stable environment during the operational phase (geomechanical stability of tunnels and shafts) as well as during the post-closure phase (mechanical integrity of the engineered barriers); to provide a transport barrier for radionuclides released from the engineered barriers.

These functions are common to most disposal programmes under development (the most prominent exception being the US design at Yucca Mountain, with a repository in the oxidising environment of the unsaturated zone where the scarcity or absence of pore water is the main feature limiting transport). However, the relative weights of the functions vary substantially due to differences in the respective safety concepts, waste types, engineered barriers and properties of the geological environment. For example, the geosphere is considered as a major transport barrier in the French, Belgian and Swiss disposal programmes in clay and shale (Ondraf/Niras, 2001; Nagra, 2002a, 2002b; Andra, 2005). On the other hand, in the current

© Woodhead Publishing Limited, 2010

224

Geological repository systems for safe disposal

concepts for crystalline-rock environments such as those considered in the Swedish and Finnish programmes (SKB, 2006a; Smith et al., 2007; Nykyri et al., 2008; Posiva, 2008) the main role of the geosphere is to provide a chemically and mechanically stable environment for the engineered barriers, whereas only limited weight is attributed to the geosphere as a barrier to radionuclide transport. The fact that the transport resistance in crystallinerock environments is typically lower and/or more difficult to predict is balanced by adaptations in the engineered barrier system, e.g. the use of highly corrosion-resistant copper instead of steel canisters that are often envisaged for clay/shale systems. The function of the host rock as a transport barrier becomes important only once radionuclides have been released from the engineered barriers (NEA, 2009). Depending on the characteristics of the latter, such release is expected to occur only in the far future. For example, in the main scenarios of repository evolution in the current Finnish and Swedish concepts, the copper canisters are considered not to fail at all over the period of interest (Miller and Marcos, 2007) or only at times beyond 100 ka (SKB, 2006a). Thus, the function of the host rock as a transport barrier remains latent over long periods of time in the main scenarios of these disposal systems. In alternative scenarios considering early failure of the canisters, its barrier function may come into play at earlier stages, and this is also the case in disposal systems considering steel canisters. In the Swiss disposal concept in Opalinus Clay, the design life of the steel canister is 10 ka, at which time all canisters are assumed to be breached (Nagra, 2002b). Given the limited retardation provided by the bentonite buffer for long-lived, non-sorbing species, these are expected to diffuse across the bentonite and to enter the host rock rapidly. Their flux is essentially controlled by the slow dissolution rate of the waste matrix.

8.1.2 Rock types considered as potential host formations An overview of rock types that are considered for geological disposal of radioactive waste worldwide is given in Witherspoon and Bodvarsson (2006). Most programmes focus on crystalline-basement rocks (14 sites) and fine-grained clastic sedimentary rocks (clay, shale, siltstone, marl; 11 sites). Rock salt, anhydrite, volcanic tuff and sandstone are considered in a small number of programmes. One of the main advantages of crystallinebasement rocks is their essentially unlimited size and availability, combined with the generally sufficient geomechanical stability (except in situations with high horizontal stresses). The main disadvantages include spatial heterogeneity (geometry of the fracture network) and the presence of transmissive fractures, even at the repository level. In contrast, argillaceous sediments can generally be considered as much more homogeneous and

© Woodhead Publishing Limited, 2010

Far-field process analysis and radionuclide transport modelling

225

spatially predictable, in addition to a typically very low hydraulic conductivity and to a high self-sealing capacity. The main disadvantages include the limited thickness and sometimes limited availability at suitable depth, the couplings of thermo-hydro-chemico-mechanical processes that render investigation programmes demanding and challenges concerning the construction of disposal vaults.

8.1.3 Time scales of concern The time scale over which repository safety needs to be explored depends on a number of factors, such as: . . .

the type of waste (e.g. spent fuel, vitrified high-level waste, low- and intermediate-level waste) and the amounts and half-lives of the relevant radionuclides; the feasibility of adequate prediction of the system behaviour in a repository environment; national regulatory frameworks.

As an illustration, Table 8.1 lists the radionuclide inventories of different types of wastes considered in the Swiss and Finnish disposal programmes. Part of the Swiss waste was reprocessed, resulting in three waste categories, whereas the Finnish concept considers direct disposal of spent fuel only. While a large part of the activity decays within geologically short periods of time, there are a number of radionuclides with half-lives longer than 1000 a. In recent years, the time frame of about 1 Ma over which repository safety must be demonstrated for all waste types except low-level waste has become accepted in several disposal programmes (overview in NEA, 2009). One of the reasons for this choice is the fact that, after some hundreds of thousands of years, most of the fission and activation products present in spent fuel have decayed to insignificance, and the remaining radiotoxicity drops significantly below that of the U ore from which the fuel was originally produced. It is generally acknowledged that the feasibility of predicting the future evolution becomes more limited the further the extrapolation extends. Human habits evolve over decades, which is one of the reasons to dispose of radioactive waste in the deep underground that is considered to be inaccessible to man. The biosphere is subjected to potential major changes over thousands to tens of thousands of years (climate change, glacial cycles), which means that only stylised quantifications of the potential radiological dose to which humans may be exposed in future are possible over these time scales. For the farther future, the radionuclide fluxes (expressed in Bq/a) can be predicted because the repository system with its ensemble of passive

© Woodhead Publishing Limited, 2010

226

Geological repository systems for safe disposal

Table 8.1 Radionuclide inventories of Finnish and Swiss waste types and sorption Kd values in the host rocks. Only nuclides with a half-life > 1000 a are shown. Data sources: Nykyri et al. (2008, Tables 6-8 and 6-17), Nagra (2002b, Tables A2.1.2, A2.1.4, A2.1.6 and A2.8), Nagra (2002c, Table A3.4-1). Activities are reported per ton of initial heavy metal (tIHM) in the Swiss case, where IHM = original mass of U or U + Pu (Pu originates from reprocessed mixed-oxide (MOX) fuel). As Finland does not reprocess spent fuel and pursues direct disposal only, there is no MOX fuel (i.e. IHM = U) and the waste categories related to reprocessing (vitrified high-level waste, intermediatelevel waste) do not exist. Activities relate to the time at which disposal is foreseen, i.e. after 30 (Finland) or 40 (Switzerland) years of decay on the surface

Radionuclide

Half-life (a)

Activity of Finnish spent fuel (OL1 and OL2 reactors) (GBq/tU)

10

Be C

1.60E+06 5.70E+03

2.78E+01

Cl Ca Ni 79 Se 93 Zr 93 Mo 94 Nb 99 Tc 107 Pd 126 Sn 129 I 135 Cs 137 Cs 166m Ho 226 Ra 229 Th 230 Th 231 Pa 232 Th 233 U 234 U 235 U 236 U 237 Np 238 U 239 Pu 240 Pu 242 Pu 243 Am 245 Cm 246 Cm

3.00E+05 1.00E+05 7.50E+04 1.10E+06 1.50E+06 4.00E+03 2.00E+04 2.10E+05 6.50E+06 2.30E+05 1.60E+07 2.30E+06 1.37E+03 1.20E+03 1.60E+03 7.90E+03 7.50E+04 3.30E+04 1.40E+10 1.60E+05 2.50E+05 7.00E+08 2.30E+07 2.10E+06 4.50E+09 2.40E+04 6.60E+03 3.80E+05 7.40E+03 8.50E+03 4.70E+03

14

36 41 59

1.04E+00 1.32E+02 3.30E+00 9.08E+01 1.13E+01 3.00E+01 6.12E+02 4.29E+00 2.18E+01 1.14E+00 2.15E+01 2.36E+06

2.41E-03 5.59E+01 7.45E-01 1.30E+01 1.30E+01 1.16E+01 1.05E+04 1.98E+04 7.61E+01 7.62E+02 6.16E+00 1.19E+00

Activity of Swiss intermediatelevel waste (ILW-1 type) (GBq/tIHM)

Sorption Kd in crystalline bedrock at Olkiluoto (dilute/ brackish water) (m3/kg) 0

Activity of Swiss spent fuel (GBq/tIHM)

Activity of Swiss vitrified high-level waste (BNFL glass) (GBq/tIHM)

1.2E-01 8.3E+01

4.34E-02

4.15E+00 1.43E-02

0

4.09E-01 9.77E-01 7.94E+01 3.05E-02

2.72E+01 2.95E-04 1.34E+00 3.17E-01 7.25E+00 3.63E-01

0.1 0.0005 0.2 0.0005 0.02 0.05 0.001 0.001 0 0.05 0.05

1.5E+00 1.9E-01 8.8E+01 1.1E+00 1.1E+02 2.8E-01 8.8E+00 7.4E+02 6.7E+00 2.0E+01 1.8E+00 2.5E+01 2.3E+06 1.4E+00 1.9E-04 9.4E-06 2.1E-02 5.1E-04 2.7E-08 4.5E-03 6.5E+01 6.1E-01 1.4E+01 2.1E+01 1.3E+01 1.5E+04 2.7E+04 1.3E+02 1.5E+03 3.7E+01 8.0E+00

5.74E+02 4.58E+00 3.12E+01 1.28E-03 2.38E+01 1.41E+06 9.16E-02 1.16E-04 3.85E-06 6.72E-03 1.22E-03

3.40E-08 7.93E-10 4.08E-06 1.88E-07

2.14E-03 6.05E-02 1.83E-04 2.26E-03 1.22E+01 3.67E-03 4.09E+01 1.41E+02 2.14E-01 5.38E+02 3.97E+00 7.33E-01

4.08E-07 1.25E-02 2.27E-04 1.72E-03 1.68E-03 2.49E-03 3.85E+00 5.44E+00 1.72E-02 8.16E-02 3.40E-04 7.03E-05

3.63E-03 9.75E-03 2.72E-02 9.75E+02

© Woodhead Publishing Limited, 2010

0.2 0.2 0.2 0.05 0.2 0.1 0.1 0.1 0.1 0.2 0.1 0.5 0.5 0.5 0.04 0.04 0.04

Sorption Kd in Opalinus Clay (reference case) (m3/kg) 0.9 0.001 (Cinorg), 0 (Corg) 0 0.001 0.9 0 10 0.01 4 50 5 100 0.00003 0.5 0.5 50 0.0007 50 50 5 50 20 20 20 20 50 20 20 20 20 10 10 10

Far-field process analysis and radionuclide transport modelling

227

barriers is expected to evolve very slowly, but the conversion to dose (in mSv/a) becomes very uncertain. On a national level, the regulatory frameworks define the time scale and degree of rigorousness over which repository safety needs to be demonstrated. For example, in Finland, the constraint of a maximum annual dose of 0.1 mSv applies in the first thousands of years after repository closure (STUK, 2001). Over this period, the biosphere can be assumed not to differ substantially from the present situation. At later stages, predictions of the evolution of the biosphere become more uncertain, and therefore the regulation is based on specifying maximum radionuclide fluxes, while no conversion to dose is required from the implementer. In addition, the regulation acknowledges the fact that the reliability of the model calculations becomes limited at more distant times. The Swiss regulation (ENSI, 2009) defines a maximum dose of 0.1 mSv/a for the most strongly exposed individuals on the basis of the expected evolution of the repository system. Unlikely disruptive scenarios with a higher radiological risk must be shown to have a probability of less than 106 per year. In contrast to the Finnish regulation, the Swiss regulation requires the calculation of an individual dose up to a time scale of 1 Ma. At the same time, it acknowledges the fact that the evolution of the biosphere cannot be predicted accurately and suggests that the calculations be based on the current state of the biosphere and human habits. For times beyond 1 Ma, the radiological consequences of surface effects such as the possible erosive exhumation of the repository must be assessed and compared to those of natural radiation.

8.2

Transport and retardation in argillaceous sedimentary formations

The majority of the formations studied in the framework of the disposal of radioactive waste (whether at potential sites or in generic rock laboratories) have a limited thickness, typically 100–200 m (Table 8.2). However, the clayrich unit is frequently embedded within a thicker but lithologically more heterogeneous low-permeability sequence, also shown in Table 8.2. Given the limited thickness, it needs to be shown for a safety case that retardation of radionuclide transport is sufficient over such short transport paths. In the Swiss safety case (Entsorgungsnachweis; Nagra, 2002a, 2002b), the geosphere barrier is represented by a vertical migration path of 40 m through Opalinus Clay. This value is less than the half formation thickness and excludes the near-repository geosphere. Similarly, the additional retardation along the rest of the migration path (upper part of the low-permeability sequence, then aquifers) is neglected in the reference case.

© Woodhead Publishing Limited, 2010

© Woodhead Publishing Limited, 2010

{

*

113.2

160.0

102.6

257.4

312.1

219.0

102.6

257.4

7.9E-11

4

5

2.1E-11–2.8E-11

4.2E-11–5.6E-11

2

1.0E-15–1.0E-14

2E-12–4.3E-11*

4E-14 (4E-15–4E-13)

6E-15–3E-14

5E-14–5E-13

<2

1.4E-10–4.0E-10* 2.3E-10–4.6E-10* 2

4.8E-11

1.9E-11

5.3E-11

1.4E-10

5.6E-11

130

256.0

Hydraulic conductivity in shaly unit, ^ (m/s)

Anisotropy of Dp in shaly unit (| | / ^)

{

Physical porosity in shaly unit (–)

0.18

0.12

2.0E-12–2.6E-12

0.43

0.54

0.5

0.5

Fraction of physical porosity accessible to anions in shaly unit (–)

0.055 – 0.085 0.3 – 0.5

4.8E-12–2.9E-10* 0.37

2E-13 (2E-14–2E-12)

1E-14–6E-14

n.d. (only minor 0.18 anisotropy)

Hydraulic conductivity in shaly unit, | | (m/s)

The latter value refers to the lowermost, sandier part of Boom Clay (Lower Belsele-Waas Member). Based on Matray, personal communication, 2008.

CallovoOxfordian at Bure Opalinus Clay at Benken Opalinus Clay at Mont Terri Boom Clay at Mol ToarcianDomerian at Tournemire

Site

Pore diffusion coefficient Dp for HTO in shaly unit at 208C, ^ (m2/s)

Pore diffusion coefficient Total Dp for thickness of lowanions permeain shaly Thickness unit at bility sequence of shaly 208C, (m) unit (m) ^ (m2/s)

Table 8.2 Comparison of geometric properties and of transport parameters for selected argillaceous formations studied in the framework of the disposal of radioactive waste. Data are from a compilation in Mazurek et al. (2009). Listed data refer to the screened data sets used for modelling, where ^ = value normal to bedding, || = value parallel to bedding

228 Geological repository systems for safe disposal

Far-field process analysis and radionuclide transport modelling

229

Owing to the limited length of the migration path within the host formation, it needs to be shown that radionuclide transport along this path is slow and predictable. A key aspect is the assumption that transport occurs through the microporous rock matrix and is dominated by diffusion, while advection along fractures is not relevant. The relative importance of diffusion and advection is expressed by the Peclet number, which is calculated from Pe ¼

va z K grad H z ¼ Dp nDp

with Pe = Peclet number (–) va = advection velocity (m/s) z = transport distance (m) K = hydraulic conductivity (m/s) grad H = hydraulic gradient (m/m) n = porosity (–) = pore diffusion coefficient (m2/s) Dp For Pe values <1, transport is dominated by diffusion, whereas advection/ dispersion is the main transport process for Pe > 10. Using the data listed in Table 8.2 and taking z as the half thickness of the low-permeability sequence, K and Dp (HTO) values relating to the direction normal to bedding (i.e. the direction towards the embedding aquifers) and an assumed hydraulic gradient of 1 m/m yield values for Pe of 0.3–2.5 (CallovoOxfordian at Bure), 0.2–0.7 (Opalinus Clay at Benken), 0.03–3.1 (Opalinus Clay at Mont Terri), 1.2–13 (Boom Cay at Mol) and 0.04–0.4 (ToarcianDomerian at Tournemire). The maximum value for Boom Clay was calculated using the hydraulic conductivity of the 12 m thick, sand-rich base of the formation, whereas conductivities for the bulk of the formation are lower. Overall, the calculated Peclet numbers suggest that transport in the mentioned formations is essentially diffusive. Another aspect is that, in clay-rich formations, the degree of spatial variability of hydraulic conductivity is limited, and there is in general sufficient evidence to exclude the possibility of advection along discrete discontinuities (e.g. faults, fractures) from the reference calculations of radionuclide transport (even though such situations may be considered in the frame of ‘what if’ cases). There is a substantial set of supporting arguments: .

A large number of hydraulic tests for the formations listed in Table 8.2 and for other argillaceous units (Boisson, 2005) are available, indicating that hydraulic conductivity is invariably very low.

© Woodhead Publishing Limited, 2010

230 .

. .

Geological repository systems for safe disposal Hydraulic tests in Opalinus Clay at the Mont Terri rock laboratory indicate that fractures and faulted intervals are hydraulically indistinguishable from the undeformed matrix (Marschall et al., 2004). Similar findings were made elsewhere. Mineralisations in discontinuities in these argillaceous formations are infrequent or absent, which is taken as evidence for their hydraulic insignificance over geological time scales. Clays and shales are known to have a large self-sealing capacity, mainly due to the presence of swelling clay minerals. The substantial amount of empirical and mechanistic evidence highlighting self-sealing phenomena is addressed in detail in Bock et al. (2010). Self-sealing operates beyond a specific depth. For Opalinus Clay, the surficial decompaction zone where open fractures occur is about 30 m thick (Hekel, 1994). Moisture zones at greater depths are infrequent and disappear completely at about 200 m below the surface (Gautschi, 2001). The decompaction zone may be thicker in more strongly consolidated and/or less clay-rich formations.

One other argument in support of diffusion as the main transport process originates from studies of pore-water composition in argillaceous rocks. Conservative constituents of the pore water, such as halogens (Cl, Br), stable water isotopes (δ18O, δ2H) and noble gases (e.g. He) typically define regular, often curved profiles across argillaceous formations sandwiched between aquifers. A study considering nine European sites (Mazurek et al., 2009) came to the conclusion that, in all cases, the observed tracer distributions can be explained and modelled as a consequence of diffusive transport alone. Laboratory-derived diffusion coefficients were used for the calculations, resulting in modelled evolution times for the profiles within geologically plausible ranges. This consistency is an indication that the laboratory-derived transport parameters can be upscaled to the formation scale, which is the relevant scale for geosphere-transport calculations. Model calculations were also made including vertical advection in addition to diffusion. Even though an additional process (with new free parameters) was brought in, the model fits turned out to be equal or worse than when diffusion alone is considered. As an example, Fig. 8.1 shows the spatial distribution of Cl in pore water in Opalinus Clay at the Mont Terri rock laboratory (see Pearson et al., 2003, for reference). Assuming an initially marine Cl concentration of 19 350 mg/L (note that current Cl/Br ratios are close to the marine value throughout the profile), the observed Cl profile can be modelled as due to out-diffusion of salinity to the overlying limestone aquifer since 6.5 Ma and a more recent out-diffusion to the underlying aquifer since 0.5 Ma. The age of thrusting and folding of the Jura Mountains in which the Mont Terri anticline is located is ≤10 Ma, so an

© Woodhead Publishing Limited, 2010

Far-field process analysis and radionuclide transport modelling

231

8.1 Measured distribution of Cl in pore water of Opalinus Clay at Mont Terri (symbols) and modelled evolution of Cl (curves). Time = 0 corresponds to the activation of the upper aquifer. Pore-water data are based on direct water sampling, leaching and squeezing tests as documented in Pearson et al. (2003). Ground-water samples taken in the adjacent aquifers (grey shading) have very low Cl contents. Numbers adjacent to model curves indicate evolution times in Ma since activation of the upper aquifer. A marine initial Cl concentration (19.35 g/L) was assumed (adapted from Mazurek et al., 2009).

erosive activation of the upper aquifer at 6.5 Ma is plausible. Similarly, the independently constrained erosion history yields an age of 0.2–0.5 Ma for the exhumation of the underlying aquifer, which is again in good agreement with the model prediction based on diffusive Cl transport. The model based on diffusion alone is capable of explaining the asymmetric profile of Cl across the formation and yields diffusion times that are consistent with palaeohydrogeological evidence. Analogous results were obtained for δ18O and δ2H as well as for He (Ru¨bel et al., 2002). Interestingly, the Cl concentration does not show any anomaly in the vicinity of the Main Fault, a major brittle structure cross-cutting Opalinus Clay. This suggests that the current observation regarding the hydraulic insignificance of this fault is

© Woodhead Publishing Limited, 2010

232

Geological repository systems for safe disposal

also valid for the longer periods in the past recorded by the Cl concentration profile. In summary, there are multiple lines of evidence indicating that – in a geologically favourable environment – fractures and faults in clays and moderately consolidated shales are infrequent and/or hydraulically insignificant. The limited structural and hydraulic heterogeneity is one of the reasons why the argillaceous host rock is considered as an important (if not the main) barrier to the release of radionuclides to the biosphere. Diffusive transport will further be retarded by sorption on clay–mineral surfaces (see Table 8.1 for sorption coefficients).

8.3

Transport and retardation in crystalline-basement environments

There is no doubt that advection through fracture networks dominates transport in most crystalline-basement environments. Because waste will be placed in low-permeability domains of a disposal site, with a minimum distance to more transmissive features, the first part of the migration path between the waste canister and the biosphere is expected to have the highest transport resistance. Among the most relevant parameters related to transport and retardation of radionuclides migrating via advection in fractures are (1) the advective flow rate q and (2) the surface area of the fracture walls along which flow takes place, called flow-wetted surface (FWS) (e.g. Neretnieks, 2002a). The spatial distribution of q is obtained from hydrogeological modelling (typically based both on an equivalent porous medium (EPM) and a discrete fracture network (DFN) approach). An example from Forsmark (Sweden) is documented in SKB (2008a). FWS is important because it relates to the area across which diffusive mass transfer between water flowing in the fracture porosity and the stagnant water in the matrix porosity can take place (matrix diffusion; Neretnieks 1980). FWS is spatially heterogeneous and not easy to quantify (see, for example, Andersson et al., 1998, and Crawford et al., 2003). Note that, from the perspective of radionuclide release to the biosphere, diffusion is a retardation process in systems where fracture flow dominates, in contrast to matrix-dominated systems, such as Opalinus Clay (and argillaceous media in general), where it represents a transport process. Even in low-porosity crystalline rocks, matrix porosity is the main reservoir of water in the system and typically greatly exceeds fracture porosity. If the large rock-matrix reservoir can be accessed by radionuclides, their concentrations in the flowing water may be greatly reduced. In addition, matrix diffusion provides access to the mineral-surface area of the matrix where sorption can occur (note that sorption on the fracture walls

© Woodhead Publishing Limited, 2010

Far-field process analysis and radionuclide transport modelling

233

alone contributes little to retardation). The surface area of the matrix exceeds the flow-wetted surface by several orders of magnitude. The sorption coefficients in hydrothermally unaltered crystalline rocks (examples are given in Table 8.1) are 1–3 orders of magnitude smaller when compared to those in shale environments, essentially reflecting the different mineralogical compositions, but nevertheless do have a marked retarding effect. Matrix diffusion alone (for non-sorbing nuclides) and the combined action of matrix diffusion and sorption (for sorbing nuclides) may greatly reduce and retard radionuclides fluxes through the geosphere. Because the transfer of nuclides from a flowing fracture to the matrix (characterised by FWS/q) is a complex process, it is not appropriate to quantify the effect by simple averaging along the flow path. Therefore, transport calculations quantifying the interaction between flowing water and stagnant matrix water are typically performed on the basis of the processed results of discrete fracture network models (DFN) or channel network models that better resolve the spatial heterogeneity of the relevant parameters. In most cases, the conceptualisation considers a dual-porosity medium where advection dominates in the fracture reservoir but only diffusion is considered in the rock-matrix reservoir. There has been a debate on whether the matrix porosity in crystalline environments is interconnected. In contrast to the early work of Norton and Knapp (1977) who, on the basis of a set of laboratory experiments, concluded that the connectivity is limited, evidence has emerged in recent times supporting the hypothesis that the pore space is connected under in situ conditions. The problem is that laboratory studies on matrix porosity and diffusion properties are difficult to extrapolate to in situ conditions due to the potential effects of stress release and sample preparation. Neretnieks (2002b) concluded that the existence of a connected porosity is generally agreed in hydrothermally altered zones that frequently occur in the first centimetres to decimetres along fractures, but no exhaustive evidence is available supporting the existence of a connected matrix porosity in the fresh rock beyond that zone. Since then, new in situ experimental results became available and shed light on this question, and a few examples are discussed below.

8.3.1 Connected porosity experiment at the Grimsel Test Site (Central Swiss Alps) A thin borehole (diameter 40 mm) was drilled from the research laboratory (ca. 400 m below the surface) into granodioritic rock. From a packed-off interval at ca. 3–5 m depth, acrylic resin was injected with a maximum pressure of 630 kPa over about 4 months. Then, the resin was polymerised

© Woodhead Publishing Limited, 2010

234

Geological repository systems for safe disposal

8.2 Connected porosity of Grimsel granodiorite visualised under UV light. Fluoresceine-doped resin was injected into the rock matrix from the central borehole. Bright spots are feldspar phenocrysts whose porosity originates from partial dissolution (etching) during the greenschist-grade metamorphic overprint of the granodiorite. The diameter of core is 30 cm (from Mo¨ri, 2009).

by heating to a maximum of 125 8C over 3 weeks. The experimental volume was then excavated by overcoring (diameter 300 mm) and studied in the laboratory. The experimental procedures and the results are documented in detail in Frieg et al. (1998) and Mo¨ri et al. (2003). Resin penetrated into the rock matrix to depths of > 10 cm, i.e. well beyond the zone that was potentially affected by mechanical damage due to drilling or stress redistribution, indicating that a connected porosity exists (see Fig. 8.2). Porosity was then measured by a suite of laboratory methods (Mo¨ri, 2009). The mass proportion of resin in subsamples (representing the in situ porosity) was quantified by measuring the carbon content and a porosity of 0.2–0.5% was obtained to depths of up to 20 cm from the borehole. Note that matrix porosity did not vary systematically as a function of distance from natural fractures, and even several metres away a connected porosity was identified.

8.3.2 Matrix fluid studies at Laxemar and Forsmark, Sweden Methods to study the chemical characteristics of pore water in crystalline rocks from Sweden (A¨spo¨, Laxemar, Forsmark) have been developed by Waber and Smellie (2008). In this framework, Waber et al. (2007) studied the Cl contents and the δ18O and δ2H values of pore water residing in the

© Woodhead Publishing Limited, 2010

Far-field process analysis and radionuclide transport modelling

235

matrix of crystalline rock at Laxemar (Sweden) in the vicinity of a waterconducting fracture at a depth of 112 m. The Cl content of the immediately adjacent matrix water corresponds to that measured in the fracture water (around 80–90 mg/L). Over a distance of 6 m away from the fracture, the Cl content increases to reach a constant value around 280 mg/L beyond that distance. Waber et al. (2009) interpreted the profile as due to diffusion in response to the evolving salinity of the water circulating in the fracture. Transport modelling considering the Pleistocene and Holocene evolution of ground waters reproduced reasonably well the observed profile considering plausible boundary conditions and diffusion as the only transport process. This evidence is taken as a supporting argument for the existence of a connected matrix porosity to a distance of at least 6 m from the fracture. Analogous results were also obtained from other Swedish study sites (see Waber and Smellie, 2008, for an overview) and from the Grimsel test site (Eichinger, 2009).

8.3.3 Constraining the formation factor by electrical conductivity logging in boreholes in Sweden Electrical conductivity (or resistivity) logging is a standard tool of geophysical borehole characterisation. Depending on the tool, the penetration depth into the formation is in the order of decimetres to metres, i.e. well beyond the zone potentially mechanically disturbed by the drilling process. Under certain assumptions (see, for example, Lo¨fgren, 2004), the measured conductivity can be used to constrain the formation factor of the rock matrix according to Ff ¼

De d kr ¼ nt 2 ¼ t Dw kw

with Ff = formation factor (–) De = effective diffusion coefficient of a species (m2/s) Dw = free-water diffusion coefficient of a species (m2/s) nt = transport porosity (–) d = constrictivity (–) t2 = tortuosity (–) kw = electrical conductivity of free water (S/m) kr = electrical conductivity of the rock matrix saturated with pore water (S/m) The assumptions underlying the applicability of electrical-conductivity logs for estimating the formation factor include: (1) the electrical conductivity of

© Woodhead Publishing Limited, 2010

236

Geological repository systems for safe disposal

the pore water within the tested volume can be constrained, either based on measurements on ground water circulating in adjacent fractures or on the basis of matrix-leaching data from the drill core (Waber and Smellie, 2008), (2) test intervals can be identified in which no fractures are present, such that the measured conductivity characterises the matrix and (3) the measured conductivity reflects ion movement in water, whereas conduction through minerals or along their surfaces is negligible or can be properly accounted for. Mean values of the formation factor for the rock matrix of various Swedish crystalline-basement rocks are in the range of 105 to 104 for insitu measurements (e.g. Lo¨fgren and Neretnieks, 2003; Liu et al., 2006; Lo¨fgren, 2007). Relating these values to the free-water diffusion coefficient of anions such as Cl or I (ca. 26109 m2/s at room temperature; Li and Gregory, 1974; CRC, 2004) yields values for De of 261014 to 261013 m2/ s. This is about a factor of 20–40 less than the range obtained for indurated clay-rich sedimentary formations (Mazurek et al., 2009), reflecting the lower porosity and different pore structure. Laboratory measurements of the formation factor yield values that are up to an order of magnitude higher when compared to in situ determinations, and Lo¨fgren (2004) attributes this difference to stress release and sample damage in the case of the laboratory samples. Overall, the in situ measurements of the formation factor indicate that a connected porosity exists in the studied formations, even though the derivation of specific parameters is subject to uncertainties, in particular due to the generally unknown chemical composition of the pore water and the distinction of ion movement in the rock matrix from that in fractures.

8.3.4 In situ diffusion experiments in the Canadian underground research laboratory (Pinawa, Manitoba) A suite of test boreholes were drilled in the URL into granitic rock to a depth of 10 m to reach rock not affected by tunnel excavation (Vilks et al., 2003, 2004). Tracer cocktails were injected into packed-off intervals, and care was taken to prevent hydraulic gradients. After 15 to 28 months, some of the test volumes were overcored and studied in the laboratory. In a first test stage, four vertical boreholes were emplaced at the 420 m level. Iodide tracer penetration to a matrix depth of 2.5 to 8 cm was identified for diffusion times of 1.3 to 2.3 years (Vilks et al., 2003). The shape of the diffusion profile could not be modelled when assuming a homogeneous medium and radial geometry, and a composite model with a 1-cm thick skin considering a lower diffusion coefficient was needed to explain the data. The skin was thought to be due to stress redistribution along the borehole, given the exceptionally high natural stresses at the 420 m level (s1 = ca. 60 MPa).

© Woodhead Publishing Limited, 2010

Far-field process analysis and radionuclide transport modelling

237

In a second test stage (Vilks et al., 2004), subhorizontal instead of vertical boreholes were drilled at the 240, 300 and 420 m levels, mostly subparallel to s1 or s2, which minimised problems with core discing. The tracer distribution was used to calculate effective diffusion coefficients for iodide of 1.461013 to 1.161012 m2/s in the distal zone unaffected by stress redistribution. While these in situ values do not vary with depth, ancillary laboratory tests suggest increasing trends for laboratory-derived diffusion coefficients and porosity with depth, which is interpreted to be due to a progressively stronger effect of stress release with increasing depth. To summarise, the experiments demonstrated (1) the existence of diffusion (and therefore a connected porosity) in situ and (2) the fact that laboratory measurements may be affected by stress release and therefore overestimate in situ transport parameters in the rock matrix.

8.3.5 Treatment of matrix diffusion in geosphere-transport calculations The in situ measurement of matrix porosity and diffusion coefficients in crystalline rocks is a non-trivial undertaking. A number of pertinent studies have been conducted in recent times, all indicating that the rock matrix contains an interconnected, diffusion-accessible pore network throughout the rock body. Adjacent to fractures, porosity and diffusion coefficients may be enhanced if hydrothermal or low-temperature alteration affected the rocks. However, diffusion has been shown not to be limited to alteration rims or a specific zone around fractures. Nevertheless, the amount of site-specific experimental data pertinent to in situ conditions is limited, and therefore geosphere-transport models in general do not consider unlimited matrix diffusion. In the Swedish longterm safety evaluation for the Forsmark and Laxemar sites, the penetration depth of matrix diffusion is in the range 0.03–10 m in the probabilistic geosphere-transport model (SKB, 2006a, 2006c). The upper limit is in the range of the typical half spacing between water-conducting fractures and therefore essentially represents full matrix connectivity, whereas the lower limit corresponds to the length over which connectedness has been undisputedly demonstrated. In the radionuclide transport model used by Posiva for Olkiluoto, the potential repository site for spent fuel in Finland, Nykyri et al. (2008) applied a simplified model considering one single pathway to quantify radionuclide transport in the geosphere, and a constant matrix-penetration depth of 10 cm was assumed. To conclude, the migration of sorbing radionuclides from defective canisters will be retarded in the bentonite and in the host rock. On the other hand, non-sorbing radionuclides released into the bentonite will diffuse out

© Woodhead Publishing Limited, 2010

238

Geological repository systems for safe disposal

into the host rock, and this occurs rapidly in comparison to the respective half-lives (Table 8.1). In the geosphere, matrix diffusion is essentially the only retardation process for long-lived radionuclides that occur mainly as anions and so interact weakly or not at all with mineral surfaces, such as 129 36 I, Cl or 79Se.

8.4

Quantifying radionuclide transport: two case studies

The role of the host rock and of the geosphere as a whole in a multi-barrier disposal system is illustrated here using the Swiss spent fuel/vitrified highlevel waste programme targeted at Opalinus Clay in northeasten Switzerland and the Finnish spent fuel programme focused on the crystalline basement at Olkiluoto (southwestern Finland) as examples. Both programmes consider the disposal of spent fuel, but in the Swiss case, vitrified high-level waste and intermediate-level waste are also part of the programme. The choice of these case studies is motivated by the very different characteristics of the geosphere at both sites, and this also affects the design of the engineered barriers. The purpose is to present and compare the approaches and conclusions as presented in the literature. The disposal concepts for the two sites are illustrated in a highly simplified form in Fig. 8.3. The Swiss concept considers a repository hosted by Opalinus Clay, a flat-lying, indurated, about 100–120 m thick shale formation. The waste will be encapsulated in steel canisters and will be embedded in a bentonite buffer (note that copper canisters are considered as an alternative option; see Nagra, 2002b). Due to the scarcity of fractures and their high self-sealing potential, there is sufficient scientific argument to exclude fracture flow from the expected evolution scenario, and diffusion is considered as the main transport process (Nagra, 2002a; Gimmi et al., 2007). In the Finnish concept, waste contained in copper canisters and surrounded by bentonite will be emplaced into Precambrian crystalline bedrock where transport is dominated by advection in fractures. In the KBS-3V design, each vertical deposition hole will be located in a rock volume devoid of major fractures and faults, even though lowertransmissivity fractures may be intersected by deposition holes. (Note that the KBS-3H design with horizontal in-tunnel emplacement is also being considered.) Along the flow path from the repository to the surface, radionuclides may be transported in progressively higher-hierarchy fractures and faults, so the first part of the flow path is thought to have the largest transport resistance. The properties of the host rock in the two disposal concepts are contrasting, and these differences affect the site-specific safety concepts.

© Woodhead Publishing Limited, 2010

Far-field process analysis and radionuclide transport modelling

239

8.3 Simplified illustrations of the disposal concepts for spent fuel in Opalinus Clay (Switzerland) and in crystalline basement at Olkiluoto (Finland). Sketch is not to scale. Arrows indicate possible migration paths of radionuclides towards the surface. For Olkiluoto, the KBS-3V concept with vertical deposition holes is shown. An alternative concept with in-tunnel deposition (KBS-3H) is currently also under consideration. The reference depths for the repositories are 650 m below the surface in Switzerland and 420 m in Finland.

8.4.1 Features, events and processes In order to understand and quantify the performance of the geosphere safety functions (see above), the relevant phenomena need to be explored. A structured way to deal with phenomena and to minimise the risk of omitting important processes and events is to compile and evaluate lists of FEPs (features, events and processes). In this context, events can be seen as discontinuous phenomena that occur over short periods of time (such as earthquakes) and features relate to qualitative attributes of the system and to parameter values needed to quantify the processes and events. Lists of geosphere-related FEPs, together with a documentation of their understanding and relevance for repository safety, are available on an international basis (e.g. general documentation in NEA, 2000, and FEP evaluation for argillaceous rocks in Mazurek et al., 2003). However, due to the specific features of national disposal concepts, of the site-specific properties of the repository system and of the safety functions of the geosphere, an understanding of features, events and processes must be obtained on a site-specific level for each disposal programme (e.g. Nagra, 2002d, for Switzerland; SKB, 2006b, for Sweden; Miller and Marcos, 2007, for Finland). It is convenient to structure geosphere-related FEPs into the following classes:

© Woodhead Publishing Limited, 2010

240 1.

2.

3.

Geological repository systems for safe disposal Undisturbed system: characterisation of the current state of the far-field geosphere. This is a snapshot of an evolving system but useful as a starting point. Repository-induced perturbations: effects on the host rock induced by the construction and operation of the repository and effects due to the presence and evolution of the waste and of the engineered barrier system over long periods of time (thermal, hydraulic, chemical, mechanical, gas-related, microbiological, coupled). Long-term evolution: effects on the repository system due to the natural evolution of the geosphere and of surface conditions over the time scale of interest (e.g. deformation events, vertical movements, climate change, human activities).

The total number of FEPs considered to be relevant within the repository system (waste, canister, bentonite buffer, host rock, geosphere) is 55 in the Swiss case (Nagra, 2002b, 2002d) and 70 in the Finnish case (Miller and Marcos, 2007). In Table 8.3, subsets of these integral lists are presented, in particular FEPs related to the host rock and to the geosphere as a whole, including the natural long-term evolution. FEPs related to the waste and to the engineered barrier system, to interactions between the engineered barriers and the host rock, as well as those considering future human actions, are not listed. Table 8.3 includes 9 (Swiss case) and 12 (Finnish case) FEPs. These represent the relevant aspects of FEP classes 1 and 3 above (with the exception of future human activities) but exclude FEP class 2. The FEPs from both programmes are thematically grouped in Table 8.3. It becomes evident that a large number of FEPs are addressed in both programmes, in spite of the large differences of the host rocks and of the disposal concepts. In particular, the main commonalties include the issue of fluid movement in the host rock (advection, diffusion, gas migration), the retardation of solute transport in the host rock by physico-chemical processes, future tectonic activity and fault displacements, and finally effects from the surface (erosion, glaciations, permafrost). However, there are also differences, and a number of FEPs are considered to be relevant in one programme but not in the other: .

.

The most direct flow path across the host rock is short (40 m) in the case of Opalinus Clay, so flow-path length is an issue, whereas this is less critical in the crystalline basement at Olkiluoto where the flow path will be at least 10 times longer. Opalinus Clay acts as a colloid filter due to its nanometric pore-space structure, so colloidal transport of radionuclides is not an issue. On the other hand, it may play a role in the fractured crystalline basement at Olkiluoto. In particular, bentonite colloids may be released from the buffer by water flowing in fractures. Various laboratory and in situ

© Woodhead Publishing Limited, 2010

Far-field process analysis and radionuclide transport modelling

.

.

.

241

studies have shown that such colloids are stable at low water salinity (e.g. Degueldre et al., 1996) and they have the potential to transport sorbing radionuclides through the geosphere (e.g. Geckeis et al., 2004; Missana et al., 2003, 2004). However, because sorbing radionuclides will be strongly retarded (if not fully contained) by sorption within the bentonite buffer and so radionuclide releases from the buffer are expected to be dominated by non-sorbing species, colloidal transport in the geosphere is not thought to be of prime importance, at least not in the expected evolution scenario of the repository system. Retardation in aquifers and in confining units embedding Opalinus Clay has no analogue in the crystalline basement at Olkiluoto because no distinction can be made between host rock and embedding units of the geosphere (the two are essentially the same). Geomorphological evolution (such as denudation and linear erosion) are important in the Swiss situation where land uplift is balanced by erosion. The crystalline rocks at Olkiluoto are more resistant to erosion, which is considered to be limited in spite of ongoing uplift and future glaciations. The possible effects of deep permafrost and deep penetration of glacial waters are studied in the Finnish case but are considered to be irrelevant for Opalinus Clay in Switzerland. In central Europe, deep permafrost is not expected to occur in future climates. Deep penetration of surficial waters is not of concern for Opalinus Clay due to its setting in a sedimentary basin with several low-permeability units that limit vertical flow.

8.4.2 Treatment of the geosphere in safety-assessment calculations Expected evolution Calculations quantifying radionuclide transport across the whole multibarrier system in the Swiss and Finnish disposal concepts were performed, and these calculations highlight the differences among the geological settings, host-rock types and the characteristics of the engineered barriers. In the reference case for Opalinus Clay, the lifetime of the steel canister containing spent fuel is assumed to be 10 ka, and thereafter all canisters are assumed to be defective (Nagra, 2002b). This means that after this time radionuclides can be released from the whole inventory of the repository. This is a gross conservative simplification of a more complex evolution over which the canisters may start to leak. In contrast, the Finnish concept considers copper canisters with corrosion rates that are much lower than those of steel. In the main scenario, the copper canisters are assumed to

© Woodhead Publishing Limited, 2010

© Woodhead Publishing Limited, 2010

Rock–water interaction; radionuclide solubility, sorption and precipitation; matrix diffusion

Geochemical retardation of radionuclide transport

Gas migration Two-phase flow

Colloidal trans- An unfractured argillaceous sedimentary rock acts as a Possible colloids originate mainly from erosion of the port colloid filter due to the pore structure with apertures in the bentonite buffer. Because sorbing radionuclides are not expected to leave the bentonite (unless it is defective), nanometre range (Voegelin and Kretschmar, 2002) sorption on colloids in the host rock may not be important

—

A free gas phase may develop because natural methane The generation of gas is a concern for intermediate-level waste (anaerobic corrosion of metals, degradation of contents in ground water are near saturation at certain levels. organics), to a lesser degree for spent fuel and vitrified high- Gas is also produced by anaerobic corrosion of metals in the

Fracture minerals such as carbonates and sulphides buffer ground-water composition and redox state. These in turn affect the speciation, solubility and sorption characteristics of radionuclides. Matrix diffusion is a retardation process operating only in fractured, advection-dominated systems. It results in a dilution of radionuclide concentrations due to mixing with pore waters in the matrix, and it provides access to sorption sites in the porous rock matrix. Co-precipitation is another process providing retardation, but it is conservatively neglected

Advective/dispersive transport along fractures dominates, and the issue is to understand the heterogeneous spatial distribution and the rate of flow. Hydraulic gradients and flow rates are expected to change in future (glaciations, land uplift, sea-level changes, water salinity)

Multiple lines of evidence suggest that flow along discrete fractures is insignificant (Nagra, 2002a). Fractures are infrequent and hydraulically not active (self-sealing). Flow through the matrix occurs but is small when compared to diffusive transport

Ground water flow (advection); dispersion

Advective flow in Opalinus Clay; heterogeneous flow

Retardation occurs mainly via sorption on mineral surfaces. Stable reducing conditions prevail and are buffered by solid phases (sulphides, organic matter). Solubility limits of radionuclides (precipitation and co-precipitation of solids) are conservatively neglected

Minimum transport path is several hundreds of metres. The first part provides most transport resistance. At larger distances from the deposition hole, solutes may be transported in more transmissive faults

Crystalline basement at Olkiluoto, Finland

Critical parameter in the case of sedimentary units of limited thickness and disposal systems in which retention in the geosphere plays an important role. The minimum vertical path length in Opalinus Clay is 40 m

Opalinus Clay in northeastern Switzerland

—

Crystalline basement at Olkiluoto, Finland

Assessment of key phenomena

Length of vertical transport path in Opalinus Clay

Opalinus Clay in northeastern Switzerland

Key phenomena (FEPs)

Table 8.3 Results of the FEP (features, events and processes) evaluation for repositories in Opalinus Clay of northeastern Switzerland and in the crystalline basement of Olkiluoto, southern Finland. Data sources: Nagra (2002d, Table A4.1.1) and Miller and Marcos (2007, Chapter 8). While these documents include a FEP analysis for the whole repository system, only FEPs that pertain to the far-field host rock, its wider frame (i.e. the geosphere) and natural effects on the geosphere originating from the surface are included here. FEPs related to all other system components (waste, canister, bentonite buffer), interactions between the engineered barriers and the host rock (thermal, chemical, hydraulic, mechanical, coupled) and the effects of future human actions (such as drilling or exploitation of resources) are excluded

System component

Host rock

System component

Geosphere

Surface environment

© Woodhead Publishing Limited, 2010

Crystalline basement at Olkiluoto, Finland

Deep permafrost is not expected to occur during future cold periods. The main issue relates to the nature and rate of erosion (in particular linear deep glacial erosion) and the evolution of the biosphere

The ongoing uplift is not balanced by erosion and results in a The expected amount of total erosion (denudation in response to uplift and linear erosion by rivers) is max. 200 m net gain in elevation. Glacial erosion is estimated at 1–4 m per in the coming 1 Ma. Deep glacial erosion will also occur but glacial cycle (Posiva, 2006) will be limited to the known overdeepened valleys where this type of erosion occurred in past glacial cycles (Nagra, 2002a)

Climatic evolu- Freezing and tion permafrost; methane hydrate formation; salt exclusion

Geomorpholo- — gical evolution

Deep permafrost is expected to occur during future cold periods and will affect the hydrogeological system. The penetration depth of permafrost is a matter of current research. Salt exclusion due to partial freezing of ground and pore water, as well as the formation of clath rates, are further potential consequences of cold climates

Not well characterised, therefore conservatively neglected in Retardation is considered along the whole flow path from the reference scenario canister to surface. However, the more distal parts of the flow path do not contribute much to retardation because flow is expected to be through major, higher-transmissivity structures

Fault activity is a concern because future glaciations may result in differential movements. Breaching of canisters is unlikely because the disposal holes will not be cross-cut by major fractures

Retardation in — local and regional aquifers

Reactivation The siting regions are in tectonically quiet areas (low displacements seismicity, no active faults). A low uplift rate is essential along existing because uplift is expected to be balanced by erosion fractures

Neotectonic activity

Microbes may be present throughout the geosphere and catalyse chemical reactions. Because mostly non-sorbing radionuclides are expected to reach the geosphere, microbial effects on their transport through the host rock are thought to be limited

level waste (anaerobic corrosion of metals). The build-up of repository. Gas transport will occur in fractures, with possible gas pressure is an issue in Opalinus Clay due to its very low effects on water flow permeability. Gas migration is expected to occur by diffusion in aqueous solution, two-phase flow and pathway dilation mainly in the horizontal direction, whereas macroscopic gas fracs are not expected (Nagra, 2004; Marschall et al., 2005) The amount of water displaced by gas migration is likely to be negligible (Rodwell, 2000).

Opalinus Clay in northeastern Switzerland

Microbes cannot metabolise in the generally nanometric Microbial populations pore network of argillaceous sedimentary rocks (Stroesand processes Gascoyne, 2002)

Crystalline basement at Olkiluoto, Finland

Assessment of key phenomena

—

Opalinus Clay in northeastern Switzerland

Key phenomena (FEPs)

244

Geological repository systems for safe disposal

remain intact over the entire period of interest. Radionuclides are only released if a canister is defective (initial welding defects, corrosion defects, defects due to rock shear), and this was quantified in the ‘defective canister scenarios’ (Nykyri et al., 2008). Here, scenario Sh1 is used for illustrative purposes. This scenario assumes that one single canister has an initial defect with a size of 1 mm. Figure 8.4 illustrates the calculated release rates of radionuclides from the bentonite buffer and from the host rock at both sites. Release is given in units of dose (mSv/a), using the site-specific dose-conversion factors. The following observations can be made: .

Given the earlier onset of release from the canister in the chosen scenario (initial canister defect), short-lived radionuclides such as 90Sr enter the geosphere at Olkiluoto but not in Opalinus Clay, where

8.4 Release of radionuclides from the bentonite buffer and from the host rock as a function of time. Left: reference case for a spent-fuel repository hosted by Opalinus Clay in northeastern Switzerland (based on Nagra, 2002b; all steel canisters are assumed to be defective). Right: case Sh1 (with assumed small defect in one single copper canister) for a KBS-3V spent-fuel repository in crystalline rock at Olkiluoto, Finland (based on Nykyri et al., 2008). Radionuclide fluxes are recalculated to units of dose using conversion factors for the reference biosphere. The recalculation is hypothetical for the releases from bentonite but facilitates the comparison with releases from the host rock. Note that the ranges of both axes are different between the two sites. Not shown: 137 Cs release from bentonite buffer at Olkiluoto (peak of 9.46107 mSv/a at 77 a; negligible release from host rock).

© Woodhead Publishing Limited, 2010

Far-field process analysis and radionuclide transport modelling

.

.

.

.

.

245

complete containment over the first 10 ka is assumed. Because Sr is sorbing, it decays completely within the host rock. In the case of Opalinus Clay, a substantial number of radionuclides migrate across the bentonite buffer and enter the host rock, including some nuclides of the U and Th decay chains. This is not the case at Olkiluoto, where all nuclides of the decay chains are contained within the bentonite. This is likely due to the fact that only one and not all canisters are assumed to have failed and due to the slow fuel dissolution and release rate of radionuclides through the small defect. Overall, the calculated release of radionuclides from the bentonite buffer into the host rock is much lower at Olkiluoto, reflecting the fact that the copper canisters are assumed to contain completely most of the activity over the entire period of interest. For Opalinus Clay, the highest release from the bentonite is predicted to occur shortly after the assumed complete failure of all steel canisters at 10 ka after the present. This peak represents the ‘instant release fraction’, i.e. the release of radionuclides residing along grain boundaries in the spent fuel and in the gap region. These radionuclides are assumed to be transferred completely into aqueous solution as soon as water penetrates the spent fuel; i.e. their release is independent of the slow fuel-dissolution rate. A corresponding peak is not seen in the Finnish case, most probably reflecting different assumptions made regarding the parameterisation of the instant release fraction and the transport from the fuel into the bentonite. Opalinus Clay is a very efficient transport barrier and contains most radionuclides. Only weakly or non-sorbing radionuclides (14C, 79Se, 36 Cl, 129I) are released into the embedding rock units. The reason for the good host-rock performance is the fact that transport occurs mainly by diffusion and because of the high sorption coefficients for many radionuclides, in particular for actinides (Table 8.1). The maximum doses are reduced by orders of magnitude even for the non-sorbing radionuclides, and the releases are shifted to substantially more distant times. The elevated releases of these radionuclides from the bentonite at early stages are attenuated. This finding contrasts with that for Olkiluoto, where retardation due to transport through the geosphere is weak for non-sorbing radionuclides such as 14C, 36Cl and 129I. The calculated release rates from the host rock into the biosphere are lower for the crystalline basement at Olkiluoto than for Opalinus Clay in Switzerland, essentially reflecting the assumption that only one copper canister (out of about 3000) fails in the former situation but all steel canisters fail in the latter. Nevertheless, both assessments yield doses that are orders of magnitude below regulatory limits. In conclusion, the weaker performance of the steel canisters in the Swiss

© Woodhead Publishing Limited, 2010

246

Geological repository systems for safe disposal concept is compensated by the high transport resistance of Opalinus Clay. In the Finnish concept, the canister, together with the surrounding bentonite buffer, is expected to retain most radionuclides over the whole period of interest (with the exception of non-sorbing nuclides, such as 14 C, 36Cl and 129I). The role of the crystalline bedrock as a transport barrier is limited in the radionuclide transport calculations, in part due to the comparatively low transport resistance (relatively high hydraulic conductivity), in part due to the difficulty to predict fully the dynamic evolution in the future (land uplift, glaciations, spatial heterogeneity), and due to the simplified and conservative approach used to represent retardation in the rock matrix.

Alternative and ‘what if’ calculations The expected evolution (or the reference case) represents the situation in which the barrier system functions according to the best current understanding. Alternative calculations explore uncertainties in scenarios, conceptualisations and parameter values. ‘What if’ calculations are meant to test the robustness of the disposal system, and they include scenarios and parameter values that are not supported by (or even in contradiction to) current scientific understanding. In the Finnish safety evaluation for a KBS-3V repository (Nykyri et al., 2008), a number of alternative and ‘what if’ calculations were performed. All underlying scenarios assume the failure of a canister in consequence of one of the following events and processes: . .

. .

An initial defect due to inadequate encapsulation of the spent fuel (hole up to 10 cm in size). A delayed defect due to corrosion (10 cm at times of 10–100 ka after the present). This scenario addresses the uncertainties regarding the stability of the geochemical environment (such as deep infiltration of oxygen-rich glacial water or corrosion due to sulphide-rich ground water). A defect growing with time (0–100 ka after the present) due to a disrupted bentonite buffer, resulting in access of sulphide-rich water and high flow rate. A delayed defect due to displacement along a fault cross-cutting the deposition hole (1–70 ka after the present). This scenario covers the issue of geomechanical (in)stability of the environment in response to future glacially triggered crustal movements.

Under these assumptions, the following cases were explored: .

Suite of ground-water compositions from glacial to saline, with effects

© Woodhead Publishing Limited, 2010

Far-field process analysis and radionuclide transport modelling

. . .

247

on anion-accessible porosity, diffusion coefficients, radionuclide solubilities and sorption characteristics. Increased water flow rates, enhancing the transfer of radionuclides to the fractured host rock and transport through the geosphere. Increased fuel-degradation rate (in particular in situations when oxygenated water penetrates to the repository depth). Formation of gas in the canister, leading to displacement of water that contains radionuclides out of the bentonite and to radionuclide migration in the gas phase.

The calculations show that, as in the base case, 129I, 135Cs and 14C contribute most to the total dose rates in the majority of the calculation cases. In ‘what if’ cases combining an initial canister defect with high flow rates and saline ground water, calculated peak doses would be reached in the range of only decades to a few centuries after emplacement, and the shortlived nuclides 90Sr and 226Ra would dominate the dose. This reflects the rapid flow and the reduced calculated retardation of these nuclides in the geosphere (transit times through the geosphere become short with respect to the decay of 90Sr and 226Ra, and matrix diffusion and sorption in the matrix are even more limited), and the higher salinity (resulting in higher assumed solubility limits and lower sorption Kd values). In the Swiss programme for deep disposal in Opalinus Clay (Nagra, 2002b), a number of parameter variations were conducted around the reference case. Variations of host-rock properties included higher/lower flow rates, higher diffusion coefficients and lower sorption coefficients, with only a limited effect (factor ≤4) on calculated dose rates from spent fuel. In the framework of alternative conceptualisations of processes in various system components (e.g. fuel dissolution, limited performance of the bentonite, recurrent flow events in the host rock during glacial cycles, radionuclide release along shaft and ramp, gas-induced displacement of pore water), the calculated dose rates originating from spent fuel are never substantially higher than in the reference case. Further, alternative scenarios explored the consequences of gas pathways on the transport of volatile radionuclides and of various kinds of human actions. Lastly, a substantial number of ‘what if’ calculations were performed. Some of these are presented here, all relating to spent fuel: .

Upward Darcy flow in Opalinus Clay is assumed to be 261012 m/s, i.e. 100 times the value in the reference case, in which case transport is advection-dominated. The resulting maximum doses are a factor 20 to 40 higher compared to the reference case and are reached at earlier times (30–80 ka after the present). The same radionuclides (129I, 36Cl, 14C and 79 Se) dominate. The increase of the maximum dose rate is lower than the

© Woodhead Publishing Limited, 2010

248

.

.

.

.

Geological repository systems for safe disposal increase of the Darcy flow because diffusion also contributes to transport in the reference case. Assuming the existence of a vertical fracture with a transmissivity of 16109 m2/s, cross-cutting all emplacement tunnels of the repository results in a maximum dose rate of 6.56104 mSv/a at a hydraulic gradient of 1 m/m. This is slightly more than one order of magnitude higher than in the reference case. The limited increase of the dose rate is partly due to the retardation by matrix diffusion that was considered to occur along the fracture. The peak dose rate occurs shortly after the assumed canister failure at 10 ka, corresponding to the contribution of the instant release fraction. In addition to the radionuclides that dominate in the reference case (Fig. 8.4, bottom left), 226Ra becomes important over the entire period of interest (i.e. up to 1 Ma). As a nuclide of the 4N+2 chain, it is continuously produced by ingrowth from 238U, so it is present in the fuel matrix at all times, in spite of its short half-life of 1.6 ka. The combination of the low sorption coefficient of 0.0007 m3/kg for Ra (Table 8.1) and the high advective flow rate assumed in this ‘what if’ case result in a transport time through Opalinus Clay that is on the order of the half-life of 226Ra, resulting in an incomplete containment of 226Ra in the host rock. Nevertheless, the peak dose rate is due to 129I and not 226Ra. An increase of the fuel-dissolution rates by factors of 10 and 100 translates into dose-rate increases by factors of about 4 and 10, indicating the non-linear relationship of these parameters (due to solubility limits and other factors). 129I,36Cl, 14C and 79Se dominate the dose rates. Assuming that a discrete gas phase is generated in the near field and then breaks through to the overlying aquifer, thereby bypassing the retardation capacity of the host rock, leads to a maximum additional dose rate from volatile 14C of about 56105 mSv/a. This value is similar to the total dose rate in the reference case, and is reached at around 10 ka after the present. Setting the sorption coefficient for 129I to zero in the bentonite and in the host rock doubles the maximum dose rate, and the maximum is reached at an earlier time. This illustrates that even the low sorption coefficients of 56104m3/kg (bentonite) and 36105 m3/kg (host rock) as used in the reference case have a marked retarding effect on iodine transport.

In conclusion, the alternative calculations in both disposal programmes were meant to quantify the consequences of various FEPs and related uncertainties on system performance, as listed in Table 8.3. The robustness of the disposal system was further explored by ‘what if’ calculations. In the Finnish analysis, the main objective of these calculations was to explore the

© Woodhead Publishing Limited, 2010

Far-field process analysis and radionuclide transport modelling

249

(in)sensitivity of the engineered barrier system to internal and external effects influencing geochemical stability and mechanical integrity. In the Swiss safety assessment, both the functionality of the engineered barriers and the performance of the host rock as a transport barrier were considered. All calculations at both sites resulted in maximum dose rates well below the regulatory limit of 0.1 mSv/a. Dose rates from spent fuel are dominated by radionuclides occurring in anionic form (129I, 36Cl, 79Se, 14C), by 14C potentially occurring in a gas phase and by the weakly sorbing cation 135Cs. With the exception of 14C, all these nuclides have half-lives > 100 ka. Actinides do not contribute to calculated doses because of their low solubilities and favourable sorption characteristics (Table 8.1), and are contained in the engineered barriers and/or in the geosphere. The only exception is 226Ra, which (together with 90Sr in the Finnish analysis) may become important in ‘what if’ calculations where unrealistically high flow rates through the geosphere are considered.

8.5

Emerging trends

8.5.1 In situ studies of porosity and transport parameters of the rock matrix In argillaceous systems, there is confidence that porosities, diffusion coefficients and sorption characteristics measured in the laboratory do not deviate substantially from the in situ values (e.g. Van Loon et al., 2004; Wersin et al., 2004, 2008). Van Loon et al. (2003) conducted through- and out-diffusion experiments with Opalinus Clay at pressures up to 15 MPa. No measurable effect on porosity was identified between experiments at 4 and 15 MPa, and the diffusion coefficient decreased only by about 20%. Diffusion coefficients measured in the laboratory and in situ were also used to model natural-tracer profiles across numerous argillaceous formations, and the resulting diffusion times were always within the range that is plausible and consistent with independent palaeo-hydro-geological arguments (Mazurek et al., 2009). In conclusion, stress release and other artefacts appear not to affect the pore structure of clays and shales substantially, and laboratory measurements can be transferred to in-situ conditions. Similarly, sorption Kd values based on laboratory measurements on powdered material are generally in agreement with in-situ values derived from modelling in situ diffusion experiments (e.g. Van Loon et al., 2005; Wersin et al., 2008). These conclusions are likely not to be true for crystalline-basement environments, where porosity is much smaller and even minor disturbances can have an impact on porosity and the tortuosity of the pore network. In recent years, efforts have been made in rock laboratories to characterise

© Woodhead Publishing Limited, 2010

250

Geological repository systems for safe disposal

matrix porosity and diffusion coefficients under in situ conditions. These efforts are continuing, with a trend towards long-term experiments, in order to achieve penetration distances of diffusing tracers well beyond the zone around the borehole where some degree of stress redistribution occurred. Examples of completed experiments include those by Vilks et al. (2003, 2004) in the Canadian underground research laboratory in Manitoba, the CP experiment at the Grimsel test site in Switzerland (Frieg et al., 1998; Schild et al., 2001; Mo¨ri et al., 2003; Ota et al., 2003). Long-term experiments are ongoing in the Swedish underground rock laboratory at A¨spo¨ (LTDE-SD; interim report in SKB, 2008b, Section 3.3) and in the Grimsel test site in Switzerland (LTD; preliminary information in Mo¨ri et al., 2007, and Eichinger et al., 2008). In the framework of these projects, diffusion of non-sorbing as well as sorbing tracers is studied, so the experiments are expected also to provide in situ information on sorption behaviour. While today Kd values for sorption are usually based on batch laboratory experiments using rock powders, there is evidence that the in situ mineral-surface area and sorption Kd values are smaller (e.g. Andre´ et al., 2008). The transferability of laboratory-derived Kd measurements to in situ conditions was first addressed by Frick (1994) and Heer and Hadermann (1996) in the framework of in situ migration experiments in granodiorite of the Grimsel test site (Central Switzerland). For Na, the Kd derived from modelling of the in situ experiments was about one order of magnitude smaller than the one obtained from batch sorption experiments on rock powders crushed to a grain size <63 μm.

8.5.2 Chemical characterisation of matrix pore fluids Pore waters in many rock types are essentially stagnant, and their chemical composition evolves only via slow exchange with waters circulating in fractures or adjacent aquifers. Therefore, the chemical and isotopic composition of pore water is an archive recording geochemical processes that have occurred over geological time scales in the past, with a much longer memory when compared with mobile ground waters. Another aspect is that the spatial distribution of chemical characteristics of pore water can be used to constrain the dominating transport process in the rock matrix. Both these aspects have been studied extensively in clay-rich formations (e.g. Patriarche et al., 2004a, 2004b; Gimmi et al., 2007; Mazurek et al., 2009). In argillaceous rocks, considerable progress has been made in recent years with respect to pore-water characterisation, and further efforts are ongoing. In order to minimise artefacts related to oxidation and bacterial activity, boreholes in rock laboratories are drilled with nitrogen and all equipment is sterilised. In situ circulation systems have been designed to measure directly the pore-water composition, including the in-line measurement of sensitive

© Woodhead Publishing Limited, 2010

Far-field process analysis and radionuclide transport modelling

251

parameters such as pH and pCO2 (e.g. Vinsot et al., 2008). On a parallel track, steps are ongoing towards the capability to constrain the full porewater chemistry on the basis of measurements on cores and geochemical modelling alone, considering carbonate and silicate minerals (e.g. Beaucaire et al., 2008; Gaucher et al., 2009). Disturbing effects and artefacts such as pyrite oxidation and mineral dissolution during aqueous leaching could be limited by new sampling techniques (e.g. sample preservation in liquid nitrogen) or corrected by geochemical modelling. The need arbitrarily to choose values for certain parameters (such as pCO2 ) has been reduced or eliminated by considering a larger number of mineral-water equilibria. The modelling approach used by Gaucher et al. (2009) for the CallovoOxfordian shale in France resulted in a pore-water composition that has narrower uncertainty ranges than previous efforts and compares well with the composition of pore water obtained from in situ experiments. Only in recent times, the techniques developed for the characterisation of pore waters in argillaceous rocks have been adapted to crystalline rocks. Because the pore-water contents of crystalline rocks are about 2 orders of magnitude smaller when compared to clay-rich formations in sedimentary basins, the study of pore waters is more demanding. The chemical characterisation of very small volumes of water results in relatively large errors, and artefacts such as contamination from fluid inclusions that were decrepitated during sample preparation may play a role. Nevertheless, there has recently been encouraging progress in developing methods to obtain constraints on pore-water chemistry in crystalline rocks (e.g. Waber and Smellie, 2008; Waber et al., 2007, 2009). Results and conclusions from these studies were presented above, and further developments are ongoing.

8.5.3 Microstructure of argillaceous rocks The microstructure of clays and shales determines several geochemical, hydraulic and geomechanical properties of the rock on a macroscopic scale. In order to obtain a better process understanding, a link between macroscopic observations and the microscopic pore structure (in the range of nm to μm) is needed. Current information on the pore structure in clays and shales is only indirect (Hg injection, ad-/desorption isotherms, etc.). With the development of new microscopic techniques, such as focused ion-beam (FIB) nanotomography, high-resolution transmission electron microscopy (TEM) and related techniques (including cryogenic systems to preserve the natural water content), more direct geometric information is becoming available at the pore scale of clays and shales. These techniques can potentially visualise at least the larger pores in shales directly.

© Woodhead Publishing Limited, 2010

252

Geological repository systems for safe disposal

8.6

References

Andersson J, Hermansson J, Elert M, Gylling B, Moreno L and Selroos J O (1998), ‘Derivation and treatment of the flow wetted surface and other geosphere parameters in the transport models FARF31 and COMP23 for use in safety assesment’, SKB Report R-98-60, SKB, Stockholm, Sweden, Available at www.skb.se. Andra (2005), ‘Dossier 2005 Argile’, Andra, Chaˆtenay-Malabry, France, Available at www.andra.fr. Andre´ M, Neretnieks I and Malmstro¨m M E (2008), ‘Measuring sorption coefficients and BET surface areas on intact drillcore and crushed granite samples’, Radiochimica Acta, 96, 673–677. Beaucaire C, Michelot J L, Savoye S and Cabrera J (2008), ‘Groundwater characterisation and modelling of rock-water interaction in an argillaceous formation (Tournemire, France)’, Applied Geochemistry, 23, 2182–2197. Bock H, Dehandschutter B, Martin C D, Mazurek M, de Haller A, Skoczylas F and Davy C (2010), ‘Self-sealing of fractures in argillaceous formations in context with the geological disposal of radioactive waste’, OECD/NEA Report, OECD Nuclear Energy Agency, Paris, France, ISBN 978-92-64-99095-1. Boisson J Y (2005), ‘Clay Club Catalogue of characteristics of argillaceous rocks’, OECD/NEA Report 4436, OECD Nuclear Energy Agency, Paris, France, Available at www.oecdbookshop.org. Crawford J, Moreno L and Neretnieks I (2003), ‘Determination of the flow-wetted surface in fractured media’, Journal of Contaminant Hydrology 61, 361–369. CRC (2004), Handbook of Chemistry and Physics, 85th edition, CRC Press, West Palm Beach, Florida. Degueldre C, Grauer R, Laube A, Oess A and Silby H (1996), ‘Colloid properties in granitic groundwater systems. II: Stability and transport study’, Applied Geochemistry, 11, 697–710. Eichinger F (2009), ‘Matrix pore water–fracture ground water interaction in crystalline bedrock based on natural tracers: an archive of long-term hydrogeological evolution’, PhD Thesis, University of Bern, Switzerland. Eichinger F, Ru¨edi J, Waber H N and Mo¨ri A. (2008), ‘Characterisation of matrix pore water – diffusion profile in GTS rock matrix’, Nagra Report NPB 08-16, Nagra, Wettingen, Switzerland. ENSI (2009), ‘Spezifische Auslegungsgrundsa¨tze fu¨r geologische Tiefenlager und Anforderungen an den Sicherheitsnachweis. Richtlinie fu¨r die schweizerischen Kernanlagen ENSI G-03’, Eidgeno¨ssisches Nuklearsicherheitsinspektorat ENSI, Villigen, Switzerland, Available at www.ensi.ch. Frick U (1994), ‘The Grimsel radionuclide migration experiment – a contribution to raising confidence in the validity of solute transport models used in performance assessment’, in NEA/SKI Symposium Proceedings GEOVAL ’94 – Validation through Model Testing, Paris, France, 11–14 October 1994, OECD Documents, Nuclear Energy Agency, pp. 245–272. Frieg B, Alexander W R, Dollinger H, Bu¨hler C, Haag P, Mo¨ri A and Ota K (1998), ‘In situ resin impregnation for investigating radionuclide retardation in fractured repository host rocks’, Journal of Contaminant Hydrology, 35, 115– 130.

© Woodhead Publishing Limited, 2010

Far-field process analysis and radionuclide transport modelling

253

Gaucher E C, Tournassat C, Pearson F J, Blanc P, Crouzet C, Lereuge C and Altmann S (2009), ‘A robust model for pore-water chemistry of clayrock’, Geochimica Cosmochimica Acta, 73, 6470–6487. Gautschi A (2001), ‘Hydrogeology of a fractured shale (Opalinus Clay): implications for the deep disposal of radioactive wastes’, Hydrogeology Journal, 9, 97–107. Geckeis H, Scha¨fer T, Hauser W, Rabung T, Missana T, Degueldre C, Mo¨ri A, Eikenberg J, Fierz T and Alexander W R (2004), ‘Results of the colloid and radionuclide retention experiment (CRR) at the Grimsel Test Site (GTS), Switzerland – impact of reaction kinetics and speciation on radionuclide migration’, Radiochimica Acta 92, 765–774. Gimmi T, Waber H N, Gautschi A and Ru¨bel A (2007), ‘Stable water isotopes in pore water of Jurassic argillaceous rocks as tracers for solute transport over large spatial and temporal scales’, Water Resources Research, 43, W04410, doi:10.1029/2005WR004774. Heer W and Hadermann J (1996), ‘Grimsel Test Site – modelling radionuclide migration field experiments’, Nagra Technical Report NTB 94-18, Nagra, Wettingen, Switzerland, Available at www.nagra.ch. Hekel U (1994), ‘Hydrogeologische Erkundung toniger Festgesteine am Beispiel des Opalinustons (Unteres Aalenium)’, PhD thesis (Tu¨binger geowissenschaftliche Arbeiten TGA C18), University of Tu¨bingen, Germany. Li Y H and Gregory S (1974), ‘Diffusion of ions in sea water and in deep sediments’, Geochimica Cosmochimica Acta, 38, 703–714. Liu J, Lo¨fgren M and Neretnieks I (2006), ‘SR-Can – data and uncertainty assessment: matrix diffusivity and porosity in situ’, SKB Report R-06-111, SKB, Stockholm, Sweden, Available at www.skb.se. Lo¨fgren M (2004), ‘Diffusive properties of granitic rock as measured by in-situ electrical methods’, PhD Thesis, Department of Chemical Engineering and Technology, Royal Institute of Technology, Stockholm, Sweden. Lo¨fgren M (2007), ‘Formation factor logging in situ by electrical methods in KFM01D and KFM08C. Forsmark site investigation’, SKB Report P-07-138, SKB, Stockholm, Sweden, Available at www.skb.se. Lo¨fgren M and Neretnieks I (2003), ‘Formation factor logging by electrical methods – comparison of formation factor logs obtained in situ and in the laboratory’, Journal of Contaminant Hydrology 61, 107–115. Marschall P, Croise´ J, Schlickenrieder L, Boisson J Y, Vogel P and Yamamoto S (2004), ‘Synthesis of hydrogeological investigations at the Mont Terri site (Phases 1 to 5)’, in Mont Terri Project – Hydrogeological Synthesis, Osmotic Flow, edited by P Heitzmann, Federal Office for Water and Geology Report 6, Bern, Switzerland, pp. 7–94, Available at www.bafu.admin.ch/publikationen. Marschall P, Horseman S and Gimmi T (2005), ‘Characterisation of gas transport properties of the Opalinus Clay, a potential host rock formation for radioactive waste disposal’, Oil and Gas Science and Technology, Rev. IFP 60, 121–139. Mazurek M, Pearson F J, Volckaert G and Bock H (2003), ‘FEPCAT Project: features, events and processes evaluation catalogue for argillaceous media’, NEA 4437, Nuclear Energy Agency, Paris, France, Available at www. oecdbookshop.org. Mazurek M, Alt-Epping P, Bath A, Gimmi T and Waber H N (2009), ‘Natural tracer

© Woodhead Publishing Limited, 2010

254

Geological repository systems for safe disposal

profiles across argillaceous formations: the CLAYTRAC Project’, OECD/ NEA Report 6253, OECD Nuclear Energy Agency, Paris, France, 358 pp., Available at www.oecdbookshop.org. Miller B and Marcos N (2007), ‘Process report – FEPs and scenarios for a spent fuel repository at Olkiluoto’, Posiva Report 2007-12, Posiva, Olkiluoto, Finland, Available at www.posiva.fi. Missana T, Alonso U and Turrero M J (2003), ‘Generation and stability of bentonite colloids at the bentonite/granite interface of a deep geological radioactive waste repository’, Journal of Contaminant Hydrology 61, 17–31. Missana T, Garcia-Guttie´rez M and Alonso U (2004), ‘Kinetics and irreversibility of cesium and uranium sorption onto bentonite colloids in a deep granitic environment’, Applied Clay Science 26, 137–150. Mo¨ri A (2009), ‘In situ matrix diffusion in crystalline rocks – an experimental approach’, PhD thesis, University of Bern, Switzerland. Mo¨ri A, Mazurek M, Adler M, Schild M, Siegesmund S, Vollbrecht A, Ota K, Ando T, Alexander W R, Smith P A, Haag P and Bu¨hler C (2003), ‘The Nagra–JNC in situ study of safety relevant radionuclide retardation in fractured crystalline rock. IV: The in situ study of matrix porosity in the vicinity of a water conducting fracture’, Nagra Technical Report NTB 00-08, Nagra, Wettingen, Switzerland, Available at www.nagra.ch. Mo¨ri A, Soler J, Ota K and Havlova V (2007), ‘Predictive modelling for the LTD monopole experiment’, Nagra Report NAB 07-42, Nagra, Wettingen, Switzerland. Nagra (2002a), ‘Projekt Opalinuston – Synthese der geowissenschaftlichen Untersuchungsergebnisse. Entsorgungsnachweis fu¨r abgebrannte Brennelemente, verglaste hochaktive sowie langlebige mittelaktive Abfa¨lle’, Nagra Technical Report NTB 02-03, Nagra, Wettingen, Switzerland, Available at www.nagra.ch. Nagra (2002b), ‘Project Opalinus Clay – safety report. Demonstration of disposal feasibility for spent fuel, vitrified high-level waste and long-lived intermediatelevel waste (Entsorgungsnachweis)’, Nagra Technical Report NTB 02-05, Nagra, Wettingen, Switzerland, Available at www.nagra.ch. Nagra (2002c), ‘Project Opalinus Clay – models, codes and data for safety assessment. Demonstration of disposal feasibility for spent fuel, vitrified high-level waste and long-lived intermediate-level waste (Entsorgungsnachweis)’, Nagra Technical Report NTB 02-06, Nagra, Wettingen, Switzerland, Available at www.nagra.ch. Nagra (2002d), ‘Project Opalinus Clay – FEP management for safety assessment. Demonstration of disposal feasibility for spent fuel, vitrified high-level waste and long-lived intermediate-level waste (Entsorgungsnachweis)’, Nagra Technical Report NTB 02-23, Nagra, Wettingen, Switzerland, Available at www.nagra.ch. Nagra (2004), ‘Effects of post-disposal gas generation in a repository for spent fuel, high-level waste and long-lived intermediate level waste sited in Opalinus Clay’, Nagra Technical Report NTB 04-06, Nagra, Wettingen, Switzerland, Available at www.nagra.ch. NEA (2000), ‘Features, events and processes (FEPs) for geologic disposal of

© Woodhead Publishing Limited, 2010

Far-field process analysis and radionuclide transport modelling

255

radioactive waste – an international database’, Nuclear Energy Agency, Paris, France, Available at www.oecdbookshop.org. NEA (2004), ‘Geological disposal: building confidence using multiple lines of evidence’, NEA 4309, Nuclear Energy Agency, Paris, France, Available at www.oecdbookshop.org. NEA (2009), ‘Considering timescales in the post-closure safety of geological disposal of radioactive waste’, NEA 6424, Nuclear Energy Agency, Paris, France, Available at www.oecdbookshop.org. Neretnieks I (1980), ‘Diffusion in the rock matrix: an important factor in radionuclide retardation?’, Journal of Geophysical Research, 85, 4379–4397. Neretnieks I (2002a), ‘A stochastic multi-channel model for solute transport – analysis of tracer tests in fractured rock’, Journal of Contaminant Hydrology, 55, 175–211. Neretnieks I (2002b), ‘Matrix diffusion’, in ‘Radionuclide retention in geologic media’, Nuclear Energy Agency Report, Paris, France, Available at www. oecdbookshop.org. Norton D and Knapp R (1977), ‘Transport phenomena in hydrothermal systems: the nature of porosity’, American Journal of Science, 277, 913–936. Nykyri M, Nordman H, Marcos N, Lo¨fman J, Poteri A and Hautoja¨rvi A (2008), ‘Radionuclide release and transport – RNT 2008’, Posiva Report 2008-06, Posiva, Olkiluoto, Finland, Available at www.posiva.fi. Ondraf/Niras (2001), ‘SAFIR 2, Safety Assessment and Feasibility Interim Report 2’, Ondraf/Niras report NIROND 2001-06 E, Brussels, Belgium, Available at www.nirond.be. Ota K, Mo¨ri A, Alexander W R, Frieg B and Schild M (2003), ‘Influence of the mode of matrix porosity determination on matrix diffusion calculations’, Journal of Contaminant Hydrology 61, 131–145. Patriarche D, Ledoux E, Michelot J L, Simon-Coinc¸on R and Savoye S (2004a), ‘Diffusion as the main process for mass transport in very low water content argilites: 2. Fluid flow and mass transport modeling’, Water Resources Research, 40, W01517, doi:10.1029/2003WR002700. Patriarche D, Michelot J L, Ledoux E and Savoye S (2004b), ‘Diffusion as the main process for mass transport in very low water content argillites: 1. Chloride as a natural tracer for mass transport – diffusion coefficient and concentration measurements in interstitial water’, Water Resources Research, 40, W01516, doi:10.1029/2003WR002600. Pearson F J, Arcos D, Bath A, Boisson J Y, Fernandez A M, Ga¨bler H E, Gaucher E, Gautschi A, Griffault L, Hernan P and Waber H N (2003), ‘Mont Terri project – geochemistry of water in the Opalinus Clay formation at the Mont Terri Rock Laboratory’, Federal Office for Water and Geology Report 5, Bern, Switzerland, Available at www.bafu.admin.ch/publikationen. Posiva (2006), ‘Expected evolution of a spent nuclear fuel repository at Olkiluoto’, Posiva Report 2006-05, Posiva, Olkiluoto, Finland, Available at www.posiva. fi. Posiva (2008), ‘Safety case plan 2008’, Posiva Report 2008-05, Posiva, Olkiluoto, Finland, Available at www.posiva.fi. Rodwell W R (ed.) (2000), ‘Research into gas generation and migration in

© Woodhead Publishing Limited, 2010

256

Geological repository systems for safe disposal

radioactive waste repository systems’, European Commission Report EUR 19133EN. Ru¨bel A P, Sonntag C, Lippmann J, Pearson F J and Gautschi A (2002), ‘Solute transport in formations of very low permeability: profiles of stable isotope and dissolved noble gas contents of pore water in the Opalinus Clay, Mont Terri, Switzerland’, Geochimica Cosmochimica Acta, 66, 1311–1321. Schild M, Siegesmund S, Vollbrecht A and Mazurek M (2001), ‘Characterization of granite matrix porosity and pore-space geometry by in situ and laboratory methods’, Geophysics Journal, International, 146, 111–126. SKB (2006a), ‘Long-term safety for KBS-3 repositories at Forsmark and Laxemar – a first evaluation. Main report of the SR-Can project’, SKB Technical Report TR-06-09, SKB, Stockholm, Sweden, Available at www.skb.se. SKB (2006b), ‘Geosphere process report for the safety assessment SR-Can’, SKB Technical Report TR-06-19, SKB, Stockholm, Sweden, Available at www.skb. se. SKB (2006c), ‘Data report for the safety assessment SR-Can’, SKB Technical Report TR-06-25, SKB, Stockholm, Sweden, Available at www.skb.se. SKB (2008a), ‘Site description of Forsmark at completion of the site investigation phase’, SKB Technical Report TR-08-05, SKB, Stockholm, Sweden, Available at www.skb.se. SKB (2008b), ‘A¨spo¨ hard rock laboratory – annual report 2007’, SKB Technical Report TR-08-10, SKB, Stockholm, Sweden, Available at www.skb.se. Smith P, Neall F, Snellman M, Pastina B, Nordman H, Johnson L and Hjerpe T (2007), ‘Safety assessment for a KBS-3H spent nuclear fuel repository at Olkiluoto – summary report’, Posiva Report 2007-06, Posiva, Olkiluoto, Finland, Available at www.posiva.fi. Stroes-Gascoyne S (2002), ‘Assessment of the likelohood of significant microbial activity in Opalinus Clay’, Nagra Internal Report, Nagra, Wettingen, Switzerland. STUK (2001), ‘Long-term safety of disposal of spent nuclear fuel’, Radiation and Nuclear Safety Authority, Helsinki, Finland, Available at www.stuk.fi. Van Loon L.R, Soler J M, Jakob A and Bradbury M H (2003), ‘Effect of confining pressure on the diffusion of HTO, 36Cl and 125I in a layered argillaceous rock (Opalinus Clay): diffusion perpendicular to the fabric’, Applied Geochemistry, 18, 1653–1662. Van Loon L R, Wersin P, Soler J M, Eikenberg J, Gimmi T, Hernan P, Dewonck S and Savoye S (2004), ‘In-situ diffusion of HTO, 22Na+, Cs+ and I in Opalinus Clay at the Mont Terri underground rock laboratory’, Radiochimica Acta, 92, 757–763. Van Loon L R, Baeyens B and Bradbury M H (2005), ‘Diffusion and retention of sodium and strontium in Opalinus Clay: comparison of sorption data from diffusion and batch sorption measurements, and geochemical calculations’, Applied Geochemistry, 20, 2351–2363. Vilks P, Cramer J J, Jensen M, Miller N H, Miller H G and Stanchell F W (2003), ‘In situ diffusion experiment in granite: Phase I’, Journal of Contaminant Hydrology, 61, 191–202. Vilks P, Miller N H and Stanchell F W (2004), ‘Phase II in-situ diffusion experiment’,

© Woodhead Publishing Limited, 2010

Far-field process analysis and radionuclide transport modelling

257

Atomic Energy of Canada Limited (AECL) Report 06819-REP-01200-10128R00. Vinsot A, Mettler S and Wechner S (2008), ‘In situ characterization of the CallovoOxfordian pore water composition’, Physical Chemistry of the Earth, 33, S75– S86. Voegelin A and Kretschmar R (2002), ‘Stability and mobility of colloids in Opalinus Clay’, Nagra Technical Report NTB 02-14, Nagra, Wettingen, Switzerland, Available at www.nagra.ch. Waber H N and Smellie J A T (2008), ‘Characterisation of pore water in crystalline rocks’, Applied Geochemistry, 23, 1834–1861. Waber H N, Gimmi T and Smellie J A T (2007), ‘Large-scale matrix diffusion in crystalline rocks revealed by natural Cl, δ18O and δ2H tracers in pore water’, Geochimica Cosmochimica Acta, 71, A1076. Waber H N, Gimmi T, Smellie J A T and De Haller A (2009), ‘Pore water in the rock matrix – site descriptive modelling, SDM-Site Laxemar’, SKB Report R-08112, SKB, Stockholm, Sweden, Available at www.skb.se. Wersin P, Van Loon L R, Soler J M, Yllera A, Eikenberg J, Gimmi T, Hernan P and Boisson J-Y (2004), ‘Long-term diffusion experiment at Mont Terri: first results from field and laboratory data’, Applied Clay Science 26, 123–135. Wersin P, Soler J M, Van Loon L, Eikenberg J, Baeyens B, Grolimund D, Gimmi T and Dewonck S (2008), ‘Diffusion of HTO, Br, I, Cs+, 85Sr2+ and 60Co2+ in a clay formation: results and modelling from an in situ experiment in Opalinus Clay’, Applied Geochemistry, 23, 678–691. Witherspoon P A and Bodvarsson G S (eds) (2006), ‘Geological challenges in radioactive waste isolation – fourth worldwide review’, LBNL-59808, Earth Sciences Division, Ernest Orlando Lawrence Berkeley National Laboratory, University of California, Berkeley, California.

© Woodhead Publishing Limited, 2010

9 Immobilisation of spent nuclear fuel and highlevel radioactive waste for safe disposal in geological repository systems E. R. VANCE and B. D. BEGG, Australian Nuclear Science and Technology Organisation, Australia

Abstract: A brief history of the technical development of immobilisation strategies for high-level nuclear wastes (HLWs) is given. The desirable performance characteristics of the waste-immobilising matrices (waste forms) are outlined. The pros and cons of different classes of waste forms, as well as spent fuel itself, are discussed, together with common production methods. While borosilicate glass is the baseline waste form to deal with the majority of HLW, ceramic and glass–ceramic waste forms can offer performance and economic benefits for the immobilisation of HLWs that are difficult to incorporate in borosilicate glass, due to limited solid solubility of HLW ions or the presence of volatile species. Key words: borosilicate glass, ceramics, glass–ceramics, spent fuel, inert matrix, radiation damage, aqueous dissolution, waste form, radioactive waste, HLW, natural analogues.

9.1

Generation of high-level waste from nuclear fuel

The first nuclear reactor was built by Enrico Fermi’s team at Chicago University in 1942. Nuclear power was first utilised for weapons production in the US during World War II and stockpiling nuclear weapons by principally US and Russia has continued in parallel with the progressive development of peaceful applications of nuclear power for commercial electricity production in many countries after the war. Nuclear power derives from the neutron-induced fission of 235U or certain (reactor-produced) transuranic nuclides such as 239Pu. Each fission event 261 © Woodhead Publishing Limited, 2010

262

Geological repository systems for safe disposal

produces nearly 200 MeV of energy, manifested as kinetic energy of two fission product (FP) nuclei of unequal atomic numbers and masses, plus several fast neutrons and gamma rays. If the neutrons are suitably moderated by slowing them down without them being captured by other nuclei, a critical mass and concentration of fissionable elements will give rise to a controlled chain reaction and produce controlled power (as distinct from an atomic bomb). Many of the FPs are highly radioactive. Figure 9.1 shows the relative distribution of the FP abundances. The buildup of actinides (transuranic elements) derived from successive neutron capture reactions depends non-linearly on the total burnup of the fuel. Table 9.1 indicates the half-lives of some of the key radioactive FPs and actinides. Only a few percent of the fissionable nuclei in nuclear fuel are actually fissioned during its useful life. This is because some of the FPs are strong absorbers of neutrons and inhibit the chain reaction, with the result that the fuel is no longer capable of producing significant amounts of nuclear energy. Consequently, there are significant advantages, from a resource recovery perspective, to reprocess the used (or spent) fuel chemically to separate out the waste FPs so that the uranium (plus the Pu produced) could be recycled

9.1 Fission yield versus mass number for uranium-235.

© Woodhead Publishing Limited, 2010

Immobilisation of spent nuclear fuel and high-level radioactive waste Table 9.1

263

Half-lives of key fission product radionuclides Isotope

Half-life (year)

Fission products

90

Sr Nb 129 I

30 2.4 6 104 1.6 6 107

Actinides

235

7 6 108 87 6.5 6 103 3 6 106

94

U Pu 240 Pu 237 Np 238

Isotope

Half-life (year)

93

Zr Tc 137 Cs

1.5 6 106 2 6 105 30

238

4 6 109 2.4 6 104 14 18

99

U Pu 241 Pu 244 Cm 239

to make more fuel. However, in the mid-1970s, the future of reprocessing for nuclear power plant fuel was thrown into doubt with cheap uranium being widely available and the US decision (unlike France) not to reprocess commercial reactor fuel, in spite of its history in reprocessing defence reactor fuel. Today many countries (including the US) are re-evaluating the role of reprocessing in their nation’s nuclear fuel cycle, although proliferation concerns are weighing heavily.

9.1.1 Types of radioactive waste There are various classifications of radioactive waste: short- and long-lived low-level waste, intermediate-level waste and high-level nuclear waste (HLW) are common, but not universal, designations. The most dangerous waste of large-scale nuclear origin is HLW, arising from reprocessing of spent fuel or the spent fuel itself. The current definition of HLW relates to its method of production and so HLW cannot be converted to low-level waste by dilution. Thus it is forbidden to dispose of it in the oceans, for instance. Activities of this kind of waste are typically 1000 Ci/l [Ci = curie, the activity of 1 gm of 226Ra, 3.761010 disintegrations/second (becquerels, or Bq)]. Other than spent fuel or reprocessing waste from power plant fuel, HLW exists in other forms. In military applications, i.e. Pu production, burnups are relatively small (otherwise the Pu is converted to higher actinides) and the wastes are often strongly diluted by process chemicals used in the chemical extraction of the Pu. Over the years, a variety of methods have been used to separate out Pu for military use, increasing the general variability of HLW from a chemical point of view. Typically the radioactivity per unit volume of military wastes is only ~ 0.1–1% of that of reprocessed nuclear power plant fuel. The descriptive definition of HLW has not been without its problems, with a number of legal challenges mounted in the US to have certain wastes currently defined as HLW, but derived from processes incidental to reprocessing, to be reclassified. A further difficulty associated with using a

© Woodhead Publishing Limited, 2010

264

Geological repository systems for safe disposal

descriptive source-based definition is that wastes that contain similarly hazardous radionuclides of concern, but derived from different sources, are often classified differently and consequently subject to different treatment standards. A strong argument exists to move away from a subjective sourcebased description of HLW to a quantitative risk-based definition. This would not only avoid many of these difficulties but more importantly ensure that components-of-concern, from a risk perspective, are treated equivalently independently of how they may have been produced. Other types of waste are low-level waste from reactor operations, decontamination of radioactive samples, hospital wastes, spent sources, mining operations, etc., which typically constitute the greatest volume of radioactive waste. Intermediate-level wastes are also recognised in some countries. The International Atomic Energy Agency is a good source of formal definitions of radioactive wastes around the world.1 Nuclear waste becomes less radioactive the longer it is stored because of radioactive decay, but some of its radioactivity persists for millions of years. Figure 9.2 shows the approximate time dependence of the activity from commercial reprocessing waste. Military wastes follow approximately the

9.2 Relative activities of fission products and transuranic elements in commercial power plant nuclear fuel after removal from reactor.

© Woodhead Publishing Limited, 2010

Immobilisation of spent nuclear fuel and high-level radioactive waste

265

same time dependence but, as mentioned above, activities are considerably less than those derived from commercial nuclear fuel. Note that these wastes are chemically and radiologically very diverse in nature. Since the activity of the waste falls with increasing time, it is technically advantageous to store as long as possible. However, the method of storage is critical. For instance, a strong initial driver of HLW cleanup in the US was that at the Hanford reservation in Washington state, the stainless steel tanks containing the old military wastes were leaking into the surrounding environment. Moreover, tanks in which the water had largely evaporated over the years of storage because of radiogenic heating gave gas evolution in the form of periodic large hydrogen bubbles which had safety implications via potential radionuclide removal from the tank into the atmosphere and ground area adjacent to the tanks, as well as ignition and explosion. All countries generating nuclear power and/or nuclear weapons produce HLWs. However, the urgency of particular countries to dispose of the HLW is a strong function of the maturity of their nuclear program.

9.2

Historical waste form development for processing

It was soon realised that the highly radioactive spent power reactor fuels (whether reprocessed or not) would need careful management to prevent the spread of the unwanted radioactivity from fission products and transuranic elements into the public domain – the biosphere. The reference disposal scenario is a deep geological repository and we will discuss this strategy in a later section. The waste form is the primary immobilisation barrier that isolates the waste from the biosphere so it is a key plank in the whole immobilisation process. As the waste form is subject to laboratory investigation, its performance can be optimised and validated. Borosilicate glasses for the immobilisation of HLW from nuclear fuel reprocessing were developed by the US Atomic Energy Commission in the 1950s and were scaled up in the late 1960s to the full size dictated by the standard US disposal canisters, which are 3 m high60.61 m outer diameter. The scientific basis for their use was not generally articulated in detail during this era, but the concept was that FPs could almost all be incorporated in the glass structure. The glass could easily be produced in large quantities by melting at modest temperatures (~1100 oC) in a Joule-heated melter and selfradiation damage from the decay of the incorporated radionuclides had little effect on the major properties of the glasses. The glasses could accommodate ~20 wt% of FPs and actinides. However, Pennsylvania State University workers in the mid-1970s noted that glasses were fundamentally unstable from a thermodynamic point of view, and devised ceramic waste forms for reprocessing HLW, based on the known natural longevity of

© Woodhead Publishing Limited, 2010

266

Geological repository systems for safe disposal

crystalline silicate, phosphate and molybdate minerals. These so-called supercalcine ceramics2 were sintered in air at ~1100 oC and had very high loadings of FPs, typically 70 wt% of FP oxides, and the chemistry of the different phases was driven by the FPs as majority components. Typical phases were pollucite, CsAl2SiO6, powellite, CaMoO4, and rare earth apatites and phosphates (e.g. monazite, REPO4, where RE = trivalent rare earth). All of these had mineral analogues that were known to be very durable in the hot, wet conditions likely to characterise a deep geological repository for the waste. Following work at Sandia National Laboratory in the US on phase assemblages occurring on heating sol-gel titania particles on which simulated HLW FPs and actinides were sorbed, Ringwood and his co-workers in Australia in the late 1970s devised multi-phase titanate ceramics in which nearly all the FPs and actinides in HLW from nuclear fuel reprocessing were incorporated substitutionally in the various mineral analogue phases.3 Typical waste loadings were 20 wt% of HLW oxides and the production technology was slurry mixing of the waste and precursor oxides, calcination of the waste/precursor mixture in a reducing atmosphere, followed by hot uniaxial pressing at ~1100 oC to make a dense ceramic. These ceramics will be discussed in a little more detail in following sections. At about the same time, there was a worldwide surge in interest in this topic. However, in the US a key decision was made in 1981–1982 to use borosilicate glass to immobilise HLW at the Savannah River, South Carolina, US site and there was a substantial decrease in US funding for HLW waste form research from then on. Nevertheless, a variety of alternative waste form development work continued around the world and the book by Lutze and Ewing4 gives an excellent survey up to nearly the end of the 1980s. Candidate materials included glasses, ceramics, glass–ceramics, cermets, coated materials and cements. However, as time went on, it has been widely (but not universally) agreed that the only real remaining candidate classes of waste forms for HLW immobilisation are glasses, ceramics and glass–ceramics. A summary of the properties of these materials is given below: 1.

2.

Glass or vitreous waste forms. As noted above, these forms have been extensively studied and are well suited for reprocessed HLW that contains a large fraction of glass-forming components. Either Joule or cold-crucible melters can produce glasses. The presence of waste components that have low solubility in glass (e.g. Zr), are volatile (e.g. Cs, Tc, I), or impact glass melt properties such as viscosity or electrical properties (e.g. Al, CaF2, sulphate) can have a significant detrimental impact on maximum achievable waste loading. Ceramic waste forms. These provide enhanced chemical durability and higher waste loadings than glass for wastes that contain a small

© Woodhead Publishing Limited, 2010

Immobilisation of spent nuclear fuel and high-level radioactive waste

3.

267

proportion of glass-forming components. Ceramic waste forms are typically produced by sintering or hot-uniaxial/hot-isostatic pressing. Glass–ceramic waste forms. This is a composite waste form that in principle combines the advantages of both glass and ceramic waste forms. Ceramic phases are crystallised by design within a glass matrix. They are well suited for wastes that contain a significant proportion of glass-formers together with components that may have limited solubility or are otherwise problematic for straight glasses. Hotisostatic pressing or cold-crucible melting may be used for production.

Waste form development is still continuing in some shape or form in different nuclear countries, although Japan chose borosilicate glass in the mid-1990s, and therefore ceased work on alternatives except in some niche areas, such as immobilisation of 129I. France instituted the ‘law of 1991’, which placed a moratorium on waste disposal until 2006, giving them 15 years of research to make a decision on the best choices of waste forms for their particular wastes. It is not yet clear whether any decision has been made at this point. Spent nuclear fuel itself has also been studied in the waste form context since the late 1970s. Very little HLW has actually been processed to date. It is appropriate at this stage to reiterate the diverse nature of HLW, depending in part on whether it derives from commercial Purex reprocessing or military Pu production. Generally speaking, the latter wastes consist of a concentrated solution of salts plus a sludge of hydroxides, and are very inhomogeneous (and largely uncharacterised) even in single tanks. Hence there is a possible need to separate individual wastes into solution and sludge fractions and a definite need to design waste forms that can cope with diversity and compositional uncertainty. Moreover, it has to be recognised that different wastes need different technical solutions, both for the class and chemical design of the waste form and its mode of processing.

9.2.1 Design drivers for high-level nuclear waste forms Three key drivers that underpin the design and selection of waste forms for HLW are: .

.

Environmental. The waste form defines the radionuclide source term for release to the geological repository. Consequently, it is imperative that the waste form exhibits very high chemical durability in terms of resistance to aqueous leaching over the life of the waste in the geological repository, to minimise environmental risk. Economic. Multi-billion dollar repository disposal cost savings can be realised by minimising waste form disposal volumes. Dense monolithic waste forms, with high waste loadings, increase the proportion of waste

© Woodhead Publishing Limited, 2010

268

.

Geological repository systems for safe disposal that can be incorporated per unit volume of the waste form, and consequently reduce the repository space needed for disposal. A cubic metre of the proposed US Yucca Mountain, Nevada, repository (albeit not favored by the current US administration) has been estimated to cost between US$600 000 and US$1 000 000. Process flexibility. The waste form needs to be processed easily and reliably in a remote environment. It is important that the waste form properties are flexible and not unduly compromised due to mismatches of waste/additives ratios and variations of waste form chemistry – noting that HLWs are almost always inhomogeneous. Waste form flexibility may derive from the use of multiple phases and chemical buffering via the presence of a phase that does not include radionuclides – such that variations of chemical composition just result in a change of the proportions of the phases present, not the identities of the phases themselves.

Balancing the opposing environmental and economic drivers is crucial to implementing a successful geological disposal system. The key to optimising that balance lies in selecting the right class of waste form for a given HLW stream. No single class of waste form can economically handle the diverse range of HLWs. The governing regulatory framework plays an integral role in achieving this balance. Repository waste form acceptance criteria need to be transparent, quantitative and performance-based to enable waste characteristics to drive selection of the appropriate class of waste form. Indeed, the performance standard should reflect the level of environmental risk posed by waste in the context of the disposal environment. One unfortunate legacy of the historical development of waste forms for HLW has been the evolution of descriptive rather than performance-based regulatory waste form acceptance criteria. Descriptive acceptance criteria, by definition, predetermine the characteristics of the final waste form, irrespective of the waste in question. Quantitative performance criteria are impartial; they encourage broader evaluation of waste forms and ultimately the pursuit of the most appropriate class of waste form for a given waste stream.

9.2.2 Waste form durability One measure of a waste form’s performance is its resistance to aqueous dissolution. The effects of self-irradiation (see below), together with the characteristics of the proposed disposal environment, should be taken into account in these determinations. The US is the leader in waste performance acceptance criteria and since

© Woodhead Publishing Limited, 2010

Immobilisation of spent nuclear fuel and high-level radioactive waste

269

the late 1970s a battery of static and dynamic aqueous durability tests designed for glass waste forms was initiated at the Materials Characterization Center associated with the Pacific Northwest National Laboratory, Washington, USA. These tests typically utilise a low waste form surface area to leach solution volume ratio to gauge the inherent durability of the waste form in the absence of solubility limit constraints. The most popular of these is the so-called MCC-1 test.5 Here a polished cylindrical or cuboid sample of ~ 2 cm2 in area is immersed in ~ 20 ml of deionised water in a closed container and leached without agitation for a given time. The leach solution may be replaced at regular intervals to minimise saturation. A ‘satisfactory’ candidate glass will yield a normalised leach rate of <1 g/m2 per day for all elements in this test, where the concept of normalisation prevents an advantage being obtained for dilution, since the gross leach rate (LR) is divided by the concentration of the species of interest to give the normalised leach rate: LR = M/Act, where M = mass of species leached into solution, A = geometrical surface area, t = time and c = atomic concentration of species in the waste form. Normally speaking, for glasses, alkalis yield the highest leach rates, followed by B and alkaline earths and Si. For a waste form that passes this test, the leach solutions are quite dilute, so that there are relatively few saturation effects. In contrast the Product Consistency Test (PCT) takes the alternative approach and utilises a high waste form surface area to volume ratio, where dissolution behaviour is clearly impacted by solubility limits and saturation effects.6 As the name implies, this test was originally designed for investigating the consistency of the leachability of actual radioactive glass produced at Savannah River or West Valley in the US. More recently, this test has found favour within the regulatory framework and applied to testing candidate waste glasses as part of the waste acceptance criteria. The saturated nature of this test perhaps more accurately reflects a realistic repository leaching scenario, where the volume of leachate is small compared to the surface area of the waste form. Consequently, this test is sensitive to precipitation effects and the formation of passivating surface layers that may restrict further dissolution. In this test the glass is crushed and the 100–200 mesh fraction (75–150 μm) is selected. After rinsing in deionised water to remove fines, 10 ml of deionised water is reacted with 1 g of powder and the test is run for 7 days at 90 oC. An acceptable glass has been defined to have normalised releases of <16, 13 and 10 g/l for B, Na and Li respectively (compared to 100 g/l if all the sample is dissolved). This glass standard is referred to as environmental assessment, or EA, glass. It is significant, and rather perplexing, to note that despite the fact that this test has been incorporated into repository waste acceptance criteria, no release limits have been defined for target radionuclides.

© Woodhead Publishing Limited, 2010

270

Geological repository systems for safe disposal

Table 9.2 Approximate ionic concentrations in water at 1 km depth in the Canadian Shield Ion

g/l

Ion

g/l

Ion

g/l

Ca Na

15 5

SO4 Mg

0.2 0.1

HCO3 Cl

0.01 35

Numerous other tests exist in which, for instance, water at 200 oC is dripped on to a candidate waste form, and the effects of these tests are measured in terms of elemental extraction in conjunction with the formation of alteration products on the glass. All too often these tests are used to give short-term leaching behaviour in an unrealistic leachant, viz. deionised water. Waters in deep repositories are often quite saline (see Table 9.2). Of greater importance is the need for longer-term (periods of years) tests to be carried out to gain a mechanistic chemical understanding of the leaching behaviour as distinct from the raw numbers in the prescribed leach tests (see below). The focus of these studies should be on the behaviour of radionuclides of concern (as determined from a risk-based assessment) within the waste form in the context of the proposed disposal environment. The interface with the near field of the geological repository should not be overlooked in these studies. The existence or otherwise of natural analogues of the phases making up HLW forms is perhaps the most significant asset in determining its longterm aqueous durability. If natural minerals can be found to exist in a wet environment, then knowledge of the local geology can give information on the time of exposure to water, and measurements of trace quantities of natural radionuclides in the mineral (such as U, Th, K, Rb) and their daughter products can allow the age of the mineral to be determined. The impact of radiation damage on the chemical durability of the mineral phase in the host environment can also be assessed. In favourable circumstances, it can be determined that mineral analogue phases can last up to millions of years in wet environments – just what is needed for the manmade phases for the sequestration of HLW. Thus there is a powerful argument to use waste forms based on natural analogue minerals that have demonstrated their survival over geological time frames.

9.2.3 Radiation damage Radiation damage is an additional factor that needs to be evaluated in the context of waste form durability. The radionuclides to be immobilised in a HLW form include alpha, beta and gamma emitters. The most serious damage to the waste form derives from alpha-decay, in which the alpha particle displaces around 100 atoms in the solid during its ~ 20 μm traverse

© Woodhead Publishing Limited, 2010

Immobilisation of spent nuclear fuel and high-level radioactive waste

271

and the heavy alpha-recoil atom that displaces ~1500 atoms over its ~ 20 nm trajectory. Beta and gamma processes produce ionisation damage but very few atomic displacements. These effects have been amply demonstrated in natural minerals that contain small amounts of U and Th and have ages of many millions of years, and they have been essentially reproduced in experiments on synthetic materials doped with a few percent of short-lived actinides (238Pu or 244Cm, which have half-lives of 87 and 18 years respectively; see Table 9.1). Thus only radiation damage processes in waste forms hosting significant amounts of actinides, especially Pu and other transuranics, need serious consideration. The variety of radiation effects include a crystalline → amorphous transformation after hundreds or thousands of years, with an associated decrease of several percent in density (e.g. ~16% in zircon, ZrSiO4), the production of lattice defects in solids that do not undergo amorphism, formation of gas bubbles, potential for enhanced leaching and radiolytic effects in which radiolysis of the water leads to the production of species such as H2O2, which may responsible for enhanced leaching. These effects have been discussed at length by many authors. Careful work at PNNL by Strachan and his co-workers7 has shown that the factor of a ~100 increase in leachability, which was thought to accompany the crystalline–amorphous transformation in zirconolite and pyrochlore-structured CaAnTi2O7 (An = actinide), is actually an artefact due to radiolysis effects in teflon components in leach vessels. Although it has also long been argued that these kinds of radiation effects on glasses are relatively trivial, it needs to be remembered that the baseline leachabilities of glasses tend to be some orders of magnitude higher than those of crystalline waste forms. Another effect is ‘transmutation damage’, due to the ionic size and valence changes that may accompany alpha or beta emissions. Particular examples are Cs+→Ba2+ and Sr2+→Y3+→Zr4+, where the size decreases are ~ 20 and 30 % respectively for the full decay schemes. These effects have not been studied in much detail because of the intense radioactivity associated with waste forms containing several percent of the parent isotopes, but sympathetic valence changes in the matrix ions and/or the production of hole centres can help to mitigate the charge changes in these decay series.

9.3

Candidate waste forms and disposition schemes

9.3.1 Glasses: borosilicate Borosilicate glasses have been studied in the radioactive waste context as the baseline for HLW immobilisation for many years. These glasses often contain seven or eight cations, apart from the waste, to try to maximise

© Woodhead Publishing Limited, 2010

272

Geological repository systems for safe disposal

Table 9.3 Typical borosilicate glass compositions designed for HLW immobilisation Oxide (wt%)

R7T7 (France)

DWPF (US)

Pamela (UK)

Belgium

SiO2 B2O3 Li2O Na2O CaO TiO2 MgO Al2O3 ZnO ZrO2

54.9 16.9 2.4 11.9 4.9 – – 5.9 3.0 –

68.0 10.0 7.0 13.0 – – 1.0 – – 1.0

58.6 14.7 4.7 6.5 5.1 5.1 2.3 3.0 – –

68.5 15.0 5.4 11.2 – – – – – –

durability while still maintaining a reasonably low melting temperature (~ 1100 oC). Some examples around the world are shown in Table 9.3. These glasses are typically produced by Joule melters of several m2 in area and production rates can be several hundreds of kg/day per m2. The glass is poured into the disposal cans. The disadvantages of Joule melters is that they have finite lifetimes due to electrode and refractory corrosion and the temperatures cannot exceed ~1150 oC. Also, the footprints associated with the offgas systems are inevitably quite large and failed melters constitute large amounts of secondary radioactive waste. As noted previously, some ions present in waste have very limited solubility in borosilicate glass or are problematic to glass production. These include Ti, Zr, Cr, Mo, sulphate, etc., and the only way to deal with this problem within the glass scenario is to decrease the waste loading and pay the economic penalty. Glass viscosity as a function of temperature needs to be a relatively flat function to facilitate pouring. This viscosity dependence is particularly important if cold-crucible melting (CCM) is used rather than a Joule melter. In CCM an oxide charge is heated by a radiofrequency (rf) field, relying on the conductivity of the charge. If the charge is a poor electric conductor in the cold state, carbon or metal is added to induce the heating from the rf field. When a melt is attained, the carbon is oxidised to gas or the metal forms the oxide and becomes part of the melt. The advantages of CCM are that much higher temperatures can be obtained to produce a wider variety of glass formulations and the water-cooled coils allow the containment of the melt by solid powder in contact with the coils, avoiding refractory corrosion. While CCM can attain higher temperatures, the process is sensitively dependent on the resistivity, viscosity and thermal conductivity of the melt, and of course at higher temperatures volatility concerns are inflated.

© Woodhead Publishing Limited, 2010

Immobilisation of spent nuclear fuel and high-level radioactive waste

273

There has been 40 years of work on the science of dissolution in water of these glasses, with the important parameters being the temperature, pH, ionic constituents in the starting water and the surface area/volume ratio, which is important in regard to saturation effects of certain species. The alteration of the glass surface is studied in conjunction with ionic extraction from the glass into solution, ion exchange processes and the appearance of colloidal species in the aqueous medium. In broad terms, it is generally agreed that the most rapid process is ion exchange of alkalis with H3O+, which tends to elevate the pH, followed by dissolution of the silicate matrix, which is enhanced in alkaline media. Surface alteration products inhibit the access of the water to the unleached glass and slow the rate of attack.

9.3.2 Glasses: phosphate Lead phosphate glass was put forward by Oak Ridge National Laboratory researchers in the mid-1980s. Although the waste form properties were generally acceptable and the melting temperatures were low (~800 oC), refractory corrosion was a significant problem. Also the use of lead per se was seen as unattractive from the environmental point of view. The University of Missouri has worked on iron phosphate based glasses that have low leach rates and little tendency for refractory corrosion in Joule melters. Russian workers have used sodium aluminophosphate glass produced by the cold-crucible melter (CCM) technique for some of their HLW.

9.3.3 Ceramics: silicate, aluminate and phosphate Single-phase ceramics have been widely advocated for both single radioactive elements formed by partitioning of reprocessing wastes or for the entire complement of waste elements. Sodium zirconium phosphate (NZP) structures have been extensively studied and/or advocated for the full range of FPs and actinides. Monazite, apatite and zircon have been studied to immobilise actinides, while pollucite and CsAlSi5O12 have been investigated for Cs immobilisation. However, ‘single-phase waste forms’ lack chemical flexibility. Thus an exact match of waste and precursor stoichiometries in multi-cation hosts, such as those mentioned above, is industrially unrealistic. What is needed is an ‘extra’ minor durable phase whose abundance may vary as the waste/precursor ratio varies while still maintaining the same qualitative phase assemblage as in the synroc-type ceramics (see above). The sintered supercalcines consisted of apatite and monazite phosphates, powellites, feldspar, pollucite, etc. There are difficulties of diluting with alumina, silicate or phosphate to deal with radiogenic heat production

© Woodhead Publishing Limited, 2010

274

Geological repository systems for safe disposal

(apart from the very inelegant approach of using cold fission products as diluents). Volatility losses would be severe. This latter factor was realised by the Rockwell Science Center (RSC), who put forward hot isostatic pressing (HIP) as the preferred consolidation method.8 The RSC ceramic was developed in 1980 and directed at the wastes at the Savannah River Laboratory (see above). It contained magnetoplumbite (Ca(Al,Fe)12O19), UO2, spinel (Mg(Al,Fe)2O4) and corundum, with the former phase nominally containing all FPs other than gaseous species. A HIP consists of a pressure vessel surrounding an insulated resistanceheated furnace. The process of HIP radioactive wastes involves a stainless steel can being filled with the calcined feed. The can is evacuated and sealed, then placed into the HIP furnace and the vessel closed, heated and pressurised. Pressure is applied isostatically via argon gas, which, at pressure, is an efficient conductor of heat. The combined effect of heat and pressure consolidates and immobilises the waste into a dense monolithic block sealed within the can. The fact that the waste is sealed within a stainless steel HIP can offer further process advantages. No high temperature off-gas treatment is required for the HIP as there are zero emissions during the consolidation process, which significantly minimises secondary wastes. Note that a small off-gas system is required for the can bake-out stage. Process flexibility is maximised, as there is no need to control viscosity or pour the waste form. In addition, as there is no direct contact between the waste and HIP process vessel itself, subsequent decontamination costs are minimised. HIP is a mature industrial process used in production environments around the world. Many industrial HIP units have been in operation for over 30 years.

9.3.4 Ceramics: titanate Multi-phase titanate ceramics based on durable natural analogue phases were called ‘synroc’ (synthetic rock) by Ringwood and co-workers3 (see above). These theoretically dense materials are made by first mixing inactive precursors of Al, Ba, Ca, Ti and Zr oxides with liquid (simulated) HLW, drying and calcining in an H2/N2 atmosphere for 1 h at 750 oC. The calcine was then mixed with 2 wt% of powdered Ti metal for redox control and then subjected to uniaxial graphite die hot pressing or hot isostatic pressing at ~ 1100 oC. The precursor composition and the titanate phases in the early synroc-C titanate ceramic designed for reprocessed commercial power reactor wastes are given in Table 9.4. Since 1984, rather than using oxides, a slurry mixture of Ba and Ca hydroxides and transesterified Al, Ti and Zr alkoxides has been used as the precursor.3 This provides better solid-state reactivity than the powdered metal oxides and hydroxides. The principal

© Woodhead Publishing Limited, 2010

Immobilisation of spent nuclear fuel and high-level radioactive waste Table 9.4

275

Composition and phase abundances in synroc-C titanate ceramics

Phase

wt%

Radionuclides in lattice

Hollandite, Ba(Al,Ti)2Ti6O16 Zirconolite, CaZrTi2O7 Perovskite, CaTiO3 Ti oxides, mostly TiO2 Alloy phases

30 30 20 15 5

Cs, Rb RE*, An* Sr, RE*, An* Tc, Pd, Rh, Ru, etc.

* RE, An = rare earths and actinides respectively.

advantage of this synroc-C ceramic was that the waste ions were dilutely incorporated in durable titanate mineral phases that were considerably more insoluble in water than the silicates and phosphates, etc., used in supercalcine (see above). The waste loading could be varied between zero and 35 wt% using the same inert additive chemistry without substantially changing the basic zirconolite + perovskite + hollandite + rutile phase assemblage, although the percentages of the different phases varied somewhat. This flexibility is seen as a large advantage. There were minor alumina-rich phases in the more dilute formulations. The grain size is ~ 1 μm, to optimise mechanical properties and prevent subsequent radiation-induced microcracking. The alloy phases derive from elements that form metals under the reducing conditions prevailing during hot pressing. The leach rates at 90 oC in water from synroc-C of the most soluble elements, alkalis and alkaline earths, are typically < 0.1 g/m2 per day for the first few days, and they decrease asymptotically to values of ~ 105 g/m2 per day after 2000 days. Leach rates of other elements are much lower. Leach rates of 105 g/m2 per day correspond to a corrosion rate of ~ 1 nm/day. In the 1980s, the inactive synroc production process was scaled up via the ANSTO Synroc Demonstration Plant to produce ~ 50 kg of monoliths containing 20 wt% of simulated HLW, with properties as good as those of gram-sized laboratory samples. The ceramic could tolerate waste loadings up to 30% HLW (neglecting radiogenic heat effects), with no changes in the phases, just their abundances. In the early 1990s, the synroc ceramics were tailored towards the study of zirconolite-rich materials for immobilisation of actinide-rich wastes such as Pu or partitioned transuranic elements. The initial work during 1991–1994 was directed at the latter application in conjunction with the Japanese Atomic Energy Research Institute. There was a strong focus on radiation damage via the incorporation of the alpha emitter 244Cm (18 year half-life), as had been done with synroc-C and an Na-doped variant thereof.9,10 Perovskite was also studied for comparison. The work on surplus Pu immobilisation, with Lawrence Livermore National Laboratory (LLNL) as the lead laboratory for the US Department

© Woodhead Publishing Limited, 2010

276

Geological repository systems for safe disposal

of Energy (DOE), moved from zirconolite- to pyrochlore-rich ceramics during 1994–1997. This was because of solid solution limits in the first instance when the target of the work changed from immobilisation of 10 wt% Pu alone to the additional inclusion of 20 wt% U. The estimated time for amorphisation to be complete is on the order of 1000 years and the resultant volume expansion would be around 6%. In addition these ceramics incorporated an atom each of neutron-absorbing Gd and Hf for each atom of Pu to deal with potential criticality in the sample. Near-field aggregation of Pu due to leaching was shown to be not a problem from the criticality aspect either, because the leach rates of Pu were spanned by those of the neutron absorbers; hence any leached Pu would be accompanied by neutron absorbers. The final baseline (no impurities) version11 of the pyrochlore-rich ceramics chosen by the US DOE in 1998 contained 95 wt% of a pyrochlore-structured Ca0.89Gd0.23Hf0.23U0.44Pu0.22Ti2O7 phase plus 5 wt% of rutile-structured Ti0.9Hf0.1O2. The form of the ceramic was to be 76 mm diameter pellets weighing ~500 g; 560 such pellets were to be enclosed in a US standard canister of Savannah River DWPF glass to provide a radioactive barrier (gamma field) to prevent diversion. This product was the first crystalline material to be validated in the US. However, in early 2002, it was decided to remove the disposal option for US/Russian surplus Pu and to proceed only with a MOX fuel option for utilisation. Other synroc derivatives have been devised for immobilisation of Tc and Cs. The use of HIP in general has large advantages in that there are essentially no volatile losses during this hot consolidation step, because the waste form is contained in a sealed metal can, as noted above.

9.3.5 Glass–ceramics Glass–ceramics seek in principle to combine the advantages of glasses and ceramics, and are well suited to waste streams that contain a significant portion of glass formers, together with components that may be problematic for a straight glass. Sphene glass–ceramics were developed for ~ 6 years in Canada for a reprocessing option, until it was eventually decided in 1984 to follow the US and concentrate on spent fuel. The Canadian glass–ceramics consisted of sphene, CaTiSiO5, in a durable aluminosilicate matrix. Additional perovskite and other phases were observed at waste loadings of > 10 wt% FP oxides. In the 1970s and 1980s a variety of other glass–ceramics was studied and these have been reviewed by Hayward.12 The major crystalline phases in these silicate-based materials were celsian (BaAl2Si2O8), fresnoite (Ba2TiSi2O8), diopside (CaMgSi2O6) and calcium aluminosilicates. More recently, glass–ceramic waste forms have been shown to be particularly advantageous for HLW calcines stored at the Idaho National

© Woodhead Publishing Limited, 2010

Immobilisation of spent nuclear fuel and high-level radioactive waste

277

Laboratory, US. The Idaho HLW calcines are heterogeneous and rich in alumina, zirconia and calcium fluoride. For borosilicate glass these components are especially problematic and difficult to incorporate at significant waste loadings as they are either refractory with low solubility in glass (zirconia) or can have a dramatic impact on glass viscosity (pourability) and melter corrosion in the case of calcium fluoride. Glass– ceramic waste forms overcome the solubility limitations of glass by allowing the controlled crystallisation of components that enhance rather than detract from the chemical durability of the waste form. Process limitations are overcome by utilising HIP technology, which is insensitive to the viscosity and electrical conductivity of the waste form and is not susceptible to melter corrosion. Consequently, while waste loadings of only ~ 20–30 wt% are achievable with glass, glass–ceramic formulations can incorporate calcine at vastly higher waste loadings of up to 80 wt%. The corresponding repository disposal cost savings resulting from these higher waste loadings (reduction in immobilized volume) have been estimated by the US DOE to be ~ $5 billion. ANSTO have also developed glass–ceramics for immobilisation of Hanford tank wastes (Washington, USA) that are rich in alkali nitrates and transition metal hydroxides. The actinides were preferentially partitioned towards synroc phases, principally zirconolite, in a boroaluminate silicate glass matrix. These glass–ceramics had waste loadings of 50–70 wt% and leach rates were often 10–100 times lower than those for standard US EA glass, the baseline glass to pass the PCT leach test. Glass–ceramic waste forms have also been shown to be an excellent class of waste form for disposition of very impure plutonium residue wastes in the UK. ANSTO has demonstrated zirconolite-glass–ceramic waste forms where actinides preferentially partitioned into the crystalline zirconolite phase by factors of 100:1 compared to glass. The glass forming impurities were readily accommodated within the glass matrix, highlighting the benefits that composite glass–ceramic waste forms can provide for these very impure actinide wastes.

9.3.6 Spent fuel The power density of nuclear fuel determines the operating temperature of the fuel. Typically, in-reactor temperatures range from ~ 600 oC at the peripheries of the fuel pellets to up to ~2000 oC in the centre. The melting point of UO2 fuel depends somewhat on the oxygen stoichiometry, but is typically ~ 2800 oC. The intense heating in the centre of the fuel pellet can lead to void formation, extended grain growth and cracking. This temperature gradient is derived from the fissiogenic heating ( ~ 200 MeV/ fission) and the poor thermal conductivity (~ 2 W/m K) of the UO2.

© Woodhead Publishing Limited, 2010

278

Geological repository systems for safe disposal

Working from the outside of the fuel bundle, the FPs distribute themselves: (a) in the gap between the cladding and the fuel pellets, (b) in the grain boundaries and (c) in solid solution in the UO2. The higher the operating temperature, the greater the likelihood that those FPs that are insoluble in UO2 will migrate to the gap or the grain boundaries. Uranium oxide can accommodate some FPs in its crystalline lattice, but some ions are too small or large to fit into substitutional sites and interstitial sites do not lend themselves to strong binding. Actinides, rare earth and Zr FPs can enter into solid solution in UO2 via substitution for U4+. Cs, I, Mo, Tc, Pd-group elements, Ba, Xe and Kr are substantially insoluble in UO2. The Pd-group elements and Tc and Mo form metal alloys under the reducing conditions prevailing during reactor operation. Ba forms BaZrO3, Xe and Kr form gas bubbles in the fuel or diffuse into the gap, while much of the Cs and I form water-soluble compounds. From a disposition perspective, another important issue is the redox condition in which UO2 finds itself. UO2 is very stable under reducing conditions, but the water radiolysis caused by fission products and transuranics tends to yield oxidising conditions that enhance the dissolution of UO2. In summary, the behaviour of spent fuel in aqueous media is reasonably well understood, and it is clear that spent fuel needs to be contained in very durable metal cans such as copper or nickel alloys to be put into a geological repository. The decision to emplace spent fuel in a repository may have broader implications as it could be argued that colocated waste forms need only be as durable as spent fuel for inclusion in a repository. The aqueous durability of spent fuel is likely to represent the bounding case for release from the repository. This notion, while not articulated into any repository acceptance criteria, has led to what has been colloquially described as the ‘spent-fuel’ durability standard.

9.4

Inert matrix fuels

Recently there was a perceived worldwide surplus of actinides arising from nuclear power production, with the main concern being diversion and proliferation of Pu. Burning the Pu in reactors and generating electricity at the same time was seen as the preferred route. So-called inert matrix fuels (IMFs) that contain Pu are being studied for this purpose. With greenhouse gases being high on the political agendas of many countries, the Generation IV strategy of increased nuclear energy production via actinide fuels in advanced fast reactors is gaining favour. However, fast reactors will need considerable development and countries that have a serious interest in this direction will be stockpiling Pu for such future reactors rather than burning it in the short term. A recent review of IMF candidate materials has been given by

© Woodhead Publishing Limited, 2010

Immobilisation of spent nuclear fuel and high-level radioactive waste

279

Kleykamp.13 As for standard nuclear fuels, the desirable features of IMFs are high melting points, good thermal conductivity and resistance to radiation damage and swelling. Thus strong candidates are stabilised zirconia, spinel, MgO, Al2O3, CeO2, ZrN, Y2O3, B4C, Si3N4 and others. The IMF fuel can in principle be made in several ways: (a) as a Pu-bearing solid solution such as stabilised (Zr, Pu, Y)O2 or Pu in CeO2, ZrN, Y2O3, etc., (b) PuO2 finely dispersed in an inert matrix in which the Pu is essentially insoluble (Al2O3, B4C, Si3N4, MgO, etc.) or (c) Pu coarsely dispersed in an inert matrix. The use of coarse dispersions allows the radiation damage to be concentrated in the radiation-resistant fluorite-structured PuO2 phase with radiation damage to the matrix being restricted to the ~ 30 μm thick region (the FP range) around the PuO2 particles. Very little work has been carried out on FP disposition and immobilisation after irradiation, with the exception of work reported in 200314 by a Japanese group looking at a zirconia-based candidate and an ANSTO/ BNFL (British Nuclear Fuels Limited plc) group looking at synroc derivatives. Another relevant set of data was obtained by a Japanese group who used a close approximation to the Rockwell Science Center (RSC) magnetoplumbite-rich ceramic, which was designed for Savannah River plant waste in the early 1980s. Here the magnetoplumbite would nominally incorporate all the FPs (but note the earlier remarks highlighting restricted flexibility of single-phase waste forms), and the actinides would inhabit the fluorite-structured actinide dioxide phase. Future work on any of these materials will involve the use of expensive in-reactor studies to look at the likely swelling phenomena due to the FP gases Kr and Xe in the first instance, so it can only be said that this aspect of IMF development is still immature.

9.4.1 Cements and geopolymers At first glance, cement may appear to be a particularly attractive means of HLW consolidation as it avoids expensive high-temperature processing and the potential for volatile losses and a costly off-gas system. Work at the end of the 1970s led to some controversy in that if the BET-type surface area was used in leaching calculations, the derived leach rate was in the zone of 1 g/m2 per day or better for most FPs, apparently showing that cements were intrinsically leach resistant. However, in the end HLW candidate materials have to be able to be produced in bulk, and the geometrical surface area is more appropriate to describe the situation for which the waste form is designed. This leads to cements failing the standard leach tests described earlier by a factor on the order of 100 and they are not these days seen as serious candidates for HLW immobilisation. However, cementitious

© Woodhead Publishing Limited, 2010

280

Geological repository systems for safe disposal

materials still have a strong potential for low- and intermediate-level waste immobilisation. Geopolymers are made by reacting, at ambient temperatures, aluminosilicates such as metakaolin, fly ash or ground blast-furnace slags with alkaline solutions. The aluminosilicates partly dissolve in the solutions and polymerise and solidify. Curing is carried out at 40–90 oC to try to finalise the reaction. Samples that pass the PCT aqueous dissolution test (see above) can be fabricated by this technique, but the science of geopolymers is still not very mature, in spite of extensive studies by solid-state nuclear magnetic resonance being carried out over the years and many empirical leach tests being performed.

9.4.2 Transmutation of long-lived FPs to stable or short-lived species The idea of using nuclear reactors or particle accelerators to irradiate longlived (104–106 year) FPs and actinides and transmute them into stable or short-lived radioactive isotopes is at first sight a very attractive option. However, the practicalities are that, for both fast and thermal neutrons, cross-sections for neutron absorption are generally small, the desired isotope production obeys a (1  ekt) law, where t is the irradiation time, and neutron capture on the desired isotope may transform it into another radioactive species. Simply making a target to hold the parent isotope for the long necessary irradiation periods is also fraught with difficulty in terms of irradiation damage. A fairly recent OECD report15 concluded that the only viable isotopes for transmutation are 129I and 99Tc and that many years of exposure to intense neutron sources from accelerators would be required. The required targetry still remains to be developed. It is difficult to see how this technology will translate into short-term benefits and even its proponents argue that decreases in radioactive inventories would take hundreds of years to realise.

9.5

Geological disposal

Many ways have been suggested over the last 50 years for dealing with radioactive waste. These range from seabed dumping to sending the waste into space by rocketships. However, there is now international consensus that disposition of high-level waste, immobilised in a waste form and placed into an engineered geological repository is the most appropriate. Such a repository relies on an engineered multibarrier approach in which the location of the repository is optimised for rainfall, lack of seismic activity, water table, isolation from population centres and, more importantly, has

© Woodhead Publishing Limited, 2010

Immobilisation of spent nuclear fuel and high-level radioactive waste

281

9.3 Schematic diagram of a deep geological repository.

the general approval of the relevant political entities and the adjacent population. The multibarrier strategy involves some or all of the following elements: (a) disposal at 0.5–1 km below the earth’s surface in a geologically stable rock or clay formation; (b) the waste form, a solid, durable material that incorporates the waste; (c) the disposal canister which also serves as containment during the transport of the waste form from the production site to the repository; (d) an overpack of clay between the disposal canisters and the walls of the geological repository and (e) a cement and/or rock backfill to the repository rooms and drill hole. Figure 9.3 illustrates the standard kind of repository concept. Sweden and Finland have now decided on HLW repository sites. Canada has looked extensively at granite repositories, having had an underground research laboratory. Switzerland is considering both granite and clay, with France and Belgium focusing on clay. At the proposed Yucca Mountain site in the US, the repository is in the middle of the mountain, ~ 2 km below the mountain top and ~ 2 km above the water table. In Japan, the candidate repository is the Soho mine. In the context of the role of a geological repository, the Oklo natural reactor found in the Gabon area of Africa is a very interesting feature. Around 2000 million years ago, enough water was present around this UO2 deposit to cause intermittent criticality (the isotopic ratio of 235U/238U in those times was ~6%, much higher than the present day value of ~0.7%). Most of the residual FPs or their stable daughters can still be found adjacent to this geological phenomenon, showing that the natural environment can exert chemical controls on the spread of radioactivity. An alternative approach to a managed underground repository is retrievable storage in an above ground location. Here the basic idea is

© Woodhead Publishing Limited, 2010

282

Geological repository systems for safe disposal

that, with further research, future generations will come up with a better way of disposal than current methods. This concept, however, brings in the notion of intergenerational responsibility – how far should the present waste-producing generation delay proceedings that would save money now via an absence of action, but load future generations with the expense and responsibility of actually effecting waste disposal? Retrievable storage also allows the occupying power to extract fissile material from the waste to prosecute any military ambitions. It can be argued that the public perception of nuclear issues is still largely driven by the popular media, which tends to exaggerate the dangers of radioactivity. Of course chemical producers are not spared either in this respect. Thus there is a strong political element in the determination of the safety to the public of stored nuclear waste. However, the safety arguments are largely driven in the technical arena by nuclear (or chemical, for chemical producers) regulators and the starting point is the natural radioactive background, due to mainly U, Th and K in soil, cosmic radiation and fallout from nuclear bomb testing in the last 50odd years. As the natural radioactive background varies in population centres around the world by as much as a factor of ten, it would seem reasonable that the maximum allowed dose to population members would be some small fraction of the average natural background. Although arguments about the effects of small amounts of manmade radiation on the public still persist, there are internationally agreed standards for radiation exposure to members of the public as well as radiation workers (scientists whose research involves radioactivity). Hence the safety of geological repositories can be defined in terms of the dose received from the repository by future earth inhabitants. These doses of course require calculation and such calculations are extremely demanding as they involve the water input to the stored waste as a function of time in the first instance. Because of the complexity of the repository, this is inevitably a very complicated calculation. Calculation of the potential migration of the radionuclides to the biosphere is equally complex (see Fig. 9.3). Current methodology involves Monte Carlo probabilistic calculations at each identifiable stage of the transport of the radionuclides. No matter how good the repository appears to be, however, the initial rate-limiting factor for radionuclide release to the environment is still dependent on the chemical durability of the waste form, and it can be argued that the best way to reduce environmental risk is to optimise the waste form by careful design for the waste in question. For Yucca Mountain, Nevada, US, the expense of generating necessary experimental data on the geology of the repository together with that of the near and far field, and conducting these calculations, has run into billions of dollars, and

© Woodhead Publishing Limited, 2010

Immobilisation of spent nuclear fuel and high-level radioactive waste

283

debate about their essential correctness continues. Moreover, stakeholders in the US courts are still debating the legal issues involved in siting.

9.5.1 Impact of waste form research on future fuels and future trends in HLW management A key factor in any industrial process is the fate of the wastes produced. Therefore it is almost axiomatic that the design of future nuclear fuels will eventually include the ease of reprocessing, disposability of the waste FPs and higher actinides, as well as the search for independent usage of the ‘waste’ ions. We have indicated some of the scientific problems in the design of such fuel in the section on inert matrix fuels. A future trend in management of HLWs is basically the increasing acceptance around the world of partitioning schemes in which the HLW is separated into heatproducing isotopes such as 90Sr + 137Cs, minor actinides such as Am and Cm, etc., and other groups such as rare earths. The separations allow not only simpler waste form chemical design because of the reduction of the number of hazardous nuclides in each group, but also more flexibility in advanced fast reactor schemes in which the minor actinides can act as fuel for conversion into less dangerous FPs. Obviously the widespread constructive use of such waste radioactivity as industrial sources would be a considerable plus.

9.6

Conclusions

In spite of more than 40 years of work, the disposition of high-level nuclear fuel wastes around the world in future is still subject to many uncertainties. Much of the debate surrounds the question of how to validate physical models that lead to the calculated maximum radiation dose to persons living close to the repository, and more particularly how to convince a lay audience that the very complex calculations, including the uncertainties, are meaningful. However, the waste form is a key primary containment barrier because it can be subjected to rigorous experimental study and optimisation of its behaviour can be studied directly at least over a few years. In this respect, more analogue studies of natural minerals are needed, where although the water/thermal history of the analogue mineral itself may be hard to derive, the history may well be derivable from the surrounding minerals in the rock formation. It seems clear that waste form development for the large spectrum of chemically distinct HLWs already in existence for many years, plus those yet to be generated by ongoing and future nuclear power programmes, will continue. The twin foci of this continuance are simply balancing the environmental risk associated with the long-term

© Woodhead Publishing Limited, 2010

284

Geological repository systems for safe disposal

durability of the waste form in the disposal environment with the economic reality that increased waste loadings can ease the amount of space required to contain the waste forms in repositories and deliver billion dollar life-cycle cost savings. The key to optimising this balance lies with the establishment of clear performance-based regulatory waste form acceptance metrics that enable the characteristics of the HLW to drive the selection of the most appropriate class of waste form. No single class of waste form can economically handle the diverse range of HLW.

9.7

Acknowledgements

The authors wish to acknowledge their numerous colleagues at ANSTO and around the world for many discussions and contributions over many years.

9.8 1.

2. 3.

4. 5. 6. 7.

8.

9.

10.

References ‘Issues and trends in radioactive waste management’, in Proceedings of an International Conference, International Atomic Energy Agency, STI/PUB/ 1175, Vienna, Austria, 2003. McCarthy G J, ‘High-level waste ceramics: materials considerations and product characterization’, Nuclear Technology, 1977, 32, 92. Ringwood A E, Kesson S E, Reeve K D, Levins D M and Ramm E J, ‘Synroc’, in Radioactive Waste Forms for the Future, edited by W Lutze and R C Ewing, North-Holland, Amsterdam. Lutze W and Ewing R C (eds) Radioactive Waste Forms for the Future, NorthHolland, Amsterdam. ‘MCC-1 Static Test, Nuclear Waste Materials Handbook’, US Report No. DOE/TIC-11400, Material Characterization Centre, Hanford, 1984. ‘Determining chemical durability of nuclear hazardous and mixed waste glasses: the product consistency test (PCT)’, ASTM Designation: C1285-97. Strachan D M, Scheele R D, Kozelisky A E and Sell R L, ‘Effects of self irradiation from 238Pu on candidate ceramics for plutonium immobilization’, Pacific Northwest National Laboratory Report PNNL-14232, 2003. Morgan P D E, Clarke D R, Jantzen C M and Harker A B, ‘High-alumina tailored nuclear waste ceramics’, Journal of American Ceramic Society, 1981, 64, 249–258. Mitamura H, Matsumoto S, Hart K P, Miyazaki T, Vance E R, Tamura Y, Togashi S and White T J, ‘Aging effects on curium-doped titanate ceramic containing sodium-bearing high-level nuclear waste’, Journal of American Ceramic Society, 1992, 75, 392–400. Mitamura H, Matsumoto S, Stewart M W A, Tsuboi T, Hashimoto M, Vance E R, Hart K P, Togashi Y, Kanazawa H, Ball C J and White T J, ‘Alpha-decay damage effects in curium-doped titanate ceramic containing sodium-free highlevel nuclear waste’, Journal of American Ceramic Society, 1994, 77, 2255– 2264.

© Woodhead Publishing Limited, 2010

Immobilisation of spent nuclear fuel and high-level radioactive waste 11.

12. 13. 14. 15. 16.

285

Vance E R, Jostsons A, Moricca S, Stewart M W A, Day R A, Begg B D, Hambley M J, Hart K P and Ebbinghaus B B, ‘Synroc derivatives for excess weapons plutonium’, in Ceramics Transactions (Environmental Issues and Waste Management Technologies IV), vol. 93, edited by J C Marra and G T Chandler, American Ceramic Society, pp. 323–329, 1999. Hayward P J, ‘Glass-ceramics’, in Radioactive Waste Forms for the Future, edited by W Lutze and R C Ewing, North-Holland, Amsterdam. Kleykamp H, ‘Selection of materials as diluents for burning of plutonium fuels in nuclear reactors’, Journal of Nuclear Materials, 1999, 275, 1–11. IMF9 Workshop on Inert Matrix Fuel, Kendal, UK, 10–11 September 2003. ‘Accelerator-driven systems (ADS) and fast reactors (FR) in advanced nuclear fuel cycles, a comparative study’, OECD Report, 2002. Halsey W G, ‘AFCI repository impact evaluation report systems: analysis progress Report-FY03’, Lawrence Livermore National Laboratory Report UCRL-ID-155287, 12 September 2003.

© Woodhead Publishing Limited, 2010

10 Development and application of low-pH concretes for structural purposes in geological repository systems M . C . A L O N S O , J . L . G A R C I´ A C A L V O a n d A . H I D A L G O , Eduardo Torroja Institute for Construction Sciences, Spain; L . F E R N A´ N D E Z L U C O , INTECIN–Universidad de Buenos Aires, Argentina

Abstract: Several construction concepts of underground repositories for high-level waste (HLW) consider the use of low pH cementitious materials and different aspects related to their development are exposed in the present chapter. The general functional requirements and the special characteristics implied in the development and production of low-pH concretes are described, mainly focused on the design of a specific application: a low-pH concrete plug using the shotcrete technique. Considering the long life expected in HLW repositories, parameters related to the durability of the low-pH concretes are analysed, such as the interaction of low-pH cementitious materials with groundwaters and the risk of corrosion if reinforcements are needed. Key words: low-pH cements, hydration, applications, functional requirements, long-term performance.

10.1

Introduction

Concrete is the most widely used construction material, employed in many types of applications and environmental conditions, tunnels, bridges, tall buildings, dams, harbours, etc. Moreover, concrete-like materials are considered to be relevant components in several European concepts for the construction of underground repositories of high-level radioactive waste (HLW). Concrete is a composite material integrated by three main components: cement (binder), aggregates and water. When only fine aggregates (particle 286 © Woodhead Publishing Limited, 2010

Development and application of low-pH concretes

287

size below 4–5 mm) are used, it is known as mortar. When the three components are put in contact, the water interacts with the anhydrous cement particles to produce new hydraulic solid phases that finally take a rigid state, but results in a porous material. The wide use of hydraulic cementitious materials, such as concrete, is due to their technological properties: . .

During fluid phase the concrete may be adapted to any specific form; this allows for its application in a wide variety of situations. As the hydration evolves, the cement shows its cementing character, which guarantees the binding of the aggregates to produce the composite material (including the aggregates surrounded by the cement paste) with variable, though predictable, mechanical properties.

Ordinary Portland cement (OPC) is the most commonly used conventional cement. This cement contains mainly particles of calcium silicates (C3S and C2S) and calcium aluminates (C3A and C4AF). The cement paste of an OPC-based concrete also shows a composite structure itself, constituted by hydrated calcium silicates (CSH with a gel structure), hydrated calcium aluminates crystals (CAH) and portlandite crystals (CH) (Taylor, 1990). These hydrated solid components contain water in their structure, but free water is also present in a liquid state. Therefore, the hardened concrete is a porous material that confines a liquid solution (as shown in Fig. 10.1) whose origin is the excess of water employed during mixing to achieve a suitable consistency (Metha, 1986). During Portland cement hydration, substantial amounts of alkalis (sodium and potassium) are released to the pore solution where they remain dissolved. After hardening, the pores are partially saturated with water that mainly contains alkalis (Na+ and K+), Ca2+ and OH, which define the characteristic alkaline pH of the conventional cementitious materials (Taylor, 1987). The ion contents related to Na+ and K+ are

10.1 Pore structure and alkaline liquid phase of concrete. Left: based on Feliu, Andrade et al. (1989). Right: based on Feldman and Sereda (1970). Types of water associated with calcium silicate hydrate and main pore water ionic composition.

© Woodhead Publishing Limited, 2010

288

Geological repository systems for safe disposal

significantly higher compared to that of Ca2+ in a very alkaline environment, OH, so it is assumed that the pore solution is always saturated with regard to Ca(OH)2. This oversaturation causes the precipitation of Ca(OH)2 within the cement paste inside pores but in crystal form. This phase constitutes the so-called portlandite or the ‘alkaline reserve’ that allows the concrete mass to maintain the pH of the pore solution above 12.6, as long as this compound is in the cement paste. The literature shows that standard concretes based on OPC have pore waters with a pH ranging from 12.6 to 14 (the level is determined by the alkaline content in the pore water), even containing pozzolanic mineral additions in proportions < 30%. The level of the pore water pH of cementitious materials varies depending both on the type and the dosage of cement used to produce the concrete. Due to their good technological properties, concretes and mortars play an important role in the engineered barrier system (EBS) of the low- and medium-level radioactive waste repository, for which the limitation of accessibility of groundwater is one requirement, because it has great influence on the chemical stability of the repository. In these types of waste repositories, the standard concrete used is usually made with OPC, or formulations based on OPC, plus a certain content of mineral addition (<30%). In the case of HLW repositories, concrete is going to be used during the construction of deep repositories for different purposes, such as grouting, rock bolting, lining of tunnels and drifts, and for sealing plugs at the mouth of the disposal drift or disposal cell, and it is also used for construction of various auxiliary structures needed for the operation of the repository (Ulm et al., 2002; Alonso et al., 2007; Buzzi et al., 2007; Rice et al., 2007): 1. 2. 3. 4.

Structural cast concrete, as road paving, floors or operational structures Shotcrete, either for tunnels linings, rock support or for plug tunnels Rock bolts Grouting to inject fractures, for sealing

Figure 10.2 shows a scheme of several locations of cementitious materials in HLW underground repositories. Some of these uses may be temporary and may be removed before the repository closure, but an important amount of concrete remains in the system along its entire life. Indeed, cementitious materials are being proposed as main components of the alternative engineered barrier systems. In the particular case of the Spanish concept for underground repositories, the total volume of concrete needed for structural purposes has been estimated in about 266.000 m3 in the case of clayish formations and 40.000 m3 in granite (Fuentes-Cantillana et al., 2005). For the repository concepts constructed in granite in Sweden and Finland, the amount of

© Woodhead Publishing Limited, 2010

Development and application of low-pH concretes

289

10.2 Location of cementitious materials in an HLW repository for rock support and plugging.

concrete used that needs to be left in the repository has been estimated as about 10 000 metric tons (Garcı´ a Sin˜eriz et al., 2004). Some of these cementitious materials are going to be in contact with the bentonite (clay mainly composed of montmorillonite) used in the EBS (engineered barrier system) of the HLW repositories; however, hyperalkaline cementitious materials have been shown to alter this clay barrier. For this reason, one of the most adopted engineering construction concepts for HLW of underground repositories requires the use of low-pH cementitious materials (different from the conventional ones), in order to avoid the formation of the alkaline plume from the cementitious materials in contact with groundwater that might perturb the barriers of the repository. The accepted solution to maintain bentonite stability, which is a function of the pH of materials in contact, is to develop cementitious materials that generate pore waters with pH ≤ 11, because the alteration of the clay components (with the subsequent loss of its swelling ability) is significantly reduced below this pH value (Ramı´ rez et al., 2002), or according to the recent report by Savage and Benbow (2007) bentonite degradation rate is greatly decreased below pH = 10. The feasibility of designing and utilizing low-pH cementitious materials in underground repositories for preventing these undesired processes is being investigated (Hidalgo et al., 2005; HugoPersson et al., 2005; Vuorinen and Lehikoinen, 2005; Cau Dit Coumes et al., 2006). In this chapter different aspects related to the development of low-pH cements for low-pH concrete production for HLW repositories are exposed. Firstly, the general functional requirements expected for this type of concrete are defined. Secondly, the design and the properties of low-pH cements are detailed. Moreover, the special characteristics implied in the development and production of low-pH concretes are described, mainly focused on the procedure followed in the design of a specific application of low-pH cements aimed at shotcreting a low-pH concrete plug. Finally,

© Woodhead Publishing Limited, 2010

290

Geological repository systems for safe disposal

taking into account the long life expected in this type of repository, parameters related to the durability of the low-pH concretes must be analysed. Therefore, this chapter also shows some recent studies that deal with the evaluation of the resistance of low-pH cementitious materials to long-term groundwater aggression and, in the case of use of a reinforced concrete, the risk of corrosion is also considered.

10.2

Functional cementitious material requirements for geological disposal

Most of the repository concepts, irrespective of the host rock, require concrete construction in the underground repository areas. The functional requirements applicable for the different low-pH concrete applications should be explicitly indicated, but in order to be applicable those requirements should be connected to measurable parameters, even though some desired properties could not be measured quantitatively, like stability and durability. Obviously, the main functional requirement demanded in any type of lowpH material used in an HLW repository is having a pore fluid pH value ≤ 11, which is considered quite acceptable for avoiding the alteration of the bentonite EBS. Therefore, pH measuring of the concrete pore fluid phase becomes a relevant requirement to be controlled. Nevertheless, some discrepancies exist with respect to the method used for measuring the pH. However, for many authors the squeezing technique (pore pressing extraction) is considered as the reference method (Longuet et al., 1973; Barneyback and Diamond, 1981) but there are other pH measuring methods calibrated with the pore pressing technique that give accurate results (Castellote et al., 2002; Ra¨sa¨nen and Penttala, 2004; Hidalgo et al., 2005; Li et al., 2005). Other functional requirements are going to depend on the specific application given to the low-pH cementitious materials. In order to get a general idea this chapter explains in depth the main low-pH cementitious materials whose applications have nowadays been better assessed for use in HLW repositories and gives the requirements needed in each case. The main applications are for plugs devised for sealing disposal galleries, for rock support and for injection grout for larger fractures. However, apart from the requirements needed in the specific material behaviour, those required by the specific construction method must be taken into account. Considering the two first applications, the use of shotcreting has proven to be an efficient and cost-saving method for rock support in underground construction. Moreover, previous experiences in plug construction showed that the shotcreting technique is also feasible with the

© Woodhead Publishing Limited, 2010

Development and application of low-pH concretes

291

same benefits. Some additional advantages of shotcreting when compared with the traditional cast concrete method for plug construction are: a better bonding with the host rock can be achieved, no keys (rock slots) in the rock are required, fast construction, no forms are required nor air voids left for later filling, and the operation can be robotized (Garcı´ a Sin˜eriz et al., 2004). Therefore, in plugs and rock support applications, the specific requirements for shotcreting must also be considered. Shotcrete, also called ‘sprayed concrete’, is a well-known construction technique based on spraying concrete on to a rigid surface (rock or soil) using a concrete pump and compressed air. The compressed air projects the concrete mix at high speed against the surface, so that it sticks to the surface. Once sprayed, the concrete starts to set immediately, thanks to an ‘accelerant’ additive, forming a self-supported layer. Due to the characteristics of the formulations used and the spraying process, shotcrete is normally a high-density and low-permeability material, compared to an average standard concrete. It also shows lower surface cracking (Fuentes-Cantillana et al., 2005). The main functional requirements that need to be complied with should be specified in every application type: hydraulic conductivity (site specific, with the same order of magnitude as that of the excavation disturbed zone (EDZ)), mechanical properties of concrete, durability (concept specific, linked to the operational life of the repository), workability and pumpability (when shotcreting), slump, peak hydration temperature, thermal conductivity (concept specific, not below that of the bentonite barrier), use of organic components (fibres or admixtures), use of other products, etc. With regards to the peak hydration temperature and thermal properties, consideration should be given to shotcrete formulation for preventing cracks induced by the heat of hydration. This is more relevant in plugs since the application of shotcrete layers within short time intervals might reduce the heat dissipation and therefore buffer the hydration temperature. Concerning thermal conductivity, it is desirable that the plug has equal or higher conductivity than the bentonite buffer. Next, the specific functional requirements of each case are detailed.

10.2.1 Functional requirements for low-pH concrete plugs Plugs are required in HLW repositories in a number of places. The plugs in the deposition galleries have basically a mechanical function, which transfers the thrust of the swelling bentonite to the rock, but in some repository designs they must also have a hydraulic seal function (especially the interruption of the EDZ). As commented above, shotcrete has a number of advantages, from an engineering point of view, which makes it a good candidate for the construction of plugs in the repository area, both for temporary and final ones.

© Woodhead Publishing Limited, 2010

292

Geological repository systems for safe disposal

Table 10.1

Functional requirements for low-pH concrete plugs

Requirement

Target

Hydraulic conductivity Final mechanical properties Young modulus Poisson ratio Tensile strength Friction angle Cohesion Compressive strength Workability Pump ability Peak of hydration temperature Thermal conductivity Maximum pressure at the plug/buffer interface

K < 1010 m/s >20 000 MPa 0.2–0.3 > 1 MPa ≥ 378 ≥ 2 MPa ≥ 10 MPa ≥2h 100–250 m ≤ 40 8C; ≤ 30 8C 1.2–1.75 W/m 8C 3–15 MPa

A basic requirement of these plugs is related to the total pressure admissible at the buffer/plug interface, so it must be taken into account that one of the main functions of the plug is to withstand the total pressure of the bentonite buffer when fully saturated and the hydraulic pressure is at the repository level. This maximum swelling pressure of the buffer will depend on the bentonite composition and the initial density and moisture content. Furthermore, the plug should not be a preferential path flow and it should not modify the confinement properties of the disposal components (Garcı´ a Sin˜eriz et al., 2004). The specified values for functional requirements for plugs defined by different European Agencies are shown in Table 10.1 (Garcı´ a Sin˜eriz et al., 2004). Some of these requirements, not considered in other applications, have to do with the construction rate, maximum total pressure or pressure at the plug/buffer interface, length of plug and the time between the start of construction and the full function of the plug. In Section 10.4, the procedure followed in the design of a specific application of a low-pH shotcrete plug for the fulfilment of the specific functional requirements is given.

10.2.2 Functional requirements for low-pH rock support The most usual application of shotcrete is rock support, in which a layer of shotcrete is sprayed on top of the rock surface immediately after excavation, to create a support arch that at the same time prevents rock expansion. Typically, shotcrete is combined with bolting and a wire mesh for improved efficiency. In comparison with standard concrete lining for roof support, shotcrete is time and cost efficient, and can be used in most host rocks considered for HLW repositories. The only difference is that the final

© Woodhead Publishing Limited, 2010

Development and application of low-pH concretes Table 10.2

293

Functional requirements for rock support

Requirement

Target

Hydraulic conductivity Final mechanical properties Young modulus

K < 1010 m/s ≈ ≈ ≈ ≈ ≈

Compressive strength

(7 days) (28 days) (8 hours) (7 days) (28 days)

≈ 0.9 MPa (7 days) ≈ 1.5 MPa (28 days) ≥2h > 100 m 15–20 cm <100 8C

Bonding Workability Pump ability Slump Peak hydration temperature Table 10.3

15 GPa 20 GPa 10 MPa 25 MPa 35 MPa

Required properties of low-pH cement-based grouts

Order of importance

Requirement

Target

Required properties

pH Penetration ability bmin Penetration ability bcrit Viscosity

≤ ≤ ≤ ≤

11 80 μm 120 μm 50 mPa

Bleed Workability time Shear strength Yield value Compressive strength

≤ ≥ ≥ ≤ ≥

10% 60 min 500 Pa 5 Pa 4 MPa

Desired properties

surface obtained is more rough and irregular than with standard lining, which could be perhaps a problem in the zones that must be backfilled with pre-shaped blocks or packages. In principle, only in the case of repositories in clay formations is shotcrete used, for in this application it would be in direct contact with the engineered barrier. In granite, basically all shotcrete used for rock support should be in access and service galleries and vaults, at longer distances from the EBS. The specified values for functional requirements for the rock support application defined by different European Agencies are shown in Table 10.2 (Garcı´ a Sin˜eriz et al., 2004).

10.2.3 Functional requirements for low-pH injection grouts The requirements shown in Table 10.3 were determined by Posiva (the waste management agency from Finland) during the planning phase of a project

© Woodhead Publishing Limited, 2010

294

Geological repository systems for safe disposal

that dealt with injection grout for deep repositories (Kronlo¨f, 2005), where the requirements worked as a guideline for laboratory determinations. In addition to the requirements listed in Table 10.3 it is desirable that the grout must be available in practice during the construction and operation of the repository. It must also have a history of use in cement technology (or practical engineering) and its durability properties need to be sufficient so that the grouted zone maintains its required properties during the expected lifetime.

10.2.4 Functional requirements for the host rock It must be said that, apart from the functional requirements defined for the material, there are also some requirements that concern the properties of the host rock in each application. As a general rule, some mechanical properties of the host rock must be defined: Young’s module, Poisson’s ratio, tensile strength, friction angle and cohesion. Additionally, in the case of the plug application, some mechanical properties of the rock–plug interface should be determined, such as friction angle, cohesion and normal and shear stiffness. Furthermore, the groundwater composition of the host rock must also be known because the durability of the low-pH cementitious materials will depend on the aggressiveness of the groundwater, which is a function of salinity and flow.

10.3

Design and properties of low-pH cements

As has already been commented in the introduction to this chapter, bentonite and concrete are essential components in building of a HLW geological repository (Gaucher and Blanc, 2006). Bentonite is used as an engineering barrier, due to its physicochemical and mechanical properties, and cementitious materials are used for several purposes. However, bentonite and concrete have geochemical characteristics that make them not fully compatible under repository constraints. When conventional ordinary Portland cements (OPCs) are used to produce concretes for these underground repositories, contact with the groundwater creates a pore water leachate with a pH as high as 13.5, the so-called alkaline plume. The alkaline solutions can react with the bentonite in the proximity of concrete, inducing dissolution and precipitation of a number of phases. The alteration of the bentonite components neutralizes the basic solutions; it is clearly a function of the solution pH, with a strong dependency on the hydroxyl concentration over pH 11 (Rozale´n et al., 2008). Below pH 11, bentonite damage is limited. Therefore, considerable effort is being made to reduce the potential

© Woodhead Publishing Limited, 2010

Development and application of low-pH concretes

295

damage of the alkaline plume over the bentonite barrier, which includes cementitious material research to develop low-pH concretes that generate pore waters with pH ≤ 11. This logically implies the development of low-pH cement formulations. The research on low-pH cementitious materials is being addressed from different approaches depending on the type of cement used: (1) calcium silicate cements (OPC based), (2) calcium aluminate cements (CAC based), (3) calcium sulphoaluminate cements, (4) phosphocalcic cements (PC) and (5) magnesia phosphate cements (MC). As the two first approaches are the most developed nowadays, they are analysed in depth in the present chapter.

10.3.1 Low-pH cements based on ordinary Portland cement (OPC) Some authors have demonstrated that the interaction of OPC concretes (traditionally used in construction) with groundwaters, producing the hyperalkaline fluids, is due to the leaching of sodium and potassium hydroxides. Once these alkaline hydroxides are released, the pH of the pore water solution in the concrete is controlled by the dissolution of one of the main phases of the OPC paste, the portlandite (Ca(OH)2), which is responsible for the duration of the alkaline plume at pH 12.6. Later on, between pH 12.6 and 10, the dissolution of the calcium from silicate hydrates of the CSH gel takes place and finally, below 10, the hydrates of Al and Fe are dissolved, as graphed in Fig. 10.3. Therefore, C–S–H gel, together with Ca(OH)2 and alkalis, dominate the observed pH properties of the pore aqueous phase of Portland cementitious materials (Taylor, 1987; Adenot and Buil, 1992;

10.3 Decrease of pH in pores of OPC paste during the neutralisation process. Curve of a cement paste neutralisation in nitric acid medium (Hidalgo et al., 2004).

© Woodhead Publishing Limited, 2010

296

Geological repository systems for safe disposal

Braney et al., 1993; Glasser and Atkins, 1994; Lovera et al., 1997; Faucon et al., 1998; Ramı´ rez et al., 2002; Hidalgo et al., 2004). It is evident that the decrease of the pore fluid pH from the initial state in the pore solution up to values close to 11 implies the use of OPCs with a low content in alkalis and a reduction of the portlandite (the main source of OH). A literature review on this topic shows that the addition of fly ash, silica fume or other pozzolan materials gradually decrease the Ca(OH)2 content as soon as they react with it, to be hydrated and form more CSH, which shortens the length of the period at pH = 12.5. Additionally, the anhydrous particles of cement react with water to be hydrated and this process contributes to the release of higher amounts of Ca2+, also enabling a kind of buffering of the pH in the solution above pH 12.5. In order to reduce the content of portlandite, mineral admixtures with a high silica content should be used (Durning and Hicks, 1991; Torii and Kawamura, 1994; Hidalgo et al., 2005; Hugo-Persson et al., 2005; Saeki and Monteiro, 2005; Vuorinen and Lehikoinen, 2005; Cau Dit Coumes et al., 2006). Therefore, the amount of OPC is reduced and substituted by mineral admixtures with a high silica content in their composition to decrease the calcium hydroxide formation in the binder paste. The fabricated blends follow the pozzolanic reaction that consumes Ca(OH)2 (Saeki and Monteiro, 2005). The high mineral admixture contents used in the low-pH binders significantly modify the microstructure and the pore fluid of these cementitious materials as well as their standard properties. Therefore, a deep understanding of the hydration of low-pH cementitious materials is needed. In low and medium waste repositories, most of the cementitious materials used are based on OPC, blended in many cases with mineral additions or blending agents. In this case, the main requirement for OPC is the content of Ca(OH)2, which is considered to increase the pH of pores up to 12.6, and the presence of the alkalis (K, Na), which increases the pH up to a maximum of 14. Blending agents can be grouped in different categories depending on their own characteristics. Some additions like the limestone filler are practically inert but others actively participate in the hydration and morphology of the reaction products. Among them, two types can be identified: pozzolanic and cementitious. Natural pozzolans (Nat. Pozz.), silica fume (SF), methakaolin (MK) and type F fly ash (FA) are typically pozzolanic while ground granulated blast furnace slag (BFS) and type C fly ash (FA) show cementitious properties. The chemical composition variations of the water pore solutions extracted from various OPC cementitious materials indicate that the dominating cations are Na+ and K+; moreover, the sulphate resistant cements, OPCSR, show a higher content of Ca2+. The next considerations are stressed: .

If OPC is employed as a cement matrix, a pore solution of pH ≤ 13.1 can

© Woodhead Publishing Limited, 2010

Development and application of low-pH concretes

. .

. .

297

be reached when a low alkaline content OPC is employed, having K+ and Na+ in pores < 300 mmol/l. If mineral additions are used, the pH in pores will remain above 13 if proportions below 15% are employed. The pH depends on the proportion and type of blended agent and its alkaline content. When mineral additions are used in a proportion below 35% the pH decreases to a value between 13 to 12.5. For identical amounts of additions such as silica, the pH decreases more with FA and SF, in comparison with BFS, Nat. Pozz. and MK. A pH around 12.5 is reached if FA > 35% or 35% > SF > 15% is employed. Then the alkaline content in pores, Na+ and K+, remains around 100 mmol/l. BFS does not reduce the pH as do FA and SF.

Therefore, the low-pH cementitious materials based on OPC must include higher contents of mineral additions, over 40%, to ensure that free portlandite is significantly reduced or even avoided. To eliminate portlandite, mineral additions are recommended, particularly those that consume portlandite during hydration via pozzolanic reaction (Alonso et al., 2006; Cau Dit Coumes et al., 2006; Lagerblad et al., 2006). Silicabased materials (silica fume or fly ash, etc.) allow the production of low-pH cementious material, either by using isolated material (binary mixes) or by mixing several mineral blends (ternary or more blended cements). In the case of low-pH cementitious material production, it has been determined that if OPC is used as the basic cement of the low-pH blend, at least 55% of SiO2 must be in the binder, or pozzolan blended additions above 40% have to be employed. Indeed, the composition of the blended agent has a marked effect on the pH, with SF being the most effective (due to its higher silica content), as shown in Fig. 10.4. Formulations with a sufficient percentage of mineral addition, > 40%, and a total SiO2 content > 55% show a pH below 11 after 90 days of curing. Evolution of the pore fluid pH of several low-pH cement formulations is shown in Fig. 10.5. Table 10.4 compiles the pore fluid composition of lowpH cement pastes from fluid expression tests after 90 days of hydration (Garcı´ a Calvo et al., 2008). The higher the increase in the silica content of the binder, the higher the decrease in the pore fluid pH of the paste. This effect is more relevant at longer curing times (> 90 days) because over short times the pH can be even higher in the low-pH cement pastes, as already described by Larbi et al. (1990). These results indicate that the percentage of silica in the cement formulation has to be above 45% to obtain a pH below 12.5 at 90 days, indicating that the portlandite has been removed from the cement paste as will be shown below, when solid phase evolution is evaluated.

© Woodhead Publishing Limited, 2010

298

Geological repository systems for safe disposal

10.4 Silicate content of cement formulation influence on pore fluid pH at 90 days.

10.5 Evolution of the pore fluid pH of several cement pastes based on OPC.

Observing Table 10.4, a decrease in the alkaline content in pore solutions is evident in binder formulations with high mineral admixtures contents. With regards to Ca2+, in up to 90 days it seems to increase in the pore fluid with respect to OPC without blending addition. The increase of the SO42

© Woodhead Publishing Limited, 2010

Development and application of low-pH concretes

299

Table 10.4 Composition of the pore fluid of low-pH pastes based on OPC at 90 days of hydration Chemical composition (ppm) +

+

Sample

pH

Na

K

OPC 40% SF 50% SF 10%SF + 10%FA 35%SF + 30%FA 50%SF + 30%FA

12.9 12.2 11.2 12.6 11.2 10.5

603.27 119.7 165.1 118.50 175.09 220.75

2129.5 321.8 368.6 459.40 387.92 219.02

Ca2+

Si4+

Al3+

SO42

474.41 612.1 650.9 689.86 486.10 1208.0

1.23 9.82 60.27 ND 47.98 72.28

ND ND ND ND ND ND

12.78 84.40 2181 31.63 1819 3105

ND: non-detected.

10.6 Evolution of Na+, K+ and Ca2+ contents in a low-pH cement paste (60%OPC + 40%SF).

content in the pastes with lower pore fluid pHs is also very significant as well as the increase in the Si4+ amount, which is supposed to come mainly from the mineral admixtures used. Figure 10.6 shows the evolution with curing time of the Na+, K+ and Ca2+ contents in the pore solution of cement paste with 40% of SF. This figure clearly shows that over short periods of time low-pH cements have a decrease in the alkali content of the pore solution but an increase in the calcium one. Concerning the evolution of the low-pH cement paste microstructures, the portlandite content decreases with curing time and has fully disappeared at 90 days in those with a percentage of SiO2 ≥ 55%, which can be confirmed from thermal analysis (DTA/TG), such as those results shown in Fig. 10.7, where the portladinte content is graphed for six cement pastes at 90 days of curing. The introduction of a high content of mineral admixtures also caused relevant changes in the CSH gel compositions of the cement pastes, which is the main binding phase in all Portland cement-based systems. The Ca/Si

© Woodhead Publishing Limited, 2010

300

Geological repository systems for safe disposal

10.7 Portlandite content of the cement pastes studied at 90 days (TG analyses).

10.8

Influence of Na2Oeq on pH.

ratio of CSH in conventional Portland cement pastes varies from 1.2 to 2.3, with a mean of 1.75 (Klur et al., 1998). On the contrary, in the low-pH cements this ratio (measured using backscattered electron microscopy with EDX analyses) varies from 1.2 to 0.8. CSH gels with these CaO/SiO2 ratios are typically considered to be formed by longer length chains of tetrahedral silica (Cong and Kirkpatrick, 1996). Indeed, Stronach and Glasser (1997) noted that to obtain a low pH in the cement materials the CaO/SiO2 ratio must be lower than 1.1. Furthermore, although in conventional OPC pastes the increase in the content of alkalis, Na+ and K+, contributes to increase the pH in the pore water, if high contents of mineral admixtures are used this general rule cannot be applied. This can be seen in Fig. 10.8, which shows the influence of alkalis in the pH of the pore fluid.

© Woodhead Publishing Limited, 2010

Development and application of low-pH concretes

301

There is another important issue regarding the alkaline content in the pore fluid of low-pH OPC pastes. A binding process during hydration that removes the alkalis from the pore solution of low-pH pastes could be taking place inside the new CSH gel. Glasser and Marr (1985) published findings that SF is very effective in removing K+ from the pore fluid, due to the formation of smaller CaO/SiO2 ratios in the cement pastes. CSH gels with high CaO/SiO2 ratios have a positively charged surface, but as the CaO/SiO2 ratio decreases the charge becomes neutral, leaving van der Waals forces to operate, and at the lowest CaO/SiO2 ratios the surface charges of the pores become negative. In fact, Hong and Glasser (1999) demonstrated that the alkaline binding process is more effective when the Ca/Si ratios of CSH gels are between 1.2 and 0.85, as occurs in low-pH cements. Therefore, to formulate a low-pH cement, adding pozzolanic materials to OPC should be advantageous in at least three respects: (1) portlandite formed by the hydration of OPC is converted into CSH by pozzolanic reaction, (2) OPC is diluted and (3) the CaO/SiO2 ratio of CSH is lowered, which enhances their sorption capacity of alkalis and reduces their equilibrium pH. Furthermore, these additions are known to improve some properties of the cementitious materials: bleeding is limited (Yogendran et al., 1987) and porosity is refined (Huang and Feldman, 1985; Durekovic, 1995), which in turn reduces permeability and diffusivity (Bentz et al., 2000; Oh et al., 2002). However, some difficulties might be expected with high contents of these mineral additions for the low-pH concrete applications, as shown in Section 10.4.

10.3.2 Low-pH cements based on calcium aluminate cement (CAC) Calcium aluminate cements (CACs) represent an interesting alternative because their pore water pH, ranging from 11.4 to 12.5, is reduced as compared to OPC (Gon˜i et al., 1991). The main difference between Portland and calcium aluminate cements lies in the nature of the active phase that leads to setting and hardening. Monocalcium aluminate (CA) is the principal active phase in the CAC, which reacts with water to give calcium aluminate hydrates. However, direct use of these binders comes up against one main difficulty. The hydration reaction of CAC at environmental temperatures produces hexagonal hydrated calcium aluminates, CaAl2O14H2O (CAH10) and Ca2Al2O13H16 (C2AH8). The stable phases are Ca3Al2(OH)12 (C3AH6) and Al(OH)3 (AH3), and the other phases will inevitably convert to these, decreasing the mechanical properties of the material (Midgley and Midgley, 1975; Fentiman et al., 1990; Xiandong and Kirkpatrick, 1993; Capmas and

© Woodhead Publishing Limited, 2010

302

Geological repository systems for safe disposal

George, 1994; Scrivener et al., 1999). The conversion reactions of CAC are shown schematically below: 3CAH10 ?C3 AH6 þ 2AH3 þ 18H

½10:1

3C2 AH8 ?2C3 AH6 þ AH3 þ 9H

½10:2

where CAH10 is CaAl2O14H2O, C2AH8 is Ca2Al2O13H16, C3AH6 is Ca3Al2(OH)12, AH3 is Al(OH)3 and H is H2O. Some authors have found that an interesting way to reduce the hydrate conversion and the decreasing strength is to replace some of the CAC by blast furnace slag (BFS) or a pozzolan such as microsilica and metakaolin (Majumdar et al., 1990a, 1990b; Collepardi et al., 1995; Ding et al., 1995; Quillin et al., 2001; Hidalgo et al., 2008, 2009). The reaction that avoids the conversion of CAC hexagonal hydrated phases could take place in the following way (Majumdar et al., 1990a, 1990b; Collepardi et al., 1995; Rivas Mercury et al., 2006): the silica content of the mineral addition would react with the calcium aluminates, initially avoiding the formation of the hexagonal form C2AH8 and, subsequently the conversion in the cubic form C3AH6. Therefore, instead this cubic phase, a hexagonal aluminate hydrate containing silica, called gehlenite (Ca2Al2SiO7·8H2O or C2ASH8), is proposed to be formed. Majumdar et al. (1990a, 1990b) considered that the amount of C2ASH8 is dependent on the capacity of a mineral addition to release silica. It is important to mention that in the literature there is not agreement between authors on the precipitation of the C2ASH8 phase (gehlenite). Although studies about stabilisation of CAC with mineral additions are scarce, this solid phase has been observed in CAC mixes with BFS, SF, metakaolin and FA. Majumdar and Singh (1992) and Fu et al. (1995) concluded that, in mixes based on CAC with a silica fume (SF) content between 30 and 50%, the gehlenite is the main hydration product before a week (T < 40 8C). Some authors (Rivas Mercury et al., 2006; Hidalgo et al., 2008, 2009) also reported that in the system CAC-SF and CAC-FA, silica reacts with the calcium aluminate phases in the cement and water to form different crystalline hydrates (with variable proportions of Ca, Al and Si) such as Ca2Al2SiO7·8H2O (C2ASH8, called gehlenite), Ca3Al2(SiO4)3-x (OH)4x (0 < x < 3) (katoite) and not very well defined and complex zeolite-type phases. The Ca(OH)2, Al(OH)3, CaAl2O4·10H2O, Ca2Al2O5·8H2O, Ca2Al2SiO7·8H2O and a series of solid solutions, with the general formula of Ca3Al2(SiO4)3-x(OH)4x, can be located in the CaO–Al2O3–SiO2–H2O system for temperatures below 100 8C. Indeed, the whole series exists with the end members, known as grossular when x = 0 and hydrogarnet if x = 3. Since these solid solutions are also present in nature, mineralogists classify

© Woodhead Publishing Limited, 2010

Development and application of low-pH concretes

303

10.9 Pore fluid pH of several cement pastes based on CAC at 90 days of hydration.

them as hibschite for the minerals with 0.2 < x < 1.5 and as katoite for the ones with 1.5 < x < 3 (Rivas Mercury et al., 2003a, 2003b, 2007). The identification of the calcium carboaluminate hydrate C4AcH11 as a precursor of hydrogarnet–garnet solid solution members with silica in their composition, C3AS3-xH2x (CASH), when SF and/or FA are used, has been also proposed in the literature (Hidalgo et al., 2008, 2009). Therefore, it seems clear that the inclusion of mineral additions in mixes based on CAC increases their microstructure stability and even slightly decreases the pore fluid pH value over long time periods, as shown in Fig. 10.9. However, in this case there is neither a minimum nor a maximum percentage of mineral addition to employ for optimum low-pH cement developing, but shows that the percentages are related to avoidance of the conversion process. It is also remarkable that, as well as occurring in low-pH cements based on OPC, the percentage of alkalis does not alter significantly the pore fluid pH of these materials, and again the alkali binding process is observed, although it is less clear than in the case of OPC low-pH mixes.

10.3.3 Low-pH cements based on other cement types Calcium sulphoaluminate cements Raw mixes for calcium sulphoaluminate (CSA) clinkers differ from those for OPC in that they contain significant amounts of sulphates. The CSA clinker is mainly composed of yeelimite (4CaO·3Al2O3·SO3), belite (C2S) and an Al-rich ferrite (Andac and Glasser, 1995). CSA cement pastes harden through the formation of an initial ettringite skeleton, and its subsequent infilling by mixtures of ettringite, calcium monosulphoaluminate hydrate,

© Woodhead Publishing Limited, 2010

304

Geological repository systems for safe disposal

alumina and ferrite gel (Kasselouri et al., 1995; Zhang and Glasser, 2002). The reported pore solution pH values of CSA mortars are highly variable, ranging from 8.5 to 13 (Andac and Glasser, 1999; Kalogridis et al., 2000; Janotka et al., 2003), but they must be studied in depth before using in HLW underground repositories. Phosphocalcic cements These cements could seem attractive at first sight due to hydroxyapatite, which is the thermodynamically stable resulting product, has a very low solubility in neutral or basic medium and leads to an equilibrium pH within the range 7–9 (El Jazairi, 1987; Popovics et al., 1987; Driessens, 1995; Pera and Ambroise, 1998; Chow, 2000; Lemaıˆ tre, 2001). However, up to now, very few materials have been designed to be workable using contemporary engineering practices (Cau Dit Coumes et al., 2006). Magnesia phosphate cements Their pore solution usually ranges between 7 and 8 (Pera and Ambroise, 1998). These cements are based on an acid–base reaction between dead burnt magnesia and a phosphate salt (El Jazairi, 1987; Popovics et al., 1987; Seehra et al., 1993; Wagh et al., 1997; Soudee and Pera, 2000).

10.4

Development and production of low-pH concretes: shotcrete plug application

This section shows the design of a concrete mix to be shotcreted at a concrete plug where a low pH is mandatory. The meaning of ‘concrete design’ is to select suitable constituents (among the available ones) and to determine the proportions to achieve the desired performance both in the fresh and hardened states. As the design of the concrete plug involves different characteristics of the concrete (mechanical strength, deformation, hydraulic permeability, heat of hydration, shrinkage, potential durability, etc.) and other technological properties, such as consistency, pumpability and a low loss in workability with time, the design and optimisation of the concrete must be carried out on a performance basis. Thus, the first step is to determine the requirements imposed by the specific application to the fresh and hardened state, usually known as ‘functional requirements’, the main one being its low pH, which must be around 11, to prevent the bentonite in contact from deteriorating. As defined in Section 10.2, functional requirements for the concrete of the shotcreted plug made of low-pH cement can be summarised as follows: at the fresh state, it has to be workable, pumpable and projectable while

© Woodhead Publishing Limited, 2010

Development and application of low-pH concretes Table 10.5

305

Functional requirements for the shotcreted concrete

Requirement

Target

Complementary considerations

Maximum pH

11

Hydraulic conductivity

K < 1010 m/s

Measurement technique Age of measurement

Final mechanical properties Young modulus Poisson ratio Tensile strength Friction angle Cohesion Compressive strength

< 20 GPa 0.2–0.3 1 MPa > 378 2 MPa 10 MPa

Workability

> 2h

Pumpability Peak of hydration

500 m < 40 8C

Final means: long-term The age of testing to assess final properties will be established from test results

Negligible workability loss for at least 2 hours A long pumping distance can be expected Temperature rise is influenced by the accelerating admixture added at the shotcrete nozzle

compressive strength, elasticity modulus, water permeability and pH can be highlighted as the most relevant requirements for the hardened state. Moreover, if the concrete maintains a relative humidity in the pores above 80%, the properties of concrete change over time due to progressive hydration of the cement and the pozzolanic reactions. The result is an increase in strength and modulus of elasticity in conjunction with a decrease in permeability and the pH has to be taken into consideration as well. On the other hand, the projection of the concrete requires the mix to be pumpable in real working conditions (equipment available, distances and heights when pumping, temperature, etc.) and to adhere to a vertical surface (the bottom of the plug). This issue requires the use of an accelerator admixture at the nozzle. Since delays must be expected during work, it would be advantageous if the mixture could preserve its characteristics in a fresh state (workability and pumpability) for at least two hours after mixing. Last, but not least, the adiabatic temperature rise must be below 40 8C. Table 10.5 summarises the main functional requirement for the low-pH shotcrete of the concrete plug.

10.4.1 Mix design procedure Concrete can be seen as a composite material composed of an aggregate skeleton bound by a paste matrix. The paste itself is composed of the low-

© Woodhead Publishing Limited, 2010

306

Geological repository systems for safe disposal

10.10

Flowchart to select the constituents of the paste.

pH cement formulation, water and the chemical admixtures. As most of the physicochemical reactions occur at the paste phase, the compatibility among different constituents can be assessed in paste evaluations, the aggregate being almost inert. Thus, the selection of the concrete components is divided into two stages: paste components and aggregates proportioning. The flowchart in Fig. 10.10 shows the selection process of the paste constituents. As can be seen from the compatibility evaluation chart shown in Fig. 10.10, different tests have to be performed successively to determine suitable

© Woodhead Publishing Limited, 2010

Development and application of low-pH concretes

307

combinations of low-pH cement formulation and the admixtures to be used. It must be kept in mind that chemical admixtures are developed and optimised for ‘standard cements’ and thus their efficiency with special cements (low-pH) cannot be neglected. Aggregates selection is a key issue in concrete design and production as about 70% of the concrete is made of aggregates and they strongly influence water demand, workability, pumpability and projectability of the concrete. To select suitable aggregates, two main considerations arise: the suitability of aggregates in terms of strength, surface hardness, dimensional stability, intrinsic durability, among others, and the aggregate grading, i.e. the particle size distribution. To produce low-rebound shotcrete, it is advisable not to use a maximum size over 16 mm and preferably not over 12 mm. Different available fractions can be mixed in order to compose an aggregate skeleton with a grading able to obtain a fresh concrete that fulfils the requirements for shotcrete. The easiest way to adjust the proportions of the aggregate fractions is to consider the limits suggested by different organisations, such as ACI (American Concrete Institute), AENOR (Spanish Standards Organisation), SCA (Shotcrete Concrete Association of the United Kingdom), etc. If the percentage passing through different sieves is plotted against the log of the sieve size, aggregate size distribution can be graphed. If the limits suggested are also indicated in the same plot, it is easy to verify the compliance with these limits, as shown in Fig. 10.11. Other aggregate characteristics, such as absorption (saturated in a dry-surface condition) and density, are also necessary to adjust concrete proportions.

10.11 Aggregate grading and bandwidth limits.

© Woodhead Publishing Limited, 2010

308

Geological repository systems for safe disposal

Chemical admixtures The importance of selecting suitable and compatible chemical admixtures when selecting the concrete constituents and designing the concrete mixture cannot be overemphasised. Once the compatibility between the low-pH cement formulation and the admixtures is verified, it is necessary to adjust the dosage of the chemical admixture in the concrete. The admixtures are used to modify the properties of the concrete, principally in the fresh state, and to accelerate hardening. The chemical admixtures used in the shotcrete plug discussed in this section were: air-entraining admixtures (used to improve the workability of the concrete mixes, which have been affected by crushed fine aggregate), superplasticisers or high-range water-reducing admixtures (they are essential when utilising ultra-fine mineral additions, such as silica fume, to properly disperse the particles; moreover they also allow for a reduction in the water content without compromising the fluidity of the mixtures) and accelerator admixtures (they are used directly at the nozzle; their effect is almost immediate and the goal is not only to achieve early strength but also to ensure that the concrete will adhere to the surface without slipping or collapsing. As their use can cause a significant reduction in the final mechanical strength of the concrete, they must be evaluated exhaustively together with the range of materials in question).

10.4.2 Concrete design Concrete mix design is based on two main parameters: water demand (for a given workability) and the water/binder ratio to attain compressive strength. For conventional concrete, using conventional cement types, there are design charts or design methods that provide preliminary data on these figures, such as Abrams law, sketched in Fig. 10.12. Each of the three curves corresponds to a nominal compressive strength of cement, according to European standards, but the nominal compressive strength of the low-pH formulation has to be determined, as well as the water demand. A rather simple procedure is to prepare mortars (no coarse aggregate in the formulation) of equivalent fluidity and to test their corresponding compressive strengths at a given age, because strength evolution with time strongly depends on the cement type. The comparison between samples cast with different water/cement (w/c) ratios and different cement types can be made from the percentage of the compressive strength at any age versus the 28-day strength. Thus, no matter what the actual strength level is, all samples exhibit the same relative value at 28 days: 100%. Figure 10.13 represents the relative strength (at 28 days) versus log age. From the figure, it can clearly be seen that the strength evolution of calcium aluminate based low-pH cement is negligible over 7

© Woodhead Publishing Limited, 2010

Development and application of low-pH concretes

309

10.12 Relationship between compressive strength (28 days) and the w/c ratio (Abrams law).

10.13 Relative strength evolution versus log time for different cement types.

day curing while Portland cement based low-pH cements exhibit an increase of about 40% at 90 days. Therefore, long-term CAC based low-pH formulation properties can be assessed after a few days of curing (typically 7) but OPC based low-pH formulations should be tested after longer curing periods (typically 90 days). Indeed, the evolution of strength (and other properties) differs considerably, depending on whether the concrete is CAC or OPC based. The composition of the mixes can be set via the absolute volume method (the sum of the volumes equals 1000 dm3) using the aggregates and the paste components already defined. A cement content of approximately 300 kg/m3 was determined and the water adjusted to achieve suitable fluidity.

© Woodhead Publishing Limited, 2010

310

Geological repository systems for safe disposal

Table 10.6 Concrete proportions and properties in the fresh and hardened state Component

Weight (dry)

Properties in the fresh state

Water Cement Water/binder Gravel Fine gravel Sand Water reducer (1.8%)

277 307 0.90 616 200 818 5.5

Unit weight Slump Cohesion Aspect

2.23 t/m3 12 cm Good Good

Compressive strength 7d 28 d

16.2 MPa 29.0 MPa

Adjustments are usually necessary when preparing the trial mix, but in just a few trials it is possible to achieve reasonably good results. Nevertheless, the properties at the hardened state have to be determined at different ages (typically 7, 28 and 90 days) and the pumpability and projectability evaluated in real scale testing. A decrease in mechanical properties of about 35 to 40% of the sprayed concrete as a result of the use of an accelerator and the spraying technique have to be considered when setting the target for strength for conventional concrete.

10.4.3 Trial mix and properties of concrete in the fresh and hardened state Once initial proportions are set, a trial mix has to be tested in order to adjust the water content according to the required consistency (slump), checking the mix for cohesion, aspect (not too sandy, nor harsh) and unit weight. Once the requirements at the fresh state are fulfilled, cylindrical or cubic samples have to be cast to determine the compressive strength at different ages and the static elasticity modulus. Values have to comply with the requirement for compressive strength plus 40% (about) to compensate for further losses when the concrete is projected. Table 10.6 shows low-pH concrete proportions and properties at the fresh and hardened states.

10.4.4 Shotcrete trials and sampling Once the basic concrete is checked against compliance of the requirements, it is necessary to make trials with shotcrete, in real scale. The trials have to be done according to the following basic work process: . .

Conditioning or preparation of the trial zone Preparation of basic concrete in the facility or in a truck mixer

© Woodhead Publishing Limited, 2010

Development and application of low-pH concretes Table 10.7

Shotcrete properties and functional requirements

Compressive strength (MPa) Static elasticity modulus (GPa) pH Water permeability K (m/s) Pumpability Slum loss (2 h) Adiabatic temperature rise (8C)

. . .

311

Shotcrete properties

Requirements for shotcrete plug

23–25 14–16 10.4 2–561011 Verified Verified < 20

> 10 < 20 < 11 < 161010 — — < 40

Test of quality of fresh and adjusted concrete Samples of basic concrete cast (cylinders) Pump and/or spray trial in the trial zone

After 28 days of curing, samples were taken from reference panels and tested. Table 10.7 shows the results obtained and the compliance with functional requirements. Therefore, according to all the results shown in this section, it could be said that it is quite possible to construct shotcrete plugs that fulfil the functional requirements needed in an underground repository of high radioactive waste using low-pH cement formulations.

10.5

Long-term durability

Due to the location and the long service life of this type of product, their durability properties must also be guaranteed. Therefore, in this section two parameters closely related to the durability of low-pH cementitious materials are evaluated: resistance to long-term groundwater aggression and the use of low-pH cements in reinforced concretes (susceptibility to corrosion). Although this chapter focuses on the degradation processes expected in low-pH cementitious materials, some ideas are now given concerning the processes expected in the region including cementitious materials and bentonite. However, it must be said that in this subject most of the studies published in the literature deal with conventional Portland concretes instead of low-pH ones (which are more recent). As already stated in previous sections, concrete pore waters originating from a Portland cement have high alkalinity and are able to react with and modify the bentonite barrier. These early cement pore waters have alkali ions that will be transported by diffusion and possibly by advection due to bentonite suction if this material is emplaced under unsaturated conditions. During a relatively short-term stage that is characterised by the leaching of dissolved alkali hydroxides, the

© Woodhead Publishing Limited, 2010

312

Geological repository systems for safe disposal

alteration of bentonite is characterised by montmorillonite dissolution and sodium and potassium zeolites (e.g. analcime, phillipsite) are formed (Bauer and Velde, 1999; Vigil de la Villa et al., 2001; Ramı´ rez et al., 2005). In a second alkaline long-term stage, the system is buffered by the portlandite dissolution and Ca2+ is the predominant cationic species in the leaching solution of the concrete. In this case, zeolites are replaced presumably by cement phases such as calcium silicate hydrates (CSHs) in the altered bentonite region. Some authors (Taylor, 1990; Berner, 1992) include a third stage where the incongruent dissolution of CSH gels and other cement phases determine the pH buffering capacity from the concrete side. Modelling simulations (Ferna´ndez et al, 2009) have shown that the mineralogical transformations in the host rock–concrete–bentonite system become significant only after 103 years. At this time, the precipitation of secondary minerals (mainly analcime, Mg-saponite, ettringite and brucite) cover most of the volume over the first centimetres of the bentonite layer on both sides of the concrete barrier and this reduces the porosity to ~5%. Total clogging is observed after 56103 years, and thereafter the concrete phases dissolve gradually and still release Ca2+, which is incorporated in the exchanger complex of bentonite, displacing the rest of the exchangeable cations. Analcime and Mg-saponite precipitate along the bentonite section over the long term (105 years).

10.5.1 Interaction of groundwaters with low-pH concretes Although concrete is stable in high humid environments, direct contact with water that is stagnant, percolating or flowing produces a diffusion of the pore solution and also alteration of the solid phases. Degradation of concrete due to leaching occurs when the hydrates in cementitious materials dissolve into the surrounding water and the precipitation of new phases takes place. This may cause a loss of strength. Although the expected degradation rate is slow, its evaluation is very important for structures near field HLW repositories, where extremely long-term stability is needed. Due to the difficulties for long timescales involved in laboratory experiments using groundwaters, accelerated tests have to be used to qualify the different concretes for deep repositories. The leaching processes of materials that contain cement pastes are a combination of chemical reactions and diffusion transport and have to be studied thermodynamically and kinetically. Cement degradation depends on physical factors including the effect of porosity, compressive strength and density, as well as on leachant characteristics including the effect of pH, flow rate, temperature and water chemical composition. In realistic conditions of HLW repositories, concrete durability is based on its interaction with clays and granite groundwaters. In general,

© Woodhead Publishing Limited, 2010

Development and application of low-pH concretes

313

groundwaters are mineralised solutions; however, their exact composition on site must be known in order to evaluate their interaction with low-pH concretes. For example, salinity of groundwater is very variable depending on the emplacement of the repository; in the south of Europe the more common groundwaters have low salinity, while in the north water with a high saline content of around 50 g/l is more probable. The presence of cementitious material may greatly alter the chemistry of water in the proposed repositories. These materials also provide a large reservoir of unstable Ca-silicate phases that will dissolve and reprecipitate at the rock–water or concrete–bentonite interfaces. Chemical interactions between water and concrete may well be dominated by dissolution kinetics of unstable amorphous and crystalline phases and precipitation kinetics of meta-stable or stable phases. The use of accelerated batch leaching tests, in order to qualify the different concretes to be used in deep repositories, enables data about the dissolution–precipitation processes to be acquired and to make comparisons between different cementitious materials. However, column leach tests are considered to be more representative of field leaching conditions because of continuous flux of the leaching solution through the monolithic material. There is enormous concern on determining the low-pH concrete behaviour under the long-term action of water in representative conditions of the real storage scenario. However, up to now, very little has been published on this subject. In general, results from the few leaching tests that have been made show good resistance of the low-pH cementitious materials against water aggression, although an altered front can be observed from the surface in all the tested samples. From leaching analyses using deionised water, Ca2+ released in the leaching solution has been described, as well as a decalcification process governed by diffusion, where the Ca2+ flux is not only balanced by the release of OH but also by that of sulphate ions (Yamamoto et al., 2007; Codina et al., 2008). However, decalcification of the low-pH cement pastes is much slower than that of OPC ones. Apart from a very low CaO/SiO2 ratio (0.3 to 0.4 for all pastes), a disappearance of ettringite and enrichment in a hydrotalcite-like phase have been reported near the leached surfaces of low-pH cement pastes (Codina et al., 2008). In leaching tests made on low-pH concretes using real groundwater (from the A¨spo¨ site) (Garcı´ a Calvo et al., 2009a) or simulated fresh groundwater (Yamamoto et al., 2007), leaching of Ca2+ is again clearly recognised. After the test periods, low-pH concretes show a small altered front (<700 μm) that can be observed from the surface. In this altered front, decalcification of the CSH gels followed by the incorporation of magnesium ions from groundwater into them (even forming MSH phases) and into the anhydrous phases (as ‘magnesia nodules’) are suggested, as graphed in Fig. 10.14, which shows

© Woodhead Publishing Limited, 2010

314

Geological repository systems for safe disposal

10.14 EDX microanalyses profiles in low-pH concrete samples based on OPC + SF after the leaching test.

the EDX microanalysis profiles obtained in the pastes of the two types of low-pH concretes. The figure shows the modifications that occur with the depth of the samples in the CaO/SiO2 ratios of the CSH gels and in the Mg2+ concentration (Garcı´ a Calvo et al., 2009a). It is quite evident that near the surface, and at a depth less than 1 mm, there is a decalcification of the CSH gels and solid phases rich in Mg2+ that comes from the groundwater used. However, after this small zone, the rest of the paste of both types of concrete has a similar composition to those obtained before testing. When low-pH cementitious materials are immersed in saline water, leaching of Ca2+ is much higher, but Si leaching was the lowest (Yamamoto et al., 2007). Calcite and smectite were shown to form a protective coating at the surface of cementitious materials in contact with this saline water. All the cases discussed above were made in low-pH cementitious materials based on OPC with mineral additions. In the case of low-pH concretes based on CAC plus mineral additions (SF and FA), results of leaching tests show a good resistance of the low-pH concretes based on CAC + SF against groundwater aggression. Moreover, calcite precipitation observed on the leached surface could be playing a protective role against water aggression. In those concretes based on CAC + FA, an altered front can be observed from the surface in the tested samples. In this altered front, a decalcification of the CASH phases and an incorporation of magnesium ions from

© Woodhead Publishing Limited, 2010

Development and application of low-pH concretes

315

10.15 Top: evolution of the rebar corrosion rate in low-pH mortar and OPC mortar. Bottom: evolution of the pore fluid pH in low-pH mortar and OPC mortar.

groundwater into them are suggested, as it occurred in low-pH concretes based on OPC (Garcı´ a Calvo et al., 2009b).

10.5.2 Use of low-pH cements in reinforced concretes The disposal of high-activity wastes in underground repositories often requires the use of concrete for structural support. Concrete should be compatible with the surrounding environment and hence its pH should be

© Woodhead Publishing Limited, 2010

316

Geological repository systems for safe disposal

low enough to prevent the clay from altering. If the concrete requires some reinforcement, its susceptibility to corrosion when embedded in a low-pH concrete should be analysed. Preliminary results of this are presented in Fig. 10.15. The figure shows the corrosion rate measured in rebars embedded in mortars fabricated using a low-pH cement formulation compared with one measured in reference reinforcement mortars (using OPC without mineral admixtures) (Garcı´ a Calvo et al., 2008). On the bottom of Fig. 10.15, the pore fluid pH values of both types of mortar are presented. It is clear that the lower alkaline pore fluid of the low-pH cementitious materials generates a significant increase in the corrosion velocity of conventional reinforcement. In fact, the increase of the corrosion rate coincides with the decrease of the pore fluid pH below 12.

10.6

Sources of further information and advice

In the following review, websites of national waste management agencies and other entities or research projects related with the chapter subject are given: .

Main waste management agencies involved in the development of high level nuclear waste repositories: ANDRA, France (www.andra.fr) ENRESA, Spain (www.enresa.es) EPA, USA (www.epa.gov) NAGRA, Switzerland (www.nagra.ch) NDA, United Kingdom (www.nda.gov.uk) NRC, USA (www.nrc.gov) NUMO, Japan (www.numo.or.jp) NWMO, Canada (www.nwmo.ca) ONDRAF/NIRAS, Belgium (www.nirond.be) POSIVA, Finland (www.posiva.fi) SKB, Sweden (www.skb.se)

. . . . .

Nuclear Energy Agency (OECD): www.nea.fr. This website contains many links related to national and international entities involved in nuclear energy management. International Energy Agency: www.iea.org The Integrated European Project ESDRED. Engineering studies and demonstration of repository designs: www.esdred.info Committee on Radioactive Waste Management (United Kingdom), CoRWM: www.corwm.org.uk Second European Project of Communities Waste Management, COWAM2: www.cowam.org

© Woodhead Publishing Limited, 2010

Development and application of low-pH concretes

10.7

317

References

Adenot F and Buil M (1992), ‘Modelling of the corrosion of the cement paste by deionized water’, Cement and Concrete Research, 22, 489–496. Alonso M C, Hidalgo A, Garcı´ a J L, Ferna´ndez Luco L, Garcı´ a-Sin˜eriz J L and Ba´rcena I (2006), ‘ESDRED project, design of low-pH concrete for the construction of shotcrete plugs’, Deliverable 2.1-3.1 WP-2.1-3.1Module 4, EC Contract FI6W-CT-2004-508851, European Commission. Alonso M C, Fernandez-Luco L, Garcı´ a J L, Hidalgo A and Huertas F (2007), ‘Low-pH cementitious materials. Design and characterization’, in Proceedings of the 12th International Congress on the Chemistry of Cement, Montreal, 15 pp. Andac O and Glasser F P (1995), ‘Microstructure and microchemistry of calcium sulfoaluminate cement’, Material Research Society Symposium Proceedings, 370, 135–142. Andac O and Glasser F P (1999), ‘Pore solution composition of calcium sulfoaluminate cement’, Advances in Cement Research, 11(1), 23–26. Barneyback R S and Diamond S (1981), ‘Expression and analysis of pore fluids from hardened cement pastes and mortars’, Cement and Concrete Research, 11, 279– 285. Bauer A and Velde B (1999), ‘Smectite transformation in high molar KOH solutions’, Clay Minerals, 34(2), 259–273. Bentz D P, Jensen O M, Coats A M and Glasser F P (2000), ‘Influence of silica fume on diffusivity in cement-based materials – I – experimental and computer modelling studies on cement pastes’, Cement and Concrete Research, 30, 953– 962. Berner U (1992), ‘Evolution of pore water chemistry during degradation of cement in a radioactive waste repository environment’, Waste Management, 12, 201–219. Braney M C, Haworth A, Jefferies N L and Smith A C (1993), ‘A study of the effects of an alkaline plume from a cementitious repository on geological materials’, Journal of Contaminant Hydrology, 13, 379–402. Buzzi O, Boulon H M, Deleruyelle F and Besnus F (2007), ‘Hydromechanical study of rock–mortar interfaces’, Physical Chemistry Earth Parts A/B/C, 32, 820– 831. Capmas A and George C M (1994), ‘Durability of calcium aluminate cement concretes’, in Proceedings of Engineering Foundation Conference on Advances in Cement and Concrete Research, Durham, New Hampshire, Materials Engineering Division, ASCE, pp. 377–405. Castellote M, Alonso M C and Andrade C (2002), ‘Determination of the OH content in the aqueous phase of hardened cementitious materials by an empirical leaching method’, Mat. de Const., 52(265), 39–56. Cau Dit Coumes C, Courtois S, Nectoux D, Leclerq S and Bourbon X (2006), ‘Formulating a low-alkalinity, high-resistance and low-heat concrete for radioactive waste repositories’, Cement and Concrete Research, 36, 2152–2163. Chow L C (2000), ‘Calcium phosphate cements: chemistry, properties and applications’, Material Research Society Symposium Proceedings, 599, 27–37. Codina M, Cau-dit-Coumes C, Le Bescop P, Verdier J and Ollivier J P (2008),

© Woodhead Publishing Limited, 2010

318

Geological repository systems for safe disposal

‘Design and characterization of low-heat and low-alkalinity cements’, Cement and Concrete Research, 38, 437–448. Collepardi M, Monosi S and Piccioli P (1995), ‘The influence of pozzolanic materials on the mechanical stability of aluminous cement’, Cement and Concrete Research, 25, 961–968. Cong X and Kirkpatrick R J (1996), ‘29Si MAS NMR study of the structure of calcium silicate hydrate’, Advances in Cement Based Material, 3 (3/4), 144–156. Ding J, Fu Y and Beaudoin J J (1995), ‘Stra¨tlingite formation in high alumina cement-silica fume systems: significance of sodium ions’, Cement and Concrete Research, 25, 1311–1319. Driessens F C M (1995), ‘Chemistry of calcium phosphate cements’, in Proceedings of 4th European Ceramic Conference, Riccione, Italy, 2–6 October 1995, vol. 8, pp. 77–83. Durekovic A (1995), ‘Cements pastes of low water to solid ratio: an investigation of the porosity characteristics under the influence of a superplasticizer and silica fume’, Cement and Concrete Research, 25(2), 365–375. Durning T G and Hicks M C (1991), ‘Using microsilica to increase concrete’s resistance to aggressive chemicals’, Concrete International, 13, 42–48. El Jazairi B (1987), ‘The properties of hardened mpc mortar and concrete relevant to the requirements of rapid repair of concrete pavements’, Concrete, 21, 25–31. Faucon P, Adenot F, Jacquinot J F, Petit J C, Cabrillac R and Jorda M (1998), ‘Long-term behaviour of cement pastes used for nuclear waste disposal: review of physico-chemical mechanisms of water degradation’, Cement and Concrete Research, 28, 847–857. Feldman R F and Sereda P J (1970), ‘A new model for hydrated Portland cement and its practical implications’, Engineering Journal, (Canada), 53(8/9), 53–59. Feliu S, Andrade C, et al. (1989). Manual de Inspeccion de Obras Dan˜adas por Corrosio´n de Armaduras, Edt ACOR, p. 23. Fentiman C H, Rashid S, Bayoux J P, Bonin A and Testud M (1990), ‘The effect of curing conditions on the hydration and strength development in Fondu: Slug’, in Proceedings of the International Symposium on Calcium Aluminate Cements, Queen Mary and Westfield College, University of London, 9–11 July 1990, Chapman and Hall, London, p. 272. Ferna´ndez R, Cuevas J and Ma¨der Urs K (2009), ‘Modelling concrete interaction with a bentonite barrier’, European Journal of Minerals, 21, 177–191. Fu Y, Ding J and Beaudoin J J (1995), ‘Mechanism of stratlingite formation in high alumina cement-siliceous material systems’, in Proceedings of the 2nd CANMET/ACI International Symposium, Las Vegas, Nevada, 11–14 June 1995. Fuentes-Cantillana J L, Huertas F and Cuevas J (2005), ‘Potential applications of shotcrete techniques in HLW repositories. Engineering aspects and chemical implications’, in R&D on Low-pH Cement for a Geological Repository, Workshop, Madrid, June 2005, pp. 129–139. Garcı´ a Calvo J L, Alonso M C, Ferna´ndez Luco L, Hidalgo A and Sa´nchez M (2008), ‘Implications of the use of low-pH cementitious materials in HARW repositories’, in Proceedings of International Event, Conrox AB, edited by G Ba¨ckblom, Prague, pp. 213–224. Garcı´ a Calvo J L, Hidalgo A, Alonso M C and Ferna´ndez Luco L (2009a),

© Woodhead Publishing Limited, 2010

Development and application of low-pH concretes

319

‘Development of low-pH cementitious materials for HLW repositories. Resistance against ground water aggression’, Cement and Concrete Research Journal, available online 16 December 2009. Garcı´ a Calvo J L, Hidalgo A, Alonso M C and Ferna´ndez Luco L (2009b), ‘Low pH concretes based on CAC for underground repositories of HLW. Resistance with respect to ground water aggression’, in Long-term Performance of Cementitious Barriers and Reinforced Concrete in Nuclear Power Plants and Waste Management, EFC Event 314 NUCPERF 2009, edited by V. L’Hostis, R Gens and C Galle´, Rilem Publications SARL, pp. 99–108, ISBN: 978-235158-072-1. Garcı´ a-Sin˜eriz J L, Alonso M C, Blu¨mling P, Pettersson S, Salo J-P and Huertas F (2004), ‘ESDRED, Input data and functional requirements’, Deliverable 1 Module 4 WP-1, EC Contract FI6W-CT-2004-508851, European Commission. Gaucher E C and Blanc P (2006), ‘Cement/clay interactions – a review: experiments, natural analogues, and modeling’, Waste Management, 26, 776–788. Glasser F P and Atkins M (1994), ‘Cements in radioactive waste disposal’, Materials Research Society Bulletin, 12, 33–39. Glasser F P and Marr J (1985), ‘The alkali binding potential of OPC and blended cements’, Il Cemento, 82, 85–94. Gon˜i S, Andrade C and Page C L (1991), ‘Corrosion behaviour of steel in high alumina cement mortar samples: effect of chloride’, Cement and Concrete Research, 21(4), 635–646. Hidalgo A, Castellote M, Llorente I, Alonso C and Andrade C (2004), ‘ECOCLAY II. Effects of cement on clay barrier performance - phase II’, Final Report EC Contract FIKW-CT-2000-00028, European Commission. Hidalgo A, Garcı´ a J L, Cruz M C, Ferna´ndez L and Andrade C (2005), ‘Testing methodology for pH determination of cementitious materials. Application to low pH binders for use in HLNWR’, in R&D on Low-pH Cement for a Geological Repository, Workshop, Madrid, June 2005, pp. 48–59. Hidalgo A, Garcı´ a Calvo J L, Garcı´ a Olmo J, Petit S and Alonso M C (2008), ‘Microestructural evolution of calcium aluminate cements hydration with silica fume and fly ash additions by scanning electron microscopy, mid and near infra red spectroscopy’, Journal of the American Ceramics Society, 91(4), 1258– 1265. Hidalgo A, Garcı´ a J L, Alonso M C, Ferna´ndez L and Andrade C (2009), ‘Microstructure development in mixes of calcium aluminate cement with silica fume or fly ash’, accepted for publication in Journal of Thermal Analytical Calorimetry. Hong S Y and Glasser F P (1999), ‘Alkali binding in cement pastes. Part I. The CSH phase’, Cement and Concrete Research, 29, 1893–1903. Huang C Y and Feldman R F (1985), ‘Influence of silica fume on the microstructural development in cement mortars’, Cement and Concrete Research, 15(2), 285– 294. Hugo-Persson T, Lagerblad B and Vogt C (2005), ‘Selective stabilization of deep core drilled boreholes using low-pH cement’, R&D on Low-pH Cement for a Geological Repository, Workshop, Madrid, June 2005, pp. 103–117. Janotka I, Krajci L, Ray A and Majumdar S (2003), ‘The hydration phase and pore

© Woodhead Publishing Limited, 2010

320

Geological repository systems for safe disposal

structure formation in the blends of sulfoaluminate–belite cement with Portland cement,’ Cement and Concrete Research, 33, 489–497. Kalogridis D, Kostogloudis G, Ftikos C and Malami C (2000), ‘A quantitative study of non expansive sulfoaluminate cement on the corrosion of steel reinforcement, Cement and Concrete Research, 30, 1731–1740. Kasselouri V, Tsakiridis P, Malami C, Georgali B and Alexandridou C (1995), ‘A study on the hydration products of a non-expansive sulfoaluminate cement’, Cement and Concrete Research, 25(8), 1726–1736. Klur I, Pollet B, Virlet J and Nonat A (1998), C–S–H Structure Evolution with Calcium Content by Multinuclear NMR, edited by P. Colombet, Springer, Berlin, pp. 119–141. Kronlo¨f A (2005), ‘Injection grout for deep repositories – low-pH cementitious grout for larger fractures: testing technical performance of materials, Working Report 2004-45, POSIVA OY. Lagerblad B, Fja¨llberg L and Vogt C (2006), ‘Low-pH shotcrete for rock support: report on selected formulation including test plan for pilot and field testing’, Appendix 2.D 2.2, 3.2 & 4.2EC Contract FI6W-CT-2004-508851, European Commission. Larbi J A, Fraay A L A and Bijeri J M M (1990), ‘The chemistry of the pore fluid of silica fume-blended cement systems’, Cement and Concrete Research, 20, 506– 516. Lemaıˆ tre J (2001), Bioce´ramiques et biociments re´sorbables pour le comblement Osseux, in Traite´ des Mate´riaux, Mate´riaux Emergents, edited by C Janot and B Ilschner, vol. 19, Presses Polytechniques et Universitaires Romandes, Lausanne, Switzerland, pp. 289–314. Li L, Jingak N and Hartt W H (2005), ‘Ex situ leaching measurement of concrete alkalinity’, Cement and Concrete Research, 35, 277–283. Longuet P, Burglen L and Zelwer A (1973), ‘The liquid phase of hydrated cement’, Revue de Mate´riaux Construction Travaux Publique, 676, 35–41. Lovera P, Adenot F, Jorda M and Cabrillac R (1997), ‘Physico-chemical transformation of sulfated compounds during the leaching of highly sulfated cemented wastes’, Cement and Concrete Research, 57(27), 1523–1532. Majumdar A J and Singh B (1992), ‘Properties of some blended high-alumina cements’, Cement and Concrete Research, 22, 1101–1114. Majumdar A J, Singh B and Edmonds R N (1990a), ‘Hydration of Secar 71 aluminous cement in presence of granulated blast furnace slag’, Cement and Concrete Research, 20, 7–14. Majumdar A J, Singh B and Edmonds R N (1990b), ‘Hydration of mixtures of ‘‘ciment fondu’’ aluminous cement and granulated blast furnace slag’, Cement and Concrete Research, 20, 197–208 Metha P K (1986), Concrete Structure, Properties and Materials, edited by W H Hall, Prentice-Hall, Inc. Midgley H G and Midgley A (1975), ‘The conversion of high alumina cement’, Magazine of Concrete Research, 27, 59–77. Oh B H, Cha S W, Jang B S and Jang S Y (2002), ‘Development of high performance concrete having high resistance to chloride penetration’, Nuclear Engineering and Design, 212, 221–231.

© Woodhead Publishing Limited, 2010

Development and application of low-pH concretes

321

Pera J and Ambroise J (1998), ‘Fiber reinforced magnesia phosphate cement composites for rapid repair’, Cement and Concrete Composition, 20, 31–39. Popovics S, Rajendran N and Penko M (1987), ‘Rapid hardening cements for repair of concrete’, ACI Materials Journal, 84(1), 64–73. Quillin K, Osborne G, Majumdar A and Singh B (2001), ‘Effects of w/c ratio and curing conditions on strength development in BRECEM concretes’, Cement and Concrete Research, 31, 627–632. Ramı´ rez S, Cuevas J, Vigil R and Leguey S (2002), ‘Hydrothermal alteration of ‘‘La Serrata’’ bentonita (Almeria, Spain) by alkaline solutions’, Applied Clay Science, 21, 257–269. Ramı´ rez S, Righi D and Petit S (2005), ‘Alteration of smectites induced by hydrolytic exchange’, Clay Minerals, 40(1), 15–24. Ra¨sa¨nen V and Penttala V (2004), ‘The pH measurement of concrete and smoothing mortar using a concrete powder suspension’, Cement and Concrete Research, 34, 813–820. Rice G, Miles N and Farris S (2007), ‘Approaches to control the quality of cementitious PFA grouts for nuclear waste encapsulation’, Powder Technology, 174, 56–59. Rivas Mercury J M, de Aza A H, Turrillas X and Pena P (2003a), ‘Hidratacio´n de los cementos de aluminatos de calcio (Parte I)’, Boletı´n de la Sociedad Espan˜ola de Cera´mica y Vidrio, 42, 269–276. Rivas Mercury J M, de Aza A H, Turrillas X and Pena P (2003b), ‘Hidratacio´n de los cementos de aluminatos de calcio. Parte II: Efecto de las adiciones de sı´ lice y alu´mina’, Boletı´n de la Sociedad Espan˜ola de Cera´mica y Vidrio, 42, 361–368. Rivas Mercury J M, Turrillas X, de Aza AH and Pena P (2006), ‘Calcium aluminates hydration in presence of amorphous SiO2 at temperatures below 90 8C’, Journal of Solid State Chemistry, 179, 2988–2997. Rivas Mercury J M, Pena P, de Aza A H, Turrillas X, Sobrados I and Sanz J (2007), ‘Solid-state 27Al and 29Si NMR investigations on Si-substituted hydrogarnets’, Acta Materialia, 55, 1183–1191. Rozale´n M L, Huertas F J, Brady P V, Cama J, Garcı´ a-Palma S and Linares J (2008), ‘Experimental study of the effect of pH on the kinetics of montmorillonite dissolution at 25 8C’, Geochim. Cosmochim. Acta, 72, 4224– 4253. Saeki T and Monteiro PJM (2005), ‘A model to predict the amount of calcium hydroxide in concrete containing mineral admixtures’, Cement and Concrete Research, 35, 1914–1921. Savage D and Benbow S (2007), ‘Low pH cements’, SKI report 2007:32. Scrivener K, Cabiron J L and Letourneux R (1999), ‘High performance concretes from calcium aluminate cements’, Cement and Concrete Research, 29, 1215– 1223. Seehra S S, Gupta S and Kumar S S (1993), ‘Rapid setting magnesium phosphate cement for quick repair of concrete pavements – characterization and durability aspects’, Cement and Concrete Research, 23(2), 254–266. Soudee E and Pera J (2000), ‘Mechanism of setting reaction in magnesia-phosphate cements’, Cement and Concrete Research, 30, 315–321. Stronach S A and Glasser F P (1997), ‘Modelling the impact of abundant geochemical components on phase stability and solubility of the CaO–SiO2–

© Woodhead Publishing Limited, 2010

322

Geological repository systems for safe disposal

H2O system at 25 8C: Na+, K+, SO42, Cl and CO32, Advances in Cement Research, 9, 167–181. Taylor, H F W (1987), ‘A method for predicting alkali ion concentrations in cement pore solutions’, Advances in Cement Research, 1, 5–16. Taylor H F W (1990), Cement Chemistry, Academic Press Ltd. Torii K and Kawamura M (1994), ‘Effects of fly ash and silica fume on the resistance of mortar to sulfuric acid and sulfate attack’, Cement and Concrete Research, 24, 361–370. Ulm F J, Heukamp F H and Germaine J T (2002), ‘Residual design strength of cement-based materials for nuclear waste storage systems’, Nuclear Engineering and Design, 211, 51–60. Vigil de la Villa R, Cuevas J, Ramirez S and Leguey S (2001), ‘Zeolite formation during the alkaline reaction of bentonite’, European Journal of Minerals, 13(3), 635–644. Vuorinen U and Lehikoinen J (2005), ‘Low-pH grouting cements – results of leaching experiments and modelling’, in R&D on Low-pH Cement for a Geological Repository, Workshop, Madrid, June 2005, pp. 79–92. Wagh A S, Cunnana J C, Singh D, Reed D T, Armstrong S, Subhan W and Chawla N (1997), ‘Chemically phosphate ceramics for radioactive and mixed waste solidification and stabilization’, in Proceedings of the Conference Technology and Programs for Radioactive Waste Management and Environmental Remediation, WM’93, Tucson, Arizona, vol. 2, pp. 1613–1617. Xiandong C and Kirkpatrick R J (1993), ‘Hydration of calcium aluminate cements. A solid state 27Al NMR study’, Journal of the American Ceramic Society, 76, 409–416. Yamamoto T, Imoto H, Ueda H and Hironaga M (2007), ‘Leaching alteration of cementitious materials and release of organic additives – study by Numo and Criperi’, in R&D on Low-pH Cement for a Geological Repository, 3rd Workshop, Paris, June 2007, pp. 52–61. Yogendran V, Langan B W, Haque M N and Ward M A (1987), ‘Silica fume in high strength concrete’, ACI Materials Journal, March–April 1987, 124–129. Zhang L and Glasser F P (2002), ‘Hydration of calcium sulfoaluminate cement at less than 24 hours’, Advances in Cement Research, 14(4), 141–155.

© Woodhead Publishing Limited, 2010

11 Development and application of smectitic buffer and backfill materials in geological repository systems R . P U S C H , SWECO AB/Geodevelopment International AB, Sweden

Abstract: High-level waste canisters in bored holes are generally proposed to be isolated from the rock by dense smectitic clay, termed buffer. Other rooms can be backfilled with less smectite-rich soil. Smectites expand on hydration and confine the canisters effectively. These clays are very tight and practically the transport of water and ions takes place by diffusion. Maturation of the buffer and backfill is slow and associated with precipitation of salt. Long-term changes include slow settlement of canisters and conversion from smectite to non-expanding minerals associated with precipitation of cementing agents like silica and iron compounds. Key words: clay, canister, creep, hydraulic conductivity, smectite.

11.1

Introduction

11.1.1 Clays for isolation of high-level waste (HLW) Most repository design concepts imply that the waste containers, termed canisters here, will be embedded by a very tight medium with sufficient bearing capacity to avoid significant movement of the canisters but with sufficient softness to even out stresses generated by tectonic movement (Pusch, 2008). It must have a potential for expansion so that tight contact with canister and rock is maintained and voids that arise from strain caused by external forces or internal processes, like temporary shrinkage due to desiccation, can self-heal. Also, it must be chemically compatible with the canister material and surrounding rock and groundwater. These properties must be preserved for tens to hundreds of thousands of years. The only

323 © Woodhead Publishing Limited, 2010

324

Geological repository systems for safe disposal

material that can provide all this is smectitic clay, like bentonites or certain in situ weathered granitic and amphibolitic rock.

11.1.2 Smectite minerals Smectite particles, which consist of very thin lamellae with a length of less than thousands of a millimetre, have a chemical composition that can be generalised as (Na, Ca, Mg, H2O)(Al, Fe, Mg)2(Si, Al)4O10(OH)2. The bonds between the lamellae, which appear as stacks of 3 to 30 units, are weak and water, inorganic cations and organic molecules can occupy the space between them (Kehres, 1983). At high densities such clays have an extremely low hydraulic conductivity because the dominant part of the porewater is nearly immobile (Pusch and Yong, 2006). The surface area of a gram of smectite can be 1000 m2, which makes the electrically charged stacks very active in forming stable clay gels, even at very low densities. The very large surface area also means that the ion exchange capacity is up to 100 meq per 100 g. The exchangeable cations can be substituted by mono-, di- and multivalent cations like Ca2+, Mg2+ and various radionuclides. Isomorphous replacement of lattice cations and the amount of interlamellar water molecules are related to the spacing of the lamellae, which gives diagnostic values of the reflection peaks of X-ray diffraction spectra (Gueven and Huang, 1990; Pusch and Yong, 2006). The particles, or rather stacks of smectite lamellae (Fig. 11.1), typically aggregate. The figure shows two coupled lamellae of the commonly favoured crystal model of montmorillonite, which is the major smectite species in most expansive clays. There are other smectite minerals but they have a lower expansion potential and one, beidellite, is sensitive to potassium, the uptake of which from the groundwater can cause collapse of the stacks of lamellae, growth of voids and loss of expandability.

11.1.3 Microstructural constitution of smectite clays The bulk physical properties of smectite clays are controlled by the microstructure. Figure 11.2 shows the typical network of particle aggregates with more water in interlamellar positions in smectites than in nonexpansive clays like illite (hydrous mica) and kaolinite (Kehres, 1983). The large specific surface area of the smectite lamellae system that binds much of the porewater, and the very small size of the voids between the particle aggregates, less than 1 thousand of a millimeter in dense clay, combine to give an extremely low hydraulic conductivity and a significant expandability (Pusch and Yong, 2006). They make smectitic clays excellent as engineered barriers in HLW repositories.

© Woodhead Publishing Limited, 2010

Development and application of buffer and backfill materials

325

11.1 Commonly assumed constitution of a stack of two montmorillonite lamellae with 10 A˚ thickness in dehydrated form. Cations and water molecules, which can form up to three hydrates, are located in the interlamellar space.

11.2 Microstructural units in clay of smectite type (left) and clays with illite or kaolinites (right) as major mineral constituents (Pusch and Yong, 2006). Networks of particle aggregates are built up with more water in interlamellar positions in smectite clays than in illites and kaolinite.

© Woodhead Publishing Limited, 2010

326

Geological repository systems for safe disposal

11.2

Types, properties and fabrication of the buffer

The waste containers need protection against external impact like tectonically induced shearing by slip along fractures that intersect the deposition holes. This requires embedment in a ductile, very tight medium – ‘buffer’ – with sufficient bearing capacity to keep the canisters in position and sufficient softness to absorb strain, and also with an ability to maintain tight contact with the canisters and the surrounding rock. The dense smectite clay used for this purpose should have a dry density – expressed as the ratio of the mass of solids and the volume including all kinds of voids – in the range of 1450 to 1900 kg/m3, which corresponds to 1900 to 2200 kg/m3 in fully water saturated form. Higher density would give a too high swelling pressure on the rock while lower density than about 1450 kg/m3 can lead to significant settlement of the canisters and to comprehensive microbial activity in the buffer (Pedersen, 1995). The major reason for selecting a high density of the buffer is, however, that the hydraulic conductivity is so low that transport of ions and water takes place by diffusion and not flow, while for lower densities they can migrate quicker because water flow contributes to the transport (Fig. 11.3). The build-up of a swelling pressure and the evolution of the hydraulic conductivity are determined in the laboratory by use of rigid ‘swelling pressure oedometers’ (Fig. 11.4). The evolution of the swelling pressure is rather slow and dependent on the rate with which water is provided by the surrounding rock.

11.3 Hydraulic conductivity of 100% smectite (montmorillonite) clay under very high hydraulic gradients for quick evaluation. The upper curve represents clay with Na as the major adsorbed cation and clay with low-electrolyte water. The lower is typical of pure smectite clay saturated and percolated by water of ocean salinity with Ca as the major cation (Pusch and Yong, 2006).

© Woodhead Publishing Limited, 2010

Development and application of buffer and backfill materials

327

11.4 Schematic section of an oedometer for determination of the hydraulic and gas conductivities and the swelling pressure. Air-dry clay grains are compacted in the cell and saturated with water while recording the swelling pressure, as illustrated by the right picture (montmorillonite-rich clay with 1700 kg/m3 density at saturation with distilled water). When equilibrium is reached, a hydraulic gradient is applied and the percolation rate measured as a function of time: (a) piston, (b) strain gauge, (c) fine filter, (d) clay sample and (e) inlet of fluid.

The diagram on the upper right of Fig. 11.4 illustrates a most important property of smectite-rich clays, i.e. to exert a swelling pressure on the confinement. This ability depends on the nature of the cations that are adsorbed to balance the negative charge of the crystal lattice of the individual lamellae. At high densities the pressure is due to the potential of ‘steric’, interlamellar water establishing hydrate layers, while at lower densities, it is caused by osmotic effects, controlled by the force fields between adjacent stacks of lamellae (Pusch and Yong, 2006). The diagram shows the typical evolution of the pressure at unlimited access to water, like in the oedometer, exhibiting an early rise to a peak value and then a slight drop followed by weeks with small changes but ending with a constant value. In practice, the evolution of the swelling pressure is controlled by the ability of the surrounding rock to give off water and, if very limited, can be a threat to acceptable performance of the buffer, as we will show in this chapter. The diagram in Fig. 11.5 illustrates the relationship between swelling pressure and density at complete saturation of smectite-rich clays with Na and Ca as major adsorbed cations. While Na-clays show some swelling pressure at densities even down to 1050 kg/m3, Ca-clays need to have a density of at least 1600 kg/m3 to exert a measurable swelling pressure. The reason is the strong coagulation of Ca-smectite, which makes it impossible

© Woodhead Publishing Limited, 2010

328

Geological repository systems for safe disposal

11.5 Typical relationship between swelling pressure and density of smectite-rich clay. Left: schematic picture of the relationship between the microstructural constitution of smectite clays and the swelling pressure for different densities. Right: the curve starting at the origin in the diagram represents Na montmorillonite (MX-80) saturated with lowelectrolyte water and the other Ca montmorillonite saturated with lowelectrolyte water (Pusch, 2008).

to prepare a coherent sample of Ca-clay of lower density than this value; it would simply lead to separation of clay and water. For the high density required for the clay surrounding HLW canisters, changes in porewater composition and total dissolved salt (TDS) content have a rather small impact on conductivity and swelling pressure. This is because the content of microstructural elements with sufficiently low density to react on

© Woodhead Publishing Limited, 2010

Development and application of buffer and backfill materials

329

such changes is very low. For clays with bulk densities lower than about 1800 kg/m3, this frequency is higher and the influence of porewater chemistry on conductivity and swelling pressure is therefore significant. It should be noticed that for densities at saturations higher than about 2150 kg/m3 the swelling pressure increases beyond 20 MPa. For smectites the mechanical properties, like shear strength and creep strain, depend very much on the nature of the particle contact (Fig. 11.5). Grain pressure is transferred via electrical double layers and merged hydrate layers on the basal surfaces of the stacks of lamellae. True mineral/mineral contacts do not exist unless the pressure is several hundred MPa. The ion diffusion properties of smectite clays are controlled by the microstructure and ion charge. Cations migrate by pore and surface diffusion as well as by transport through the interlamellar space, while anions are largely excluded from the interlamellar space for charge reasons (‘Donnan exclusion’). The anion diffusion capacity is therefore very low at high densities while the cation diffusion capacity is less dependent on density (Fig. 11.6). In performance analysis and assessment one needs to consider not only the hydraulic properties but also the issues referred to as thermal (T), hydraulic (H), mechanical (M), chemical (C), biological (B) and radiological (R) processes and properties (Svemar, 2005). The first stage of evolution of the buffer comprises several complex processes that can take a dozen years in richly water-bearing rock to hundreds of years in a tight environment like

11.6 Measured ion effective diffusivities for smectite clay (Pusch, 2008). Upper: monovalent cations. Lower: monovalent anions like chlorine.

© Woodhead Publishing Limited, 2010

330

Geological repository systems for safe disposal

argillaceous rock. Later stages involve mineral changes in the few thousand years of enhanced temperature.

11.2.1 Thermal impact The heat produced by the radioactive decay of the HLW counteracts the uptake of water by the buffer clay, complete hydration of which may take tens to hundreds of years depending on the capacity of the surrounding rock to provide the clay with water. The hydration is most delayed in sedimentary rock, like clay shales, and in fracture-poor crystalline rock. In a long perspective the heating of the clay has an impact on its mineralogical composition and microstructural constitution. For temperatures lower than about 90 oC temperature-generated changes are limited even in a 100 000 year perspective (Pusch and Yong, 2006).

11.2.2 Hydraulic processes Once the buffer clay has become fully water saturated groundwater can percolate it and transport possibly released radionuclides. The hydraulic conductivity of the clay is therefore of fundamental importance and the aim is to keep it lower than that of the surrounding rock, which is on the order of 1010 m/s for crystalline rock and considerably lower for clay shales (Pusch, 2008).

11.2.3 Mechanical properties A sufficiently high shear strength is required for the canister-bearing capacity of the buffer clay but too high a strength means that the stiffness is high and that shearing along fractures that intersect the deposition holes can transfer strong forces to the HLW canisters and damage them. The density must therefore be lower than about 2100 kg/m3 but higher than about 1900 kg/m3 at complete water saturation (Pusch, 2008).

11.2.4 Chemical, biological and radiological impact The groundwater chemistry affects the mechanical strength and hydraulic conductivity as well as the long-term integrity of the buffer, as is well known and documented by comprehensive international research projects (Svemar, 2005). Microbes have a negligible impact on the performance of smectite buffer clay as long as the density exceeds about 1900 kg/m3, and even strong gamma radiation does not cause practically important damage (Pusch, 2008).

© Woodhead Publishing Limited, 2010

Development and application of buffer and backfill materials

11.3

331

Design and performance of the buffer

11.3.1 Design principle We will take here the so-called KBS-3V concept as an example when considering the first phase of maturation of the buffer (Fig. 11.7). It is favoured by the Swedish Nuclear Fuel and Waste Co. (SKB) and implies that a system of blasted deposition tunnels is constructed at 400–500 m depth and that the canisters are placed in bored deposition holes with about 8 m depth and 1.85 m diameter (Svemar, 2005). The buffer that embeds the canisters interacts with the surrounding rock and the overlying backfill,

11.7 SKB’s concept KBS-3V. The 50 mm wide gap between the buffer and the rock is filled with clay pellets while the 10 mm space between the buffer and canister is left open.

© Woodhead Publishing Limited, 2010

332

Geological repository systems for safe disposal

which both supply it with water. Backfilling of the tunnels does not need to be made in conjunction with the waste disposal but should be made relatively soon afterwards in order to make it possible for the piezometric pressure to rise and supply the buffer with water.

11.3.2 Experimental The buffer must be manufactured and placed so that the homogeneity of the ultimately matured clay becomes high and the density so high that transport of ions will take place by diffusion and not by flow. This has led to the criterion that the minimum density of the matured buffer should be 1950 kg/ m3 (Svemar, 2005; Pusch and Yong, 2006; Pusch, 2008). Figure 11.8 shows how the buffer is constructed. Big blocks of smectite-rich clay are placed to form a column with space reserved for the canister that is inserted from a radiation-protecting transport container. Figure 11.9 can be taken as a basis for describing the maturation process. The various processes leading to maturation can be described as follows: .

Porewater in the buffer migrates from the hot part of the buffer towards the colder walls of the deposition hole. The outer, colder part of the buffer expands and exerts a swelling pressure that tends to compress the warmer, less wetted buffer.

11.8 Large block of highly compacted bentonite powder being placed in a bored deposition hole with 1.85 m diameter and 8 m depth for surrounding an HLW canister (Pusch and Yong, 2006).

© Woodhead Publishing Limited, 2010

Development and application of buffer and backfill materials

333

11.9 Engineered barrier components.

.

.

Water enters the deposition holes, which causes additional wetting of the buffer from the rock/buffer contact and successive migration of water towards the hot canister. A swelling pressure is first built up in the porous pellet fill next to the blocks and in the coldest part of the dense clay blocks by which the blocks are pressed against the canister. Water is successively sorbed also by the hotter part of the blocks. The block system expands and starts compressing the then fully water-saturated pellet fill. The wetting involves transport of water in liquid form from the cold towards the hot part of the buffer where partial vaporisation takes place, followed by flow back in vapour form towards the colder part where condensation takes place. This is in turn followed by migration of water towards the hot side where vaporisation takes place, etc.

Certain dissolved components precipitate in the hot part of the buffer, like sodium chloride and calcium sulphate (Pusch, 2008). They contribute to corrosion of the canister and loss in expandability of the buffer (Fig. 11.10).

© Woodhead Publishing Limited, 2010

334

Geological repository systems for safe disposal

11.10 Scanning electron micrographs of precipitated salt in hydrothermal experiment (micrographs by J. Kasbohm, Greifswald University, Geography and Geology Department). Left: normal MX-80 clay with typical interwoven stacks of very thin montmorillonite lamellae. Centre: precipitated NaCl crystals in the clay matrix. Right: precipitated CaSO4 (gypsum) in the matrix (Pusch et al., 2005).

The evolution of the buffer involves a number of dissolution/precipitation processes that affect the physical performance of the buffer. Thus, all silicate minerals in the buffer clay undergo slight dissolution in the period about 1000 years long, the ‘thermal period’, by which silica is released and precipitates, causing cementation that becomes obvious at cooling (Grindrod and Takase, 1993). This can reduce the expandability of the clay and preserve microstructural changes, like fissuring, causing stiffening and enhanced hydraulic conductivity (Pusch, 2008). The smectite powder grains have a water content that is related to the relative humidity of the air in which they are stored. At 50–70% relative humidity the water content of the grains is about 10%, implying that the interlamellar space holds 1–2 hydration layers. Compaction of the grains under 100 MPa pressure to yield blocks welds the grains together leaving only small voids between the stacks of lamellae. The voids have the form of channels of varying aperture, the tightest parts being completely filled with capillary water while wider ones contain air. This ‘isothermal’ state is shown in the upper part of Fig. 11.11, which also indicates the changes in the microscopic channels caused by a thermal gradient before water uptake at the cold side is initiated. At this stage the drying of the hottest buffer makes stacks of lamellae contract, causing widening of the channels and voids and formation of steep, radial fractures at a distance of several centimetres from the hot canister. The accumulation of water in the cold part of the buffer increases the degree of water saturation and causes expansion. Assuming that buffer blocks fully occupy a deposition hole and that the rock provides unlimited amounts of water for uptake of the buffer, it does so by exerting a tremendous suction, up to 100 MPa, on the rock. The suction of dense smectite clay is known to be this high if the initial degree of water saturation of the clay is below 20–30%, while it drops to very low

© Woodhead Publishing Limited, 2010

Development and application of buffer and backfill materials

335

11.11 Schematic picture of microstructural changes in the initial process of porewater redistribution under a thermal gradient. Drying with expansion of channels and voids takes place in the hot part.

values when the clay is nearly saturated. The suction is the driving force of the wetting, which takes place by vapour flow and migration of water molecules along particle surfaces by diffusion (Kro¨hn, 2003; Yong et al., in preparation). The density of the pellet fill at the buffer/rock contact is low and the degree of water saturation high. The dense blocks suck water from the pellet fill, which becomes compressed. Still, its density will stay lower than that of the blocks and it will therefore make up the most permeable part of the buffer, disregarding, for the moment, the buffer closest to the hot canister, where physico/chemical processes will yield a relatively high porosity and hydraulic conductivity. The denser the compressed fill, the lower is the hydraulic conductivity, which therefore partly determines the rate of water transport from the rock via the fill to the blocks. With an increasing degree of water saturation follows a drop in suction and in the latest saturation phase it is the water pressure in the rock that determines the rate of water uptake by the whole buffer. One realises from all this that the ability of the rock to give off water to the contacting buffer has a strong effect on the wetting rate. Hence, if the conductivity of the rock is sufficiently high, there is unlimited access to water for wetting the buffer, while a low rock conductivity means that less water is available to the buffer, which thus retards the water uptake and maturation of the buffer (Pusch and Yong, 2006). In fact, for very tight rock, the buffer dries rather than becomes wetted in the first decades.

© Woodhead Publishing Limited, 2010

336

Geological repository systems for safe disposal

11.12 Test arrangement for investigating the evolution of a compacted buffer contacting a soft pellet fill (Pusch, 2008).

The mechanical and hydraulic interaction of the pellet fill and the dense blocks have been studied in laboratory experiments using the test arrangement in Fig. 11.12 with compacted MX-80 contacting MX-80 pellets. Distilled water was let into the pellet-filled space, thereby simulating the true conditions in a deposition hole. The water uptake and development of swelling pressure are shown in Fig. 11.13. The hydration rate is typically represented by a straight line in a log-time diagram indicating diffusion-controlled saturation. The evolution of the swelling pressure is typical with an initial small peak and a second main peak appearing a few weeks later, followed by a slight drop and finally a continuous slight rise in pressure caused by delayed expansion of the very dense pellets. Several large-scale tests have been made that simulate the conditions in a repository. The so-called Prototype Repository Project, conducted by SKB at A¨spo¨ Underground Laboratory and planned to run for 20 years, involved construction of a full-scale deposition tunnel with six deposition holes in which big blocks and true copper/canisters were placed (Gunnarsson et al., 2002). The canisters had 600 W heaters for providing heat corresponding to what is caused by the decay of radioactive waste. Figure 11.14 shows that the swelling pressure at mid-height of the canister stabilised at different levels at different distances from the canister after about 2 years. The differences can partly be explained by internal friction in the clay mass and partly by loss of expandability in the hot part. The temperature still rose

© Woodhead Publishing Limited, 2010

Development and application of buffer and backfill materials

337

11.13 Maturation of highly compacted MX-80 powder contacting pellets of MX-80 clay. Upper: hydration rate. Lower: development of swelling pressure (Pusch, 2008). Notice the similar development of the swelling pressure as in Fig. 11.4 which refers to a clay consisting of small grains.

after more than 4 years and reached about 72 oC in the clay adjacent to the canister surface and around 60 oC at the rock after about 2 years.

11.3.3 Modelling The thermal evolution of the buffer can be predicted by use of common codes for numerical calculation and even simple estimates are usually sufficient in practice. For the water uptake it is different because the wetting rate has a strong impact on the physical performance of the buffer and several attempts have been made to work out coupled THM codes for accurate prediction of the sorption of water by the buffer (Svemar, 2005). The problem is that the involved transport processes are still rather poorly known and that the hydraulic interaction between the buffer and surrounding rock is hard to define and concretise. The accuracy is illustrated by the comparison of predicted and recorded wetting of smectite clay (MX-80, Wyoming bentonite) with a dry density of 1600 kg/m3 contained in a tube for water uptake from one end. The tube was exposed to a thermal gradient of 10 oC/cm with the highest temperature, 80 oC,

© Woodhead Publishing Limited, 2010

11.14 Evolution of total pressure at mid-height of the buffer in the wettest hole. The highest pressure, about 6.7 MPa, was reached after about 2.3 years at the periphery of the buffer, while the lowest (4 MPa) was recorded close to the canister (Svemar, 2005).

338 Geological repository systems for safe disposal

© Woodhead Publishing Limited, 2010

Development and application of buffer and backfill materials

339

11.15 Comparison of recorded and predicted wetting rate of a 5 cm long sample of smectite-rich clay.

maintained at the opposite end. The initial water content was 10% corresponding to 50% degree of saturation. A laboratory experiment of the same type was conducted and several samples were taken for determination of the water content at different periods of time. The comparison between the theoretical and actual water content distributions are shown in Fig. 11.15. The code was the ‘Barcelona basic model’ that has been frequently used in small-and large-scale tests (Pusch, 2008). One finds that the measured values are consistently lower than the predicted ones, meaning that the saturation process was in fact significantly slower than predicted. The same experience has been gained from large-scale experiments, indicating that the model is not quite adequate (Pusch, 2008).

11.3.4 Other design principles The KBS-3 concept has two versions, the KBS-3V design that we have had a look at and the KBS-3H design that is characterised by horizontally oriented canisters contained in perforated ‘supercontainers’ filled with compacted smectite-rich blocks in which the canisters are fitted (Fig. 11.16). It is similar to canister/clay units proposed by SKB’s sister organisations in Belgium and France with supercontainers and canisters of steel. The KBS-3H design presumes copper-shielded iron canisters for spent fuel and supercontainers of steel, titanium or possibly Navy Bronze. The perforation of the supercontainer lets clay through from the expanding buffer inside it and complete embedment is ultimately reached. There will be some heterogeneity because access to water from the rock will be different and causes non-

© Woodhead Publishing Limited, 2010

340

Geological repository systems for safe disposal

11.16 The components of a supercontainer of KBS-3H type (provided by SKB).

11.17 Appearance of 24-hour old clay plug of ‘basic’ type after cutting off clay specimens for investigation with respect to water content and density (Pusch, 2008).

© Woodhead Publishing Limited, 2010

Development and application of buffer and backfill materials

341

uniform swelling of the buffer. Also, internal friction will counteract homogenisation. Such heterogeneity can be lessened by pumping in a clay mud to fill the space between rock and supercontainer, and this would also speed up the maturation of the buffer. Laboratory experiments demonstrate the performance, as shown in Fig. 11.17. The perforated tube of copper contained densely compacted smectite clay grains and the mud around it was of the same type but had a density of only 1100 kg/m3.

11.4

Types, properties and fabrication of backfill

The primary purpose of backfilling tunnels and shafts is to seal them and to provide support to the rock. Materials for backfilling should meet the following general requirements: . . . .

Lower hydraulic conductivity than the surrounding rock Exertion of a swelling (effective) pressure that prevents rock fall Capability for placement even under the condition of significant water inflow Rational placement

These criteria make it necessary to use smectitic material for backfilling. From a safety assessment point of view the hydraulic conductivity of backfills can be accepted to be higher than that of the buffer but it must be lower than that of the surrounding rock. Candidate backfill materials that have been investigated with respect to hydraulic conductivity, swelling pressure and placeability are mixtures of finely ground smectite-rich clay (bentonite) and suitably graded silt/sand/gravel of granitic origin, or natural smectitic clay (Gunnarsson et al., 2002). The principle of preparing tight backfills by mixing silty sand and gravel

11.18 Schematic picture of the microstructure of clay/ballast mixture with low compressibility and low hydraulic conductivity. G = ballast grains, D = clay powder grains.

© Woodhead Publishing Limited, 2010

342

Geological repository systems for safe disposal

11.19 Grain size distribution of a mixture of moraine and 4% very fine Na bentonite yielding the very low hydraulic conductivity of 1011 m/s (Pusch and Yong, 2006).

11.20 Compaction curves of a mixture of 20% MX-80 clay and crushed rock using different compaction techniques. The dry density can be about 2300 kg/m3 (2.3 g/cm3) of the mixture but the dry density of the clay component does not exceed 800 kg/m3.

with smectite powder would work well if the microstructure has the form shown in Fig. 11.18, with clay aggregates fully occupying the voids between the larger, non-expandable grains. If so, the hydraulic conductivity of the backfill can be very low (Fig. 11.19).

© Woodhead Publishing Limited, 2010

Development and application of buffer and backfill materials

343

11.21 Placement (a, b, c) and compaction (d, e, f) of backfill using the ‘inclined layer principle’. A special vibratory roof compactor for densifying material close to the roof is not shown (Gunnarsson et al., 2002).

11.22 Clay block masonry with smectite pellets or clay mud (Pusch and Yong, 2006).

There are two major problems with backfills of this type. One is that the density of the clay component cannot be raised to more than about 1600 kg/ m3 at complete water saturation even if the density of the mixture is as high as can possibly be reached (Fig. 11.20). This means that saturation with salt calcium-rich groundwater, which often prevails at more than a few hundred metres depth, may increase the conductivity by orders of magnitude as indicated by the diagram in Fig. 11.3. The other problem is to place the backfill in homogeneous form and with sufficient density. Placement and compaction of inclined layers forming a

© Woodhead Publishing Limited, 2010

344

Geological repository systems for safe disposal

forward moving slope using equipment of the type shown in Fig. 11.21 has been tried in SKB’s Underground Laboratory (Gunnarsson et al., 2002). The density has been found to be relatively low near the rock, which, in combination with softening caused by water flowing from the rock, may cause heterogeneities. Other materials and techniques for placement have therefore been tested, like placement of precompacted blocks of smectitic material and filling the remaining space by blown-in dense pellets of smectite-rich clay or pumped-in smectitic clay mud (Fig. 11.22). A masonry of blocks of natural mixed-layer smectite/illite/mica clay material (see Friedland Ton in Pusch and Yong, 2006) with a net dry density of 1700 kg/m3 will ultimately homogenise and mature with a density at water saturation of about 2000 kg/m3. The net hydraulic conductivity is less than 1010 m/s even for very salt porewater and the swelling pressure is more than enough to support the surrounding rock (>>100 kPa). None of the techniques can be used if the inflow of water from the rock is high since piping and erosion will make the backfill heterogeneous and fluid.

11.5

Long-term performance

11.5.1 Issues The most important issue for long-term isolation of HLW is the chemical stability of the smectite under hydrothermal conditions created in repositories. Another matter is the deformation of the buffer that it will undergo under stresses induced by the load of canisters, earthquakes and tectonics. We will consider mechanical strain first.

11.5.2 Time-dependent mechanical strain Theory The Terzaghi concept of consolidation and expansion has a quite obvious physical meaning for soils containing non-expanding clay mineral particles, like illite and kaolinite (Pusch and Yong, 2006). Thus, an increased total pressure Δp of water-saturated clay of these latter types generates a pore pressure of exactly the same magnitude, but if water is allowed to be drained the pressure Δp is successively transferred to the particle network, which is thereby compressed and densified. The same process takes place in smectiterich clay, leading to a closer distance between some of the stacks of lamellae and closer spacing of the lamellae in some stacks, by which the repulsion between the lamellae and stacks of lamellae increases. Unloading has the opposite effect, i.e. water is taken up in the interlamellar space and on the basal surfaces of the stacks of lamellae by the very strong hydration

© Woodhead Publishing Limited, 2010

Development and application of buffer and backfill materials

345

potential, yielding expansion. The load of a canister on the buffer hence causes compression of the buffer in conjunction with expulsion of porewater until equilibrium is reached between the imposed stresses and the mobilised stress resistance. Such equilibrium is reached relatively quickly but in a longterm perspective, considering a period of time of hundreds or thousands of years, another mechanism, i.e. shear-induced creep, is expected to dominate and cause settlement. It has the form of slip within and at the contacts of adjacent stacks of lamellae, taking place under constant volume conditions or in conjunction with compression. The accumulated time-dependent strain is termed ‘creep’. For smectite clay with insignificant friction between particles, creep is stronger than for non-expandable clays, but the microstructural strain mechanisms have the same character of stochastically distributed slip occurring where energy barriers are overcome. The empirically derived relationship in the following equation is being used in current buffer design work for describing time-dependent strain as a function of the principal stress state and stress conditions at failure (Pusch and Yong, 2006):  n t g_ ¼ g_ 0 eaðs1 s3 Þ=ðs1 s3 Þfe ðaÞðs1 s3 Þ0 =ðs1 s3 Þf ½11:1 tr where t = time after stress change tr = reference time (105 s) (σ1σ3)0 = reference deviator stress (0.5 (σ1σ3)) (σ1σ3)f = deviator stress for actual condition (σ1σ3)fe = deviator stress at failure g_ = creep rate g_ 0 = creep rate at time tr n and α = parameters derived from laboratory tests The following equation expresses the shear strength q as a function of the mean effective stress p: ½11:2

q ¼ apb where a = q for p = 1 kPa b = inclination of curve in log p/log q diagrams

Equations [11.1] and [11.2] are completely empirical and not based on any conceptual microstructural model and are therefore not of general validity. For deriving reliable theoretical expressions of creep and creep rate as functions of stress and clay density the true involved strain mechanisms must be identified and taken as a basis of formulation of conceptual and

© Woodhead Publishing Limited, 2010

346

Geological repository systems for safe disposal

theoretical models (Pusch and Yong, 2006; Pusch 2008). Such work involves definition and concretisation of the barriers on the microstructural scale to slip, which are represented by van der Waals forces, hydrogen bonds and strong primary valence bonds. They form a spectrum with the weakest forces, van der Waals forces, at its lower end and strong primary valence bonds at its upper end (Pusch and Feltham, 1980). The response of the structure to a macroscopic shear stress is that the overall deformation of the entire network of particles changes by disintegration, translation and rotation of small, weaker aggregates while larger ones are less affected and stay strong. The breakdown of weak aggregates involves transformation to a laminated system of flaky particles. This microstructural organisation is believed to be the reason for the Newtonian rheological behaviour of smectite clay that has undergone a large strain in one or two directions. Using stochastic mechanics the bulk shear strain rate for low bulk shear stresses, i.e. lower than about 1/3 of conventionally determined shear strength, is found to be as defined below with t
½11:3

The appropriate constant B and the value of t0 depend on the deviator stress, temperature and structural details of the slip process. The creep can be expressed as g ¼ at  bt2

ðt < a=2bÞ

½11:4

meaning that the creep starts off linearly with time and then dies out. For higher bulk loads the strain on the microstructural level yields some irreversible changes associated with local breakdown and reorganisation of structural units. There is repair by inflow of new low-energy barriers parallel to the strain retardation caused by the successively increased number of slip units being halted by meeting higher energy barriers. This type of creep can go on for very long periods of time without approaching failure. For thermodynamically appropriately defined limits of the barrier spectrum the strain rate appertaining to logarithmic creep takes the form dg=dt ¼ BTt=ðt þ t0 Þ

½11:5

where B is a function of the shear stress t, t0 is a constant of integration that leads to a creep relation closely representing the commonly observed logarithmic type, implying that the creep strain is proportional to log (t + t0).

© Woodhead Publishing Limited, 2010

Development and application of buffer and backfill materials

347

11.23 Model test with loaded canister confined in buffer clay with a density at water saturation of 2000 kg/m3 and exposed to a load of 800 N simulating the weight of a canister with a weight of 80 kg.

Practice No large-scale field tests have yet been made for measuring the settlement of the heavy canisters embedded in buffer or for simulating large-scale shearing caused by earthquakes. A small-scale test illustrates the expected performance in practice (Pusch and Adey, 1986). The diagram in Fig. 11.23 shows that the settlement of a model canister was less than 8 μm after 3 months under a constant load of 80 kg placed after complete water saturation, and that it was approximately proportional to log time. One realises that if the load on the buffer below and around the lower part of the canister makes it sink, and the expansion upwards of the upper part of the buffer lifts the canister, it will be exposed to tension. Boundary element calculations of the settlement rate of a real 20 000 kg canister using laboratory-derived parameter values indicates that the settlement caused by creep is on the order of 1 millimetre in 1 million years (Pusch and Adey, 1986). The total settlement will, however, be several times larger because compression will contribute and because the creep rate will be significantly higher in the period of time when the temperature is higher. The average buffer temperature in the first 100 years will be 50–90 oC and 15–50 oC in the subsequent 1000 years for SKB’s concepts.

© Woodhead Publishing Limited, 2010

348

Geological repository systems for safe disposal

Table 11.1 Changes in one year long hydrothermal tests of MX-80. M = montmorillonite, F = feldspars, G = gypsum, Q = quartz, K = kaolinite, Chl = chlorite, I = illite; +++ means strong increase, ++ significant increase, + slight increase, - - - strong loss, - - significant loss, - slight loss, 0 means no change Treatment

125–130 oC

115–120 oC

105–110 oC

90–95 oC

Hydrothermal without gamma radiation

MF --Chl+ G ++ K ++ Q+ I+

M -F -Chl + G +++ K+ Q+ I0

M0 FChl 0 G+ K0 Q0 I0

M0 FChl 0 G+ KQ0 I0

11.5.3 Chemical stability There are two different chemical processes that need to be considered, firstly cation exchange related to the electrolyte composition of the groundwater with which the clay will be saturated, and possible changes in the groundwater regime with time. The impact on the hydraulic conductivity, swelling pressure and creep of the chemical composition of the groundwater is very small for dense buffer but can be considerable for backfills with a small amount of smectite clay, as discussed earlier. Secondly, chemical changes will take place in the form of conversion of smectite to non-expanding forms, like illite and chlorite, and in the form of precipitation of silica and other compounds. They can have a strong influence on the practical performance of smectitic buffers and backfills (Grindrod and Takase, 1993). The following total reaction affecting smectite has been proposed (Pusch and Yong, 2006): S þ ðFk þ MiÞ ¼ I þ Q þ Chl

½11:6

where S = smectite Fk = K-feldspar Mi = K-mica I = illite Q = quartz Chl = chlorite Quartz has been assumed to originate from decomposed feldspars and from silicons lost from the SiO4 tetrahedrons in montmorillonite in conjunction with conversion of this mineral to beidellite as a first step in conversion to illite. The major mineralogical changes identified in hydrothermal tests are summarised in Table 11.1. The results are in agreement with analyses of natural analogues (Pusch, 1993, 2008).

© Woodhead Publishing Limited, 2010

Development and application of buffer and backfill materials

349

11.24 Diagram showing predicted conversion of smectite to illite for the activation energy 27 kcal/mole.

Comparison of the chemical composition of untreated MX-80 clay and hydrothermally treated clay with and without radiation showed that there were nearly no differences except that Fe migrated from a confining iron plate into the clay quicker under radiation (Pusch et al., 1993). An often used expression of the rate of illitisation of montmorillonite (Pytte and Reynolds, 1989) is    m ½11:7 dS=dt ¼ AeU=RTðtÞ ðKþ =Naþ Þ Sn where S = mole fraction of smectite in I/S assemblages U = activation energy R = universal gas constant T = absolute temperature t = time m, n = coefficients Use of equation [11.7] with appropriate parameter values gives diagrams like the one in Fig.11.24, showing the rate of illitisation of smectite for the activation energy of 27 kcal/mole. The graph shows that heating to 50 oC would cause only insignificant loss of smectite in 106 years, while about 100% of the original smectite will turn into illite in this period of time at 100 oC.

© Woodhead Publishing Limited, 2010

350

Geological repository systems for safe disposal

11.25 Nitrogen gas movement in bentonite clay with 1680 kg/m3 density at saturation with seawater (Pusch, 2008).

11.5.4 Impact of gas under pressure Gas can be of importance in two respects: (1) hydrogen gas from corrosion of steel canisters or from the iron component of canisters of the kind presently favoured by SKB can create channels that serve as paths for the quick release of radionuclides from the canisters to the biosphere via fracture systems (Horseman and Harrington, 1997) and (2) water vapour formed in the hotter parts of the buffer in conjunction with water saturation can pressurise and displace water from colder parts of the buffer, forming channels (Pusch and Yong, 2006). Considering the matter of hydrogen gas first, a number of tests have been made for finding out how and at what pressure gas can penetrate smectite clay confined between filters. These tests have shown that pressurising it through a filter causes displacement of a small fraction, i.e. 0.02 to 0.2%, of the porewater before gas passes through the clay (Fig. 11.25). The studies indicate that the gas pressure has to exceed a threshold value for making the gas proceed through the clay. The critical pressure is on the same order of magnitude as the swelling pressure, which is explained by the fact that this latter pressure has to be overcome for letting gas displace interconnected clay particle assemblies sufficiently much to form channels.

11.5.5 Influence of radiation A number of laboratory experiments have indicated that γ radiation has two major effects: (1) radiolysis of the porewater, by which the radical OH and various other products like negatively charged O2 are produced, and (2) breakdown of the crystal lattice of clay minerals (Pusch et al., 1993). The first process yields decomposition of the porewater that can result in gas accumulation at the surface of HLW canisters, while the second mechanism

© Woodhead Publishing Limited, 2010

Development and application of buffer and backfill materials

351

yields mechanical disintegration of the crystallites by which the average particle size is reduced and the crystal lattices fragmented. At present, the general understanding is that strong gamma radiation will not cause substantial degradation of smectitic buffer while it can contribute to the formation of channels serving as paths for contaminated water and gas.

11.5.6 Microbial effects Microbial species detected in the groundwater in deep vault repositories include aerobic and heterotrophic microorganisms and anaerobic ironreducing and sulphate-reducing bacteria (Pedersen, 1995). The capability of microorganisms such as facultative anaerobic bacteria, fungi and even anaerobes in reducing iron is well known but their performance in dense buffer is not known in detail. The presence of water in the buffer material allows both acid–base reactions and also oxidation–reduction reactions. The latter can be abiotic and/or biotic. In respect to biological issues, microorganisms are significant participants in catalysing redox reactions. The activity of electrons is a significant factor in oxidation–reduction reactions since the transfer of electrons between the clay minerals and biotic issues can cause a reduction of structural Fe3+ in the octahedral and tetrahedral sheets to Fe2+, which can significantly alter the short-range forces between the lamellae and result in: (a) lower specific surface area, (b) cementation by precipitation, (c) decreased water-holding capacity, (d) reduced swelling capability and (e) increased hydraulic conductivity.

11.6

References

Grindrod P and Takase H (1993). ‘Reactive chemical transport within engineered barriers’, in: Proceedings of the 4th International Conference on The Chemistry and Migration Behaviour of Actinides and Fission Products in the Geosphere, Charleston, South Carolina, 12–17 December 1993, Oldenburg Verlag, 1994, pp. 730–779. Gueven N and Huang W-L (1990), ‘Effects of Mg2+ and Fe3+ substitutions on the crystallization of discrete illite and illite/smectite mixed layers’, Department of Geosciences, Texas Technical University, Exxon Production Research Company, Houston, Texas. Gunnarsson D, Bo¨rgesson L, Ho¨kmark H, Johannesson L-E and Sande´n T, (2002), ‘Installation of the backfill and plug test’, in Clays in Natural and Engineered Barriers for Radioactive Waste Confinement, Proceedings of the International ANDRA Meeting, Reims, 9–12 December 2002, ANDRA, Chatenay-Malabry, France. Horseman S T and Harrington J F (1997), ‘Study of gas migration in MX-80 buffer bentonite’, Report WE/97/7, National Environmental Research Council, British Geological Survey. Kehres A (1983), Isotherms de deshydratation des argiles. Energies d’hydratation –

© Woodhead Publishing Limited, 2010

352

Geological repository systems for safe disposal

diagrammes de pores surfaces internes et externes, Dr thesis, Universite´ Paul Sabatier, Toulouse, France. Kro¨hn K-P (2003), ‘New conceptual models for the resaturation of bentonite’, Applied Clay Science, 23, 25–33. Pedersen K (1995), ‘Survival of bacteria in nuclear waste buffer materials. The influence of nutrients, temperature and water activities’, SKB Technical Report TR 95-27, SKB, Stockholm, Sweden. Pusch R (1993), ‘Evolution of models for conversion of smectite to non-expandable minerals’, SKB Technical Report TR-93-33, SKB, Stockholm, Sweden. Pusch R (2008), Geological Storage of Radioactive Waste, Springer-Verlag, ISBN: 978-3-540-77332-0. Pusch R and Adey R (1986), ‘Settlement of clay-enveloped radioactive canisters’, Applied Clay Science, 1, 253–365. Pusch R and Feltham P (1980), ‘A stochastic model of the creep of soils’, Ge´otechnique, 30(4), 497–506. Pusch R and Yong R N (2006), Microstructure of Smectite Clays and Engineering Performance, Taylor & Francis, London and New York, ISBN: 10-0-41536863-4. Pusch R, Karnland O, Lajudie A and Decarreau A (1993), ‘MX-80 exposed to high temperatures and gamma radiation’, SKB Technical Report TR-93-03, SKB, Stockholm, Sweden. Pusch R, Kasbohm J, Pacovsky J and Cechova Z (2005), ‘Are all smectite clays suitable as ‘buffers’’?, in Clay in Natural and Engineered Barriers for Radioactive Waste Confinement – Part 1. Physics and Chemistry of the Earth, vol. 32/1-7, Elsevier Publishing Company, pp. 116–122. Pytte A M and Reynolds R C (1989), ‘The thermal transformation of smectite to illite’, in Thermal History of Sedimentary Basins, edited by N D Naeser and T H McCulloh, Springer-Verlag, New York, pp. 133–140. Svemar C (2005), ‘Prototype Repository Project’, Final Report of European Commission Contract FIKW-2000-00055, Brussels, Belgium. Yong et al. (in preparation), Environmental Soil Properties and Behavior, Taylor & Francis group, LLC.

© Woodhead Publishing Limited, 2010

12 Near-field processes, evolution and performance assessment in geological repository systems W . Z H O U , Rensselaer Polytechnic Institute, USA; R . A R T H U R , INTERA Inc., USA

Abstract: The near field is defined as the portion of a geologic repository for nuclear waste that contains the waste form and other components of the engineered barrier system (EBS), as well as a volume of immediately surrounding host rock that is significantly affected by repository construction and waste emplacement. An analysis of how the near field is likely to evolve over long periods of time considered in a safety assessment is complicated by interactions that will occur within and among the different components of the EBS and near-field rock, and by perturbations to the local thermal, hydrologic, mechanical and chemical (T-H-M-C) conditions arising from emplacement of the waste. This chapter provides a summary of current understanding regarding these various interactions and conditions, and how modeling can be used to help account for them in a safety case. Key words: near field, safety indicators, barrier functions, transport of radionuclides, diffusion, advection, porous media, fractured rock, coupled processes.

12.1

Introduction

It is convenient to think of geologic repository systems for nuclear wastes as consisting of both a near field and a far field. As can be seen in Fig. 12.1, for a repository for high-level waste (HLW) and spent nuclear fuel (SNF), and Fig. 12.2, for a low-level waste (LLW) repository, the near field comprises the EBS and that portion of the immediately surrounding host rock that is significantly affected by repository construction and waste emplacement. The far field, on the other hand, includes the remaining volume of rock

353 © Woodhead Publishing Limited, 2010

354

Geological repository systems for safe disposal

12.1 Illustration of the near field of a deep geological repository for high-level radioactive wastes.

12.2 Illustration of a near-surface disposal facility for low-level radioactive wastes (based on NCRP, 2005).

separating the near field from the biosphere. The far field is for all intents and purposes unaffected by repository construction and waste emplacement. Performance assessment considers the capabilities of both the near field and far field in containing radionuclides and, after containment failure, in

© Woodhead Publishing Limited, 2010

Near-field processes, evolution and performance assessment

355

providing long-term isolation by limiting radionuclide migration and release to a future biosphere environment (see Chapter 1 of this book). While the actual performance contributions of different engineered and natural barriers vary among different geological formations and repository design concepts, the performance of the near field is a vital consideration, both because of key containment and isolation processes occurring in the near field and because of the opportunity to enhance containment and isolation by design modification of the engineered barriers. The remainder of this chapter considers the functions of the EBS, perturbations to the near-field rock due to the emplacement of waste, nearfield processes, the evolution of the near field and modelling of near-field processes.

12.2

Near-field component: engineered barrier system (EBS)

The function of the waste form, which is the innermost component of the EBS, is to provide a solid matrix for encapsulation and immobilization of radionuclides, and to aid transportation and emplacement of the waste into a geological repository. After an initial containment period, water migrating through the repository will contact the waste form and initiate dissolution of the matrix (see Chapter 14). The dissolution behavior and rate will be a function of the form and composition of the waste form (e.g. thermodynamically stable mineral versus thermodynamically unstable glass), and environmental conditions (especially chemical and thermal) prevailing at the waste form–water interface after containment failure. Waste-form dissolution is a basic process initially limiting the rate of radionuclide release. Waste forms with extremely low dissolution rates can significantly enhance the isolation performance of a geological repository. Additional near-field factors and processes, however, may minimize the importance of waste-form dissolution with respect to long-term repository safety. First, the inventory of radionuclides within a waste form (or ‘source’) may be limited; the extreme case is the so-called ‘instant release fraction’ of radionuclides in spent fuel, which is typically assumed to be completely released immediately after containment failure. Second, the rate of mass transport of dissolved radionuclides from the surface of the dissolving waste form may be slow relative to the rate of waste-form dissolution. In such cases, the concentration of dissolved radionuclides will increase at the wasteform surface until saturation is reached with respect to the precipitation of a new radionuclide-bearing solid phase (see Chapter 1). The dividing line between dissolution-rate controlled release and solubility-limited release depends on the ratio of the rates of dissolution and transport, as well as on

© Woodhead Publishing Limited, 2010

356

Geological repository systems for safe disposal

12.3 Example of HLW tunnel backfill. (a) Sketch of the deposition tunnel and the deposition hole. The buffer (bentonite blocks) is preplaced before the canister is deposited. (b) Backfilling process after deposition of the canisters (based on SKB, 2004).

the solubility of potential radionuclide-bearing solids precipitating at the waste-form surface. The solid waste form, in turn, is encapsulated by one or more containers (also called ‘overpacks’ or ‘canisters’ in different repository programs) that can be made of metals, ceramics or concrete. The assembled waste form and canister are typically referred to together as a ‘waste package’. The containers or canisters* initially prevent groundwater from contacting the waste form immediately following emplacement. Container materials are generally highly corrosion resistant, or have a certain corrosion allowance, and are also structurally stable. It should also be noted that even corroded metals can delay radionuclide release by sorption. Details of canister performance are described in Chapter 13. The EBS often includes a ‘buffer’ that separates the waste package from the host rock (Fig. 12.3(a)). Bentonite is assumed to be the buffer material in many EBS designs (e.g. SKB, 1999). It is composed mainly of montmorillonite, which is a swelling clay and has extremely low permeability. This is advantageous because any radionuclide transport through the buffer will be controlled by diffusion only. Additional functions of the buffer are to filter and remove any radionuclide-bearing colloids that might form during waste-form dissolution, to sorb and delay the diffusive transport of dissolved radionuclides, to restrict the viability of microbes that could *

‘Containers’ mean generally the engineered barriers that are made of any materials and have any shapes. ‘Canisters’ means particularly the cylindrical, metal containers. ‘Overpacks’ are typically metal or ceramic vessels that are placed around the stainless steel ‘pour canisters’ containing vitrified HLW borosilicate glass or assemblies of solidified waste forms.

© Woodhead Publishing Limited, 2010

Near-field processes, evolution and performance assessment

357

12.4 L/ILW disposal silo in Project SAFE: (a) view from side and (b) view from top (SKB, 2001a).

possibly exist close to the waste package, to accommodate small rock displacements that might occur and to dissipate efficiently heat resulting from radioactive decay of the nuclear waste. In L/ILW disposal facilities, such as the Swedish Project SAFE (Fig. 12.4), concrete is used as a buffer material. Concrete provides structural support and has an extremely low intrinsic permeability (1012 to 1010 m2) that can block or reduce groundwater flowing through the waste packages. In addition, concrete is an effective sorbent for some radionuclides. Deposition tunnels and other excavated regions in a repository will be backfilled, and the backfilling material is often regarded as being part of the EBS. The primary functions of backfilled tunnels are to keep waste packages in place, to inhibit fast groundwater flow and to keep the tunnels stable. Figure 12.3(b) shows a disposal tunnel and backfilling process after emplacement of waste packages for a KBS-3 type repository (SKB, 2004). In this case, the tunnel is backfilled with a mixture of crushed rock and bentonite. Figure 12.4 provides a similar example for a Swedish ILW

© Woodhead Publishing Limited, 2010

358

Geological repository systems for safe disposal

repository, where crushed rock, gravel and a mixture of crushed rock and bentonite are planned for use as backfill at various locations throughout the facility. The EBS design for the proposed nuclear waste repository at Yucca Mountain, Nevada, which is located in unsaturated host rocks, includes two unique components not described above (US DOE, 2007). One is a drip shield that will exist between the host rock and waste package. The drip shield is intended to block water from dripping on to the waste packages from the upper drift wall (US DOE, 2007). The other unique EBS component at Yucca Mountain is an invert composed of crushed tuff that will lie below the waste packages. The invert is intended to provide structural support and can also retard radionuclide transport by sorption (Fig. 12.5). An alternative and potentially more effective barrier than a drip shield in unsaturated rocks is the so-called Richards barrier, or ‘flowdiversion barrier’, ‘capillary barrier’, as illustrated in Fig. 12.6(a) (see Zhou et al., 1996; Conca et al., 1998). A Richards barrier could consist of a highly conductive sand layer overlying a crushed gravel inner layer that surrounds the waste packages. Such a barrier could in principle divert groundwater flow and maintain a dry inner gravel layer having an exceptionally low (100– 1000 times lower) effective diffusion coefficient compared to saturated

12.5 Illustration of the drip shield used in the US Yucca Mountain repository (US DOE, 2007).

© Woodhead Publishing Limited, 2010

Near-field processes, evolution and performance assessment

359

12.6 (a) Illustration of the Richards barrier. (b) Near-surface disposal facility utilizing the similar barrier concept (adapted from NCRP, 2005).

media. Many near-surface LLW disposal facilities actually employ the same concept as the Richards barrier, as shown in Fig. 12.6(b). Previous studies (Zhou et al., 1996; Conca et al., 1998) have confirmed the robust performance of the Richards barrier under various undesirable conditions. There are also numerous archaeological evidences from ancient burial practices that demonstrate the unique function in preserving relics using a similar structure (e.g. Watanabe, 1989).

12.3

Near-field component: host rock

As defined here, the near field includes a volume of host rock that has been disturbed by the construction and operation of a nuclear waste disposal facility. The disturbed host rock behaves differently than the rock at ambient conditions, and hence must be analyzed differently. It is, of course, subjective as to how much of the surrounding host rock is ‘significantly impacted’ or perturbed by repository construction and waste emplacement. Nevertheless, such a treatment offers advantages in understanding and analyzing the evolution of the environment surrounding a nuclear waste repository after closure. Disturbances to the near-field host rock are generally considered in terms of perturbations to the ambient thermal, hydraulic, mechanical and chemical conditions in the rock.

© Woodhead Publishing Limited, 2010

360

Geological repository systems for safe disposal

Thermal disturbances in water-saturated media can extend many tens of meters into the host rock depending on the heat-generation rate, repository design and thermal conductivities of EBS materials. Peak temperatures are typically reached within 50–100 years after repository closure and temperatures return to near ambient conditions within about 1000–3000 years. In unsaturated rock, the extent of thermal disturbance also depends on the degree of water saturation of the EBS and host rock. As an example of an unsaturated repository, such as the proposed Yucca Mountain repository where SNF and HLW will be buried in unsaturated tuff, high areal heat loads will be intentionally imposed to dry out the near field. In this approach to thermal management, a boiling front will propagate from the waste package up to about ten meters into the host rock above and surrounding the emplacement drift during the initial thermal transient (Buscheck et al., 1996). The front will prevent liquid water from penetrating into a ‘dry-out’ zone in the near-field rock and EBS. Temperature changes impacting moisture redistribution will also extend several hundred meters into the host rock above and below the repository horizon (e.g. Zhou et al., 2008). Thus the heat-affected, near-field rock in unsaturated repository environments will extend over a much greater region than for the low heat-loading repository concepts in saturated environments. The most important disturbance due to heating is the resultant vaporization and condensation of porewater in the host rock, as illustrated in Fig. 12.7. Heating from SNF waste packages creates a favorable dry-out zone that may persist over many hundreds to thousands of years (Buscheck et al., 1996). Liquid water cannot contact waste packages during this time, which delays the onset of canister corrosion and subsequent release of radionuclides. Zhou et al. (2008) have shown that a tripling of the SNF

12.7 Dry-out zone in the near field predicted to occur in the Yucca Mountain repository (US DOE, 2008).

© Woodhead Publishing Limited, 2010

12.8 Extended dry-out zone shown as liquid saturation contours can be created if SNF storage is increased by a factor of three compared with the current design (63 000 MTHM) in the Yucca Mountain repository (Zhou et al., 2008).

Near-field processes, evolution and performance assessment

© Woodhead Publishing Limited, 2010

361

362

Geological repository systems for safe disposal

12.9 Example of the near-field stress distribution after waste package emplacement (SKB, 1999).

storage capacity at Yucca Mountain over the current design of 63 000 MTHM (metric tons heavy metal) would create an extended dry-out zone that could persist for 10 000 years (Fig. 12.8). Thermal disturbances in the near-field rock also affect the local flow field, stress field and chemistry. An increase in temperature will pressurize porewater in the rock, which could enhance the flow rate and drive water toward cooler regions away from the repository. In unsaturated repositories, flow of liquid water and gases will be closely coupled with the thermal evolution of the near field. Heating can also alter the stress field, as shown in Fig. 12.9. In fractured rock, in particular, heating causes expansion of fractures near the hotter places and contraction of fractures near the colder places. Hydrogeological changes resulting from repository construction and

© Woodhead Publishing Limited, 2010

Near-field processes, evolution and performance assessment

363

12.10 Water saturation contours as a result of flow simulation near an emplacement drift, which shows a shadow effect in which water saturation is reduced in unsaturated tuff below the drift (US DOE, 2002).

waste emplacement will depend on the design of the disposal facility and the extent and duration of radiogenic heating. As mentioned earlier, flow can be affected by heating in unsaturated rocks if the heating results in vaporization and condensation of pore water. Flow can also be affected in such systems by the physical obstruction of the disposal facility itself, resulting in a ‘shadow’ zone, as illustrated in Fig. 12.10. A shadow zone is generated because water flowing around an open cavity tends to be retained within pores where the capillary pressure is much greater than the pressure in the cavity. This leaves a relatively dryer zone, the shadow zone, underneath the cavity. When evaluating radionuclide transport in such systems, advective transport through the shadow zone is assumed to be either zero or proportional to a flow rate that is lower than the ambient flow rate. Hence, the shadow zone is considered to be a part of the near field.

© Woodhead Publishing Limited, 2010

364

Geological repository systems for safe disposal

12.11 Examples of EDZ for different excavation methods and mechanical response of the near-field host rock (SKB, 2001b).

Shadow zones may also be present in LLW near-surface disposal facilities that are located in underground vadose zones. Flow in the vicinity of repositories located in saturated host rocks can be altered by the excavation of disposal tunnels and subsequent backfilling. In particular, flow will be diverted by low-permeability backfilled tunnels or buffers composed of bentonite containing swelling smectitic clay. Such a flow diversion effectively decouples the diffusive release of radionuclides from the EBS from potential future changes in hydrological conditions at the site. Mechanical disturbances arise primarily from repository construction, especially by formation of the so-called excavation damaged zone (EDZ), as shown in Fig. 12.11. Rock within the EDZ is more fractured than the intact host rock, and therefore has a higher hydraulic conductivity. This may result in a relatively higher flow rate through the EDZ compared to the ambient flow rate of groundwater in the undamaged far-field rock. The thickness of EDZ depends on the excavation methods used as well as on the stress state and other intrinsic properties of the host rock. Drilling and blasting methods may create a thicker EDZ than that generated using a boring machine, for example. Values for the thickness EDZ typically range from a few centimeters to about 0.4 m. In performance

© Woodhead Publishing Limited, 2010

Near-field processes, evolution and performance assessment

365

assessment, the EDZ can be explicitly represented by giving the appropriate amount of rock a higher porosity and permeability than the intact host rock. In disposal systems that deploy a bentonite buffer between the waste canister and host rock, after bentonite is saturated with water the bentonite will swell and intrude against the surrounding host rock, and potentially into the surrounding host rock fractures. The hydrological and transport characteristics of the rock with intruded fractures can behave differently compared with the host rock that is not intruded (Borrelli and Ahn, 2008). In summary, the rock immediately surrounding the EBS will experience perturbations resulting from repository construction and waste emplacement that can be described in terms of thermal, hydrological, mechanical and chemical processes. This distinguishes the near-field rock as a more dynamic system than the far-field rock where relatively constant conditions prevail. Although the near-field environmental conditions generally return to ambient conditions after a relatively brief transient period, there can be irreversible changes in the properties of the near-field rock and EBS. Such changes must be accounted for in assessments of the long-term T-H-M-C evolution of the disposal system.

12.4

Summary description of near-field containment and isolation

The containment and isolation performance of the near field can be understood by identifying key features, events and processes (FEPs) of the near field: 1. 2.

3. 4. 5. 6.

Water contact with the near-field components. Water contact causes corrosion of canisters/containers/overpacks, which subsequently results in containment failure and the initiation of radionuclide release from the near field. Concrete barriers can crack due to chemical attack from surrounding soils or groundwater, as well as due to geomechanical stress. Water enters the breached canister and container(s) and contacts the waste form. Waste-form dissolution begins, releasing radionuclides into the water from the various radionuclide-bearing sources of the waste form. For the spent-fuel waste matrix, radionuclides residing within grain boundaries and in the gap between the spent fuel and cladding are released into the water instantly (so-called ‘instant release fraction’). Radionuclides within the UO2 matrix will be released into the water at a rate proportional to the UO2 matrix dissolution rate. For other waste forms (e.g. borosilicate glass), the release of radionuclides will be proportional to the waste-form dissolution rate.

© Woodhead Publishing Limited, 2010

366

Geological repository systems for safe disposal

7.

Because of mass-transport constraints, dissolved radionuclides at the waste-form surface may reach solubility limits for radionuclide-bearing secondary solids. 8. The dissolved radionuclides migrate from the breached containers through failed canisters and into the buffer surrounding the canister. Radionuclides may be sorbed by canister corrosion products. 9. The dissolved radionuclides migrate through the buffer into the surrounding host rock. Radionuclides may be sorbed by minerals within the buffer. In addition, the buffer assures that any radionuclidebearing colloids potentially formed from waste-form dissolution are physically filtered and prevented from being released to the far field. 10. The dissolved radionuclides migrate through the near-field subsurface or geological materials and may be sorbed along the migration pathway. 11. The dissolved radionuclides are released into the far field. The above FEPs define the modeling context used in assessments of nearfield performance. Quantitative modeling and simulation require input for a number of key parameters, including barrier dimensions, container failure time, waste-form dissolution rate, radioelement solubility limits, groundwater flow rate, sorption coefficients, diffusion coefficients, etc. Selection of credible and defensible values for these parameters can be supported by an integrated program of site characterization, laboratory testing, consideration of natural and archaeological analogs, and design analyses.

12.5

Overview of near-field process modeling

The fundamental equations governing the main processes of the near field include mass continuity of fluids, equations of motion of fluids (or momentum conservation), energy conservation, momentum conservation equation for the whole porous medium and mass balance equations for all the chemical species of concern. In this chapter, a general case of two-phase systems is considered. The equations presented will apply to unsaturated zones. In the case of saturated zones, water saturation in the equation is equal to one at all times and all places, and gas phase equations and terms disappear. The mass conservation equations for the fluids in the porous medium consider liquid and gas phases in pore volumes. Assuming no sinks or sources, the mass conservation equations are:   qðS er Þ l l ¼0 ½12:1 H  rl q l þ qt

© Woodhead Publishing Limited, 2010

Near-field processes, evolution and performance assessment

367

for the liquid phase and   q½ð1  Sl Þer  g H  rg q g þ ¼0 qt

½12:2

for the gas phase where r = density of fluids (kg/m3) q = specific discharge or Darcy velocity (m/year) Sl = water saturation e = porosity t = time (year) and subscripts ‘l’ and ‘g’ denote ‘liquid’ and ‘gas’ phases respectively. In a two-phase system, the saturation summation is unity: Sl þ Sg ¼ 1

½12:3

For flowing fluids, momentum conservation must be satisfied. For most cases, Darcy equations, or equations of fluid motion in porous media, are sufficient to describe the balance between momenta and forces acted on the fluids. These equations are: ql ¼ 

kl ðSl Þ ðHpl þ rl gkÞ ml

for the liquid phase and  kg ðSl Þ  Hpg þ rg gk qg ¼  mg

½12:4

½12:5

for the gas phase, where k = intrinsic permeability (m2) p = pressure (Pa) g = gravity acceleration (m/s2) and m = dynamic viscosity (N s/m2) Note that equations [12.4] and [12.5] are vectors and thus have up to three components. In the Cartesian coordinate system, Darcy fluxes and pressure gradients can be resolved into i, j and k components. For unsaturated media, permeability is a function of liquid saturation and is usually expressed as the intrinsic permeability of porous media multiplied by relative permeability for a given phase: kl ¼ kkrl ðSl Þ

½12:6

for the liquid phase, where krl is relative permeability for the liquid phase.

© Woodhead Publishing Limited, 2010

368

Geological repository systems for safe disposal

Similarly, for the gas phase, the permeability is a function of gas saturation, which can also be expressed as a function of liquid saturation by using equation [12.3]: kg ¼ kkrg ðSl Þ

½12:7

In addition, pressures of the two phases can be related by the capillary pressure of the porous medium: pc ðSl Þ ¼ pg  pl

½12:8

Capillary pressure varies with saturation and depends on both the porous medium and fluids. The energy conservation equation takes account of internal energy changes of solid, liquid and gas in a porous medium, as well as heat conduction and convection. Assuming local thermodynamic equilibrium, the energy conservation equation reads i qh ð1  eÞrR cv T þ eSl rl ul þ eð1  Sl Þrg ug qt   ¼ H  ðKR HTÞ  H  rl hl ql þ rg hg qg ½12:9 where u = specific internal energy (J/kg) h = specific enthalpy (J/kg) KR = thermal conductivity for the porous medium (W/m K) and subscript ‘R’ denotes rock To evaluate stress in the near field, the conservation of momentum for the whole porous medium is needed. The whole porous medium, however, is at rest but can be subject to forces and can deform under the resulting stress. This process may be described by the equilibrium of forces, the so-called ‘equilibrium equation’. Assuming that the medium is isotropic and poroelastic, and taking into account the effects of heat transfer and pore pressure change, one form of the stress equilibrium equation can be expressed as (Tsang and Pruess, 1987)  ¼ ¼ 1 ¼ ¼ ¼T  C 6 u þ u  b I ðT  T0 Þ þ w I rf c þ rb ¼ 0 H ½12:10 2 where ¼ C = elastic tensor ¼ m = deformation tensor ¼ uT = transpose of deformation tensor

© Woodhead Publishing Limited, 2010

Near-field processes, evolution and performance assessment

369

¼

I =unit tensor rf = fluid density (kg/m3) r = porous medium density (kg/m3) y = pressure head (m) T and T0 = temperature and temperature at the reference state (K) and b = body force vector In equation [12.10], c is defined as

1 for saturated zone w¼ ½12:11 wðSl Þ for unsaturated zone For the unsaturated zone, c is a non-linear function of water saturation Sl. This parameter is normally determined in a laboratory. Another parameter β is defined as b ¼ að3l þ 2mÞ

½12:12

where

a = expansivity coefficient of the porous medium (1/K) and l and m = Lame’s constants, which are derived from the porous medium’s Young’s modulus and Poisson’s ratio Recent years have seen the development of quantitative reactive-transport models appropriate for near-field systems, i.e. multicomponent, multiphase systems composed of porous or fractured media (Lichtner, 1985, 1996; Steefel and MacQuarrie, 1996; Xu and Preuss, 2001; Bethke, 2008). The relevant governing equations can be summarized by (e.g. Lichtner, 1996) ‘i Ci ¼

Nr X

vir Ir

ði ¼ 1; . . . ; NÞ

½12:13

r¼1

where ‘i is an operator for which ‘i Ci ¼

qðeSpi Ci Þ þ H  Ji qt

½12:14

for an aqueous or gaseous species, i, and ‘i Ci ¼

qCi qt

½12:15

for a mineral species. In these equations Ci stands for concentration νir represents the stoichiometric coefficient for the ith species in the rth reaction involving that species as a reactant or product

© Woodhead Publishing Limited, 2010

370

Geological repository systems for safe disposal

Nr refers to the total number of such reactions Ir denotes the reaction rate Sπi stands for the saturation of the πth fluid phase containing i t refers to time and Ji represents the flux The latter term includes contributions from advection and diffusion and is given by Ji ¼ etSpi Di HCi þ nCi

½12:16

where τ stands for tortuosity Di refers to the diffusion/dispersion coefficient and v denotes fluid velocity The parameter Ir in equation [12.13] essentially represents a source/sink term resulting from any equilibrium or kinetically controlled homogeneous or heterogeneous reaction (e.g. mineral dissolution/precipitation, gas dissolution/exsolution, sorption, ion exchange, oxidation–reduction, etc). Equation [12.13] is not in a form amenable to solution because it requires knowledge of kinetic rate laws governing Ir as a function of temperature, pressure and species concentrations. For many reactions, however, the intrinsic rate is sufficiently fast that a condition of instantaneous local equilibrium can be assumed and the reaction rate is then controlled by the transport rate of species to and from the site of reaction. Under such conditions equation [12.13] can be reformulated using mass-action constraints in place of the kinetic term Ir (Lichtner, 1996; Steefel and MacQuarrie, 1996). Local equilibrium may not be appropriate for all reactions, however, especially those involving mineral dissolution/precipitation and aqueous oxidation/reduction. For mineral reactions the reaction rate is often assumed to follow a rate law that is compatible with transition-state theory (e.g. Lasaga, 1998). For example:  Z R ¼ +ks1  Oy  ½12:17 where positive values refer to dissolution and negative values indicate precipitation. In this equation k stands for the rate constant, which may vary with temperature and aqueous parameters such as pH, s refers to the reactive surface area, W denotes the saturation index, and θ and η are dimensionless quantities (it is often assumed that θ = η = 1). The saturation

© Woodhead Publishing Limited, 2010

Near-field processes, evolution and performance assessment

371

index is given by O¼

Q K

½12:18

where Q represents the ion-activity product and K refers to the equilibrium constant. Bethke (2008) describes analogous rate laws for oxidation– reduction reactions, including the effects of heterogeneous catalysis and microbial catalysis. Solutions to equation [12.13] are further complicated by the fact that the transport and reaction terms are strongly coupled (e.g. Steefel and MacQuarrie, 1996). The reaction term changes the local concentrations of aqueous species and this affects their flux (equation [12.16]). Changes in flux in turn affect the reaction rate, e.g. by changing the saturation index (equation [12.18]). The transport and reaction terms for systems involving extensive mineral dissolution and precipitation are also coupled through the relation e¼1

M X

fm

½12:19

m¼1

where fm represents the volume fraction of the mth mineral and M refers to the total number of minerals in the system. Cumulative changes in the volume fractions of minerals resulting from all dissolution and precipitation reactions may thus affect solute transport through associated changes in the porosity. Steefel and MacQuarrie (1996) and Xu and Pruess (2001) describe numerical approaches that have been used to obtain solutions to equation [12.13]. The approaches include a one-step, or global implicit, method, which solves the reaction and transport equations simultaneously, as well as alternative approaches that use operator-splitting techniques to solve the reaction and transport equations sequentially. The latter approaches are further categorized based on whether they allow for iterations to occur between the reaction and transport calculations during a single time step. Computer programs based on these techniques include OS3D/GIMRT (Steefel and Yabusaki, 1996), 3DHYDROGEOCHEM (Cheng and Yeh, 1998), TOUGHREACT (Xu et al., 2004), CrunchFlow (see Steefel, 2001) and the Geochemist’s Workbench (GWB) (Bethke, 2008). Examples illustrating how these numerical methods and programs have been applied to near-field problems are described by Steefel and Lichtner (1994), BSC (2004, 2005) and Arthur and Zhou (2005) (see Fig. 12.12). Solutions to each conservation or equilibrium equation with proper initial and boundary conditions provide spatial and temporal distributions of fluid pressures (or head), temperature and stress or compositions of chemical

© Woodhead Publishing Limited, 2010

372

Geological repository systems for safe disposal

12.12 Calculated changes in the porosity of the buffer at various times during the thermal period in a KBS-3 repository (Arthur and Zhou, 2005; see text). The canister–buffer bounday (‘hot end’) is located 0.525 m from the canister center and the buffer–rock bounday (‘cold end’) is 0.875 m from the canister center. Porosity variations reflect the cumulative effects of all mineral dissolution/precipitation reactions. A net reduction in porosity, for example, corresponds to a net increase in mineral volumes (mainly anhydrite in the present model).

species in aqueous, gaseous and solid phases. These equations are mostly non-linear, which makes it difficult to solve them analytically. Geometric considerations can further complicate the solution process. In certain cases, reliable, albeit approximate, solutions can be obtained using simplifying assumptions such as constant porosity, non-deformable porous media, isotropic and uniform properties, etc. In most cases, however, the equations have to be solved numerically. Numerical methods are needed especially if the processes considered are strongly coupled (e.g. BSC, 2005). In such cases, obtaining solutions to the governing equations may be complicated by differences in the duration of time-steps that are relevant to different processes, non-linearity of the equations, spatial discretization, etc. (e.g. Xu et al., 2004). An example of how these numerical methods have been applied to nearfield problems is described by Arthur and Zhou (2005). These authors considered whether the bentonite buffer in a KBS-3 repository for SNF could experience significant cementation during the early ‘thermal’ period of repository evolution. The buffer will be subjected to heating and cooling during this time as radiogenic heat dissipates into the surrounding host rock. Heat conduction will establish temperature gradients between the relatively hot canister–buffer boundary and the relatively cool buffer–rock boundary. The temperature gradients will in turn generate concentration gradients among aqueous solutes because equilibrium and kinetic constraints on reactions involving buffer minerals and pore fluids are temperature

© Woodhead Publishing Limited, 2010

Near-field processes, evolution and performance assessment

373

dependent. Diffusional mass transport of aqueous ions and neutral species responding to the thermally induced concentration gradients may cause secondary minerals to precipitate on to the surfaces of the primary clay minerals in the buffer. Should these complex, and possibly coupled, reactivetransport processes lead to significant mineral mass transfer, individual clay particles could become cemented together by the secondary solids. This is a concern because such cementation could alter the swelling pressure and other properties of the buffer irreversibly, thereby affecting the primary performance function of this EBS component to provide a stable diffusional transport pathway between the canister and rock. Arthur and Zhou (2005) used TOUGHREACT (Xu et al., 2004) to solve the system of reactive-transport equations governing buffer cementation. Their model accounts for the imbibition of groundwater from a granitic host rock into initially unsaturated buffer materials under capillary and hydraulic pressure gradients, and uses realistic time–temperature constraints on the thermal evolution of the near field. Model results suggest that the total extent of mineral mass transfer, represented by spatial and temporal variations in porosity, is likely to be very small throughout the thermal period (Fig. 12.12). Assessing the effects of such changes on buffer performance is problematic, however, because it is unclear how mineralogical changes might affect the physical, mechanical and rheological properties of these materials. The results also suggest that a complex interplay between time, temperature and reaction rate will control the extent of any cementation that does occur, and that such effects may be as important during the waning stages of the thermal period as they are initially. Cooling beyond the period of peak buffer temperatures may stimulate dissolution of neoformed minerals having retrograde solubilities, but this may not be sufficient to reverse completely the effects of cementation before and after this period. The US repository program has offered an unique opportunity to advanced simulation of coupled T-H-M-C processes. During the early stage of the program, both semi-analytical (Doughty and Pruess, 1988) and numerical models (Tsang and Pruess, 1987) were developed to study the heat-pipe effect surrounding an emplacement drift. Later a multiscale T-H modeling methodology has been developed that integrates the results from one-, two-, and three-dimensional drift-scale models and a three-dimensional mountain-scale model to calculate the near-field T-H variables affecting the performance of the EBS (Buscheck et al., 1998). The program has also developed a numerical drift-scale model to study the near-field T-HC coupled processes (BSC, 2005). The three-dimensional coupled T-H model has been developed to investigate the impact of natural convection on near-field T-H processes by including vapor diffusion within an empty space of the emplacement drift (Birkholzer et al., 2006). In the recent enhanced

© Woodhead Publishing Limited, 2010

374

Geological repository systems for safe disposal

SNF storage capacity study, both two- and three-dimensional drift-scale numerical models were developed to simulate the coupled T-H-M processes (EPRI, 2007), in which the T-H processes are fully coupled, with results at several given times fed to a T-M coupled model to simulate permeability and porosity variations. The latter were then delivered back to the T-H coupling model. Such a treatment of coupling is able to capture the important coupling moments at a low computing expense. In fact, in this particular problem where the maximum rock temperature has been found to be below 200 8C, direct or full coupling of T-H-M is unnecessary.

12.6

Future trends in near-field analysis

The ultimate goal of near-field analysis is to assist evaluations of the safety case in disposal concepts for nuclear wastes. We conclude this chapter with a discussion of R&D topics that we believe will improve linkages between near-field analysis and safety assessment. A safety case includes a description of functional requirements for individual components of the EBS and near-field rock. For example, a functional requirement of the buffer in a KBS-3 repository for spent nuclear fuel is to ensure that transport of corrodants, to the canister, and of potential radio-nuclide releases from the canister, is diffusion limited (e.g. see Vieno and Ikonen, 2005). It is convenient first to divide the assessment time frame considered in a safety case into separate phases, each of which can be distinguished by certain features, events and processes controlling the evolution of the near field as it responds to human-induced and natural changes in the geological environment. Three such phases have been defined for a possible spent-fuel repository at Olkiluoto, Finland, for example (Pastina and Hella, 2006): . .

.

an ‘operational phase’, starting with emplacement of the first waste package and ending with repository closure, about 100 years later; a ‘post-closure thermal phase’, starting at the time of repository closure and ending at the onset of the next glaciation (this phase is further subdivided into two periods, an ‘early post-closure period’, characterized by elevated gradients in physiochemical variables attributable to repository construction and emplacement of radiogenically heated waste packages, and a ‘quasi-steady-state period’, in which gradients in these physiochemical variables return to near-ambient values); a ‘glacial phase’, starting with the formation of the first layer of permafrost and ending when temperatures in the repository return to near-ambient values such as exist in the present interglacial period.

An additional ‘initial states phase’ can be added to the above to permit

© Woodhead Publishing Limited, 2010

Near-field processes, evolution and performance assessment

375

consideration of design criteria, quality assurance/quality control protocols and methods of EBS fabrication and emplacement (SKB, 2006). Identification of distinct time phases, such as the above, serves to focus attention on coupled processes and associated issues appropriate for each phase. We believe key issues requiring further evaluation, within the context of the time phases noted above, include: .

. . . . . .

the exchange of atmospheric gases (N2, O2) in excavated tunnels and dissolved gases (CO2, CH4) in pressurized groundwater within the rock during the extended operational phase, and possible chemical buffering within the rock that would tend to resist changes induced by such gas exsolution/dissolution reactions; redistribution of water by evaporation/condensation during the operational phase driven by radiogenic heating of emplaced waste packages; possible introduction of aerobic bacteria into anaerobic host rocks during the operational phase, which may further perturb local geochemical conditions; possible operational phase disturbances to the geological environment caused by pumping and removal of inflowing groundwater and by upconing deeper groundwater; mass redistributions among and between EBS components induced by thermal gradients during the early post-closure thermal phase; possible mechanical effects on the host rock resulting from the generation of H2 gas or expansion of canister corrosion products during the post-closure thermal phase; effects on EBS stability and performance of changes in the hydrogeology and hydrochemistry of the far field during periods of glacial advance and retreat.

Insights regarding each of these processes can be obtained using an approach that combines numerical modeling, laboratory experimentation and observations of natural systems. General topics requiring additional research to support this three-pronged approach include: .

. .

the provision of reliable thermodynamic data, as necessary, for key minerals, aqueous species and reactions characteristic of specific EBS components (e.g. clay buffer/clay host rocks, cementitious materials), radioelements and disposal environments; improvements to kinetic models and data, especially with regard to crystal nucleation and growth, and the characterization of reactive surface areas; improvements to models or empirical correlations relating changes in chemical/mineralogical properties to corresponding changes in the

© Woodhead Publishing Limited, 2010

376

. .

.

Geological repository systems for safe disposal mechanical/rheological properties of EBS components (especially bentonite buffers); a need for laboratory testing of T-H-M-C processes conducted with full control of relevant repository conditions, especially for the post-closure thermal period; better use of data interpreted from observations of natural/archaeological analogues to test the accuracy of predictions made using coupled T-H-M-C models over time scales that are experimentally inaccessible; application of the multiscale numerical modeling approach so that sufficient computing resources are allocated to spatial and temporal domains of the model with strong coupling effects of T-H-M-C processes.

12.7

References

Arthur R and Zhou W (2005), ‘Reactive-transport model of buffer cementation’, SKI Report 2005:59, Swedish Nuclear Power Inspectorate, Stockholm, Sweden. Bethke C M (2008), Geochemical and Biogeochemical Reaction Modeling, Cambridge University Press, New York. Birkholzer J T, Halecky N, Webb S W, Peterson P F and Bodvarsson G S (2006), ‘The impact of natural convection on near-field TH processes at Yucca Mountain’, in Proceedings of the 11th International High-Level Radioactive Waste Management Conference, April 2006, Las Vegas, Nevada. Borrelli R and Ahn J (2008), ‘Numerical modeling of bentonite extrusion and radionuclide migration in a saturated planar fracture’, Physics and Chemistry of the Earth, Parts A/B/C, 33, S131–S141. BSC (2004), ‘Engineered barrier system: physical and chemical environment model’, Bechtel SAIC Company Report to DOE, ANL-EBS-MD-000033 Rev. 02, February 2004. BSC (2005), ‘Drift-scale THC seepage model’, MDL-NBS-HS-000001 Rev. 04, February 2005, Bechtel SAIC Company, Las Vegas, Nevada. Buscheck T, Nitao J and Ramspott L (1996), ‘Localized dry-out: an approach for managing the thermal-hydrological effects of decay heat at Yucca Mountain’, in Materials Research Society Symposium Proceedings, vol. 412, Material Research Society, Pittsburgh, Pennsylvania, pp. 715–722. Buscheck T A, Gansemer J, Nitao J J and DeLorenzo T H (1998), ‘Multi-scale nearfield thermohydrologic analysis of alternative designs for the potential repository at Yucca Mountain’, in Materials Research Society Symposium Proceedings, vol. 556, Material Research Society, Pittsburgh, Pennsylvania, pp. 615–622. Cheng H P and Yeh G T (1998), ‘Development of a three-dimensional model of subsurface flow, heat transfer, and reactive chemical transport: 3DHYDROGEOCHEM’, Journal of Contaminant Hydrology, 34, 47–83. Conca J L, Apted M J, Zhou W, Arthur R C and Kessler J H (1998), ‘Flow barrier

© Woodhead Publishing Limited, 2010

Near-field processes, evolution and performance assessment

377

system for long term high-level waste isolation: experimental results’, Nuclear Technologies, 124(October 1998), 88–100. Doughty C and Pruess K (1988), ‘A semi-analytical solution for heat pipe effects near high level nuclear waste packages buried in partially saturated geologic media’, International Journal of Heat and Mass Transfer, 31 (1), 79–90. EPRI (2007), ‘Program on technology innovation: room at the mountain – analysis of the maximum disposal capacity for commercial spent nuclear fuel in a Yucca Mountain Repository’, TR-1015046, Electric Power Research Institute, Palo Alto, California. Lasaga A C (1998), Kinetic Theory in the Earth Sciences, Princeton Series in Geochemistry, Princeton University Press, Princeton. Lichtner P C (1985), ‘Continuum model for simultaneous chemical reactions and mass transport in hydrothermal systems’, Geochimica et Cosmochimica Acta, 49, 779–800. Lichtner P C (1996), ‘Continuum formulation of multicomponent–multiphase reactive transport’, in Reactive Transport in Porous Media, edited by P C Lichtner, C I Steefel and E H Oelkers, Reviews in Mineralogy 34, Mining Society of America, Washington, DC, pp. 1–81. NCRP (2005), ‘Performance assessment of near-surface facilities for disposal of lowlevel radioactive waste’, National Council on Radiation Protection and Measurements Report 152, NCRP, Bethesda, Maryland, December 2005. Pastina B and Hella P (eds) (2006) ‘Expected evolution of a spent nuclear fuel repository at Olkiluoto’, POSIVA 2006-05, Posiva Oy, Olkiluoto, Finland. SKB (1999), ‘SR-97 Post-closure safety, vol. 1, SKB Technical Report TR-99-06, Swedish Nuclear Fuel and Waste Management Company, Stockholm, Sweden, October 1999. SKB (2001a), ‘Project SAFE: radionuclide release and dose from the SFR repository’, SKB Report R-01-18, Swedish Nuclear Fuel and Waste Management Company, Stockholm, Sweden, October 2001. SKB (2001b), ‘Project JADE: long-term function and safety comparison of repository systems’, SKB Report TR-01-18, Swedish Nuclear Fuel and Waste Management Company, Stockholm, Sweden, December 2001. SKB (2004), ‘Choice of rock excavation methods for the Swedish deep repository for spent nuclear fuel’, SKB Report R-04-62, Swedish Nuclear Fuel and Waste Management Company, Stockholm, Sweden, September 2004. SKB (2006), ‘Long-term safety for KBS-3 repositories at Forsmark and Laxemar – a first evaluation’, SKB Technical Report TR-06-09, Swedish Nuclear Fuel and Waste Management Company, Stockholm, Sweden. Steefel C I (2001), ‘GIMRT, Version 1.2: software for modeling multicomponent, multidimensional reactive transport, User’s Guide’. Report UCRL-MA143182, Lawrence Livermore National Laboratory, Livermore, California. Steefel C I and Lichtner P C (1994), ‘Diffusion and reaction in rock matrix bordering a hyperalkaline fluid-filled fracture’, Geochimica et Cosmochimica Acta, 58, 3595–3612. Steefel C I and MacQuarrie K T B (1996), ‘Approaches to modeling reactive transport in porous media’, in Reactive Transport in Porous Media, edited by P C Lichtner, C I Steefel and E H Oelkers, Reviews in Mineralogy 34, Mining Society of America, Washington, DC, pp. 83–125.

© Woodhead Publishing Limited, 2010

378

Geological repository systems for safe disposal

Steefel C I and Yabusaki S B (1996), ‘OS3D/GIMRT, software for multicomponent– multidimensional reactive transport, User Manual and Programmer’s Guide’, PNL-11166, Pacific Northwest National Laboratory, Richland, Washington. Tsang Y W and Pruess K (1987) ‘A study of thermally induced convection near a high-level nuclear waste repository in partially saturated fractured tuff’, Water Resources Research, 23(10), 1958–1966. US DOE (2002), ‘Yucca Mountain Science and Engineering Report, technical information supporting site recommendation consideration Revision 1’, DOE/ RW-0539-1, US Department of Energy Office of Civilian Radioactive Waste Management, February 2002. US DOE (2007), ‘EBS radionuclide transport abstraction’, ANL-WIS-PA-000001 Rev. 03, US Department of Energy, October 2007. US DOE (2008), ‘Total system performance assessment model/analysis for the license applications’, MDL-WIS-PA-0000005 Rev. 00, US Department of Energy. Vieno T and Ikonen A T K (2005), ‘Plan for safety case of spent fuel repository at Olkiluoto’, Posiva 2005-01, Posiva Oy, Olkiluoto, Finland. Watanabe K (1989), ‘Archeological evidence supports a preferred soil cap for LLRW disposal’, in Proceedings of the 1989 Joint International Waste Management Conference, edited by F Feizollahi, R Kohout and A Susuki, Kyoto, Japan, The American Society of Mechanical Engineers. Xu T and Pruess K (2001), ‘Modeling multiphase non-isothermal fluid flow and reactive geochemical transport in variably saturated fractured rocks: 1. Methodology’, American Journal of Science, 301, 16–33. Xu T, Sonnenthal E, Spycher N and Pruess K (2004), ‘TOUGHREACT User’s Guide: a simulation program for non-isothermal multiphase reactive geochemical transport in variably saturated geologic media’, LBNL-55460, Lawrence Berkeley National Laboratory, Berkeley, California. Zhou W, Conca J Arthur R and Apted M (1996), ‘Analysis and confirmation of robust performance for the flow-diversion barrier system within the Yucca Mountain Site’, TR-107189, Final Report, Electric Power Research Institute, Palo Alto, California, December 1996. Zhou W, Apted M J and Kessler J H (2008), ‘Thermohydrological behavior in rock pillars between emplacement drifts at Yucca Mountain Repository’, in Proceedings of the 12th International High-Level Radioactive Waste Management Conference, 7–11 September 2008, Las Vegas, Nevada.

© Woodhead Publishing Limited, 2010

13 Nuclear waste canister materials, corrosion behaviour and long-term performance in geological repository systems F . K I N G , Integrity Corrosion Consulting Ltd, Canada; D . W . S H O E S M I T H , University of Western Ontario, Canada

Abstract: The nuclear waste canister is an integral component of the engineered barrier system and is the only absolute barrier in the overall multi-barrier concept. Corrosion of the canister material can lead to failure and the release of radionuclides. Here the corrosion behaviour of five classes of metal alloys that have been considered as canister materials is reviewed, along with the environmental conditions to which they are exposed. The five classes of material are: carbon steel and cast iron, stainless steels, copper and copper alloys, titanium alloys and nickel-based alloys. Key words: canisters, containers, waste package, copper, carbon steel, cast iron, titanium alloys, nickel alloys, stainless steel, corrosion, localised corrosion, pitting, crevice corrosion, microbiologically influenced corrosion, MIC, stress corrosion cracking, SCC, hydrogen damage, failure.

13.1

Introduction

The nuclear waste canister is an integral component of the engineered barrier system (EBS), itself part of the multi-barrier system on which the disposal of nuclear waste is based. The canister is the only absolute barrier in the overall waste disposal system and, as such, is often a focus of attention. From the public’s viewpoint, the concept of a ‘container’ is more tangible than that of a redox, chemical or diffusion barrier. From a technical viewpoint, the prospect of very long canister lifetimes (greater than, say, 100 000 a), due either to thermodynamic immunity or kinetic constraints, can compensate for the lack of, or uncertainty in, the barrier capability of other components of the system. 379 © Woodhead Publishing Limited, 2010

380

Geological repository systems for safe disposal

The focus here is on the corrosion behaviour of candidate materials for the disposal of spent fuel (SF) or high-level waste (HLW). For the disposal of SF or HLW, the canister is required to provide absolute containment for a given period of time. Less emphasis is given to the disposal of low- and intermediate-level waste (LILW) since there is often no absolute containment requirement in this case and the canister design may include a vent to release gases from inside the container. Although the canisters may ultimately fail due to mechanical overload, it is assumed here that early structural failure can be avoided through prudent design. Consequently, most of the discussion here is on the corrosion processes that may lead to structural failure, rather than on the mode of mechanical failure itself.

13.2

Environmental aspects important for nuclear waste canister performance

The corrosion behaviour of the canister cannot be understood or predicted without a detailed knowledge of the environmental conditions to which it will be exposed. These environmental conditions include not only the characteristics of the host geological formation but, more importantly, those in the near field and how they change with time.

13.2.1 Environmental factors There are many environmental factors of importance for the corrosion of the canister, including: temperature; the ground water and pore water chemistry, especially the pH and the concentrations of Cl, S species and carbonate; the redox conditions, including the effect of gamma radiation; microbial activity; the degree of saturation; the mass transport of reactants to, and of products away from, the canister; and the level of residual stress and external load. The location of the repository with respect to the level of the local water table has a significant impact on the corrosivity of the environment. With one exception, all underground repository sites are located in the saturated zone underneath the water table. Such environments tend to be cooler, because of the relatively high thermal conductivity of the saturated backfill and host rock, with a limited supply of oxidant (McMurry et al., 2003; Pastina and Hella¨, 2006; SKB, 2006a). The one exception is the proposed Yucca Mountain (YM) repository site in the US (DOE, 2008), which is located in the unsaturated zone approximately halfway between the surface of the mountain and the water table. This repository environment is permanently aerobic and, by design, is hotter and drier than repository designs below the water table.

© Woodhead Publishing Limited, 2010

Nuclear waste canister materials

381

13.1 Schematic illustration of the evolution of the near-field environment for a repository located below the water table (McMurry et al., 2003).

Temperature Heat generation from radioactive decay of the SF/HLW results in an initial period of elevated temperature (sometimes referred to as the ‘thermal pulse’), followed by a period of slow cooling as the waste decays (Fig. 13.1). In saturated repository designs, the maximum canister surface temperature is typically ca. 100 oC, a limit set more by the thermal stability of the buffer and backfill materials than by the corrosion behaviour of the canister. In the unsaturated YM repository, the maximum canister surface temperature is on the order of 200–220 oC. The maximum temperature is, of course, a design parameter that can be varied by adjusting the waste loading, the type of waste, the nature of the buffer and backfill materials used (if any), and any period of forced or natural ventilation following canister emplacement. Ground water and pore water composition Corrosion of the canister is a result of the ground water contacting the canister surface. (The maximum canister temperature is too low for there to be any significant gas-phase oxidation.) In backfilled repository designs, the ground water will be modified by reactions in the backfill so that the

© Woodhead Publishing Limited, 2010

382

Geological repository systems for safe disposal

13.2 Comparison of the maximum temperature and ground water chloride concentration in various national nuclear waste management programmes (Shoesmith, 2006). (1) tuff (United States), (2) clay (Belgium), (3) granite (Sweden), (4) granite (Canada), (5) salt (Germany).

composition of the pore water that actually contacts the canister can be quite different (McMurry et al., 2003; Pastina and Hella¨, 2006). As a general guide, Fig. 13.2 shows the ranges of ground water Cl concentration for various proposed host-rock formations (Shoesmith, 2006). These range from relatively dilute ground waters in sedimentary deposits, through the more saline ground waters found in crystalline rock, to the extremely saline inclusions found in salt deposits. Because the canister is hot, the water contacting the surface will be concentrated by evaporation. Evaporative concentration will not only occur for any dripping vadose zone water that drips on to the hot canisters in the YM repository (DOE, 2008) but it will also occur initially in repository concepts in the saturated zone as the moisture in the buffer and backfill materials is re-distributed under the thermal gradient (King, 2006). Apart from chloride, the other ground water and pore water species important to the corrosion of the canister include: sulphide, which can adversely affect the corrosion of copper, carbon steel (C-steel) and Ni-based alloys (Kursten et al., 2004); sulphate, the reduction of which by sulphatereducing bacteria results in the formation of sulphide (SKB, 2006b); carbonate/bicarbonate, which can both buffer the pH and promote passivation of certain materials (JNC, 2000); pH, especially the alkaline pH resulting from the use of cementitious backfills or overpacks (King, 2002); cations such as Ca2+ and Mg2+, which may either hydrolyse to produce aggressive environments (as in certain salt brine inclusions) or which may promote the formation of protective mineralised surface films

© Woodhead Publishing Limited, 2010

Nuclear waste canister materials

383

(Kursten et al., 2004); and certain other specific species, such as ammonium ions, which can support the stress corrosion cracking of copper alloys (King et al., 2001, 2002). Redox conditions Corrosion involves the oxidation of the canister material, so the redox conditions in the repository are of crucial importance. In saturated repository designs, the amount of O2 available to support corrosion is limited to that initially present in the buffer and backfill sealing materials at the time of closure (McMurry et al., 2003; Pastina and Hella¨, 2006; Johnson and King, 2008). Although H2O is also an oxidant for some canister materials, the nature of the corrosion processes in anoxic conditions is typically less severe than in aerobic environments (Shoesmith, 2006). In the YM repository, however, there is a virtually unlimited supply of atmospheric O2 and the conditions are permanently aerobic. Radiolysis of ground water or pore water by gamma (or neutron) radiation can potentially generate oxdising (and reducing) molecular and radical species. However, most candidate canister materials exhibit little effect of radiation at absorbed dose rates of <1 Gy/h (Shoesmith and King, 1999). Since the surface dose rate of most canister designs is below this level, because of the trend towards thick-walled designs for structural purposes, the impact of radiolysis on the corrosion behaviour of canister materials is minimal. Microbial activity Microbes can affect the corrosion behaviour of metals by altering the nature of the environment through the production of aggressive metabolic byproducts or through the creation of locally occluded regions under biofilms (King, 2009a). Microbes can also be beneficial, by consuming a fraction of the initially trapped O2 in saturated repository designs or by forming protective biofilms. When considering the impact of microbial activity on the corrosion behaviour of the canister, it is necessary to distinguish between microbial activity at the canister surface from microbial activity further away. Surface microbial activity can lead to colonisation and biofilm formation, which in turn can lead to the spatial separation of anodic and cathodic processes and the possibility of localised corrosion (Little et al., 1991). On the other hand, for remote microbial activity to impact the canister, the aggressive species must first be transported to the surface, often through low-permeability buffer and backfill material (King, 2009a). In general, the repository environment is not hospitable to microbial

© Woodhead Publishing Limited, 2010

384

Geological repository systems for safe disposal

activity (King and Stroes-Gascoyne, 1997). Elevated temperatures, desiccation, radiation fields, the lack of nutrients or terminal electron acceptors, saline ground and pore water, mechanical forces from swelling bentonite and the evolving redox conditions all contribute to this inhospitable environment for microbes. Microbes can always be identified that can survive, but not necessarily be active, in any of these extreme conditions. However, taken together, these actions of the environmental stressors reduce the overall activity and diversity of the microbial population (King, 2009a). Saturation The time-dependent degree of saturation of the repository is an important factor in determining when, and in what form, corrosion can occur (King, 2006). The YM repository is specifically designed to remain sufficiently dry to avoid aqueous corrosion for an initial period of a few thousand years (Shoesmith, 2006; DOE, 2008). As the temperature cools and the relative humidity (RH) increases, there comes a time at which the surface of the canister will wet and aqueous corrosion becomes possible (Fig. 13.3). The same type of processes will occur for repository designs in the saturated zone, albeit somewhat sooner (King, 2006). Predictions of the time for complete saturation of a repository in the saturated zone are typically in the range of 100–1000 a (McMurry et al., 2003; Pastina and Hella¨, 2006; SKB, 2006a; Johnson and King, 2008). Prior to saturation, localised corrosion is possible because surface salts

13.3 Time dependence of the canister temperature and the relative humidity in the drift for the Yucca Mountain repository (Shoesmith, 2006).

© Woodhead Publishing Limited, 2010

Nuclear waste canister materials

385

will wet (deliquesce) prior to the development of full saturation, leading to spatial separation of anodic and cathodic sites (King, 2006). With increasing RH more of the canister surface will wet, until such time that the surface is uniformly wetted. The rate of supply of gaseous reactants (e.g. O2) will be higher under unsaturated conditions, but the rate of removal of dissolved corrosion products will diminish, increasing the propensity for film formation. Mass transport The rates of mass transport of reactants to, and of products away from, the canister surface will have a strong influence on the corrosion behaviour (King et al., 1996). Mass transport is restricted in most repository designs, primarily because of the need to retard the migration of radionuclides once the canister has failed. Restrictive mass-transport conditions are imposed either through the use of low-permeability buffer and backfill materials, such as highly compacted bentonite, or because of the low permeability of the host rock (e.g. saturated sedimentary deposits, such as Opalinus Clay, or unsaturated porous materials, such as tuff). Highly compacted bentonite (HCB) is proposed as a buffer material in a number of repository designs (McMurry et al., 2003; Pastina and Hella¨, 2006; SKB, 2006a). The hydraulic conductivity of HCB is sufficiently low (<1013 m/s when saturated) that mass transport occurs by diffusion only. Diffusivities are on the order of a factor of 100 lower than in bulk solution (King et al., 1996). Bentonite also adsorbs cationic species, such as Cu(II), which can impact the corrosion behaviour. Of concern, however, is that saturated bentonite is so impermeable that H2 produced by corrosion of base materials, such as C-steel, may not be able to be transported away from the canister surface fast enough to prevent an H2 gas phase from forming (Nagra, 2002). Residual and applied stress Canisters will be subject to both residual and applied stresses and, hence, the possibility of environmentally assisted cracking (Shoesmith, 2006). The applied stress, in particular, will be largely compressive but inevitably there will be regions of tensile stress on the surface of the canister. Because the final closure weld cannot be thermally heat treated because of the temperature limits of the waste form, there will be residual stress in that region up to the yield stress of the weld material. Some mitigation of the surface residual stress may be possible using non-thermal techniques, such as laser peening or low-plasticity burnishing (DOE, 2008). The magnitude of the applied stress depends on the repository design and

© Woodhead Publishing Limited, 2010

386

Geological repository systems for safe disposal

host rock formation. In strong rock (i.e. rock that is self-supporting), the external load will comprise the swelling pressure of the HCB backfill (if used) and the hydrostatic pressure once the repository has saturated (McMurry et al., 2003; Pastina and Hella¨, 2006; SKB, 2006a; Johnson and King, 2008). The latter will be augmented during glacial cycles by a hydraulic head equivalent to the thickness of the ice shield. In non-selfsupporting rock, such as salt domes and sedimentary deposits, there will be an additional lithostatic load. The total external load is typically of the order of 20–50 MPa – hence the use of thick-walled canister designs to withstand the applied load. In the unsaturated zone, there will be no hydraulic load, of course. In the unbackfilled YM design, the only sources of external load are the impact from falling or ejected rocks and the impacts of canisters with each other and with other EBS components in the event of seismic activity (DOE, 2008).

13.2.2 Evolution of the repository environment An important characteristic of the repository environment mentioned above is that it will evolve over time (McMurry et al., 2003; Pastina and Hella¨, 2006; SKB, 2006a; Johnson and King, 2008). The most obvious example of this evolution over time is the decrease in canister surface temperature (Figs 13.1 and 13.3). For saturated repositories, an equally important aspect of the evolving environment is the consumption of the initial inventory of atmospheric O2 by a combination of corrosion of the canister, aerobic respiration by microbes and the oxidation of Fe(II) minerals (Fig. 13.1). Other environmental parameters that will change over time are the degree of saturation, the applied load, the magnitude of the external radiation field (if significant), the type and distribution of microbial activity and the composition of the aqueous phase contacting the canister (McMurry et al., 2003; Pastina and Hella¨, 2006; SKB, 2006a; Johnson and King, 2008). In the latter case, bentonite pore waters will tend to become more saline with time, eventually resembling the composition of the ground water in the surrounding rock. If a cementitious backfill is used, the pore water pH will evolve as the alkaline mineral phases are leached out over time (Bel et al., 2006; Kursten and Druyts, 2008). In the case of the YM repository, however, the aqueous phase contacting the canister becomes less concentrated with time as the canister cools and there is less chance of evaporative concentration (DOE, 2008). For most canister materials and for most potential corrosion processes, the evolution in the repository environment is characterised by a change from initial aggressive conditions to a subsequent benign environment (McMurry et al., 2003; Johnson and King, 2008). This evolution is illustrated for a saturated repository in Fig. 13.1 as an evolution from the

© Woodhead Publishing Limited, 2010

Nuclear waste canister materials

387

initial warm, oxidising period to the longer-term cool, anoxic period. Recognition of this evolution of conditions is important for justification of long-term predictions of canister lifetimes (Johnson and King, 2008). The more-rapid and, arguably, the less-predictable forms of localised corrosion tend to occur during the initial warm, oxidising period. Slower, morepredictable forms of general corrosion tend to dominate the long-term cool, anoxic period. Since the aerobic period may only last a few tens of years, this means that the difficult task of predicting the corrosion behaviour over periods of thousands or tens of thousands of years is greatly simplified. If the most-aggressive forms of corrosion only occur over a period of a few tens of years, it almost becomes possible to perform experiments over meaningful lengths of time. It certainly becomes easier to justify the predictions of, for example, localised models or models for the prediction of stress corrosion cracking. A similar evolution from initially aggressive conditions to a long-term benign environment is also expected for the unsaturated YM repository, although in this case the evolution is almost entirely in terms of the evolution of temperature, rather than a combination of temperature and redox conditions as in a saturated repository (DOE, 2008; King et al., 2008a). There are some environmental conditions and corrosion processes, however, for which this evolution from ‘aggressive’ to ‘benign’ does not apply. In general, the conditions for microbial activity tend to improve with time as the temperature cools (King, 2009a), although in a saturated repository aerobic respiration will obviously cease once the initially trapped O2 has been consumed. In terms of corrosion processes, the H-related degradation of C-steel (and possibly of Ti alloys) becomes more likely as the temperature decreases (King, 2009b; Turnbull, 2009), primarily as a consequence of the decrease in H solubility and the threshold conditions for cracking with decreasing temperature.

13.2.3 Importance of the near-field environment The corrosion behaviour of the canister is determined primarily by the environmental conditions of the near field. Although this may seem an obvious statement, it is worth emphasising that the near-field environment may be quite different from that in the host rock. Thus, although deep ground waters may contain significant concentrations of, say, sulphide or ammonia, it does not necessarily disqualify the use of copper as a canister material because the interfacial concentration of these species may be much lower as a result of reactions in the buffer or backfill material. An obvious example of this principle is the microbiologically influenced corrosion of copper canisters surrounded by HCB. It has been demonstrated that microbial activity is suppressed in HCB (King and Stroes-Gascoyne, 1997;

© Woodhead Publishing Limited, 2010

388

Geological repository systems for safe disposal

Stroes-Gasoyne et al., 2007; Masurat et al., 2009) so, although conditions may be suitable for the microbial reduction of sulphate to sulphide in the ground water, the impact on the canister is minimal in this case because the sulphide must first diffuse to the interface through the HCB (King and Kolar, 2006). It is important, therefore, to select a canister material based on the near-field environmental conditions rather than on those in the host rock.

13.2.4 Managing the near-field environment through engineering design There are opportunities to manipulate the near-field environment to either improve, or in some cases to simplify the prediction of, the corrosion behaviour of the canister. Examples of such engineering design include: .

.

The use of cementitious backfill or a cement overpack to promote passivation (Bel et al., 2006; Kursten and Druyts, 2008). This strategy is used in the Belgian ‘supercontainer’ concept in which a C-steel overpack is surrounded by an OPC-based concrete buffer (Fig. 13.4). The use of low-pH cement to prevent passivation. An alkaline plume from cement backfill or seals could promote the passivation of C-steel or copper and, hence, increase the probability of localised corrosion. This possibility can be avoided through the use of low-pH cement.

13.4

Schematic illustration of the Belgian ‘supercontainer’ concept.

© Woodhead Publishing Limited, 2010

Nuclear waste canister materials .

.

389

The use of HCB to prevent microbial activity near the canister. Microbial activity is suppressed in HCB due to the low water activity and spatial and mechanical restrictions imposed by the swelling pressure (Stroes-Gascoyne et al., 2007; Masurat et al., 2009). The use of HCB, therefore, is an effective way to prevent biofilm formation on the canister surface and to eliminate the attendant uncertainties in predicting the extent of corrosion (King, 2009a). Varying the waste loading to either increase or decrease the temperature. There are obvious advantages in lowering the temperature as it minimises the opportunity for localised corrosion and other rapid forms of attack. On the other hand, increasing the temperature could delay the onset of aqueous corrosion, as in the YM strategy.

13.3

Selection of nuclear waste canister materials

There are a number of considerations when selecting a canister material for a particular repository design. The overriding consideration, however, is that the canister is just one of a number of barriers in a multi-barrier system. Safety should not be reliant on any single barrier, but those barriers, including the canister, should be selected to be compatible with other components of the system.

13.3.1 Target lifetimes The choice of canister material will be influenced by the target canister lifetime(s). That lifetime may be specified by a regulatory requirement (e.g. complete containment for a period of 1000 a) or by a functional requirement to enable safe handling or ease of retrieval. From an overall system performance perspective, there may be a desire to avoid the uncertainties in having to predict the rates of waste alteration and radionuclide transport during the initial transient period before the establishment of more stable environmental conditions (Johnson and King, 2008). Alternatively, there may be a desire to claim substantial compliance with the overall dose limits through the use of a long-lived canister, i.e. 100 000 a or longer (SKB, 2006a). Conversely, in low-permeability sedimentary host rock, there is less need for a long-lived canister since the rock itself is a significant barrier (Nagra, 2002). Having defined the target canister lifetime(s), there is then the need to select a canister material to meet that target for the given environment. In saturated-type repositories in which the availability of O2 is limited, copper is capable of providing very long canister lifetimes (King et al., 2001, 2002; Shoesmith, 2006). On the other hand, if the target lifetime is only 1000 a, then C-steel is a perfectly adequate choice (Johnson et al., 1994; Shoesmith,

© Woodhead Publishing Limited, 2010

390

Geological repository systems for safe disposal

2006). Passive materials, such as Ti and Ni–Cr–Mo alloys, also offer the possibility of very long lived canisters, provided an alloy immune or highly resistant to localised corrosion in the repository environment is selected.

13.3.2 Corrosion-allowance versus corrosion-resistant materials Candidate canister materials generally fall into one of two categories: corrosion-allowance materials, such as copper and C-steel, and corrosionresistant materials, such as the stainless steels and Ti- and Ni-based alloys (Shoesmith, 2006). Typically, corrosion-allowance materials corrode uniformly and are less susceptible to localised corrosion and environmentally assisted cracking. Lifetime prediction and, more importantly, justification should be relatively straightforward, especially with the availability of natural and archaeological analogues. Corrosion-resistant materials tend to be more susceptible to localised attack in the form of pitting or crevice corrosion. Despite efforts over the past 20 a or so, there is still a degree of skepticism in the corrosion community about our ability to make long-term predictions of localised corrosion. However, there may be reasons of economics or resource usage that will also influence the choice between a corrosion-allowance or corrosion-resistant material.

13.3.3 Compatibility with other engineered and natural barriers The canister material should not adversely affect the performance of other barriers in the system. Of the candidate canister materials, C-steel has the greatest impact on other parts of the system, both adversely and advantageously. Carbon steel corrosion products may adversely affect HCB, both through the generation and transport of gaseous H2 (Nagra, 2002) and through the alteration of swelling smectite clays by interaction with Fe(II) (Carlson et al., 2006). On the other hand, the H2 produced by corrosion will suppress the alteration of SF (Shoesmith, 2008) and the precipitated Fe(II) corrosion products will both maintain reducing conditions in the near field and provide significant sorption capacity for radionuclides released from the waste form (Pastina and Hella¨, 2006).

13.3.4 Definition of canister failure There is no universally accepted definition of what constitutes canister failure. Two useful definitions are: 1.

Loss of absolute containment, corresponding to the time at which

© Woodhead Publishing Limited, 2010

Nuclear waste canister materials

2.

391

through-wall penetration occurs and at which it might be conservatively assumed that alteration of the waste form (spent fuel or high-level waste) begins due to contact by groundwater. Loss of barrier function, corresponding to the time at which the canister no longer provides significant mass-transfer resistance to the transport of radionuclides.

Various processes may lead to canister failure. First, manufacturing defects that are not detected during canister inspection could lead to early throughwall penetration. In performance assessment calculations, it is typical to assume that on the order of 1 in 1000 to 1 in 10 000 of the canisters will contain undetected manufacturing defects leading to immediate or early failure. Second, as discussed extensively below, corrosion of the canister material can lead to failure, either locally as a result of pitting, crevice corrosion, stress corrosion cracking or more generally. A thorough understanding of the mode(s) and mechanism(s) of the corrosion process(es) is essential for predicting failure times due to corrosion. Third, mechanical overload of the canister can also lead to failure. Although immediate structural failure of the canister can be avoided through a combination of prudent engineering design and an understanding of the service loads to which the canister will be subjected, time-dependent phenomena such as corrosion and creep complicate the analysis. For example, a canister undergoing general corrosion will inevitably fail by mechanical overload prior to penetration of the canister wall by corrosion.

13.4

Corrosion behaviour of candidate nuclear waste canister materials

13.4.1 Carbon steel and cast iron Carbon steel (C-steel) refers to alloys of iron and carbon with a C content <2 wt%, typically containing Mn (<1.65 wt%), Si (<0.60 wt%), Cu (<0.60 wt%) and with minor amounts of other alloying elements such as Cr, Ni, Mo, W, V and Zr (ASM, 1987). Low carbon and mild steels, which are the alloy types proposed for use as canister materials, have C contents of 0.05–0.15 wt% and 0.16–0.29 wt% respectively. Cast iron, which is proposed as a structural insert for a copper outer shell in the Swedish and Finnish programmes (SKB, 2006a), is an alloy of iron and C that contains more than 2 wt% C and 1–3 wt% silicon. Carbon or mild steels have been proposed as canister materials in Switzerland (Johnson and King, 2003, 2008), France (Fe´ron et al., 2009), Belgium (with a cementitious buffer) (Kursten and Druyts, 2008), Japan (JNC, 2000), Canada (King, 2007), Germany (Kursten et al., 2004), the UK for both HLW/SF (Marsh and Taylor, 1988) and ILW

© Woodhead Publishing Limited, 2010

392

Geological repository systems for safe disposal

(Smart et al., 2004), and, at one time, in the US Yucca Mountain Project (Brossia and Cragnolino, 2000). Carbon steel has a number of obvious benefits as a canister material, including: (1) good combination of strength and ductility, (2) generally good corrosion properties in the expected repository environments, (3) extensive experience with fabricating and sealing large cylindrical objects and (4) the relative abundance and cost of the material. The disadvantages of C-steel are the production of H2 and Fe(II) under anaerobic conditions, which may adversely impact other barriers, particularly HCB. Carbon steel is classified as a corrosion-allowance material and will corrode uniformly at a rate that depends on the redox conditions, pH, temperature and, to a lesser degree, the composition of the aqueous phase. Carbon steel is not suitable for use in permanently aerobic environments because of the possibility of localised corrosion and the formation of aggressive Fe(III) species that can impact the waste form or other barriers. Under anaerobic conditions, the corrosion rate decreases with time (Fig. 13.5) due to the build-up of a protective corrosion product film (Smart et al., 2001), eventually attaining a steady-state rate that is a function of pH and of whether the test is done in bulk solution or in the presence of compacted bentonite (Fig. 13.6) (Johnson and King, 2008). Steady-state corrosion rates of the order of a few μm/a are expected in bentonite-backfilled repositories. The corrosion rate of cast iron is similar to that of C-steel (Fig. 13.5). In cement or in alkaline solutions simulating cement pore water, the steadystate corrosion rate of C-steel is <0.1 μm/a (Smart et al., 2004). Localised corrosion of C-steel may occur as a result of (King, 2007):

13.5 Time dependence of the anaerobic corrosion rate of carbon steel and cast iron in a simulated ground water solution (Smart et al., 2001).

© Woodhead Publishing Limited, 2010

Nuclear waste canister materials

393

13.6 Differences between the anaerobic corrosion behaviour of carbon steel in bulk solution and in compacted bentonite (Johnson and King, 2008).

1. 2.

3.

Localised breakdown of a passive film leading to crevice corrosion or pitting. Non-uniform wetting of the surface due to deliquescence of surface contaminants, resulting in the spatial separation of anodic and cathodic sites. Reductive dissolution of Fe(III) corrosion products during the transition from aerobic to anaerobic conditions.

In bentonite-backfilled repositories, evidence from JNC (2000) suggests that C-steel is not passive in the moderately alkaline pore water. This is an important observation since it implies that localised corrosion of C-steel canisters due to the breakdown of a passive film should not occur. Carbon steel will be passive in cementitious environments and remain so as long as the cement pore water remains alkaline (Bel et al., 2006; Kursten and Druyts, 2008). As the cement degrades, however, and the pore water pH diminishes, the surface will become susceptible to localised corrosion at some stage, depending on the Cl concentration. In non-alkaline environ-

© Woodhead Publishing Limited, 2010

394

Geological repository systems for safe disposal

ments, any localised attack takes the form of surface roughening rather than discrete pitting (JNC, 2000). Environmentally assisted cracking could take the form of stress corrosion cracking (SCC) or various forms of hydrogen-related cracking, including hydrogen-induced cracking and blister formation (King, 2009b; Turnbull, 2009). The various forms of SCC are unlikely to affect C-steel canisters in a repository because of either the absence of the necessary chemical environment (in the cases of phosphate, nitrate and caustic cracking) or the absence of cyclic loading (in the cases of the near-neutral pH and highpH forms of cracking that are observed for pipeline steels in carbonate/ bicarbonate environments) (King, 2007). Hydrogen generated during anaerobic corrosion of C-steel will be absorbed by the canister material and may lead to various forms of hydrogen-related damage (King, 2009b; Turnbull, 2009). In general, the expected repository environments are benign compared with the acidic sour (H2S-containing) environments encountered in the oil and gas industry and in which H-related damage is observed. In repository environments, the lattice H concentration due to the cathodic reduction of H2O should be well below that expected to cause failure. In bentonite-backfilled repositories, H will also be absorbed from the gas phase which, depending upon the maximum H2 pressure (approximately equivalent to the sum of the hydrostatic and bentonite breakthrough pressures), could result in a somewhat higher lattice H concentration. Diffusion of H through the canister wall will result in an H2 pressure developing inside the canister that will eventually equal the external pressure. Therefore, the initiation of cracking is also possible from the inside of the canister, depending upon the distribution of tensile residual stress (Turnbull, 2009). The minimal threat of H damage posed by the benign environment can be reduced even further through the use of a low-strength material (JNC, 2000; Johnson and King, 2003). The susceptibility to H damage increases with the strength of the material, with only hydrogen-induced cracking and blister formation of concern for lower-strength materials (yield stress <500–700 MPa). As with other canister materials, the degree of microbiologically influenced corrosion (MIC) depends on where the microbial activity occurs (King, 2009a). If a biofilm forms on the canister surface then various forms of general and localised attack are possible (Little et al., 1991), particularly during the aerobic phase. If, however, microbial activity is restricted to locations away from the canister surface, either because of the presence of HCB or a cement-based grout or overpack, then the consequences should be minimal and limited to the possibility of enhanced H absorption if sulphide produced by sulphate-reducing bacteria reaches the canister surface.

© Woodhead Publishing Limited, 2010

Nuclear waste canister materials

395

13.4.2 Stainless steels Stainless steels are iron-based alloys characterised by a minimum chromium content of 11–13 wt% (Sedriks, 1996). Different families of stainless steel are classified based on their crystallographic structure, e.g. austenitic, ferritic, martensitic and duplex alloys, the latter containing approximately equal fractions of austenite and ferrite (ASM, 1987, 2003, 2005). Stainless steels exhibit good resistance to general corrosion but can be susceptible to various forms of localised corrosion and SCC. Localised corrosion is often a precursor to SCC and can be controlled to some degree through the addition of Mo, W and N. Low carbon grades reduce the susceptibility to SCC and intergranular attack. Stainless steels are classified as corrosion-resistant materials. Because of their susceptibility to localised attack (pitting and crevice corrosion) and, especially for the austenitic grades, SCC in Cl environments at modest temperature (Sedriks, 1996), stainless steels are not suitable candidates for SF/HLW canisters for which containment is crucial, unless the environment has a low Cl content (less than a few ppm) and/or there are significant concentrations of inhibiting species, such as nitrate, carbonate or sulphate. Austenitic and duplex stainless steels (Table 13.1) are under consideration as canister materials for LILW with or without a cementitious backfill, for which the provision of absolute containment is less of a concern. Table 13.1

Compositions of common austenitic and duplex stainless steels Composition (wt%)*

UNS Common number name

Other

Cr

Ni

C

Mn Si

P

S

Austenitic alloys S30400 304 S30403 304L S30900 309 S31600 316 S31603 316L S31703 317L S32100 321 S34700 347

18–20 18–20 22–24 16–18 16–18 18–20 17–19 17–19

8–10 8–12 12–15 10–14 10–14 11–15 9–12 9–13

0.08 0.03 0.20 0.08 0.03 0.03 0.08 0.08

2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0

0.045 0.045 0.045 0.045 0.045 0.045 0.045 0.045

0.030 0.030 0.030 0.030 0.030 0.030 0.030 0.030

Duplex alloys S32304 SAF 2304

21.5–24.5 3–5.5

S31803 S32404 * †

1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0

Mo 2–3 Mo 2–3 Mo 3–4 Ti 56C2 (Nb + Ta) 106C2

0.03 2.5 1.0 0.040 0.040 N 0.05–0.2, Mo 0.05–0.6 2205 21–23 4.5–6.5 0.03 2.0 1.0 0.030 0.020 N 0.08–0.2, Mo 2.5–3.5 Uranus 50 20.5–22.5 5.5–8.5 0.04 2.0 1.0 0.030 0.010 N 0.2, Mo 2–3, Cu 1–2

Maximum unless otherwise indicated, balance Fe. Minimum.

© Woodhead Publishing Limited, 2010

396

Geological repository systems for safe disposal

Table 13.2 General corrosion rates for austenitic stainless steels as a function of pH, temperature, chloride concentration and redox conditions Grade pH

Temperature [Cl] (oC) (μg/g)

304

30, 45

304

12.8 10.5 13.3

304

13

304

10 12.5 13.5 12.8

30 50 80 50

304

304L 304L

Ambient

30

316

Ambient 25–100 Ambient 27 90 Ambient 25 50 75 13.3 Ambient

316L

> 13

316L

Ambient 30 50–100 Ambient 27 Ambient 25 50 75

304

316L 316

Ambient

18,400

Redox Other conditions

Aerated

200 days 60 days 28 days

Deaerated Deaerated 230 days Deaerated Calculated 1 year 100 years 104 years ‘Freshwater’ Aerated ‘Saltwater’ Aerated

Rate (μm/a)

Reference

0.0003 0.01 0.3

Fujisawa et al., 1999 Mcdonald et al., 1995 Blackwood et al., 2002

0.06 0.18 0.82 0.009 0.0055 0.0063 0.055 0.002 0.00006 0.21 11.4 5.82 0.2–0.96 0.22–0.23 0.3–0.35 0.6

Interstitial clay water

Aerated

18 400

Aerated

10 000

Deaerated 0.1 MPa H2 0.03

28 days

‘Freshwater’ Aerated ‘Saltwater’ Interstitial clay water

Aerated Aerated

0.01 0.25 1.94 0.1–0.24 0.1–0.34 0.1–0.17

Wada and Nishimura, 1999 Fukaya and Akashi, 2003 BSC, 2004 BSC, 2004 Casteels et al., 1986 Mcdonald et al., 1995 Smart et al., 2004 BSC, 2004 BSC, 2004 Casteels et al., 1986

Being passive materials, stainless steels exhibit low rates of general corrosion, especially in alkaline conditions (Table 13.2). Rates vary from a few μm/a in aerated solution at neutral pH to a few nm/a under deaerated conditions at alkaline pH. As with C-steel, the corrosion rate decreases with time due to thickening of and the elimination of defects from the passive film. Canisters may also be exposed to atmospheric conditions, but here the database of general corrosion rates is smaller (Table 13.3). Austenitic and duplex stainless steels are susceptible to localised corrosion in chloride environments in the form of pitting and crevice corrosion. The susceptibility to localised attack can be assessed in a number of ways (Szklarska-Smialowska, 2005): . .

based on the pitting resistance equivalent number (PRE or PREN), which accounts for the effect of alloy composition on localised corrosion susceptibility; based on the critical pitting (CPT) or crevice (CCT) temperature, an

© Woodhead Publishing Limited, 2010

Nuclear waste canister materials Table 13.3

397

Rates of atmospheric corrosion of austenitic stainless steels

Grade

Temperature Redox Other (oC) conditions

Rate Reference (μm/a)

304

Ambient

<0.03 Johnson and Pavlik, 1982 0.022 0.05–2 0.01

304

Ambient

Urban, 5–15 years Urban, 5–15 years Marine, 5–15 years Industrial/urban, 5–15 years Industrial/urban

316

Ambient

Aerated

Urban, 5–15 years

<0.03

Johnson and Pavlik, 1982

Stainless Ambient steel

Aerated

Various atmospheres

0.05

Dechema, 1990

Aerated

0.03–3 Kearns et al., 1984

13.7 Comparison of the critical crevice and pitting temperatures for a range of austenitic and duplex stainless steels in 10 wt% ferric chloride solution.

.

experimentally determined threshold temperature dependent on the composition of the test solution; and comparison of the breakdown or repassivation potential for pitting or crevice corrosion and the corrosion potential ECORR.

The PREN method is useful for comparing the effect of composition, particularly the beneficial effects of increasing Cr, Mo and N content of the alloy. The general form of the expression is PREN ¼ %Cr þ a %Mo þ b %N

© Woodhead Publishing Limited, 2010

398

Geological repository systems for safe disposal

where the values of a and b vary slightly between different authors. Figure 13.7 shows the variation of CPT and CCT for various austenitic and duplex alloys (IMOA, 2001), from which it is apparent that the initiation and/or propagation of localised corrosion in an occluded region (crevice) occurs at a lower temperature than pitting of the exposed surface. Finally, the criterion for localised corrosion based on electrochemical measurements of susceptibility is ECORR > ECRIT

½13:2

where ECRIT is the critical potential for pitting or crevice corrosion defined as the film breakdown potential or the potential at which a propagating pit or crevice repassivates. Figure 13.8 illustrates this concept for AISI Type 316L stainless steel at 95 oC for a range of Cl concentrations. In this case, the critical potential is based on the pit breakdown (Ep) or repassivation (Erp) potential (Dunn et al., 1996). Pitting would not be predicted in either aerated or deaerated solutions based on the filmbreakdown criterion, regardless of the Cl concentration. However, if the repassivation potential is used as the criterion for pitting, localised corrosion is predicted to occur in aerated solution (since ECORR > Erp) but not under deaerated conditions (since ECORR <Erp). This example illustrates an important feature of the localised corrosion of stainless steels, namely that

13.8 Comparison of the pitting and pit repassivation potentials for Type 316L stainless steel to the corrosion potential ECORR in aerated and deaerated chloride solutions at 95 oC (Dunn et al., 1996).

© Woodhead Publishing Limited, 2010

Nuclear waste canister materials

399

attack is only possible in O2-containing (or acidic) conditions. Thus, if the initiation of localised corrosion can be prevented during the aerobic phase, for instance through the use of cement backfill, then potentially long canister lifetimes are possible. Austenitic stainless steels are also susceptible to SCC in chloride environments (Sedriks, 1996). Cracks typically initiate from pits so cracking can, in theory, be avoided by selecting an alloy that is resistant to localised corrosion. Alternatively, susceptibility can be reduced by using a cement backfill or overpack or by using a duplex alloy. As in the case of C-steels discussed above, the susceptibility of stainless steel canisters to MIC depends on the location of microbial activity. If microbial activity is suppressed at the canister surface due to the use of a cement or highly compacted bentonite backfill then MIC may not adversely affect the canister. However, if a biofilm forms, stainless steel is susceptible to MIC (Little et al., 1991), with welds being particularly sensitive.

13.4.3 Copper and copper alloys Oxygen-free copper was first proposed as a candidate canister material for the disposal of SF in Sweden in 1978 (Swedish Corrosion Institute, 1978). Since then, copper has been selected as either the reference or as an alternative canister material in Finland (Pastina and Hella¨, 2006), Canada (Maak, 1999), Switzerland (Johnson and King, 2003) and Japan (JNC, 2000). In each of these countries the repository would be located several hundred metres below the water table in stable host rock and sealed using bentonite-based backfill. Under these conditions, copper has been found to provide very-long-lived containment with predicted canister lifetimes in excess of 100 000 a (King et al., 2001, 2002). Two copper alloys (70–30 Cu– Ni, UNS 71500 and an Al bronze, UNS 61300) were also briefly considered in the YM programme (Sridhar and Cragnolino, 1993), but copper alloys are not suitable as canister materials in a permanently aerobic repository. The great advantage of the use of copper as a canister material in saturated repositories is its thermodynamic stability in O2-free environments (in the absence of sulphide). Figure 13.9 shows the potential (E)–pH (Pourbaix) diagram for the Cu–H2O–Cl system at 25 oC (King et al., 2001, 2002). The range of ECORR values expected in natural environments and in bentonite pore water solutions is shown by the shaded box. Thus, in Cldominated environments copper is thermodynamically stable in the absence of O2 and corrosion will, theoretically, stop once the initially trapped O2 has been consumed. In the presence of sulphide, ECORR drops below the H2/ H2O equilibrium line (line (a) in Fig. 13.9) and H2O becomes an oxidant for copper. It has also been suggested (Hultquist, 1986; Szaka´los et al., 2007) that H2O is an oxidant for copper in pure water and that H2 is evolved as a

© Woodhead Publishing Limited, 2010

400

Geological repository systems for safe disposal

13.9 Potential-pH diagram for copper for the Cu–H2O–Cl system at 25 oC with 1 mol/dm3 Cl. The shaded box represents the range of corrosion potential values observed in laboratory experiments in O2-containing solutions (King et al., 2001, 2002).

consequence. These experimental observations, which apparently contradict conventional thermodynamic thinking, have not been reproduced by other researchers and the scientific basis for the claim is unclear. Even if the proposed mechanism is correct, however, the apparent equilibrium partial pressure of H2 of ~100 Pa (0.001 atmospheres) is so small that even if corrosion occurred in the repository it would stop since the bentonite buffer will retard the transport of H2 away from the interfacial region. The general corrosion behaviour of copper under repository conditions has been investigated in some detail (King et al., 2001, 2002). Figure 13.10 shows the proposed mechanism for the general corrosion of copper in O2containing Cl solution in the presence of compacted bentonite. The mechanism comprises various electrochemical, precipitation/dissolution, adsorption/desorption, redox, microbial and mass-transport processes, each of which has been studied and the respective rate constants or diffusion coefficients determined. This mechanism has been verified by comparison of the results from both a comprehensive one-dimensional transient reactivetransport model and a steady-state mixed-potential model to experimental data, as well as by evidence from archaeological artefacts. Since corrosion will cease once all the initially trapped O2 (and the Cu(II) that is produced by the homogeneous oxidation of Cu(I) by O2) has been consumed, the

© Woodhead Publishing Limited, 2010

Nuclear waste canister materials

401

13.10 The reaction mechanism for the general corrosion of copper in O2-containing chloride solutions in compacted bentonite (King et al., 2001, 2002).

corrosion rate of copper is less important than the total inventory of the oxidant. In the case of general corrosion in the presence of sulphide, the rate of corrosion is determined by the rate of supply of HS (Smith et al., 2007), which can be estimated based on the known transport properties of the HCB (SKB, 2006a, 2006b). Copper canisters are not expected to be susceptible to localised corrosion in the repository environment (King et al., 2001, 2002). When exposed to simulated repository conditions including compacted bentonite, the surfaces of copper coupons develop a roughened appearance (Fig. 13.11). This nonuniform corrosion is not pitting in the classical sense of the term because all areas of the surface have been attacked. A better description is underdeposit attack, since the deepest corrosion was found under precipitated mineral phases on the surface. Although the use of copper with a cement backfill or overpack has not been proposed, the canister would also be immune to pitting under these conditions, despite passivation of the surface by a Cu2O or duplex Cu2O/Cu (OH)2 layer. Figure 13.12 shows a comparison of the pitting potential and ECORR in aerated 0.5 mol/dm3 NaCl as a function of pH (King, 2002). Based on the criterion in equation [13.2], pitting would not occur since the value of ECORR is always less than the critical potential. In alkaline cement pore water, the difference between the pitting potential and ECORR increases, indicating that localised corrosion becomes less likely with increasing pH. Copper is susceptible to SCC in certain well-known environments, including ammonia, acetate and nitrite solutions (King et al., 2001, 2002).

© Woodhead Publishing Limited, 2010

402

Geological repository systems for safe disposal

13.11 The roughened surface observed after exposure of copper to simulated repository environments (King et al., 2001, 2002).

Cracking in these environments occurs at potentials close to the Cu2O/Cu (OH)2 equilibrium potential (King and Newman, 2009), suggesting that dissolution plays a role in the cracking mechanism and that, therefore, SCC is only possible during the early aerobic phase. However, it can be shown that the necessary combination of potential, interfacial pH and a sufficient concentration of the SCC agent will not occur simultaneously so that cracking of the canister will not occur in the repository environment (King and Kolar, 2005). Recently it has been suggested that cracking can also occur in concentrated (> 103 mol/dm3) sulphide solutions (Taniguchi and Kawasaki, 2008) and this has raised the more general question of whether SCC can occur under anaerobic conditions. Certainly the sulphide concentration apparently required to cause SCC will not occur at the canister surface and there seems to be no other mechanistic reason why cracking should occur under anaerobic conditions (King and Newman, 2009). As noted above, the use of compacted bentonite prevents significant microbial activity close to the canister surface. The effect of the HCB appears to be to reduce the water activity sufficiently that microbes are not active, since the same effect can be produced by using saline solutions (which

© Woodhead Publishing Limited, 2010

Nuclear waste canister materials

403

13.12 Comparison of the corrosion potential of copper in aerated solution and the pitting potential as a function of pH in 0.5 mol/dm3 NaCl at 25 oC (King, 2002).

produces an osmotic potential) or compacted bentonite (which produces a suction potential) (Stroes-Gascoyne et al., 2007). There are suggestions that a very low level of microbial activity is possible in HCB under some circumstances, but the rate is so low that the depth of attack would be minimal even over repository timescales (Masurat et al., 2009). Remote microbial activity may produce sulphide, which will contribute to the flux of HS to the canister surface and the transport-limited general corrosion of the canister (SKB, 2006a, 2006b).

13.4.4 Titanium alloys Titanium alloys can be divided into single-phase α (and near-α) alloys and two-phase α–β alloys (Fig. 13.13). The introduction of the β-phase increases the strength of the alloy (important for aerospace applications), but can increase the susceptibility to HIC (Shoesmith, 2006). The other important alloying distinction is the amount of Pd or Ru, since these elements significantly improve the crevice-corrosion resistance of Ti alloys. Commercially pure (CP) Ti has been considered as a canister material in Canada, as well as the more crevice-corrosion-resistant alloys Grade 12 and Grade 16 (Johnson et al., 1994). The Pd-containing alloy Grade 7 has been considered for use in salt host rock (Kursten et al., 2004), in a bentonitefilled repository in granite (Mattsson and Olefjord, 1984, 1990), in the Boom Clay host rock in Belgium (Kursten et al., 2004) and as the drip shield material in the unsaturated Yucca Mountain repository (DOE, 2008).

© Woodhead Publishing Limited, 2010

404

Geological repository systems for safe disposal

13.13

Relationship between various grades of titanium alloys.

Commercially pure Ti was also considered in Sweden (Mattsson and Olefjord, 1984, 1990) and in Belgium (Kursten et al., 2004), the latter country also considering Grade 5 (Ti–6Al–4V) and a Ti–5Al–2Sn alloy. In the Japanese programme, JNC (2000) have considered the use of Grades 1, 2 and 12. The important corrosion modes for Ti are general corrosion, not because the rate is high but because it is a source of hydrogen, crevice corrosion and hydrogen-induced cracking (Hua et al., 2005; Shoesmith, 2006). Pitting will not occur because the film breakdown potentials are several V. Alone among the common structural materials, Ti alloys are believed to be immune to MIC, possibly because within the stability field of water it is present almost exclusively as the Ti(IV) species (and, hence, cannot participate in the redox reactions that are a key component of microbial processes). The corrosion resistance of Ti alloys is due to the formation of a stable TiO2 passive film (Fig. 13.14). This film is thermodynamically stable over the entire range of redox potentials and pH possible in the repository, with the exception of increased solubility in alkaline solutions (> pH 12) and in acidic crevices (< pH 2) (Pourbaix, 1974). The mean corrosion rate of Ti Grade 7 in aerated aqueous solutions is 20 nm/year, with the rate reported to be independent of temperature (within the range 60–90 oC) and salinity (up to several mol/l) (Hua et al., 2005). Mattsson and Olefjord (1984, 1990) and Mattsson et al., (1990) reported a mean corrosion rate of 1 nm/a after six years of exposure to a bentonite clay environment. Figure 13.15 shows how the film structure evolves with potential and temperature, along with the associated corrosion processes (Shoesmith, 2006).

© Woodhead Publishing Limited, 2010

Nuclear waste canister materials

405

13.14 Potential–pH diagram for the Ti–H2O system at 25 oC (Pourbaix, 1974).

Commercially pure grades of Ti alloys are susceptible to crevice corrosion (Shoesmith, 2006). Crevice corrosion occurs through the classical mechanism namely: (1) the depletion of O2 within the occluded region, (2) the initiation of localised film breakdown by Cl ions or other aggressive species, (3) separation of the anodic and cathodic processes, with the anodic dissolution of Ti as Ti(III) or Ti(IV) occurring within the crevice supported by the reduction of O2 (or of other oxidants) on external surfaces, (4) the hydrolysis of dissolved Ti(IV) within the crevice leading to local acidification and further destabilising the passive film and (5) propagation of the crevice supported by the reduction of H+ inside the crevice and the external reduction of O2 (or other oxidant). Although arguments can be made that a propagating crevice will stifle under repository conditions (Shoesmith, 2006), a better strategy when

© Woodhead Publishing Limited, 2010

406

Geological repository systems for safe disposal

13.15 Schematic diagram showing the properties of the passive film on titanium as a function of potential (Shoesmith, 2006).

localised corrosion is of concern is to use a crevice-corrosion-resistant grade, such as the Pd-containing Grades 7 and 16 or the Ru-containing Grade 29 (Fig. 13.13). Prevention of crevice corrosion has a second benefit of preventing rapid H pick-up due to proton reduction in acidic crevices. Titanium forms hydrides with hydrogen and alloys are susceptible to hydrogen-induced cracking in a manner similar to zirconium alloys (Hua et al., 2005; Shoesmith, 2006). The solubility of hydrogen in α-phase Ti is on the order of 20–150 wppm (Hua et al., 2005), barely above the level often found in as-received material. The rate of hydrogen absorption under passive conditions depends on a number of factors, including: (1) potential, (2) temperature, (3) redox transformations within the oxide and (4) the presence of intermetallics. The passive film acts as a significant barrier to H absorption by the metal (Shoesmith, 2006). Below a threshold potential of about 0.65 VSCE hydrogen is absorbed by the oxide, and enters the underlying metal at potentials below about 1 VSCE (Fig. 13.15). Temperature increases the rate of H diffusion in α-Ti and promotes hydride formation, with a temperature of 80 oC considered as a threshold for the onset of H effects. Intermetallics, such as Ti2Ni in Ti Grade 12 or Pd particles in the Pd-containing alloys, not only act as cathodes for the reduction of H+ but can also act as preferential sites for H absorption. Hydrogen absorption efficiencies (i.e. the fraction of H generated that is absorbed by the material) of a few % have been reported for passive films on Ti (Hua et al., 2005). Hydride phases are more brittle than the metal substrate and lead to a loss in ductility and the possibility of fast fracture. Shoesmith and co-workers

© Woodhead Publishing Limited, 2010

Nuclear waste canister materials Table 13.4 Alloy

407

Composition of selected nickel alloys*

Alloy class

Ni

Cr

Fe

Mo

56.0 Balance

22.0 14.0– 18.0 14.5– 16.5

3.0 3.0 4.0– 7.0

13.0 0.015 0.08 — 14.0– 0.015 0.08 1.0 17.0 15.0– 0.02 0.08 1.0 17.0

58.0 min 20.0– 5.0 23.0 Alloy 825 Ni–Fe–Cr–Mo 36.0–46.0 19.5– 24– 23.5 40

8.0– 0.10 0.5 0.5 10.0 2.5– 0.05 0.5 1.0 3.5

Alloy 22 Ni–Cr–Mo Hastelloy Ni–Cr–Mo C-4 Hastelloy Ni–Cr–Mo C-276

Balance

Alloy 625 Ni–Cr–Mo

C

Si

Mn

Ti

Other

— 3.0 W 0.70 2.0 Co, 0.03 S, 0.04 P 3.0–4.5 W, 2.5 Co, 0.35 V, 0.03 S, 0.03 P 0.4 0.4 Al, 0.015 S, 3.15–4.15 Nb 0.6– 1.5–3.0 Cu, 0.2 1.2 Al, 0.03 S

* wt% maximum, unless stated otherwise.

(Hua et al., 2005) model hydrogen-induced cracking in terms of a thresholdabsorbed hydrogen content (HC), above which fast fracture may occur. The value of HC is alloy-dependent, with values of 500–800 wppm for Ti Grade 2, 400–600 wppm for Grade 12 and 1000–2000 wppm for Grade 16. The time to failure can then be predicted based on a knowledge of the rate of H absorption (i.e. the general corrosion rate multiplied by the H absorption efficiency). Nakayama and co-workers (2008) base predictions on the thickness and mechanical properties of the hydrided layer, with failure occurring at a critical hydride layer thickness.

13.4.5 Nickel-based alloys There is a wide range of Ni-based alloys with an equally wide range of corrosion properties (Table 13.4). Nickel alloys are generally more resistant to SCC in Cl environments than austenitic stainless steels. Some alloys are designed for use in reducing acids, others in oxidising acids and others in alkaline media. The breadth of corrosion properties is a result of the high solubility of other metals in Ni, resulting in single-phase alloys and a wide choice of possible alloying elements. Nickel alloys have been considered as candidate canister materials in Canada (Hastelloy C-276 and Inconel 625; Johnson et al., 1994), Germany (salt brines and Hastelloy C-4; Kursten et al., 2004), Belgium (Boom Clay, Hastelloy C-4, Inconel 625 and Hastelloy C-22; Kursten et al., 2004) and in the US YM project (Alloys 625, 825 and 22; Brossia et al., 2001; DOE, 2008). These alloys are classified as either Ni–Cr–Mo or Ni–Fe–Cr–Mo alloys (Table 13.4). The Ni–Cr–Mo and Ni–Fe–Cr–Mo alloys are protected from corrosion by a Cr-based passive film. Both Cr(OH)3 and Cr2O3 exhibit similar ranges of stability in water, spanning from acidic to highly alkaline pH and virtually the entire potential range of the stability of H2O (Pourbaix, 1974). However, transpassive dissolution as Cr(VI) occurs at potentials more negative than those of some other passive materials, resulting in a tendency

© Woodhead Publishing Limited, 2010

408

Geological repository systems for safe disposal

towards localised corrosion. The general corrosion rate of Ni–Cr–Mo alloys in aerobic and anaerobic environments is very low. The mean corrosion rate of Alloy 22 in a range of simulated concentrated YM pore waters at a temperature of 60 oC is 5–10 nm/year (DOE, 2008), based on 5- and 10-year exposure tests. For the more aggressive ‘Q-brines’ studied in the German programme, 3-year mean corrosion rates from 200 nm/year at 90 oC to 900 nm/year at 200 oC are reported (Kursten et al., 2004). The candidate Ni-based alloys are susceptible to both crevice corrosion and pitting, but because crevice corrosion in occluded regions occurs under less-severe conditions than the pitting of exposed services, most attention has been focused on the former form of localised corrosion (Brossia et al., 2001; DOE, 2008). Figure 13.16 shows the temperature dependence of the crevice repassivation potential for Alloy 22 for different chloride concentrations (Cragnolino et al., 1999). The repassivation potential shifts to more active values with increasing temperature indicating that localised corrosion is of most concern during the initial thermal transient (and then only if O2 is present). Different alloys exhibit different susceptibilities because of the differing amounts of Cr, Mo and W (Fig. 13.17) (Brossia et al., 2001). Alloy 22 is the most crevice-corrosion-resistant alloy of those considered for SF/HLW canisters. The criterion in equation [13.2] is typically used to assess the susceptibility to crevice corrosion. In the event that localised corrosion initiates, Ni-based alloys show a tendency for propagating pits or crevices to repassivate or ‘stifle’ (Mon et al., 2005). The stifling mechanism is uncertain, but may result from:

13.16 Temperature dependence of the crevice repassivation potential of Alloy 22 in various chloride solutions (Cragnolino et al., 1999).

© Woodhead Publishing Limited, 2010

Nuclear waste canister materials

409

13.17 Chloride concentration dependence of the crevice repassivation potential for various Ni-based alloys (Brossia et al., 2001).

. . . . .

iR (potential) drop down the crevice; mass-transport effects; loss of critical crevice chemistry by catalysis of H+ reduction; loss of critical crevice chemistry by the reduction of an inhibiting anion, e.g. NO3 in NO3:Cl mixtures leading to an increase in pH in the crevice; negative shift in ECORR upon the initiation of localized corrosion.

Nickel-based alloys are more resistant to SCC than austenitic stainless steels (Sedriks, 1996). Nickel alloys are generally immune to cracking in Cl-based ground waters, but can be susceptible in the presence of sulphur species and lead. Nickel alloys are not immune to MIC but their susceptibility appears to be limited (Lloyd et al., 2004). Even in the presence of large quantities of nutrients, Farmer et al., (2000) only reported a doubling of the general corrosion rate. This lack of susceptibility has been attributed to the wide range of stability of the passive film (Lloyd et al., 2004). Again, the greatest concern would be if a biofilm forms on the canister surface. Nickel alloys do exhibit some susceptibility to sulphide, elemental sulphur and oxysulphur anions (ASM, 2005). Adsorption of S on Ni inhibits passivation because of the formation of an Ni3S2 layer (Marcus, 1995). A critical S surface coverage of a 0.7–0.8 monolayer is required to suppress passivation. This surface coverage can arise from the segregation of S impurities in the alloy or from S species in the environment. Nickel alloys do exhibit some sensitivity to gamma radiation at high absorbed dose rates (Kursten et al., 2004). An increasing rate of general corrosion and a susceptibility to pitting were observed for dose rates of

© Woodhead Publishing Limited, 2010

410

Geological repository systems for safe disposal

between 10 and 1000 Gy/h in the aggressive ‘Q-brines’ in the German programme.

13.4.6 Ceramics There has been little development of ceramics as a potential canister material, despite the obvious advantages of their stability in a range of aqueous environments coupled with the lack of H2 generation. However, there are concerns about the fracture toughness of ceramics and the sealing of vessels is problematic.

13.5

Long-term performance of nuclear waste canisters

13.5.1 Comparison of candidate materials The materials discussed in the previous section offer different advantages and disadvantages which are summarised in Table 13.5.

13.5.2 Lifetime predictions As with many other aspects of the repository system, it is necessary to make long-term predictions of the corrosion behaviour of the canister material and, in particular, to predict the canister lifetime. From an overall system safety perspective, the two important factors are the time of failure and the spread in failure times. Various approaches have been used to predict the impact of various corrosion processes on canister lifetimes, including: . . . . . . . .

Simple linear extrapolation of an empirically determined general corrosion rate. Mass balance arguments based on the amount of available oxidant (especially for materials, such as copper, that are thermodynamically stable in the absence of O2). Mass-transport calculations, for example, in the case of the long-term general corrosion of copper due to sulphide. Comparison of ECORR to the repassivation potential for the crevice corrosion of Alloy 22. Use of a slip-dissolution model for predicting SCC of Alloy 22. Reasoned arguments based on non-susceptibility to exclude various corrosion processes for a variety of materials. Use of the point defect model for predicting the long-term behaviour of passive materials. Time-dependent crevice propagation to account for the stifling of localised corrosion of Alloy 22.

© Woodhead Publishing Limited, 2010

© Woodhead Publishing Limited, 2010

Robustness of lifetime prediction

Flexibility of repository design

Impact on other barriers

Overall corrosion performance

Canister design and fabrication

Copper

Excellent. Canister lifetimes on the order of 100 000 a are possible in the appropriate environment Good–very good. Requires internal structural support. Need to weld and inspect thick wall sections Adequate. Suitable Very good. Corrodes Excellent corrosion primarily in cementitious behaviour, especially in actively in bentonitebackfill. Concerns with based backfill or under Cl dominated SCC and localised passive conditions in environments that corrosion in Cl cementitious backfill promote active dissolution environments Adequate. Potential Excellent. Minimal Very good. Minimal impacts of Fe(II) on impact on other barriers impact on other barriers, bentonite, H2 gas on although will require cast iron or steel insert, which bentonite and/or lowmay have adverse effects permeability host rock Adequate. Copper should Excellent. C-steel can be Adequate. Best performance with cement not be used without a used with bentonite or backfill to minimise the cementitious backfills or backfill or overpack rate of mass transport. with no backfill There is no advantage to using a cementitious backfill for copper Excellent. Simple lifetime Excellent. Simple lifetime Uncertain. Less experience with lifetime predictions based on predictions based on mass-balance and massprediction than with mass-balance transport calculations. other alloy systems calculations and Support from natural and empirical data. Support archaeological artefacts from natural and archaeological artefacts

Stainless steels Unknown. Generally specified as an LILW canister material with less emphasis on lifetime prediction Excellent. Actual experience in manufacturing and sealing SS LILW canisters

Carbon steel

Good–very good. Long canister lifetimes of the order of tens of thousands of years are possible Very good. Single-shell designs possible. Need to weld and inspect thick wall sections

Predicted canister lifetime

Advantages and disadvantages of the various candidate canister materials

Parameter

Table 13.5

Excellent. Ni alloys can be used with bentonite or cementitious backfills or with no backfill

Good–very good. Need to predict long-term behaviour of passive film. Need to predict localised corrosion behaviour for crevice-corrosionsusceptible alloys

Very good. Ti alloys are suitable for use with bentonite backfill or with no backfill

Good–very good. Need to predict long-term behaviour of passive film and HIC. Need to predict localised corrosion behaviour for crevicecorrosion-susceptible alloys

Excellent. Minimal impact on other barriers

Very good–excellent. Corrosion resistance can be tailored to environment through appropriate alloy selection Excellent. Minimal impact on other barriers Excellent. Crevicecorrosion-resistant alloys are subject only to general corrosion and HIC

Nickel alloys Very good–excellent. Potential for very longlived containment with appropriate alloy selection Very good. Requires internal support. Thin wall simplifies joining and inspection

Titanium alloys Very good–excellent. Potential for very longlived containment with crevice-corrosionresistant grades Very good. Requires internal support. Thin wall simplifies joining and inspection

412 . . . . . . . . . . .

Geological repository systems for safe disposal Mass-balance arguments to estimate the maximum amount of MIC due to sulphate-reducing bacteria. Detailed mechanistically based reactive-transport modelling. Assessment of the susceptibility to localised corrosion by comparison of ECORR to a critical potential. Use of a pitting factor for ‘pitting’ of copper or C-steel. Use of extreme-value analysis to predict the maximum pit depth. Comparison of the rate of H uptake to the critical H concentration for the hydrogen-induced cracking of Ti alloys. Use of empirically determined damage functions for the crevice corrosion of Ti alloys. Comparison of the time dependence of ECORR and the interfacial pH to determine the SCC susceptibility of copper. Use of an enhancement factor for the rate of general corrosion to account for microbial effects. Reactive-transport modelling of microbial processes to predict the consequences of microbial activity remote from the canister surface. The use of decision trees and reasoned arguments to assess the possibility of different corrosion processes.

The predicted canister lifetimes are a function of the material of construction and of the nature of the environment. The general ranges of expected lifetimes are given in Table 13.5 for the various classes of material. Further details of modelling approaches and predicted lifetimes can be found in publications from each of the nuclear waste management programmes, some of which are listed in Section 13.7.

13.6

Future trends

It is always difficult, of course, and dangerous to predict future trends. Nevertheless, here is an attempt to predict the future or, perhaps, to capture the authors’ hopes for the future. There is clearly scope for optimisation of various canister designs. There has already been a trend in this direction with, for example, the reference thickness of the copper canister in the Swedish programme being reduced, first from the original thickness of 20 cm to a thickness of 10 cm and now to the current reference thickness of 5 cm. Even this thickness is more than is required as a corrosion barrier. It is impractical to fabricate a separate copper shell of this size with a significantly smaller wall thickness, but perhaps the copper corrosion barrier could be applied as a thinner cladding instead. Even if there is a reluctance to reduce the canister wall thickness, it is likely that there will be a trend towards more-realistic (i.e. less-conservative)

© Woodhead Publishing Limited, 2010

Nuclear waste canister materials

413

predictions as the level of mechanistic understanding, and confidence in making such long-term predictions, grows. An example of this trend can be seen with the predicted lifetimes of C-steel canisters. Initially, C-steel was seen as a ‘1000-year option’. More recently, however, as more measurements of the steady-state anaerobic corrosion rate have become available and as the understanding of the limited degree of localised attack has developed, lifetimes of the order of tens of thousands of years have been predicted for C-steel canisters. Finally, we hope that there will be greater acceptance of the use of passive alloys as candidate canister materials. The thinner-walled canister designs that the use of a passive material permits offer certain advantages with respect to ease of fabrication and inspection and resource availability. However, there has been a reluctance, especially from regulators and external review groups, to accept that it is possible to predict the long-term behaviour of passive materials, especially if there is the possibility of localised attack. We consider this an unnecessarily cautious approach and hope that more programmes will look at the potential advantages of using corrosion-resistant canister materials.

13.7

Sources of further information and advice

By necessity, we have only been able to provide a brief discussion of the long-term performance of canister materials here. The following is a list of more-detailed presentations for each of the classes of material considered. Carbon steel and cast iron Akashi et al. (1990), Andra (2001), Asano et al. (1992), Bel et al. (2006), Blackwood et al. (2002), Casteels et al. (1985), Carlson et al. (2006), Crossland (2005), DeBruyn et al. (1991), Fe´ron et al. (2009), JNC (2000), Johnson and King (2003, 2008), Johnson et al. (1994), King (2007, 2008), Kursten and Druyts (2008), Kursten et al. (2004), Marsh and Taylor (1988), Neff et al. (2006), ONDRAF-NIRAS (2004), Shoesmith (2006), Smart et al. (2001, 2004) Stainless steels Blackwood et al. (2002), Johnson et al. (1994), Sedriks (1996), Smart (2000, 2002, 2005), Smart and Wood (2004), Smart et al. (2004, 2006)

© Woodhead Publishing Limited, 2010

414

Geological repository systems for safe disposal

Copper and copper alloys Crossland (2005), JNC (2000), Johnson and King (2003), Johnson et al. (1994), King (2006), King and Kolar (2000), King and LeNeveu (1992), King et al. (2001, 2002, 2008b), Kursten et al. (2004), Shoesmith (2006), SKB (1983, 2006b), Swedish Corrosion Institute (1978) Titanium alloys DOE (2008), Hua et al. (2005), JNC (2000), Johnson et al. (1994), Kursten et al. (2004), Shoesmith (2006) Nickel alloys Brossia et al. (2001), Cragnolino et al. (1999), DOE (2008), Farmer et al. (2000), Johnson et al. (1994), King et al. (2008a), Kursten et al. (2004), Mon et al. (2005), Rebak et al. (2000), Shoesmith (2006)

13.8

References

Akashi M, Fukuda T and Yoneyama H (1990), ‘A corrosion localization assessment of the mild steel used for nuclear waste package’, in Scientific Basis for Nuclear Waste Management XIII, edited by V M Oversby and P W Brown, Materials Research Society Symposium Proceedings 176, Materials Research Society, Pittsburgh, Pennsylvania, pp. 525–532. Andra (2001), ‘Materials baseline, vol. 4, Corrosion of metallic materials’, Andra Report C.RP.AMAT.01.060. Asano H, Wakamatsu H and Akashi, M (1992), ‘Corrosion lifetime assessment for candidate materials of geological disposal overpack for high-level nuclear waste canisters – perspective of R&D in Japan’, in Proceedings of the High Level Radioactive Waste Management Conference, Las Vegas, Nevada, 12–16 April 1992, American Nuclear Society, La Grange Park, Illinois, and American Society of Civil Engineers, New York, pp. 1658–1669. ASM (1987), Metals Handbook, 9th edition, vol. 13, Corrosion, American Society for Metals International, Metals Park, Ohio. ASM (2003), ASM Handbook, vol. 13A, Corrosion: Fundamentals, Testing, and Protection, American Society for Metals International, Metals Park, Ohio. ASM (2005), ASM Handbook, vol. 13B, Corrosion: Materials, American Society for Metals International, Metals Park, Ohio. Bel J J P, Wickham S M and Gens R M F (2006), ‘Development of the Supercontainer design for deep geological disposal of high-level heat emitting radioactive waste in Belgium’, in Scientific Basis for Nuclear Waste Management XXIX, edited by P Van Iseghem, Materials Research Society Symposium Proceedings 932, Materials Research Society, Warrendale, Pennsylvania, pp. 23–32. Blackwood D J, Gould L J, Naish C C, Porter F M, Rance A P, Sharland S M, Smart N R, Thomas M I and Yates T (2002), ‘The localised corrosion of

© Woodhead Publishing Limited, 2010

Nuclear waste canister materials

415

carbon steel and stainless steel in simulated repository environments’, AEAT/ ERRA 0318, December 2002. Brossia C S and Cragnolino G A (2000), ‘Effect of environmental variables on localized corrosion of carbon steel’, Corrosion 56, 505–514. Brossia S, Browning L, Dunn D S, Moghissi O C, Pensado O and Yang L (2001), ‘Effect of environment on the corrosion of waste package and drip shield materials’, Center for Nuclear Waste Regulatory Analysis Report CNWRA 2001-03. BSC (Bechtel SAIC Company) (2004), ‘Aqueous corrosion rates for waste package materials’, Prepared for US DOE, ANL-DSD-MD-000001, October 2004. Carlson L, Karnland O, Olsson S, Rance A and Smart N (2006), ‘Experimental studies on the interactions between anaerobically corroding iron and bentonite’, Posiva Working Report 2006-60. Casteels F, Dresselaars G and Tas H (1985), ‘Corrosion behaviour of container materials in a clay formation’, Commission of European Communities Report EUR-9570/1985. Casteels F, Dresselaars G and Tas H (1986), ‘Corrosion behaviour of container materials for geological disposal of high level waste’, Commission of the European Communities Report EUR 10398 EN, pp. 3–40. Cragnolino G, Dunn D S, Brossia C S, Jain V and Chan K S (1999), ‘Assessment of performance issues related to alternate engineered barrier system materials and design options’, Center for Nuclear Waste Regulatory Analysis Report CNWRA 1999-003. Crossland I (2005), ‘Long term corrosion of iron and copper’, in ICEM’05: 10th International Conference on Environmental Remediation and Radioactive Waste Management, 4–8 September 2005, American Society of Mechanical Engineers, New York, paper ICEM05-1272. Debruyn W, Dresselaers J, Vermieren Ph, Kelchtermans J and Tas H (1991), ‘Corrosion of container and infrastructure materials under clay repository conditions’, Commission of the European Communities Report EUR 13667 EN. Dechema (1990), Corrosion Handbook: Corrosive Agents and Their Interaction with Materials, vol. 7, Atmosphere. DOE (2008), Yucca Mountain Repository license application. US Department of Energy, DOE/RW-0573. Dunn D S, Cragnolino G A and Sridhar N (1996), ‘Localized corrosion initiation, propagation, and repassivation of corrosion resistant high-level nuclear waste container materials’, in Proceedings of CORROSION/96, NACE International, Houston, Texas, paper 97. Farmer J, McCright D, Gdowski G, Wang F, Summers T, Bedrossian P, Horn J, Lian T, Estill J, Lingenfelter A and Halsey W (2000), ‘General and localized corrosion of outer barrier of high-level waste container in Yucca Mountain’, in Proceedings of Transportation, Storage, and Disposal of Radioactive Materials – 2000, PVP-vol. 408, American Society of Mechanical Engineers, New York, pp. 53–69. Fe´ron D, Crusset D and Gras J-M (2009), ‘Corrosion issues in the French high-level nuclear waste program’, Corrosion, 65, 213–223. Fujisawa R, Kurashige T, Inagaki U and Senoo M (1999), ‘Gas generation behavior

© Woodhead Publishing Limited, 2010

416

Geological repository systems for safe disposal

of transuranic waste under disposal conditions’, in Materials Research Society Symposium Proceedings, vol. 556. Fukaya Y and Akashi M (2003), ‘Estimation of the cathodic hydrogen evolution rate on radioactive waste packaging materials’, in Proceedings of CORROSION/ 2003, NACE International, Houston, Texas, paper 03680. Hua F, Mon K, Pasupathi P, Gordon G and Shoesmith D (2005), ‘A review of corrosion of titanium grade 7 and other titanium alloys in nuclear waste repository environments’, Corrosion, 61, 987–1003. Hultquist G (1986), ‘Hydrogen evolution in corrosion of copper in pure water’, Corrosion Science, 26, 173–177. IMOA (International Molybdenum Association) (2001), ‘Practical guidelines for the fabrication of duplex stainless steels’, International Molybdenum Association, London, UK. JNC (2000), ‘H12: project to establish the scientific and technical basis for HLW disposal in Japan’, Japan Nuclear Cycle Development Institute, Supporting Report 2, Repository Design and Engineering Technology. Johnson L H and King F (2003), ‘Canister options for the disposal of spent fuel’, Nagra Technical Report 02-11. Johnson L H and King F (2008), ‘The effect of the evolution of environmental conditions on the corrosion evolutionary path in a repository for spent fuel and high-level waste in Opalinus Clay’, Journal of Nuclear Materials, 379, 9– 15. Johnson L H, Tait J C, Shoesmith D W, Crosthwaite J L and Gray M N (1994), ‘The disposal of Canada’s nuclear fuel waste: engineered barriers alternatives’, Atomic Energy of Canada Limited Report AECL-10718, COG-93-8. Johnson M J and Pavlik P J (1982), ‘Atmospheric corrosion of stainless steels’, in Atmosphere Corrosion, edited by W H Ailor, p. 461. Kearns J R, Johnson M J and Franson I A (1984), ‘The corrosion of stainless steels and nickel alloys in caustic solutions’, in Proceedings of CORROSION/84, NACE International, Houston, Texas, paper 146. King F (2002), ‘Corrosion of copper in alkaline chloride environments’, Swedish Nuclear Fuel and Waste Management Company Report SKB TR–02–25. King F (2006), ‘Review and gap analysis of the corrosion of copper containers under unsaturated conditions’, Ontario Power Generation, Nuclear Waste Management Division Report 06819-REP-01300-10124-R00. King F (2007), ‘Overview of a carbon steel container corrosion model for a deep geological repository in sedimentary rock’, Nuclear Waste Management Organization Report NWMO TR-2007-01. King F (2008), ‘Corrosion of carbon steel under anaerobic conditions in a repository for SF and HLW in Opalinus Clay’, Nagra Technical Report 08-12, Nagra, Wettingen, Switzerland. King F (2009a), ‘Microbiologically influenced corrosion of nuclear waste containers’, Corrosion, 65, 233–251. King F (2009b), ‘Hydrogen effects on carbon steel used fuel containers’, Nuclear Waste Management Organization Report, to be published. King F and Kolar M (2000), ‘The copper container corrosion model used in AECL’s second case study’, Ontario Power Generation, Nuclear Waste Management Division Report 06819-REP-01200-10041-R00, Toronto, Ontario.

© Woodhead Publishing Limited, 2010

Nuclear waste canister materials

417

King F and Kolar M (2005), ‘Preliminary assessment of the stress corrosion cracking of used fuel disposal containers using the CCM-SCC.0 model’, Ontario Power Generation, Nuclear Waste Management Division Report 06819-REP-0130010103-R00. King F and Kolar M (2006), ‘Consequences of microbial activity for corrosion of copper used fuel containers – analyses using the CCM-MIC.0.1 code’, Ontario Power Generation Nuclear Waste Management Division Report 06819-REP01300-00120-R00. King F and LeNeveu D (1992), ‘Prediction of the lifetimes of copper nuclear waste containers’, in Proceedings of Conference on Nuclear Waste Packaging, FOCUS ‘91, American Nuclear Society, La Grange Park, Illinois, pp. 253–261. King F and Newman R C (2009), ‘Stress corrosion cracking of copper canisters’, Swedish Nuclear Fuel and Waste Management Company Report, in press. King F and Stroes-Gascoyne S (1997), ‘Predicting the effects of microbial activity on the corrosion of copper nuclear waste disposal containers’, in Microbial Degradation Processes in Radioactive Waste Repository and in Nuclear Fuel Storage Areas, edited by J H Wolfram, Kluwer Press, Dordrecht, Netherlands, pp. 149–162. King F, Kolar M and Shoesmith D W (1996), ‘Modelling the effects of porous and semi-permeable layers on corrosion processes’, in Proceedings of CORROSION/96, NACE International, Houston, Texas, paper 380. King F, Ahonen L, Taxe´n C, Vuorinen U and Werme L (2001), ‘Copper corrosion under expected conditions in a deep geologic repository’, Swedish Nuclear Fuel and Waste Management Company Report SKB TR 01-23. King F, Ahonen L, Taxe´n C, Vuorinen U and Werme L (2002), ‘Copper corrosion under expected conditions in a deep geologic repository’, Posiva Oy Report POSIVA 2002-01. King F, Kolar M, Kessler J H and Apted M (2008a), ‘Yucca Mountain engineered barrier system corrosion model (EBSCOM)’, Journal of Nuclear Materials, 379, 59–67. King F, Kolar M and Maak P (2008b), ‘Reactive-transport model for the prediction of the uniform corrosion behaviour of copper used fuel containers’, Journal of Nuclear Materials, 379, 133–141. Kursten B and Druyts F (2008), ‘Methodology to make a robust estimation of the carbon steel overpack lifetime with respect to the Belgian Supercontainer design’, Journal of Nuclear Materials, 379, 91–96. Kursten B, Smailos E, Azkarate I, Werme L, Smart N R and Santarini G (2004), ‘COBECOMA, State-of-the-art Document on the COrrosion BEhaviour of COntainer MAterials’, European Commission, Contract FIKW-CT-2001420138 Final Report. Little B, Wagner P and Mansfeld F (1991), ‘Microbiologically influenced corrosion of metals and alloys’, International Materials Reviews, 36, 253–272. Lloyd A C, Schuler R J, Noe¨l J J, Shoesmith D W and King F (2004), ‘The influence of environmental conditions and passive film properties on the MIC of engineered barriers in the Yucca Mountain Repository’, in Scientific Basis for Nuclear Waste Management XXVIII, edited by J M Hanchar, S StroesGascoyne and L Browning, Materials Research Society Symposium

© Woodhead Publishing Limited, 2010

418

Geological repository systems for safe disposal

Proceedings 824, Materials Research Society, Warrendale, Pennsylvania, pp. 3–9. Maak P (1999), ‘The selection of a corrosion-barrier primary material for used-fuel disposal containers’, Ontario Power Generation, Nuclear Waste Management Division Report 06819-REP-01200-10020-R00. Mcdonald D B, Sherman M R, Pfeifer D W, and Virmani Y P (1995), ‘Stainless steel reinforcing as corrosion protection’, Reprint from Concrete International, Reprint Number 14034, May 1995. McMurry J, Dixon D A, Garroni J D, Ikeda B M, Stroes-Gascoyne S, Baumgartner P and Melnyk T W (2003), ‘Evolution of a Canadian deep geologic repository: base scenario’, Ontario Power Generation, Nuclear Waste Management Division Report 06819-REP-01200-10092-R00. Marcus P (1995), ‘Sulfur-assisted corrosion mechanisms and the role of alloyed elements’, in Corrosion Mechanisms in Theory and Practice, edited by P Marcus and J Oudar, Marcel Dekker, New York, pp. 239–263. Marsh G P and Taylor K J (1988), ‘An assessment of carbon steel containers for radioactive waste disposal’, Corrosion Science, 28, 289–320. Masurat P, Eriksson S and Pedersen K (2009), ‘Microbial sulphide production in compacted Wyoming bentonite MX-80 under in situ conditions relevant to a repository for high-level radioactive waste’, Applied Clay Science, doi:10.1016/ j.clay.2009.01.004. Mattsson H and Olefjord I (1984), ‘General corrosion of Ti in hot water and water saturated bentonite clay’, Swedish Nuclear Fuel Supply Company Report SKB TR-84-19. Mattsson H and Olefjord I (1990), ‘Analysis of oxide formed on Ti during exposure in bentonite clay. I. The oxide growth’, Werk. Korros, 41, 383–390. Mattsson H, Li C and Olefjord I (1990), ‘Analysis of oxide formed on Ti during exposure in bentonite clay. II. The structure of the oxide’, Werk. Korros, 41, 578–584. Mon K G, Gordon G M and Rebak R B (2005), ‘Stifling of crevice corrosion in Alloy 22’, in Proceedings of the 12th International Conference on Environmental Degradation of Materials in Nuclear Power System–Water Reactors, edited by T R Allen, P J King and L Nelson, The Minerals, Metals and Materials Society, Warrendale, Pennsylvania, pp. 1431–1438. Nagra (2002), ‘Project Opalinus Clay. Safety Report’, National Cooperative for the Disposal of Radioactive Waste, Nagra Technical Report NTB 02-05. Nakayama G, Sakakibara Y, Taniyama Y, Cho H, Jintoku T, Kawakami S and Takemoto M (2008), ‘The long-term behaviors of passivation and hydride layer of commercial grade pure titanium in TRU waste disposal environments’, Journal of Nuclear Materials, 379, 174–180. Neff D, Dillmann P, Descostes M and Beranger G (2006), ‘Corrosion of iron archaeological artefacts in soil: estimation of the average corrosion rates involving analytical techniques and thermodynamic calculations’, Corrosion Science, 48, 2947–2970. ONDRAF-NIRAS (2004), ‘A review of corrosion and material selection issues pertinent to underground disposal of highly active nuclear waste in Belgium’, ONDRAF Report NIROND 2004-02.

© Woodhead Publishing Limited, 2010

Nuclear waste canister materials

419

Pastina B and Hella¨ P (2006), ‘Expected evolution of a spent fuel repository at Olkiluoto’, POSIVA 2006-05, Posiva Oy, Olkiluoto, Finland. Pourbaix M (1974), Atlas of Electrochemical Equilibria in Aqueous Solutions, 2nd edition, NACE International, Houston, Texas. Rebak R B, Koon N E, Dillman J R, Crook P and Summers T S E (2000), ‘Influence of aging on microstructure, mechanical properties, and corrosion resistance of a Ni–22Cr–13Mo–3W alloy’, in CORROSION/2000, NACE International, Houston, Texas, paper 00181. Sedriks A J (1996), Corrosion of Stainless Steels, 2nd edition, John Wiley & Sons, Inc., New York. Shoesmith D W (2006), ‘Assessing the corrosion performance of high-level nuclear waste containers’, Corrosion, 62, 703–722. Shoesmith D W (2008), ‘The role of dissolved hydrogen on the corrosion/dissolution of spent nuclear fuel’, Nuclear Waste Management Organization Report NWMO TR-2008-19. Shoesmith D W and King F (1999), ‘The effects of gamma radiation on the corrosion of candidate materials for the fabrication of nuclear waste packages’, Atomic Energy of Canada Limited Report AECL-11999. SKB (1983), ‘Final storage of spent nuclear fuel – KBS-3’, Swedish Nuclear Fuel Supply Company Report KBS-3, vols I–IV. SKB (2006a), ‘Long-term safety for KBS-3 repositories at Forsmark and Laxemar – a first evaluation’, Main Report of the SR-Can Project, Swedish Nuclear Fuel and Waste Management Company Report, Technical Report TR-06-09. SKB (2006b), ‘Fuel and canister process report for the safety assessment of SR-Can’, Swedish Nuclear Fuel and Waste Management Company Report SKB TR 0622. Smart N R (2000), ‘Atmospheric corrosion of stainless steel waste containers during storage’, AEA Technology Report AEA-TPD-262, Issue C. Smart N R (2002), ‘Review of effect of chloride in cementitious environments on corrosion of stainless steels’, Serco Assurance Report for UK Nirex Limited, SA/SIS/14921/R001. Smart N R (2005), ‘Atmospheric pitting corrosion of stainless steel radioactive waste containers’, Serco Assurance Report to Nirex, Report SA/EIG/14921/C0050. Smart N R and Wood P (2004), ‘Corrosion resistance of stainless steel radioactive waste packages’, UK Nirex Limited, Nirex Report N/110. Smart N R, Blackwood D J and Werme L O (2001), ‘The anaerobic corrosion of carbon steel and cast iron in artificial groundwaters’, SKB Technical Report TR-01-22. Smart N R, Blackwood D J, Marsh G P, Naish C C, O’Brian T M, Rance A P and Thomas M I (2004), ‘The anaerobic corrosion of carbon and stainless steels in simulated repository environments: a summary review of Nirex research’, Report AEAT/ERRA-0313. Smart N R, Naish C C and Pritchard A M (2006), ‘Corrosion principles for the assessment of stainless steel radioactive waste containers’, Serco Assurance Report to Nirex, Report SA/EIG/14921/C010. Smith J, Qin Z, King F, Werme L and Shoesmith D W (2007), ‘Sulfide film formation on copper under electrochemical and natural corrosion conditions’, Corrosion, 63, 135–144.

© Woodhead Publishing Limited, 2010

420

Geological repository systems for safe disposal

Sridhar N and Cragnolino G A (1993), ‘Effects of environment on localized corrosion of copper-based, high-level waste container materials’, Corrosion, 49, 967–976. Stroes-Gascoyne S, Hamon C J, Dixon D A, Kohle C and Maak P (2007), ‘The effects of dry density and porewater salinity on the physical and microbiological characteristics of highly compacted bentonite’, in Scientific Basis for Nuclear Waste Management XXX, edited by D Dunn, C Poinssot and B Begg, Materials Research Society, Pittsburgh, Pennsylvania, paper 0985NN13-02. Swedish Corrosion Institute (1978), ‘Copper as canister material for unreprocessed nuclear waste – evaluation with respect to corrosion’, Swedish Nuclear Fuel Supply Company Report KBS-TR-90. Szaka´los P, Hultquist G and Wikmark G (2007), ‘Corrosion of copper in water’, Electrochemical Solid-State Letters, 10, C63–C67. Szklarska-Smialowska Z (2005), Pitting and Crevice Corrosion, NACE International, Houston, Texas. Taniguchi N and Kawasaki M (2008), ‘Influence of sulphide concentration on the corrosion behaviour of pure copper in synthetic seawater’, Journal of Nuclear Materials, 379, 154–161. Turnbull A (2009), ‘A review of the possible effects of hydrogen on lifetime of carbon steel nuclear waste containers’, National Cooperative for the Disposal of Radioactive Waste, Nagra Technical Report NTB 09-04, in preparation. Wada R and Nishimura T (1999), ‘Experimental study of hydrogen gas generation rate from corrosion of Zircaloy and stainless steel under anaerobic alkaline condition’, Radioactive Waste Management and Environmental Remediation, ASME.

© Woodhead Publishing Limited, 2010

14 Post-containment performance of geological repository systems: source-term release and radionuclide migration in the near- and far-field environments Ch. POINSSOT and C. FILLET, CEA, Commissariat a` L’Energie Atomique et aux Energies Alternatives, France; J . - M . G R A S , JMG Consulting, France

Abstract: This chapter describes the processes that will occur in the nearfield and far-field environments after container corrosion and the ingress of water within the container. This stage will be dominated by the slow, progressive alteration of the waste form and the subsequent release of radionuclides first in the near-field environment, followed by migration towards the geological environment. This chapter describes the processes governing the long-term alteration of the different waste forms, including spent fuel, and the anticipated alteration rates. It then discusses the processes governing radionuclide migration in porous media such as those encountered in the near and far fields of a repository. Conclusions are drawn about the key radionuclides that dominate the long-term impact of a potential deep repository. Key words: spent fuel dissolution, glass dissolution, long-term behaviour, concrete alteration, bitumen alteration, radionuclide source term, radionuclide migration, diffusion, retention.

14.1

Introduction

The safety of a geological repository for high- or medium-level long-lived nuclear waste is based on a multi-barrier concept designed to limit and delay for as long as possible any transfer of radiotoxicity to the biosphere. The successive containment barriers (waste conditioning matrix, container, possible engineered barrier, geological formation) have complementary and 421 © Woodhead Publishing Limited, 2010

422

Geological repository systems for safe disposal

sometimes redundant roles; the function assigned to each barrier depends on the repository status at a given period in its lifetime, on the nature of the radionuclides and on the predicted evolution of the site over time. For high-level waste (HLW) and spent fuel the container must be capable for a specified time period of isolating the source term and preventing any radionuclide release. The objective is notably to benefit from the decay of short-lived fission products, which are the most radioactive (137Cs, 90Sr), and to avoid any radionuclide release while the packages and their environment are still hot. The main function of the other barriers is to limit the rate of release and to delay radionuclide migration. This is the role in particular of the waste matrix when the container is no longer capable of ensuring its function and of the near-field material and host rock when radionuclides begin to be released from the waste package. The durability of the container depends mainly on the nature of the material and its corrosion resistance after disposal (refer to Chapter 13); this depends on the repository design options. Some countries, such as Sweden and Finland, have selected copper as the spent fuel container material, assuming that in the normal disposal scenario the waste will not be exposed to water during the repository design life. Waste degradation is considered in these countries only in a modified repository scenario arising from a container welding defect. Other countries, such as France and Japan, have adopted carbon steel as the long-lived high-level waste container material and assigned it a limited lifetime (a few thousand years): waste degradation is considered simply as a stage in the normal disposal scenario. The degradation of long-lived high-level waste is thus assumed to occur after a few thousand years at the earliest. In most disposal sites, the repository will then be situated in a reducing environment; the oxygen in the air occluded in the repository after closure will have been consumed by corrosion and by oxidation of the minerals (pyrite, etc.) and organic matter present in the nearfield material and surrounding rock. The Yucca Mountain site considered by the US DOE for a repository is an exception; the particular configuration of this site would imply operation under oxidizing conditions and in unsaturated conditions. The chemical nature of the porous media surrounding the waste (e.g. clay or cement) is another important parameter that will maintain a neutral or alkaline pH. All the concepts under consideration today for longlived high-level waste around the world assume that vitrified waste forms will be placed in a clay environment at near-neutral pH. This is generally also the case for direct disposal of spent fuel, although some concepts have considered the use of structural concrete in the immediate vicinity of the spent fuel packages. Conversely, hydraulic binders are widely taken into account in intermediate-level waste disposal concepts and are routinely used for shortlived waste disposal. During leaching, long-lived high- and intermediate-level waste will no

© Woodhead Publishing Limited, 2010

Post-containment performance of geological repository systems

423

longer be in the same state as at the moment of disposal. The glass and the spent fuel will evolve during the time necessary for corrosion of the metal containers protecting them from aqueous leaching. Because of alpha decay, for example, helium will form in the glass matrix or in the spent fuel during the first few thousand years in the repository; its effect on the matrix microstructure must be taken into account because it could affect the reactive surface subjected to leaching. Once they are released from the waste matrix, the radionuclides will be contained by the near-field materials, and eventually the far field around the repository. Radionuclide migration is a subject of considerable research to identify, understand and model the reactions and physical and chemical processes involved (speciation, complex formation, sorption, precipitation); many of these processes are related to or are coupled with other processes. Likewise, the near-field environmental conditions will evolve after placement of the waste and can affect the long-term performance of the near field. Before considering the geochemical fate of the radionuclides released in the repository, this chapter reviews the present state of knowledge concerning waste matrix leaching mechanisms after disposal, focusing on long-lived high-level waste and spent fuel.

14.2

Waste form degradation

Predicting the long-term behaviour of the waste forms primarily involves materials science. Various types of waste forms can be differentiated: .

.

Materials and matrices optimized for waste conditioning, with the objective of high-performance long-term properties, have been developed mainly to meet the needs arising from spent fuel treatment. This is the case of vitrified waste, which today constitutes the industrial reference matrix for long-lived high-level waste: they are capable of simultaneously conditioning fission products and minor actinides (more than 35 radioelements in waste containment glass produced after reprocessing UO2 fuel) and exhibit very good long-term durability. In the future, further progress is likely to be made in loading increasing amounts of fission products and actinides. French vitrified glass is referred to as R7T7 by reference to the industrial production facility in the La Hague plant (France). More generally, we will refer to vitrified waste as nuclear glass in the rest of this chapter. Other matrices that are specifically designed for presenting long-term confinement are insoluble crystalline mineral matrices without constituent water, such as ceramics, or with constituent water such as cement, and composite matrices such as glass–ceramics. Other materials and matrices should theoretically be accepted as they are and are not subject to fuel cycle back-end operations. This is the case

© Woodhead Publishing Limited, 2010

424

Geological repository systems for safe disposal especially for the UO2 and MOX (UO2–PuO2) spent fuel matrix, which was designed for use in the reactor and which will be packaged in its irradiated state for direct disposal.

Moreover, some waste packages have already been produced, including vitrified (especially R7T7-type) waste, packages consisting of compacted hulls and end pieces, bituminized waste packages and packages made from hydraulic binders (cement, mortar, concrete). Others, notably spent fuel packages, are still at the design stage. The main aggressive factor to consider is deep groundwater, which will saturate the repository site after closure and can progressively alter the containment properties of the barriers, first corroding the container, then leaching the waste matrix and transporting the released radionuclides into the engineered barriers and geological environment. The most difficult scientific issues raised by waste containment concern long-term behaviour predictions for the containment materials. The problem is to demonstrate the durability of containment of radioactive products during the repository design life, at time scales measured in thousands, tens of thousands or even hundreds of thousands of years – in any case well beyond the time accessible to humans (Poinssot and Toulhoat,

14.1 Schematic diagram describing the specificity of predicting the long-term evolution of waste forms in a repository. In this figure, past events are plotted on the same timescale as the half-lives of some of the long-lived radionuclides present within the waste form. The estimated confinement period of the waste form is also given and the period of radionuclide migration within the near and far fields. This figure clearly illustrates that the timeframes considered are well beyond the time accessible to human perception and experiments. It therefore requires a specific approach to ensure the reliability of the predictions (Poinssot and Toulhoat, 2007).

© Woodhead Publishing Limited, 2010

Post-containment performance of geological repository systems

425

2007) (Fig. 14.1). Establishing these predictions raises two major difficulties that are specific to waste confinement studies. .

The first lies in the fact that the predictions extend over very long time periods that are inaccessible to laboratory experimentation. For lowlevel short-lived waste placed in surface storage, the performance requirements are limited in time (30 to 50 years of operation, followed by a 300-year surveillance period before releasing the site). The materials problems have now been overcome: predictions of material ageing remain circumscribed within a time period compatible with reasonable extrapolation of available data. The difficulty is of a different order of magnitude in the case of geological disposal of high-level and/or longlived waste. The package materials can be subjected to degradation experiments lasting a few years at the most, whereas the predictions must cover several centuries. The ageing kinetics can be accelerated by enhancing the effect of one of the influential parameters (the temperature or the leaching solution flow rate, for example) but such changes often raise the risk of modifying the nature of the very phenomenon being observed. A specific approach has therefore been defined which allows the development of robust and reliable performance models for predicting the long-term waste form evolution (Fig. 14.2). This approach requires first developing a detailed and

14.2 Integrated approach of the science of long-term waste form behaviour (Poinssot and Toulhoat, 2007). In order to ensure the reliability and relevance of long-term predictions, they are postulated using dedicated simplified operational models (top of the pyramid) based on mechanistic modelling of the most relevant processes involved in the long-term waste form evolution (after studying and ranking all the processes involved as a function of the evolution scenario).

© Woodhead Publishing Limited, 2010

426

.

Geological repository systems for safe disposal reliable understanding and model of all the actual physical and chemical mechanisms involved in waste form evolution, based on specific experimental data or molecular modelling (ab initio, molecular dynamics, Monte Carlo, etc.) (Bureau et al., 2008). This involves developing complex scientific models. From this knowledge and based on the reference evolution scenario, the various processes can be ranked and hierarchized, and the potential couplings determined. Finally, by considering the sole governing processes and couplings, a simplified dedicated performance model can be developed for further long-term performance assessment. Such models are classically referred to as operational models by opposition to the complex scientific models that integrate all the available knowledge at a given time and evolve as science progresses. Two complementary approaches can be used to validate predictive models: (1) corroboration of behaviour models by analogy with natural systems and archaeological materials (Petit, 1992; Miller et al., 1994; Mazurek et al., 2008) or (2) development of in situ validation experiments, especially in underground research laboratories (e.g. tracing experiments; Dewonck et al., 2006). The second difficulty is related to the fact that most of the phenomena observed are coupled, and that in the sequence of geochemical processes involved, we are faced with a combination of mechanisms interacting among themselves. The waste matrix leaching mechanism and kinetics may depend on the groundwater leaching solution composition, whose chemistry is the result of equilibrium with the minerals in the surrounding geological formation and can be modified by canister corrosion products. In spent fuel, for example, hydrogen evolving from canister corrosion could inhibit the radiolytic dissolution of UO2. It is therefore indispensable to allow for coupling between alteration processes and parameters.

In order to address these questions, the long-term behaviour of waste packages must be considered in three types of generic environmental conditions (Poinssot et al., 1999b, 2002) (Fig. 14.3): .

.

In a closed system. During this period the package container is intact and the waste is fully confined. The package is subject to internal evolution and to energy exchanges in the form of heat (thermal impact of high-level waste and spent fuel) or in the form of radioactivity (residual a, bg radioactivity and neutrons in spent fuel). However, no mass transfer occurs between the system and the external environment. In a water-saturated open system. After the leaktight container is breached, the waste matrix is subjected to aqueous alteration and can exchange matter with other near-field materials; radioelement migration then becomes possible.

© Woodhead Publishing Limited, 2010

Post-containment performance of geological repository systems

427

14.3 Generic boundary conditions relevant to the study of long-term waste form evolution in a repository. Classically, waste forms initially evolve for hundreds to thousands of years in a closed system before evolving to open-system conditions, rarely unsaturated and more frequently saturated, which correspond to waste form alteration leading to radionuclide release. Q stands for energy exchange, σ for any mechanical stress and α, β, γ for the radioactive fields (Poinssot et al., 1999b).

.

Between these two situations, a transient state can be distinguished: an open unsaturated medium if the near-field environment is not yet resaturated by groundwater.

This overall approach defines a new scientific field that has been developed in recent decades and that we could refer to as long-term behaviour science (Gras et al., 2007; Poinssot and Toulhoat, 2007). The following sections describe more specifically the way it has been applied to the two HLW waste forms: nuclear glass and spent nuclear fuel.

14.3

Long-term nuclear glass performance

Vitrification is a reference process for high-level waste today in the countries that have opted for spent fuel reprocessing. Borosilicate glass formulations constitute the most widely adopted solution for conditioning this type of waste (USA, Japan, Great Britain, Russia, France, etc.). In France, for example, spent light water reactor fuel is reprocessed in the La Hague plant, where the fission products and minor actinides are immobilized in a borosilicate glass medium (R7T7 glass) industrially produced in the R7 and T7 facilities. The vitrified waste packages are currently stored on the production site. Disposal of these waste packages in a deep geological formation is being examined by the French National Radioactive Waste Management Agency (ANDRA) under the terms of the French Waste Management Act of June 2006, which calls for a staged investigation leading to an application in 2015 for the creation of a repository scheduled to become operational in 2025. In several countries including France, glass canisters may be placed in a metallic overpack to ensure a closed system stage of sufficient duration. The long-term glass

© Woodhead Publishing Limited, 2010

428

Geological repository systems for safe disposal

behaviour must therefore be assessed first during the closed-system stage, and then within the water-saturated open-system phase, as described above.

14.3.1

Glass behaviour in a closed system

Thermal stability Thermal stability is a prerequisite for the conservation of a homogeneous glass over time. From a theoretical standpoint, that glass can evolve naturally to a thermodynamically more stable crystalline state. This thermally activated transformation becomes extremely slow or is even inhibited when the glass is maintained at temperatures below the glass transition point. Prediction of long-term thermal stability at low temperature is based on investigations of devitrification in a supercooled liquid and modelling. For the French nuclear glass three major phases have been identified: powellite (CaMoO4), CeO2 and zinc chromite spinel (ZnCr2O4). The nucleation and growth curves for these three phases reveal several phenomena: .

. .

Nucleation occurs suddenly during the first few hours of treatment, and then ceases. Nucleation is heterogeneous and favourable to crystallization at existing active sites. The nucleation curves are of larger amplitude and are shifted towards lower temperatures when insoluble particles of platinum-group metals are present. The crystal nuclei grow very slowly, reaching saturation after a few tens of hours. The combination of extensive nucleation and very limited growth produces a material that is relatively insensitive to devitrification.

The glass transformation kinetics versus time and temperature have been described on the basis of the theory advanced by Kolgomorov (1937), Johnson and Mehl (1939) and Avrami (1939, 1940, 1941). Atomic diffusion appears to be the main factor limiting crystallization. The viscosity thus determines the nucleation and growth kinetics in the glass; this parameter determines the diffusive transport of atoms in the liquid silicate. At low temperatures the nucleation-growth kinetics can be determined by viscosity measurements independently over a wide temperature range. After validation on a simple barium disilicate glass known to crystallize rapidly and homogeneously (Orlhac et al., 2001), the R7T7 glass model shows that several million years are necessary for crystallization of the three major phases, demonstrating that the vitreous state is conserved in the glass over the long term. More generally, although uncontrolled crystallization must of course be avoided, crystallization is not necessarily an unacceptable

© Woodhead Publishing Limited, 2010

Post-containment performance of geological repository systems

429

phenomenon: everything depends on its volume fraction, on the homogeneity of its distribution, on the location of the radionuclides and on the chemical durability of the crystals and the residual vitreous phase. Self-irradiation behaviour Under interim storage and disposal conditions the glass is subjected to bg and a decay (the latter predominates after several hundred years). Numerous studies have been carried out to assess the impact of a bg irradiation on glass properties (Weber and Roberts, 1983; Ewing et al., 1995; Weber et al., 1997). b decay, g transitions and helium nuclei arising from a decay interact mainly electronically with atoms in the glass network. Recoil nuclei produced by a decay result, mainly in elastic interactions between nuclei, and are responsible for most of the atomic displacements sustained by the glass in a nuclear waste repository. Irradiation damage studies combine several approaches: .

.

.

The study of glass samples doped with short-lived radioactive elements to accumulate the highest possible decay doses in the glass in a short time, at a laboratory time scale (Matzke, 1997; Fares et al., 2009). This methodology has the advantage of being representative of reality because it irradiates the entire glass volume. However, since the doses are integrated at a higher rate than in repository conditions, care must be taken to ensure the absence of any dose rate effect. Glass specimens subjected to external irradiation. These methods are used to accumulate high doses in a thin glass layer near the surface, and allow a broader range of characterization methods than doped glass (Raman spectroscopy, XANES) (De Bonfils, 2007; Bureau et al., 2008). Atomistic modelling (molecular dynamics). Computation of cumulative displacement cascades is a means of estimating the impact of irradiation on macroscopic properties such as the density or the modulus of elasticity (Delaye and Ghaleb, 2005a, 2005b, 2006; Deladerrie`re et al., 2008).

For the French R7T7 nuclear glass, the three approaches all yield consistent results: due to the effect of α decay the glass density diminishes slightly (Fig. 14.4) and its mechanical properties appreciably improve, especially its resistance to cracking. The variations in these properties reach a saturation level and stabilize beyond 261018 a/g (Matzke, 1997; Peuget et al., 2006, 2010). Nuclear interactions caused by recoil nuclei induce slight structural changes, especially a drop in the boron coordination number and residual depolymerization of the borosilicate network (~1%). These changes are

© Woodhead Publishing Limited, 2010

430

Geological repository systems for safe disposal

14.4

Density of R7T7-type glass versus alpha decay dose.

comparable to the effects of local thermal quenching of zones partially disorganized by recoil nuclei. A local thermal quenching accumulation model has been developed to describe the origin of these structural changes: each alpha disintegration locally damages the glass, which stabilizes in a new glass structure corresponding to a hypothetical high-temperature state. The accumulation of these events throughout the glass volume gradually produces a new glass corresponding to this slightly modified structure. This description accounts for the stabilization of the macroscopic properties observed beyond 261018 a disintegrations per gram of glass by postulating a damage threshold affecting the entire material volume (Bureau et al., 2008). The results of these studies reveal no measurable specific effect of helium generation in the glass up to the maximum dose of 1019 a/g. The data acquired establish that the properties of R7T7-type glass will not be modified (Chamssedine et al., 2010).

14.3.2

Glass behaviour in a water-saturated open system

Glass alteration regimes and predominant mechanisms Irrespective of the glass composition, the leaching mechanisms can be described as follows (Frugier et al., 2008): . . .

Exchange and hydrolysis reactions involving the mobile glass constituents (alkalis, boron, etc.) occur rapidly during the initial instants. Slower hydrolysis, especially of silicon, results in the establishment of an initial glass dissolution rate. The difference between these two kinetics results in the creation of an

© Woodhead Publishing Limited, 2010

Post-containment performance of geological repository systems

. .

.

431

amorphous layer at the glass/solution interface regardless of the alteration conditions. This layer is gradually reorganized by hydrolysis and condensation mechanisms (Ledieu et al., 2003; Arab et al., 2008). The amorphous layer dissolves as long as the solution is not saturated with respect to its constituent elements (Si, Zr, Al, Ca, etc.). Renewal of a pure water solution sustains the dissolution process. The amorphous layer constitutes a barrier against the transport of water towards the glass and of solvated glass ions into solution (Cailleteau et al., 2008; Jollivet et al., 2008). This transport-inhibiting effect rapidly causes this layer to control glass alteration. It is referred to as the ‘passivating reaction interphase’ (PRI) in accordance with its properties. Some glass constituent elements precipitate as crystallized secondary phases. The precipitation of these crystallized phases on the external surface or in solution can sustain glass alteration (Frugier et al., 2005, 2006).

Based on these dominant mechanisms and considering a fresh glass sample placed in pure water, five kinetic regimes can be distinguished (Fig. 14.5): Initial interdiffusion The term interdiffusion refers to the exchange between glass networkmodifying cations and protons in solution (Geneste et al., 2006). This mechanism has been identified experimentally during leaching of many minerals and natural glasses, especially in acidic media (Doremus, 1975). The interpretation is based on observation of anticorrelated sigmoidal

14.5 Time sequence of different rate regimes and sodium and silicon concentrations in solution: I, initial diffusion; II, initial rate; III, rate drop; IV, residual rate; V, resumption of alteration under particular conditions.

© Woodhead Publishing Limited, 2010

432

Geological repository systems for safe disposal

concentration profiles in the hydrated glass between alkali ions and protons. Its diffusive nature makes this the overriding phenomenon during the initial moments of leaching of a glass or mineral. In a diffusive process the concentration variations of aqueous species can often be expected to follow a square root time-dependence relation, with a rate derived from the concentrations exhibiting inverse square root time dependence (Ojovan et al., 2006). For a very brief period of time these kinetics exceed the hydrolysis rate of the silicate network. Initial dissolution rate Hydrolysis of the glass network is the mechanism responsible for the initial dissolution rate, also called the forward rate. Hydrolysis profoundly modifies the silicate network by attacking bridging bonds (Si–O–Si, Si–O– Al, Si–O–Zr, etc.) in the interphase created by the release of mobile elements. Water molecules thus act on the glass by directly affecting network connectivity, leading to matrix dissolution according to the reaction: Si–O–Si + H2O ® Si–OH. The rupture of the four bridging bonds likely to surround a silicon atom in the glass ultimately leads to the release of orthosilicic acid, H4SiO4 (Bunker, 1994; Conradt, 2008). This reaction depends on the nucleophilic character of water and is thus facilitated in basic media. The term initial dissolution rate, r0, is used when the leaching solution is sufficiently dilute to prevent any feedback effect of aqueous species from the glass on the glass hydrolysis kinetics. Dissolution is then congruent (all the elements are leached at the same rate) and no alteration products are formed. This situation is, in fact, never observed in nuclear glasses. In practice, few experimental conditions can prevent precipitation of the leastsoluble phases or hydrolysis of the most-rigid bonds (such as Si–O–Zr). The initial dissolution rate depends essentially on the temperature, the pH and the glass composition (Grambow, 1985; Advocat et al., 1991). Rate drop The rate drop is without any doubt the most complex kinetic regime, and its potential effect in predictive models is very significant, as it ranges over several orders of magnitude. The rate drop is first observed only when the dissolved silicon in solution reaches a sufficient concentration, about 1 mg/l at 90 8C. Under these conditions, a fraction of the dissolved silicon recondenses to form an amorphous, porous, hydrated phase commonly referred to as the gel (Gin, 2000; Angeli et al., 2006, 2008). Until recently, two opposing approaches dominated in the literature to account for the rate drop, one based on the chemical affinity expressed with respect to the initial

© Woodhead Publishing Limited, 2010

Post-containment performance of geological repository systems

433

glass and the other on the passivating effect of the alteration gel (Grambow and Strachan, 1998; Vernaz, 2002). It has now been established that the rate drop can be attributed to the combined effects of affinity and passivation related to the gel (rate-limiting effects due to solution saturation and water transport within the gel layer). Residual alteration rate in a closed system The residual rate can either refer to the intrinsic glass behaviour following the rate drop regime in a closed system or can be the final rate controlled by the boundary conditions in an open system (including flow rate, near-field materials, etc.). This regime, still under discussion at the international level, appears to be predominant in geological disposal conditions (see below). In this kinetic regime the rate is generally slowly decreasing or remains constant. Two concomitant mechanisms account for this phenomenology in the French R7T7 nuclear glass: .

. .

Reactive diffusion of the aqueous species in the protective gel. This mechanism involves reactive diffusion of water and glass elements through the gel (Ferrand et al., 2006). When the activities of the gelforming elements (Si, Al, Ca, etc.) tend towards steady-state conditions in solution, the glass network hydrolysis rate diminishes exponentially. Under these conditions the mechanisms of interdiffusion (see regime 1), which had become negligible after a few instants, again become predominant. Dissolution of the gel layer by surrounding porewater and migration of silica towards the near-field environment. Precipitation of secondary crystalline phases that consume elements from the protective gel. In a closed system, reactive diffusion results in a continuous rise in the concentrations of mobile glass elements in solution. This raises the issue of long-term pH variations, and more generally of the geochemical evolution of the entire system comprising the solution, the gel and crystallized secondary phases. The precipitation of crystallized secondary phases can sustain glass alteration.

Resumption of alteration A resumption of alteration refers to a significant increase in the alteration rate following the rate drop or the residual rate. Irrespective of the type of borosilicate, the resumption of alteration appears to be related to massive precipitation of crystallized zeolite secondary phases (Gin and Mestre, 2001; Ribet and Gin, 2004). This precipitation occurs suddenly and after an extended time; it must therefore be kinetically limited or highly activated

© Woodhead Publishing Limited, 2010

434

Geological repository systems for safe disposal

when an activity or pH threshold in solution is exceeded. The precipitation of this phase must also be self-sustaining. This destabilizes the protective gel, glass alteration resumes and, given the alkali concentration of the glass, the solution pH increases, maintaining the precipitation of zeolites. The rate during the resumption of alteration does not exceed the initial rate at the same pH. Modelling Demonstrating the safety of the geological disposal concept implies assessing the long-term behaviour of the glass in contact with groundwater and environmental materials during the thousands of years necessary for decay of the radionuclides incorporated in the glass structure. Over timescales inaccessible to laboratory experimentation, and faced with the chemical complexity of the glass and its environment, modelling is the principal means of assessing the source term, i.e. the radionuclides released from the glass over time. Two families of models have been developed and are presented below. The GRAAL mechanistic model Progress in recent years, especially in identifying the mechanisms responsible for the residual rate, has made it possible to propose new theoretical foundations for modelling the different kinetic regimes of R7T7 nuclear glass dissolution. The GRAAL model considers that water diffusion in the passivating reaction interphase (PRI, corresponding to the gel formed under saturation conditions) is a rate-limiting step in the overall glass dissolution kinetics (Frugier et al., 2008, 2009). Moreover, this passivation zone is a soluble phase whose stability is directly dependent on the nature of the secondary phases likely to precipitate and on the solution renewal conditions. The following simplifying hypotheses are considered in the GRAAL model: .

. .

Only water diffusion in the PRI is considered to be rate-limiting. A single apparent diffusion coefficient is used to simulate water diffusion in the PRI and diffusion of hydrolysed and solvated glass constituent elements in solution. Specific zones like those corresponding to water diffusion in the glass, proton/alkali ion exchange and depleted gel are ignored. The reactivity of the PRI with the leaching solution is described by a thermodynamic equilibrium.

The simplified system is described by five basic equations:

© Woodhead Publishing Limited, 2010

Post-containment performance of geological repository systems 1.

2.

An equation describing the kinetics of dissolution of the PRI,  dE CSi ðtÞ ¼ rdisso 1  dt Csat

4.

½14:2

where e stands for the PRI thickness at time t, rhydr for the hydrolysis rate of soluble glass constituents and DPRI for the water diffusion coefficient in the PRI. An equation describing the kinetics of precipitation of neoformed phases,  dMpr 0 0 0 CSi ðtÞ ½14:3 ¼rkS Csat dt where Mpr stands for the mass of the neoformed phase at time t, r0 for the density of neoformed phase, k0 for the precipitation rate parameters of the neoformed phase and S0 the surface area of the neoformed phase. A silicon mass balance equation, O

5.

½14:1

where E stands for the total dissolved PRI thickness at time t, CSi(t) for the Si aqueous concentration, Csat for the Si concentration at which dissolution ceases and rdisso for the dissolution rate observed in pure water. An equation describing the kinetics of formation of the PRI, de rhydr dE ¼  dt 1 þ erhydr =DPRI dt

3.

435

dCSi dE dMpr  QCSi ðtÞ  ¼ SCvSi dt dt dt

½14:4

where W stands for the reactor volume, S for the surface area of the PRI, CvSi for the silicon mass concentration of the PRI and Q for the solution volume flow rate. A boron mass balance equation, O

dCB dðE þ eÞ  QCB ðtÞ ¼ SCvB dt dt

½14:5

where CB stands for the boron aqueous concentration and CvB for the boron mass concentration of the PRI. The key model parameters, Csat (silicon concentration at saturation in an equilibrium relation between PRI and aqueous solution), rdisso (PRI dissolution rate in pure water) and DPRI (water diffusion coefficient in PRI), are independent and can be measured. As an example, the simulations presented in Fig. 14.6 illustrate the good

© Woodhead Publishing Limited, 2010

436

Geological repository systems for safe disposal

14.6 Altered glass mass calculated by the model compared with experimentally measured values (350 samples from 31 experiments at 90 8C in pure water at different S/V and Q/S ratios).

agreement with the experimental data (Frugier et al., 2009). This type of model will allow simple coupling of glass alteration with reactivity and transport in the complex chemical environment of a geological repository. It is also suitable for describing any composition effect involving the solution chemistry or the glass composition. This model therefore allows consistent and extensive modelling of the existing set of experimental data and can be used for long-term modelling as it is based on a mechanistic understanding of the long-term glass behaviour. The V0 ® Vg operational model This performance assessment model was developed in 2004 to quantify the source term by calculating the quantity of altered glass (Ribet et al., 2004). It therefore postulates the very conservative hypothesis that all the radionuclides are released at the same rate as the glass alteration tracers, such as boron. The following additional hypotheses are assumed: .

As long as near-field materials remain reactive with respect to silicon from the glass matrix, the glass alteration rate is assumed equal to the maximum observed rate, i.e. the initial rate r0, which depends only on the temperature and pH. The reactive surface area during this alteration

© Woodhead Publishing Limited, 2010

Post-containment performance of geological repository systems

.

437

phase is only a fraction of the total consisting of the external (geometric) surface area (S0) and the surface area of large-aperture cracks. The cracking factor is designated t0. The quantity of glass altered per unit time during this phase is equal to the product of r06S06 t0. After the initial phase of rapid alteration due to chemical reactions in the immediate environment of the glass (Si sorption on metal canister corrosion products), the alteration kinetics reach a residual rate (rr) under conditions in which the water renewal rate is very low, as in the case of a geological repository. The reactive surface area comprises all the external surfaces and the crack surfaces. The cracking factor is designated tr. The quantity of glass altered per unit time during this phase is equal to the product of rr6S06tr.

The model parameters (alteration rates r0 and rr, cracking factors t0 and tr) were determined as a function of the temperature (between 25 and 100 8C), the pH (between 7 and 10) and the glass composition throughout the R7T7 composition range. The uncertainties on the parameter values were also determined. The model can be used to calculate the lifetime of the glass package from the time/temperature profile, the pH of the medium, the date of water ingress in contact with the glass and the quantity of accessible silicon sorption sites on the canister metal products. The graph illustrates the importance of the residual rate phase in determining the total package lifetime, and the need to better understand the mechanisms responsible for this rate regime (Fig. 14.7). Both models predict a very long lifetime for nuclear glass even though it will progressively be altered by the repository groundwater.

14.4

Long-term behaviour of spent nuclear fuel (SNF)

By opposition to nuclear glass, spent nuclear fuel was not designed and optimized for the very long term, but rather for optimum behaviour in the reactor and to produce electricity. Its long-term behaviour is therefore dependent on its prior evolution during irradiation in the reactor.

14.4.1

Fuel pellet condition after irradiation

This paragraph briefly synthesizes the physical and chemical condition of spent fuel pellets on removal from the reactor, focusing on PWR fuel. More details can be found in Johnson and Shoesmith (1988) and Poinssot et al. (2001).

© Woodhead Publishing Limited, 2010

438

Geological repository systems for safe disposal

14.7 Calculated glass lifetime plot for two hypotheses concerning the quantity of unsaturated sorption sites during the initial rate phase, assuming water ingress in contact with the glass after 4000 years. Dotted lines correspond to the lower and upper estimates of the model considering the parameters uncertainties (Ribet et al., 2004).

Physical condition The fuel pellet undergoes transformations in the reactor due to the effects of external irradiation and fission reactions. The main effects include: .

.

A significant temperature rise in the material (up to 1200 8C at the pellet core) accompanied by a radial temperature gradient (500–700 8C) between the pellet rim and core. The temperature gradient induces pellet cracking in the reactor. The formation of fission products and additional U and Pu isotopes and actinides through neutron capture. The production of rare gases, Xe and Kr in particular, which accumulate as bubbles inside and between the grains, modifies the mechanical properties of the fuel pellet.

The following changes occur in the physical state of the spent fuel pellet on removal from the reactor: . .

At macroscopic scale, the pellet is fragmented into about ten to fifteen pieces (Fig. 14.8). At microscopic scale, the pellet is embrittled at the grain boundaries by the accumulation of pressurized fission gas bubbles and metallic precipitates, and restructured zones of high porosity appear at the periphery of the pellet in UOX fuel (or in Pu-rich aggregates in MOX

© Woodhead Publishing Limited, 2010

Post-containment performance of geological repository systems

439

14.8 State of the fuel pellet and fuel microstructure after irradiation in the reactor (spent UO2 fuel, burnup: 60 GWd/t).

fuel) as a result of much higher local fission rates which lead to the restructuration of the grain (decrease of size and increase of porosity). Chemical composition Concerning the chemical composition of spent fuel removed from the reactor, the elements can be classified according to their chemical form: . . . .

elements soluble in the fluorine lattice of the matrix, including the lanthanides (La, Ce, Pr, Nd, etc.) and other rare earth elements, together with elements having soluble oxides (Zr, Nb, Sr); elements forming oxide precipitates: Rb, Cs, Ba, Zr, Nb, Mo, Te; fission products forming metallic precipitates: Mo, Tc, Ru, Rh, Pd, Ag, Cd, In, Sn, Sb, Te; fission gases (Kr, Xe, He), helium formed by alpha decay and volatile FP in the reactor (I, Br, Rb, Cs, Te).

The first class of elements accounts for about 95 wt% of the fuel (excluding oxygen). The oxygen potential is practically unmodified by the changes in the fuel chemical composition during its reactor residence time and thus remains near-stoichiometric after irradiation. Radionuclide location within the fuel rod The radionuclide distribution in the spent fuel rod on removal from the reactor depends on the mobility of the species in the pellet during irradiation. The most mobile elements are the rare gases, for which the fractions released into the free volumes in the rod exceed those of the other

© Woodhead Publishing Limited, 2010

440

Geological repository systems for safe disposal

14.9 Schematic representation of the nature and location of the main radionuclides in a UO2 fuel with a low burnup (no rim formation) (IAEA, 1991).

elements. A large intergranular fraction is also observed in the grain boundaries and pores on removal from the reactor: about 10% in UOX fuel irradiated to 60 GWd/t, and 35% in MOX fuel with the same burnup. Given their mobility in UO2, iodine and cesium are also released from the grains during in-pile irradiation, although the fraction found in the gap and grain boundaries is generally less than the fission gas release. The corresponding fractions for the other elements found as precipitates are extremely low. The location of the radionuclides in the rod is schematically represented in Fig. 14.9. When elements are sparingly soluble in UO2, such as the rare gases, most of the locally created fraction could be released in the pores during restructuring of the rim or the large Pu-rich aggregates. The corresponding fraction, which has been estimated for the rare gases, increases with the burnup in UOX fuel, and with the number of large Pu-rich aggregates in MOX fuel. The location of the radionuclides in the rod can vary over time depending on the intrinsic evolution of the fuel as the temperature and the a and b activity diminish (see Section 14.4.2).

© Woodhead Publishing Limited, 2010

Post-containment performance of geological repository systems

441

14.10 Residual activities in a UO2 fuel of 60 GWd/t versus time (Piron et al., 2000).

14.4.2

Evolution of the spent nuclear fuel in a closed system (before the container is breached)

Spent fuel is not in a state of equilibrium on removal from the reactor. It evolves mainly because of decay of the radionuclides it contains: irradiation damage accumulates, the chemical inventories are modified and significant quantities of helium are produced by a decay. Chemical evolution of spent fuel Spent fuel is highly radioactive on removal from the reactor; b decay predominates for about 200 years, during which the main emitters are 137Cs, 90 Sr and 241Pu. The subsequent activity arises mainly from a decay of the actinides (Fig. 14.10). The chemical composition of the fuel varies over time as the radioactive isotopes decay, although there is little change within each category of elements. The results show that the oxygen potential in the spent fuel is relatively unaffected by b decay and diminishing temperature, but is controlled by oxidation of Mo, one of the most abundant fission products, to MoO2 (Martin et al., 2004; Ferry et al., 2006b). The oxidation state of the fuel matrix and the chemical form of the elements other than molybdenum should thus not be subject to significant change as long as the SNF is not exposed to an oxidizing atmosphere.

© Woodhead Publishing Limited, 2010

442

Geological repository systems for safe disposal

14.11 Helium produced over time by alpha decay within UOX and MOX fuels (47.5 GWd/t) (Poinssot et al., 2000). Only part of it is expected to be released within the free volumes.

Helium behaviour Alpha decay produces large quantities of helium in the spent fuel, especially in fuel rich in actinides such as MOX fuel (Fig. 14.11). This raises the issue of the fate of the gas, which is produced mainly within the pellet grains, and of its consequences on the mechanical strength of the grain boundaries already embrittled by irradiation. Unlike the fission gases, helium is soluble in UO2 to some extent. Observations of flaking effects indicate a solubility limit below 1 at% evaluated between 0.1 and 0.5 at% (Guilbert et al., 2003, 2004). The presence of irradiation defects, particularly vacancies, in the spent fuel should increase the helium solubility. For example, the quantity of helium that infuses under irradiation appears to be considerably greater than the published helium solubility limits in UO2. The helium diffusion coefficient in UO2 is several orders of magnitude higher than that of xenon, but remains low below 400 8C: it varies from 1026 to 1023 m2/s for temperatures between 300 and 400 8C (Roudil et al., 2004, 2008). Thermal diffusion of helium is significantly affected by the presence of irradiation defects such as fission gas bubbles, which tend to trap helium and slow down its diffusion, and by damage arising from a selfirradiation, which appears to accelerate its diffusion. The study by Ronchi and Hiernaut (2004) indicates a diffusion coefficient ranging from 1024 to 1021 m2/s in (U0.9, Pu0.1)O2 for temperatures between 300 and 400 8C. The apparent activation energy of helium diffusion in the UO2 matrix is 2 eV (Martin et al., 2006), but a fraction of the helium inventory in the samples (5 to 20%) is released by intergranular diffusion (directly via the

© Woodhead Publishing Limited, 2010

Post-containment performance of geological repository systems

443

grain boundaries for nearby helium), as confirmed by comparisons of helium release in mono- and polycrystalline UO2 samples. This explains the fact that significant helium release is measured during vacuum annealing of MOX fuel at temperatures above 350 8C. This release cannot be attributed to diffusion in the UO2 grains; the helium released (in a vacuum) corresponds to a fraction of the helium already present at the grain boundaries on removal from the reactor. The helium diffusion coefficient at the grain boundaries is several orders of magnitude higher than in the grains. The fission gas diffusion coefficient accelerated by a self-irradiation based on the value measured in a reactor was initially estimated at about 1025 m2/s during the first hundred years after removal from the reactor (Poinssot et al., 2002). Although no reactor data are available for helium, the behaviour observed under thermal conditions suggests that the coefficient under a selfirradiation should be higher for helium than for the fission gases (Olander, 2004). Furthermore, recent observations on natural samples confirm that helium mobility is quite large at high defect concentrations (Roudil et al., 2008), but results are still lacking for intermediate concentrations such as those expected for SNF evolution in a repository. In the present state of knowledge regarding helium behaviour, various scenarios can still be considered (Ferry et al., 2005): (1) helium released from the grains diffuses via the grain boundaries to reach free volumes in the rod; (2) helium released from the grains accumulates at the grain boundaries in the form of bubbles or in pre-existing fission gas bubbles; (3) a fraction of the helium precipitates in the form of intragranular bubbles that should inhibit release from the grains. The consequences of these various scenarios on the SNF pellet microstructure, and particularly on the grain boundary stability, must still be assessed (Gras et al., 2007). Physical evolution Pellet swelling observed in the reactor is due to the cumulative irradiation damage (slight), to the production of fission products and, above 45 GWd/t, to the formation of fission gas bubbles. Once unloaded from the reactor, the cumulative effect of self-irradiation damage together with the changes in the chemical composition and helium production can also lead to swelling and modify the mechanical condition of the pellet. The irradiation damage generated in the UO2 matrix is due mainly to displacement cascades created by a radioactivity, especially by the recoil nucleus emitted with each disintegration. The cumulative effect of irradiation damage is limited in the UO2 matrix because of extensive recombination of the defects. The results have shown that (1) cumulative a self-irradiation damage does not lead to amorphization of the material,

© Woodhead Publishing Limited, 2010

444

Geological repository systems for safe disposal

14.12 Physical state of a 30-year-old PuO2 pellet (containing 90% 238Pu) stored in inert atmosphere (Gras et al., 2007). The fractures located at the grain boundaries are thought to be created by alpha decay helium release towards grain boundaries.

which conserves its cubic crystallographic structure; (2) the value of the lattice parameter remains constant for doses exceeding about 1 dpa; (3) this saturation value corresponds to a relative increase of about 0.4%, i.e. to a 1% volume increase (Ferry et al., 2006b). Following irradiation and a self-irradiation damage in the reactor, saturation of the lattice parameter should occur very rapidly after the fuel is unloaded from the reactor. No further microscopic swelling should be expected. Nevertheless, macroscopic swelling due to the formation of helium bubbles cannot be disregarded (Roudil et al., 2006). The study of consequences of the formation of intra- and intergranular helium bubbles on the mechanical strength of the grain boundaries, embrittled by irradiation in the reactor, is currently in progress. Although no evolution of grain boundaries is anticipated in dry storage and for the first 10 000 years, it is still difficult to be completely conclusive for all types of fuel for the very long term since helium release towards grain boundaries may increase the pre-existing bubble pressure and contribute to the destabilization of the grain boundaries (Fig. 14.12). The long-term fate of helium and the long-term stability of the grain boundaries are still key issues for long-term spent fuel performance as they may determine the surface area accessible to water (Gras et al., 2007).

© Woodhead Publishing Limited, 2010

Post-containment performance of geological repository systems

445

14.13 Radionuclide release mode from spent nuclear fuel (modified from Johnson et al., 1985). Initial rapid release (labile fraction) and intermediate release of grain boundaries are often collectively known as the instant release fraction (IRF).

14.4.3

Evolution of the spent fuel in the repository (after breaching of the container)

Radionuclide release from spent fuel in water can conceptually be broken down into two main fractions (Johnson and Shoesmith, 1988; Lutze and Ewing, 1988) (Fig. 14.13): .

.

The first involves the rapid release of the labile activity corresponding to the radionuclide inventory on the free surfaces of the fuel and the release of the grain boundary inventory over a period of several months to several years. This fraction is often referred to as the instant release fraction (or IRF). The second corresponds to the slow release of activity as the UO2 matrix dissolves and may last several thousand years. The release of most of the activity from the spent fuel rod, i.e. the activity contained in the grains, is thus controlled by the fuel dissolution kinetics.

The scientific challenges associated with each of these fractions are quite different: whereas the only issue for IRF is to assess the inventory, understanding the mechanisms governing the release of the matrix fraction requires significant R&D. The next section describes these two contributions.

© Woodhead Publishing Limited, 2010

446

Geological repository systems for safe disposal

14.14 Evolution of published IRF data for Cs as a function of time of publication and for different types of burnup and fuels. The inconsistency of the experimental data justifies a new approach based on a more mechanistic understanding of the IRF inventory (from Gras et al., 2007).

The instant release fraction The IRF was historically defined and characterized through short-term leaching experiments: it was assumed to be the inventory of radionuclides released in the first few hundred or thousand days of leaching, and for which the release rate significantly decreases with time (Johnson and Shoesmith, 1988). Until the early 2000s, most studies sought to collect new leaching results and attempted to derive empirical correlations with irradiationrelated parameters such as linear power or burnup. However, it was demonstrated that the consistency of the existing experimental data was very poor (Fig. 14.14) and that any empirical model based on experimental measurements will never be able to account for the potential spent fuel evolution prior to the water access (cf. previous paragraph) (Poinssot et al., 2003, 2004, 2006). In the framework of the European ‘Spent Fuel Stability under Repository Conditions’ project (SFS project; Poinssot et al., 2005b), the IRF concept was renewed and defined as the radionuclides inventory located within microstructures with low confinement properties: the gap interface, the rim pores and, optionally in the model, the rim grains and grain boundaries (Johnson et al., 2005). This model explicitly accounts for the potential evolution of the fuel microstructure and the potential influence of fuel burnup. The IRF at any given time is defined as the sum of the initial IRF determined on ‘freshly-irradiated spent fuel’ completed by any radionuclides subsequently diffusing towards one of the previous low-

© Woodhead Publishing Limited, 2010

Post-containment performance of geological repository systems

447

Table 14.1 Estimated IRF (% of total inventory) for various radionuclides in PWR UO2 fuel: best estimates, with the pessimistic estimate in parentheses (Ferry et al., 2007) Burnup (GWd/tU)

41

48

60

75

RN Fission gas 14 C 36 Cl 90 Sr 99 Tc, 107Pd 129 135 I, Cs, 137Cs

IRF 1 (2) 10 5 1 (2) 0.1 (1) 1 (3)

IRF 2 (4) 10 10 1 (3) 0.1 (3) 2 (4)

IRF 4 (8) 10 16 1 (5) 0.1 (5) 4 (8)

IRF 8 (16) 10 26 1 (9) 0.1 (9) 8 (16)

confinement microstructures. Recent results demonstrate that alpha selfirradiation enhanced diffusion is not significant enough to result in a quantitative migration of radionuclides (Ferry et al., 2005, 2006b). Ferry et al. (2006a; 2007) developed a micromechanical model which also demonstrates that helium ingrowth within the irradiated UOX pellet should not lead to any grain boundary cracking in the rim zone, contrary to the initial estimates. An updated and less conservative IRF assessment was therefore recently published which significantly decreases the previous estimated IRF (Ferry et al., 2007) (Table 14.1). Although the IRF is currently well defined, the remaining uncertainties must be addressed: .

.

.

First, the initial radionuclide distribution between the various microstructures is not fully understood, in particular for high burnup and MOX fuels. This is particularly the case for highly mobile elements such as 36Cl, for which recent studies seem to indicate high mobility in the reactor (Pipon et al., 2007). Since these mobile elements are also highly mobile in a geological environment and responsible for most of the repository impact, it is of prime importance to better understand their in-reactor behaviour in the fuel pellet and their subsequent distribution after irradiation (Roudil et al., 2007). Second, the behaviour of the rim zone is still poorly understood. Despite the large pore size and the small grains, its resistance to alteration seems to be much higher than the central part of the rod (Grambow et al., 2008). This may be related to the high fission product content of this zone as for oxidation. In any case, this result should be confirmed to conclude definitely that the rim does not significantly contribute to the IRF. This is potentially of great importance since a significant part of the radionuclide inventory is located in the rim zone. Finally, the micromechanical model, which demonstrates the long-term stability of the grain boundary (in bulk and rim), should be confirmed

© Woodhead Publishing Limited, 2010

448

Geological repository systems for safe disposal (1) by new independent experimental validation of the most sensitive parameters, including the critical bubble pressure and the stress intensity factor, and (2) by a better micromechanical understanding of spent nuclear fuel.

Basically, the level of confidence of IRF in geological disposal for irradiated PWR fuel has been greatly enhanced by studies within the last decade. The figures are below 10% for every radionuclide except 36Cl due to its enhanced in-reactor mobility. Although it represents only a fraction of the inventory, further work is necessary because it represents the main source of repository impact. Lack of data on MOX fuels also leads to very conservative figures. R&D must be pursued on this type of fuel either to confirm these figures or to decrease the conservatism to reach more realistic figures. Fuel matrix release Although uranium is sparingly soluble under reducing conditions similar to those encountered in a repository site (see next section), its solubility can increase significantly at the UO2/water interface because of the abg irradiation field. Indeed, water radiolysis produces both oxidizing and reducing primary species as radicals (OH·, HO2·, eaq, H·) or in molecular form (H3O+, H2, H2O2 and, by recombination, O2) at concentrations that depend on the nature of the radiation and on the dose deposited in the water. Radiolysis can therefore lead to the onset of oxidizing conditions at the UO2/water interface (redox disequilibrium with the environment) and accelerate the dissolution of the spent fuel matrix under disposal conditions (Shoesmith, 2000; Sattonay et al., 2001; Je´gou et al., 2005a, 2005b). The bg radioactivity that predominates during the first two hundred years subsequently gives way to a radioactivity (cf. Fig. 14.10). Radiolytic dissolution Under reducing conditions as in the main repository sites, alpha radiolysis of water was long assumed to be the governing alteration process by locally producing significant amounts of radiolytic oxidants (for instance Christensen and Bjergbakke, 1987; Christensen et al., 1994; Corbel et al., 2006; Muzeau et al., 2009) (Fig. 14.15). Alpha particles therefore lead to the production of radiolytic oxidants (step 1, generation of oxidants) which oxidize U(IV) at the fuel surface to U(VI) (step 2, oxidation). U(VI) is then solubilized and complexed by aqueous ligands (step 4, dissolution) before potentially reprecipitating (step 5, secondary phases). Several types of models have been developed to simulate this simplified reaction scheme and relate the fuel alteration rate to the residual alpha activity:

© Woodhead Publishing Limited, 2010

Post-containment performance of geological repository systems

449

14.15 Processes governing matrix radiolytic dissolution in a reducing environment (Poinssot and Gras, 2008).

. .

.

Electrochemical models correlate the corrosion potential with the corrosion rate and focus on the oxidation step (e.g. Shoesmith, 2000; Shoesmith et al., 2003; Sunder et al., 2004; Wren et al., 2005). Surface chemistry models that focus on the kinetics of surface complex formation succeed in rationalizing most of the available data on unirradiated UO2 but fail to reproduce radiolytic dissolution at low carbonate and oxygen content (De Pablo et al., 1999). Radiolytic models attempt to detail all the chemical reactions occurring in the different zones at the fuel/water interface: production of oxidants considering the radiolytic primary yield, the recombination of primary radiolytic species, their diffusion around the interface, etc.

Beyond these detailed models, more performance-oriented models have been developed in recent years. A simplified performance model developed by the French team (CEA) only accounts for primary radiolytic yields and mass balance calculations (Poinssot et al., 2005a). It gives a reasonable upper estimate of the available experimental alteration rates and is very robust due to the small number of parameters. Another approach was developed by the Spanish team (referred to as the Matrix Alteration Model, MAM; Poinssot et al., 2005b; Quinones et al., 2005, 2009) and basically couples an extended radiolytic model (Christensen et al., 1994; Christensen, 1998) with a surface complexation approach and dissolution reactions. This model is specific to a given geometry and was calibrated on data obtained on unirradiated samples. It successfully reproduced the anticipated decrease of

© Woodhead Publishing Limited, 2010

450

Geological repository systems for safe disposal

14.16 Spent fuel alteration rates versus specific alpha activity. Results are selected both from the literature and the SFS project and are obtained both on 233U- and 238Pu-doped samples and 225Ac-doped colloids (Poinssot et al., 2005a).

alteration rate with time but does not overestimate the complete set of experimental data. Both of these models have been conservatively used for several performance assessment studies in France (ANDRA), in Spain (ENRESA) or in Sweden (SKB), yielding fuel lifetimes compatible with the required performance (e.g. ANDRA, 2006; de Windt et al., 2006). Transition to corrosion and chemical dissolution, environmental coupling Radiolytic dissolution has recently been demonstrated to be effective only above a dose threshold (e.g. Poinssot et al., 2005b) (Fig. 14.16). This effective dose threshold is a purely empirical measurement of the competition between radiolytic oxidant production and subsequent matrix alteration, on one side, and the consumption of oxidants by the other aqueous ions, on the other side (step 3 in Fig. 14.15). It has been estimated to be in the range 18–33 MBq/g, which corresponds to ages between 3500 and 55 000 years depending on the fuel types: between 4500 and 15 000 years for a 60 GWd/t UOX fuel and between 42 000 and 55 000 years for a 60 GWd/t MOX fuel (Poinssot et al., 2006; Poinssot and Gras, 2008). For activity lower than this dose threshold, which means times longer than 15 000 years for UOX and 50 000 years for MOX, spent nuclear fuel is thought to alter by corrosion (electrochemical processes) and, when ECORR is lower than EOX, by pure chemical dissolution (controlled by U(IV) aqueous solubility). In addition, hydrogen activation has been demonstrated under the

© Woodhead Publishing Limited, 2010

Post-containment performance of geological repository systems

451

conditions expected at the fuel/water interface, and it behaves as a strongly reducing species (Spahiu et al., 2000, 2004). This behaviour is very surprising at low temperature and would probably mean that a catalyst is active in the system, the origin of which is not yet fully understood. Several explanations have been proposed, among which is the potential reactivity of the epsilon particles. These submicrometric particles correspond to metallic alloy precipitates (Tc, Ro, Pd, Ru, etc.) dispersed within the fuel pellet. These particles are poorly known and attempts have been made to separate them accurately and characterize their reactivity (Wronkiewicz et al., 2001). Another explanation may be the direct reactivity of hydrogen at the UO2 surface. Over the long term under reducing conditions such as those prevailing in a deep repository, the amount of reducing species is expected to be high enough to counteract completely the production of oxidants and therefore inhibit radiolytic dissolution. However, a more mechanistic understanding of the different processes occurring at the fuel/water interface, in particular the redox balance, is still required to rule out definitely the potential occurrence of radiolytic dissolution. Once radiolytic dissolution has been suppressed, other processes will govern fuel alteration: (1) at intermediate potentials, alteration would proceed by fuel corrosion, whereas (2) in a more strongly reducing environment, purely chemical alteration is expected. Predicting the fuel lifetime in such conditions would therefore require the development of an electrokinetic model describing the oxidation/reduction reactions at the fuel/ water interfaces coupled with the near-field environment. Considerable progress has been made by several teams (e.g. Grambow et al., 2004; Brocskowski et al., 2006) in this direction to account for the large influence of container corrosion. These models must be completed at the scientific level to account for the complexity of the natural environment, and they must also be simplified to derive more operational models capable of predicting the fuel alteration rate with a reasonable degree of confidence and a small number of parameters. Under these conditions, any reaction influencing either the redox potential or the uranium concentration may influence the fuel alteration rate, in particular the potential secondary phases. Secondary phases can also affect the hydrodynamics (pore clogging) and radionuclides behaviour (sorption). Among the various potential secondary phases, the U(IV) silicate USiO4· (H2O), coffinite, is a serious candidate, as has been observed in the Oklo natural reactor and at many uranium ore mining sites (Gurban et al., 2003). Recent studies should produce more reliable thermodynamic data and better assessment of the potential for coffinite precipitation (Pointeau et al., 2009). Finally, the effective surface area of the fuel/water interface will also significantly influence the fuel alteration rate. This parameter is poorly

© Woodhead Publishing Limited, 2010

452

Geological repository systems for safe disposal

understood since it is directly related to the fuel microstructure after irradiation (in particular the first cycles of irradiation when the temperature gradient is still high) and the fuel/cladding/canister interface in the repository. It still represents a major uncertainty regarding spent fuel long-term alteration as it directly influences the mass of spent fuel altered (Poinssot and Gras, 2008). The evolution of spent fuel in a geological repository has been much better understood since the late 1990s and the current level of knowledge is sufficient conservatively to model the radionuclide release from the fuel in geological repository conditions. However, a suitable mechanistic understanding is still lacking to support the expected overall fuel performance and to extend the current performance model by accounting for all the coupling phenomena occurring in the near-field environment (hydrogen, redox balance, secondary phases, etc.).

14.5 Brief overview of low- and intermediate-level waste (L/ILW) performance 14.5.1

Cement and concrete

Cementitious binder materials (grout, mortar, concrete) are widely used for conditioning and disposal of low- and intermediate-level waste: as a waste encapsulation matrix, as grout for immobilizing waste in drums or as the waste container itself (Neilson and Dole, 1986; Frizon and Cau-dit-Coumes, 2006). These materials are thus subjected to a wide range of physical, chemical and radiological environments (Glasser, 2002). Waste conditioned in a cement matrix may be classified into two broad categories (Fig. 14.17): . .

heterogeneous waste, especially miscellaneous compacted or bulk technological waste, encapsulated in a cementitious binder inside concrete or steel containers; homogeneous waste, especially sludge and concentrates, encapsulated in a cementitious binder inside concrete or steel containers.

Portland cement is the most commonly used variety for waste encapsulation. In contact with water it forms hydrated calcium silicates (70%), calcium hydroxide or portlandite (20%), hydrated calcium aluminates and sulphoaluminates (ettringite and monosulphoaluminate) (10%). Hardened cement paste constitutes a heterogeneous porous solid; the interstitial solution is highly basic (pH between 12.5 and 13.6) and its composition varies with the age of the material. For economic reasons and to improve the cement properties, a fraction of the clinker may be replaced by industrial by-products such as silica fume, fly-ash or blast-furnace slag. More details dealing with the manufacturing of cement and concrete can be found in

© Woodhead Publishing Limited, 2010

Post-containment performance of geological repository systems

453

14.17 Examples of B-type waste (intermediate-level waste): (top) heterogeneous technological waste embedded in cement paste and (bottom) homogeneous ionic exchanger resins embedded in cement paste.

Baron and Ollivier (1996), Glasser (1997) and Taylor (1997). Some new binders are then currently under development for specific applications: sulphoaluminate cements (Cau-dit-Coumes et al., 2009; Berger et al., 2009a) or geopolymers (Berger et al., 2009b) could be used in the future for low- or

© Woodhead Publishing Limited, 2010

454

Geological repository systems for safe disposal

medium-level wastes, for which commonly used Portland cement does not give the expected performance. During interim storage and the first phase of the repository operation, under normal or altered conditions (‘closed system’ or ‘water-unsaturated open system’), the following basic phenomenology must be taken into account (Bary and Sellier, 2004; Galle´ et al., 2006): . .

Cracking of the concrete as a result of its physical and chemical evolution. For the package itself, this risk can be mitigated by the use of suitable concrete formulations and fibre or structural reinforcement. Interactions between radiation and the cement matrix in the case of irradiating waste. This phenomenon results in radiolysis of the free pore water and hydrogen production at a rate that varies with the type of irradiation (abg), the dose rate, the degree of pore saturation with water and the presence of solutes (NO3, S2, organic matter).

In a nuclear waste repository over the long term (i.e. in an ‘open watersaturated’ system), the major phenomenon affecting the behaviour of cement materials is chemical degradation in water (hydrolysis and decalcification), which is highly dependent on the abundance of sulphate and carbonate ions (Atkins et al., 1994; Adenot and Richet, 1995; Adenot et al., 1997; Mainguy et al., 2000; Kamali et al., 2003). Concrete degradation by a neutral or basic solution (pH < 11.5) results in decalcification of the minerals, especially the calcium silica hydrate (CSH), controlled by diffusion; a local equilibrium occurs between the solid and its interstitial solution. The phenomenology can be simplified by considering the sole Ca diffusion. Based on this assumption, The DIFFU-CA model allows cement paste alteration to be successfully described as a function of time by solving the following diffusion equation (Mainguy et al., 2000): @ðo½CaÞ @SCa ¼ DivðDe grad½CaÞ  @t @t

½14:6

where [Ca] stands for the Ca concentration, w the porosity, De the effective diffusion coefficient and SCa the Ca concentration in the solid phase. The first term on the right-hand side stands for the diffusion process of the calcium in the liquid phase, which is assumed to be governed by Fick’s law. The second term on the right-hand side of Equation [14.6] accounts for the dissolution process, which leads to an arrival of calcium in the liquid phase. The degradation is therefore proportional to the square root of the time as long as there is an unaltered zone and the aggressive solution composition is constant. The main consequences on conventional materials are increased porosity (by about a factor of 2) and a higher diffusion coefficient (by a factor of 5 to 20) in the decalcified zone, and thus diminished containment

© Woodhead Publishing Limited, 2010

Post-containment performance of geological repository systems

455

properties. Various models have been developed to describe the coupling between chemical degradation of the material, radionuclide migration within the material and the mechanical strength of the waste package (Peycelon et al., 2006). These models are capable of predicting the evolution of radionuclide confinement in the event of external alteration of a concrete container by slightly ionized water containing carbonates, and with allowance for cracking.

14.5.2

Bitumen

Bituminization is used to encapsulate coprecipitation sludge containing either insolubles arising from liquid effluent treatment or evaporator concentrates from spent fuel reprocessing (CEA, 1983; IAEA, 1993). The first studies of bitumen for use as a nuclear waste conditioning medium were carried out in 1958 at Oak Ridge in the USA. The same year this idea was examined by the CEN/SCK in Belgium, where this process was first developed in a pilot facility in the 1960s. In 1964 a bituminization batch process was implemented at Mol and in 1966 an industrial effluent bituminization process was commissioned at Marcoule (France) (Dejonghe, 1999; Meeus and Luycx, 1999; Redonnet et al., 1999). Although bituminization is now being replaced by cement encapsulation or vitrification, the number of bituminized waste drums already produced throughout the world (70 000 packages in France alone) warrants a short discussion. A bituminized sludge package consists of a solid salt mixture dispersed in a bitumen matrix (Camaro et al., 1999). Bitumen is a mixture of heavy hydrocarbons, consisting mainly of aliphatic, naphthenic and aromatic compounds. It is produced industrially by vacuum distillation of crude oil. From a chemical standpoint, the encapsulated waste consists mainly of salts that are insoluble in water (barium sulphate, ferrocyanide, cobalt sulphide) and soluble salts (sodium nitrate, sodium sulphate). The main factors affecting the bituminized waste package are irradiation (in a closed system) and water and possibly microorganisms (in an open system). The bitumen matrix is sensitive to the effects of self-irradiation (Chaix et al., 2001a). Because of its organic composition it emits radiolysis gases – mainly hydrogen – via a radical mechanism affecting CH bonds. Depending on the activity it contains, a freshly produced package generates between 1 and 10 L of radiolysis gases a year. Because of radioactive decay the gas production becomes negligible after a thousand years. The cumulative gas release over a thousand years is about a cubic metre per package. The gas produced throughout the payload material is first solubilized in the matrix up to saturation (about 1 vol%), after which the hydrogen forms gas bubbles whose growth can lead to swelling (Riglet-Martial et al., 2005).

© Woodhead Publishing Limited, 2010

456

Geological repository systems for safe disposal

This volume expansion must be taken into account at the package design stage by providing an apical void, for example; however, it does not affect the radioactive confinement properties of the bitumen. Another way to control the volume expansion of the encapsulated material is to include a hydrogen-trapping salt among the bituminized salt contents. A salt capable of binding radiolytic hydrogen is obtained by precipitation of cobalt sulphide, which reacts chemically with hydrogen and ultimately forms water (Camaro et al., 2003). After disposal, the bitumen will eventually be leached by water. Although pure bitumen is almost impermeable to water, the presence of salts favours water absorption by diffusion and osmosis. Water diffusion in the bitumen matrix occurs at a very slow rate, but is the fundamental source of salt release from bitumen-encapsulated waste. Water absorption and release of the most soluble salts follows kinetic laws based on the square root of time. Models based on Fick’s laws have been developed to describe the phenomena from which the lifetime of a bituminized waste package can be predicted (Sercombe et al., 2004, 2006; Gwinner et al., 2006). The time necessary for solubilization of all the NaNO3 in encapsulated waste with a soluble salt content of about 15% is estimated at about 26105 years. For actual bituminized waste the radionuclide release kinetics are lower by 2 to 4 orders of magnitude than for a tracer salt such as NaNO3.

14.5.3

Compacted hulls and end-pieces

Standard compacted waste packages enclose metal canisters containing stacks of pucks (compacted drums) filled with zirconium alloy (Zircaloy, etc.) hulls and stainless steel or alloy nickel structural materials arising from spent fuel treatment (Fig. 14.18). The packages are heterogeneous and contain no immobilization materials. The container is made of austenitic stainless steel. The packages have an initial bg activity of a few hundred TBq and also contain alpha emitters. They contain three types of radionuclides: . . .

activation products arising from impurities in the structural materials, distributed throughout the thickness of the metal parts; fission products: about 0.2% of the fission products contained in the spent fuel are found in the cladding and in the zirconia layer; actinides: about 0.03% of the actinides contained in the spent fuel are found in the clads or zirconia.

The two last categories have mainly been implanted in the reactor by recoil during the fission reactions or the actinide alpha decay. The confinement performance of the metallic materials is therefore dependent (1) on the implantation depth of the various radionuclides and (2) the durability of the

© Woodhead Publishing Limited, 2010

Post-containment performance of geological repository systems

457

14.18 Hulls and end-pieces are compacted to create slices that are subsequently inserted into welded drums.

materials carrying radionuclides within the package (metal, oxide, etc.). In the disposal concept under consideration in France, the packages will be placed in concrete structures; the expected long-term environment will thus be an alkaline medium. During roughly the first 300 years the waste package and its confinement properties will remain intact. Any subsequent release of radionuclides will depend on their location in the package. From a safety standpoint, any radionuclides that are not loaded in a solid material – or that are incorporated in a material whose alteration is not known in sufficient detail or occurs too rapidly – are considered labile, i.e. capable of immediately entering solution at the moment of water contact with the package (Chaix et al., 2001b). Their release therefore depends only on their physicochemical state and on the external conditions (groundwater composition and renewal). This is the case for the radionuclides present in the fines and in zirconia. Conversely, the release of radionuclides loaded in durable bulk materials is controlled by the degradation of the materials themselves, which serve as a containment matrix; this degradation depends on the external environment. Only zirconium alloy, stainless steel and nickel alloy are assumed to have

© Woodhead Publishing Limited, 2010

458

Geological repository systems for safe disposal

a potential immobilization role in the operational model currently selected by ANDRA, making it a conservative model. The generalized corrosion rates of these materials expected under repository conditions are very low (≤ 10 nm/year). Slightly different values can be used to test more or less conservative conditions. Typical confinement times are about 10 000 years for Zircaloy, 100 000 years for stainless steel end-pieces and 1000 years for small nickel alloy springs.

14.6 Conclusion on the relative performance of the different waste forms A reliable and scientifically sound approach has been defined in the last 15 years to deal with the issue of predicting the long-term performance of nuclear waste forms. The research performed in this area is often at the cutting edge of the science of multi-physics/multi-scale issues, capable of meeting the challenge of predicting hundreds of thousands of years of evolution in systems where the characteristic time of some key processes is only a few nanoseconds (radiolysis for instance). From the preceding overview, it is clear that nuclear waste forms are well suited for the radionuclides they are confining. Relatively long-term performance has been demonstrated for nuclear waste glass and the spent fuel matrix, which contain more than 99% of the total waste radioactivity. However, spent nuclear fuel was not designed for long-term performance and is penalized by its instant release fraction, which rapidly releases a significant part of the total radionuclide inventory into the near field, in particular the mobile radionuclides 36Cl and 129I (see the next section). Conversely, ILW waste forms have a much shorter lifetime, which is compatible with their lower radionuclide content and the shorter half-life of most of them.

14.7 Radionuclide fate after release Previous chapters focus on the long-term evolution of waste forms under intrinsic driving forces or in interaction with their environment. Once the canister has been breached, this long-term evolution gives way to the progressive release of radionuclides into the near-field environment and subsequently into the geological far field. Once released, the radionuclides will undergo several processes that will determine their mobility and therefore the potential long-term impact of any repository. This chapter synthetically describes these processes with emphasis on the governing mechanisms. The fate of radionuclides in the environment will be governed by two main types of processes that occur simultaneously and interact together (Fig. 14.19):

© Woodhead Publishing Limited, 2010

Post-containment performance of geological repository systems

459

14.19 Schematic representation of processes governing radionuclide mobility and migration in the geosphere after their release from the waste form.

.

.

The chemical reactivity of the radionuclides in the liquid phase and at the solid/liquid interfaces. Two main issues have to be addressed: ○ First, determine the chemical speciation of the radionuclides, which means determining which species will actually be present for each radionuclide (aqueous complexes, redox state, etc.). This is important since it determines the overall reactivity and mobility of the radionuclides. ○ Second, determine the amplitude of the physical–chemical interactions of the radionuclides with the surface of the surrounding materials. This process, referred to as retention or sorption, can concern most of the radionuclide inventory and therefore also contributes to lowering their availability. The transport processes affecting the element distribution within the dispersed phase of the given system or at the interface between phases (e.g. surface or bulk diffusion). Several transport regimes may be encountered, basically dominated by advection or diffusion, although some secondorder processes can also be significant in very specific cases (the so-called off-diagonal processes such as diffusion driven by a thermal gradient).

The following sections describe in greater detail the main mechanisms governing each of the previous processes and their relative significance in the specific case of a geological repository.

© Woodhead Publishing Limited, 2010

460

Geological repository systems for safe disposal

Table 14.2 Composition of Callovo-Oxfordian argillite porewater (Leroy et al., 2007), Boom Clay porewater (Beaucaire et al., 2000) and Yucca Mountain site groundwater (Rosenberg et al., 2001). All concentrations are in mmol/l Callovo-Oxfordian argillites (mmol/l)

Boom Clay porewater (mmol/l)

Yucca Mountain tuff (J13) (mmol/l)

pH Eh (mV)

7.3 180 to 150

8.2–8.6 240

7.41 +430

Na+ K+ Mg2+ Ca2+ SiO2 HCO3 SO42 Cl NO3 F

32 7 14 15 — 1.2 34 30 — —

12.6 0.2 0.1 0.05 0.15 — 0.04 0.5 — —

1.99 0.13 0.08 0.33 1.02 2.11 0.19 0.20 0.14 0.11

14.8 Role of aqueous chemical processes in defining the relevant aqueous chemical species The first chemical reactions that the radionuclides will undergo after their release from the waste form involve water molecules and other chemical species that are present in the porewater. Natural porewater is mainly dominated by alkali and alkaline-earth elements, by carbonates or silicates (Bradbury and Baeyens, 1998; Rosenberg et al., 2001; Leroy et al., 2007; Beaucaire et al., 2008). In a deep environment the pH is usually not far from neutrality (in the range 7–8) whereas the Eh is generally negative (reducing conditions). The geologic situation of the Yucca Mountain site is specific since it is located in an unsaturated oxidizing environment. Typical deep porewater compositions are indicated in Table 14.2.

14.8.1

Hydrolysis reactions

The first reactions to consider are interactions with the water molecules. The charge, size and electronegativity of the radionuclides will determine whether they will only be hydrated (surrounded by water molecules) or if they will destabilize the water molecules and form more complex species. This behaviour can easily be evaluated by considering the Z2/R ratio, where Z stands for the charge and R the ionic radius (Toulhoat, 2002; Toulhoat et al., 2005): .

For very low Z2/R values, radionuclides will be mainly in the aquo-

© Woodhead Publishing Limited, 2010

Post-containment performance of geological repository systems

.

.

461

form, simply surrounded by a layer of undissociated water molecules. This is typically the case for Cs+ or Sr2+. For high Z2/R values, radionuclides will link to several oxygen atoms and therefore decrease the apparent charge and the interaction with water molecules. This is specifically the case for hexavalent actinides, which are present in solution as UO22+, PuO22+, NpO22+ or for fission products such as Tc or Se, which are present as oxoanions: TcO4–, SeO4–. For intermediate Z2/R values, radionuclides will significantly destabilize the water molecules and bind with one or more hydroxides. Increasing the number of hydroxides therefore decreases the charge of the aqueous species towards neutrality, initially favouring the formation of colloidal aggregates and then for the longer term, the precipitation of mineral hydroxides (Neck et al., 2007; Geckeis and Rabung, 2008). This behaviour is typical for metals such as Co2+, Ni2+, Pd2+, Sn4+ or Zr4+, or for tetravalent actinides U4+, Np4+, Pu4+. It explains the very low solubility of tetravalent actinides.

Figure 14.20 illustrates these different behaviours.

14.8.2

Redox reactions

Many radionuclides exist at different redox states and can therefore be oxidized or reduced depending on the chemical conditions. These reactions are referred to as redox reactions and correspond to the exchange of electrons as follows: 1 1 oxidized þ e , reduced n n

½14:7

with an equilibrium constant KR: KR ¼

Ox1=n ðe Þ ðRedÞ1=n

½14:8

The deep geological environment is usually strongly reducing, the oxygen being trapped in the upper layer of the earth’s crust and internal layers being depleted in oxygen (Stumm and Morgan, 1996). Redox-sensitive species will therefore preferentially be in their most reduced states, especially the actinides (Geckeis and Rabung, 2008; Hu et al., 2008). The Yucca Mountain site in the USA has quite different properties (Table 14.2). The unsaturated pore spaces and the presence of a gas phase result in oxidizing conditions and different actinide chemistry compared with a clay or granite environment as in Europe (Runde et al., 2002). Figure 14.21 details the predicted chemistry of neptunium as a function of pH and redox conditions (represented here as a

© Woodhead Publishing Limited, 2010

462

Geological repository systems for safe disposal

function of the potential E) considering the redox, hydrolysis and carbonates complexation reactions (Vitorge and Poinssot, 2005). It is noteworthy that redox reactions may be quite slow and that equilibrium may not be reached. Kinetics are therefore also to be accounted for and direct thermodynamic modelling may not be representative of what will actually happen (Stumm and Morgan, 1996).

14.8.3

Aqueous complexation

The main chemical reactions occurring in the aqueous phase are complexation, which can be written as follows: RNnþ þ nL , RNLn

½14:9

14.20 Aqueous form of selected radionuclides as a function of their charge (Z), radius (r) and electronegativity (Pauling scale) (modified from Toulhoat et el., 2007).

© Woodhead Publishing Limited, 2010

Post-containment performance of geological repository systems

463

14.21 Eh–pH diagram of neptunium describing the dominant aqueous species as a function of pH and redox potential (E is measured by comparison with hydrogen-saturated electrodes) (Vitorge and Poinssot, 2005). The upper grey field, white field, lower grey field and dashed field correspond to the respective stability fields of hexavalent, pentavalent, tetravalent and trivalent Np. The open circle describes the chemical conditions measured in the French underground research laboratory (Meuse/Haute-Marne) and the full circle corresponds to the chemical conditions measured at Yucca Mountain. The dominant Np aqueous species is therefore different in both cases, which mainly explains the difference of Np mobility in both cases.

with a complexation constant b¼

ðRNLn Þ ðRNnþ ÞðL Þn

½14:10

where ( ) stands for the aqueous activity of the species. Most of the radionuclides are cationic and therefore will mainly be sensitive to the presence of anions. Among others, carbonates are of major importance since they are predominant in many surface environments and

© Woodhead Publishing Limited, 2010

464

Geological repository systems for safe disposal

14.22 Eh–pH diagram for Pu in groundwater at 25 8C. Conditions are representative of J-13 water found near the proposed Yucca Mountain nuclear waste repository: [Pu]tot = 105 M, [CO3]tot = 102.6 M, P = 1 atm (from Reilly et al., 2007).

are highly reactive. The importance of carbonate complexes increases with increasing pH and under repository conditions, in particular for the actinides (Runde et al., 2002; Reilly et al., 2007). Figure 14.22 presents the speciation diagram of Pu in representative conditions of the Yucca Mountain site (Reilly et al., 2007) and evidences the large number of carbonate complexes for higher pH ranges. Other ligands of interest are fluorides and phosphates, which can be encountered in specific environment and are highly attractive for many radionuclides. Silicates must also be taken into account. A major obstacle is the current lack of knowledge concerning mixed complexes. Finally, natural organic matter (NOM) such as humic or fulvic substances can also bind to actinides and modify their speciation. This issue is more significant in surface or near-surface environments and has been the subject of numerous studies (e.g. Choppin, 1992, 2006; Artinger et al., 2003; Geckeis and Rabung, 2008; Reiller et al., 2008). In particular, NOM can occur as natural colloids and therefore contribute to the general mobility of some radionuclides such as actinides (see below).

© Woodhead Publishing Limited, 2010

Post-containment performance of geological repository systems

14.8.4

465

Precipitation and coprecipitation

Interaction with other aqueous species may also lead to the precipitation of newly formed phases. The reaction can be written as follows: bRNnþ þ nBb , ðRNb Bn Þs

½14:11

Precipitation of solid phases is described by the ionic product Qs, which is defined as n b QS ¼ ðRNnþ Þ Bb ½14:12 If Qs > Ks + e, the solid phase precipitates (Ks stands for the solubility product and e represents the uncertainty on the thermodynamic data). In a repository, most of the radionuclides will be released at very low concentrations and are therefore not expected to precipitate in a solid phase. The main exceptions are the actinides, the solubility of which can be very low, in particular in reducing environments (Johannesson et al., 1995; Neck et al., 2007). They will therefore not be present at significant concentrations in solution and this will therefore decrease their mobility. This point is clearly illustrated by Fig. 14.23, in which the neptunium solubility is plotted versus the redox potential (Vitorge and Poinssot, 2005). It can easily be seen that although neptunium has very low solubility in a reducing environment, such as those encountered in European repositories, its solubility is much higher in oxidizing conditions, such as those measured at the Yucca Mountain site (Runde et al., 2002). Neptunium will therefore be more mobile in oxidizing conditions than in a reducing environment. Similar conclusions can be drawn for actinides that will be present at a lower oxidation state in a deep reducing environment and will have a very low solubility, in the range 109–1010 mol/l (Runde et al., 2002; Neck et al., 2007).

14.23 Evolution of neptunium solubility as a function of the redox potential by accounting for the hydrolysis and carbonate complexation reactions. This figure clearly shows that the solubility increases from 109.5 to 106 M when the redox potential increases from 0 to 0.3 E/ESH (adapted from Vitorge and Poinssot, 2005).

© Woodhead Publishing Limited, 2010

466

Geological repository systems for safe disposal

Furthermore, radionuclides will be present at trace concentrations in the surrounding environment of a repository. Under these conditions, they can also coprecipitate with other major species and be incorporated as traces in solid solution minerals. This process is ubiquitous and leads to lower solubility than those of pure end-members as calculated previously (Bruno et al., 2007; Neck et al., 2007). However, thermodynamic data are still lacking to correctly model such processes and they are often neglected in migration simulations (which is obviously conservative). Finally, most of these chemical reactions can be modelled by equilibrium thermodynamics. Implementing reliable and scientifically sound thermodynamic databases is therefore of prime importance (Voigt et al., 2007). Significant effort has been devoted to this area in recent decades. Beyond the national thermodynamic databases developed for each national repository project, special mention is required for the OECD-NEA International Thermochemical DataBase (TDB) project, which aims to define reference thermodynamic data that can be reliably used to model radionuclides behaviour in a repository (cf. http://www.nea.fr/html/dbtdb/info/publications/welcome.html; Grenthe et al., 1992; Silva et al., 1995, Rard et al., 1999; Lemire et al., 2001; Guillaumont et al., 2003; Brown et al., 2005; Gamsja¨ger et al., 2005; Olin et al., 2005; Bruno et al., 2007; Hummel et al., 2007; Rande et al., 2009). In any case, users must keep in mind that databases only collect existing information and cannot provide an exhaustive picture of the relevant chemical reactions to consider when performing geochemical modelling. Uncertainties on experimental and theoretical data must therefore also be taken into account and lead to ranges of potential solubilities (Chen and Pearson, 2008).

14.9 Significance of retention processes as a net retardation effect Retention is often measured by the distribution coefficient Kd, which is a measure of the proportion of radionuclides sorbed: ðRNsorbed Þðmol/kgÞ  Kd (L/kg) ¼  RNeq aq ðmol/LÞ

½14:13

Kd is dependent on the chemical conditions and therefore cannot be directly extrapolated to other conditions than those measured. Classical experimental characterizations usually include batch measurements of Kd = f(pH) (the pH-edge) and Kd ¼ fðRNeq aq Þ (the sorption isotherm) (see, for example, Poinssot et al., 1999a; Bradbury and Baeyens, 2005).

© Woodhead Publishing Limited, 2010

Post-containment performance of geological repository systems

467

14.24 Evolution of Cs, Np, Am and Se sorption on a montmorillonite clay mineral versus pH for a constant NaClO4 background electrolyte concentration of 0.1 M. Sorption of cations increases as a function of pH due to deprotonation of surface sites whereas anion sorption (here SeO42) decreases as a function of pH (Gorgeon et al., 1994).

14.9.1

Amplitude of retention processes

Clay minerals are ubiquitous in nature, and notably in a geological repository. They are characterized by a high specific surface area (up to several hundred m2/g) and by large chemical flexibility and reactivity. They are therefore highly interactive with aqueous elements, in particular with any potentially released radionuclides. The magnitude of retention depends not only on the chemical conditions but also on the nature of the cation. Figure 14.24 shows that the amplitude of sorption can vary by orders of magnitude and can reach very high efficiency (in the present case, for example, at pH~10 more than 99.999% of Am is sorbed on the clay surface) (Gorgeon, 1994). More generally, cation sorption increases with increasing pH due to the deprotonation of surface sites, whereas anion sorption decreases (Stumm and Morgan, 1996). An example of distribution coefficients on smectite is given in Fig. 14.25.

14.9.2

Origin of retention processes

Retention processes are due to the presence either of broken bonds at the mineral borders or of permanent structural charges (in particular for the clay minerals) (Tournassat et al., 2003, 2004). The permanent structural charge is related to isomorphic substitutions in the crystallographic lattice

© Woodhead Publishing Limited, 2010

468

Geological repository systems for safe disposal

14.25 Classical distribution coefficient at neutral pH for different radionuclides on a smectite-rich material used for a bentonite engineered barrier, MX80 (modified from ANDRA, 2006).

(e.g. Si4+ exchanged by Al3+) and therefore does not depend on the chemical conditions. On the contrary, the broken bonds can be protonated or deprotonated and their charge will change with the pH of the solution, as shown by the following reactions: þ Si  OHþ 2 , Si  OH þ H

½14:14

Si  OH , Si  O þ Hþ

½14:15

This charge is compensated by cations or anions located in the immediate vicinity of the mineral surface, depending on the pH. This process is referred to as retention. It can be either a pure electrostatic interaction (outer-sphere complex) or involve chemical interaction (inner-sphere complex) (Stumm and Morgan, 1996). Many studies since the mid-1990s have developed a molecular understanding of retention processes and produced numerous papers in the scientific literature. They usually involve characterization of the molecular environment of the molecules by spectroscopic techniques such as time-resolved laser-induced fluorescence (TRLIF) (Rabung et al.,

© Woodhead Publishing Limited, 2010

Post-containment performance of geological repository systems

469

2005) or extended X-ray absorption fine structure spectroscopy (EXAFS) (Schlegel et al., 1999a, 1999b, 2001; Dahn et al., 2003; Tertre et al., 2006) and/or atomistic and molecular modelling (Drot et al., 2007; Liu et al., 2008; Hattori et al., 2009; Rotenberg et al., 2009). These studies reveal that different types of molecular environments exist at the surface/water interface and attempt to use this information in the subsequent mechanistic modelling (Bradbury et al., 2005; Rabung et al., 2005). Several studies have also indicated the existence of a continuum between sorption and surface precipitation for higher surface loading (for instance see Schlegel et al., 2001).

14.9.3

Empirical modelling of retention

Soil scientists since the early 20th century have developed empirical models to describe the amount of trace elements that can be trapped by mineral surfaces (Stumm and Morgan, 1996; recent review of Limousin et al., 2007). These models are still often used in the coupled modelling of the repository near field for their simplicity and efficiency. They basically relate the amount of radionuclides sorbed at the surface of the materials as a function of the radionuclide concentration in solution. The first model of this kind is the Langmuir isotherm which assumes a 1:1 equilibrium between the surface and the radionuclides according to the equation: RNeq þ S , S-RN

½14:16

where S-RN stands for the sorbed radionuclides. It can be easily demonstrated that the amount of radionuclides sorbed can be calculated according to the following equation:



K RNaq

S-RN ¼ CE ½14:17 1 þ K RNaq where CE stands for the exchange capacity (maximum surface loading) and K the mass action law equilibrium constant for the sorption reaction. This parameter is difficult to assess since it strongly depends on the chemical conditions used to measure it (ionic strength, background electrolyte, sorbing cation, etc.). Another model is the Freundlich isotherm, which assumes a more complex reaction:

n

S-RN ¼ m RNaq ½14:18

© Woodhead Publishing Limited, 2010

470

Geological repository systems for safe disposal

where m stands for the Freundlich constant and n the non-linearity parameter. It is important to remember that these models, although they are widely used, are only empirical correlations unrelated to any mechanistic understanding. Any extrapolation out of the initial set of experimental parameters is therefore neither justified nor reliable.

14.9.4

Attempts to derive more mechanistic and predictive models

Since the 1960s, development has attempted to describe more precisely the mechanisms occurring at the solid/water interface. The best-known example is the surface complexation approach (Schindler and Stumm, 1987; Dzombak and Morel, 1990). It basically assumes that the sorbing species are located in the near-field environment of the surface where an electrostatic field occurs (the Stern compact layer). The sorbing species are therefore submitted not only to chemical but also to electrostatic interactions that must be taken into account. An electrostatic term must therefore be added to the classical chemical potential equation, as indicated below:   ½14:19 mRN ¼ m0RN þ RT Ln RNaq þ zFc where μ stands for the chemical potential, μ0 the standard potential, z the charge, F the Faraday constant and y the electrostatic field. The difficulty with this modelling approach lies in estimating the electrostatic field, which is not directly measurable. Several models have been developed to correlate the electrostatic field with the known surface charge and capacitance (or exchange capacity): constant capacitance, double diffuse layer, triple layer, etc. The number and charge of sorbed species must be fitted to the experimental sorption data. This is the most common modelling approach and has been successfully used to model sorption on clay minerals as well as clay-rich rocks. For example, Bradbury and Baeyens developed models to predict the evolution of the sorption properties of clay-rich rocks from measurements on pure minerals (Bradbury and Baeyens, 1999, 2005a, 2005b, 2006). Another approach based on ion exchange has also been developed (Motellier et al., 2003; Jacquier et al., 2004; Beaucaire et al., 2008; Tertre et al., 2009). By comparison with surface complexation, it considers not only the solid/solution interface but the {solid + interface} system, which is by definition always neutral (surface charges are compensated by counterions). Equilibrium thermodynamics can therefore be applied (mass action laws) to derive a thermodynamic constant for the retention reaction from the

© Woodhead Publishing Limited, 2010

Post-containment performance of geological repository systems

471

14.26 Am sorption of bentonite as a function of pH at constant ionic strength (0.1 M NaClO4). Experimental data are successively modelled using ion exchange theory (unpublished data from Ly, 2006; CEA internal report, 2006).

experimental distribution ratio. Furthermore, the number and charge of the sorbed species can directly be derived from the experimental data, which therefore reduce the uncertainties and the adjustable parameters in the model. Figure 14.26 illustrates the comparison of experimental results and predictive modelling of Am sorption on bentonite as a function of pH (Ly, 2006, unpublished work).

14.10 14.10.1

Coupling with transport processes Diffusion and advection processes, general transport equation

Transport processes in the geological environment can proceed by two ways: either the aqueous or particulate radionuclides are transported with the water flux itself (advection driven by the hydraulic potential) or they diffuse due to their heterogeneous distribution within the environment (Fick

© Woodhead Publishing Limited, 2010

472

Geological repository systems for safe disposal

diffusion driven by the concentration gradient) (a general presentation can be found in Horseman et al., 1996). Advection The advection flux can be calculated by Darcy’s law and is proportional to the hydraulic gradient H and the permeability K: ?

  ?

U ¼ K grad H

½14:20

The heterogeneity of advective flow also leads to an apparent diffusion process, which is referred to as mechanical dispersion and is expressed in a way similar to diffusion. Low-permeability media therefore lead to low advective flow. The permeability of the geological environment of a repository is selected to be reasonably low. For example, a permeability of less than 1014 m/s has been measured in the French Callovo-Oxfordian argillite (Distinguin and Lavanchy, 2007) whereas it ranges from 1010 to 1012 m/s in the Yucca Mountain tuff rock (Zhang et al., 2006). Nanometric-scale particles called colloids can also be transported within an advective flow and contribute to radionuclide migration (Hwang et al., 1992; Kersting et al., 1999; Degueldre and Bolek, 2009). Two situations can be encountered: either the radionuclides themselves precipitate as colloids (which can occur when nucleation dominates over particles growth) or they can sorb on existing natural colloids (either minerals or natural organic matter). In both cases, colloidal transport can be responsible for significant radionuclide displacement. The importance of this process is obviously increased by increasing permeability and pore sizes. It therefore mainly occurs in surface environment or in fractures, but is not expected to have a significant amplitude in bulk compacted rock such as deep argillites (Alonso et al., 2007). Diffusion The diffusion flux can be calculated by Fick’s law and is related to the effective diffusion coefficient De and the concentration gradient according to the equation: ?

  ?

J ¼ De grad C

½14:21

The diffusion length x can be approximated to the diffusion time t through the equation: rffiffiffiffiffiffiffiffi De t x! ½14:22 o

© Woodhead Publishing Limited, 2010

Post-containment performance of geological repository systems

473

Diffusion coefficients are slightly dependent on the radionuclides considered whereas they strongly depend on the rock properties, as follows: De ¼ o

d D0 t2

½14:23

where w stands for the porosity, d the tortuosity, t the constrictivity and D0 the diffusion coefficient in porewater, which can be calculated by the classical Stokes–Einstein relation. Diffusion coefficients are classically measured on thin rock samples in an experimental diffusion cell (for instance see Tevissen et al., 2004). Diffusion coefficients are usually in the range between 1010 and 1012 m2/s. Transport equation The evolution of the radionuclides concentration as a function of time at a given point can therefore be calculated by considering both processes and is expressed by the transport equation (in one dimension): o

@C @C @2C @2C ¼ U þ aU 2 þ De 2  G @t @x @x @x

½14:24

where G is a source/sink term representing either radioactive decay or mineral precipitation/dissolution, and a is the dispersivity of the rock. This equation can be simplified for preliminary assessment by considering only the dominant process, which is often diffusion in the repository nearfield environment. Otherwise simulation codes based on finite elements (or volumes) can be used to solve the equation in a two-dimensional or threedimensional simulation. These calculations are useful to assess the long-term amplitude of the transport processes.

14.10.2

Coupling with chemical reactivity: ionic retardation and anionic exclusion

The present transport equation was generic and does not directly account for the actual radionuclides species or for the sorption processes. However, these processes have a significant influence on the overall mobility and two generic behaviours can be distinguished. Ionic sorption Most ions, and especially cations, sorb strongly on mineral surfaces. This process decreases the mobile radionuclides inventory in solution. It qualitatively represents a migration-inhibiting phenomenon. More precisely,

© Woodhead Publishing Limited, 2010

474

Geological repository systems for safe disposal

Table 14.3 Half-life period, porosity accessible to diffusion, effective diffusion coefficient and retardation factor for various radionuclides within a bentonite engineered barrier composed of MX80 clay (data taken from ANDRA, 2006). Basically, anions have access to a small part of the porosity and are repelled by the negatively charged clay mineral surfaces (no sorption, smaller porosity, R = 1), which leads to a decrease in their effective diffusion coefficient by two orders of magnitude by comparison to the cations. Most of the cations are characterised by relatively high retardation factors, which strongly contribute to reduce their mobility

10

Be C 36 Cl 41 Ca 59 Ni 79 Se 93 Zr 93 Mo 99 Tc 107 Pd 126 Sn 129 I 166m Ho 135 Cs 14

Period (years)

Ionic form

Porosity accessible to diffusion

De (m2/s)

R

1 600 000 5 730 302 000 103 000 75 000 65 000 1 530 000 3 500 213 000 6 500 000 100 000 15 700 000 1 200 2 300 000

Cation Anion Anion Cation Cation Anion Cation Anion Cation Cation Cation Anion Cation Cation

0.36 0.05 0.05 0.36 0.36 0.05 0.36 0.05 0.36 0.36 0.36 0.05 0.36 0.36

561010 561012 561012 561010 561010 561012 561010 561012 561010 561010 561010 561012 561010 561010

973 1 1 6 2 430 1 486 000 1 146 000 4 380 53 500 1 58 300 487

it can be demonstrated, by assuming a linear sorption isotherm and fast and reversible sorption, that the general transport equation becomes   rd Kd @C @C @2C @2C ¼ U þ aU 2 þ De 2  G o 1þ ½14:25 @t @x @x @x o where rd is the dry density of the rock. The retardation factor R is defined thus: R¼1þ

rd Kd o

½14:26

It can also be demonstrated that the breakthrough time of an RN, tRN, through a given thickness of rock can be approximated by that of water, tw, corrected by the retardation factor: tRN ¼ Rtw

½14:27

Classical retardation factor or values for bentonite are given in Table 14.3. For instance, the Np retardation factor is estimated to be in the range 103–104 for the Meuse/Haut-Marne site, whereas it is near 10 for the Yucca

© Woodhead Publishing Limited, 2010

Post-containment performance of geological repository systems

475

Mountain site. The mobility of Np will therefore be reduced by 3 to 4 orders of magnitude in the Callovo-Oxfordian argillites by comparison to water, whereas the reduction will only be by a factor of 10 for Yucca Mountain tuff. This directly evidences the very significant role of sorption in the overall migration process. The greater the sorption, the longer the migration time. Selecting rocks with high sorption properties such as clay-rich rocks therefore significantly affects the overall impact of a repository, since the radionuclides will also decay during the same time. Anionic exclusion Clay particles present negative surface charges due both to isomorphic lattice substitutions and to deprotonation of their surface sites. Anions will therefore be repelled from the mineral surfaces and will only have access to a restricted part of the porosity compared with cations (Bazer-Bachi et al., 2006; Rotenberg et al., 2007a, 2007b; Tournassat et al., 2007;) (Fig. 14.27). This leads to a specific migration behaviour: when the clay content increases in rock, the apparent diffusion coefficient significantly decreases (Descostes et al., 2008). Such behaviour is clearly illustrated in Fig. 14 28. Typical radionuclides affected by such behaviour include 129I, 36Cl, 79Se2O42, 99 TcO42 and 14C. Depending on the chemical conditions occurring in the near field of a repository, other radionuclides can also be present as anions and could therefore exhibit similar behaviour if the size of the aqueous complex is small enough to migrate through the small pores of the rock. This is a clear illustration of the effect of speciation on the overall migration properties of radionuclides.

14.27 Origin of anionic exclusion within clay-rich rocks. Due to the presence of negatively charged surfaces, anions have access only to a restricted part of the porosity (here shown by the band in the middle).

© Woodhead Publishing Limited, 2010

476

Geological repository systems for safe disposal

14.28 Evolution of Cl and I experimental diffusion coefficient normalized to the tritium diffusion coefficient as a function of depth on the Meuse/Haute-Marne French site. Within the first 350 metres the values are close to the diffusion coefficient ratio measured in free water (dotted line), ~0.9. Below 400 m the values decrease sharply to 0.2. This evolution is related to the increasing clay content in the rock, up to 50% in the argillites and is experimental evidence of the occurrence of anionic exclusion (adapted from Descostes et al., 2008).

14.10.3

Migration simulations and confrontation with experimental results

The approximations presented above are useful in order to understand the influence of the different processes on the overall migration phenomenon and its amplitude. However, any robust and reliable assessment of radionuclide migration from a repository requires developing a full set of numerical simulations that accounts for all the processes involved and their parameter ranges. Migration calculation basically requires simultaneously modelling the transport equation and the chemistry reactions in a heterogeneous environment (for instance see MacQuarrie and Mayer, 2005; Montarnal et al., 2007; Mathieu et al., 2008; Mukhopadhyay et al., 2009). This can be performed through two-dimensional or three-dimensional finite element (or volume) simulations. In such calculations, transport

© Woodhead Publishing Limited, 2010

Post-containment performance of geological repository systems

477

equation and chemical reaction systems (mass action laws and sorption processes) are coupled and solved in each part of the mesh. This implies very extensive calculations, which require significant computing power. This approach is used for quantitative assessment of the overall impact of any repository, e.g. to calculate the migration plume. However, the results of such calculations must still be validated on natural systems to verify the relevance and reliability of the processes and parameters. Basically, two complementary approaches are used. The first approach consists in performing in situ tracing experiments with adequate sampling and measurements (for instance see Dewonck et al., 2006; Wersin et al., 2008). However, such experiments are limited in time and can hardly give results for strongly sorbing cations that migrate very slowly. The second approach consists in modelling the past evolution of natural analogue systems in order to verify whether current models are able to reproduce past events. Regarding radionuclide migration, the most extended study was performed on the Oklo natural nuclear reactors (Gabon) in the framework of several European-funded projects (Pourcelot and Gauthier-Lafaye, 1999; Bruno et al., 2002; Gauthier-Lafaye, 2002; Gurban et al., 2003).

14.11

Summary of the behaviour of the main radionuclides

Based on the previous insights, significant conclusions can be drawn concerning the radionuclide migration from a potential deep repository. First, chemical conditions have a major influence on radionuclide mobility through several processes. Locating the repository in a deep reducing environment favours the formation of reduced species. This feature is particularly important for actinides since it decreases their solubility by 2 to 3 orders of magnitude to a very low level (of the order of 1010 M). Furthermore, cations strongly sorb on mineral surfaces, in particular on clay particles. This process is enhanced by the increasing charge of the cations. Actinides and most of the metallic fission products will present such behaviour. It is therefore of prime importance to identify the aqueous speciation of radionuclides, which will partly determine their solubility limit and the amplitude of their retention. Second, transport processes depend on the material hydrodynamic properties, i.e. engineered material properties for the near field and rock properties for the far field. Low-permeability environments favour diffusion whereas advection is predominant as soon as the permeability is high enough. Most repository projects are planned to be located in thick layers of very low-permeability rock where diffusion predominates, and are surrounded by a more hydraulically conductive layer such as limestone.

© Woodhead Publishing Limited, 2010

478

Geological repository systems for safe disposal

Concerning the near field, most projects also plan to embed the waste canister with low-permeability backfill material, in many cases clay-rich material as bentonite. Highly sorbent radionuclides such as actinides would therefore only slowly diffuse from the waste form within this lowpermeability engineered barrier whereas more mobile elements such as the anions will reach the far-field environment and, subsequently, potentially reach the surrounding conductive rock formation. Therefore, the anticipated impact of the repository will be dominated by the mobile elements, mainly the anions. More precisely, in a deep repository in a reducing environment, actinides and metallic fission products are basically not expected to migrate very far from the waste form and, in any case, not to diffuse out of the lowpermeability layer where the repository is located. A rough calculation based on equations [14.19] and [14.24] allows orders of magnitude to be given of representative diffusion length for different radionuclides: nonsorbing anions will first diffuse through the engineered barrier (representative times are on the order of 100 years), then the slightly sorbing anions (~thousands of years), then the slightly sorbing cations (more than ten thousands of years) and, much later, the strongly sorbing cations as the actinides. Clay-rich material with diffusive properties acts as an efficient and selective filter that only allows the anions to diffuse. Strongly sorbing cations will therefore mainly be confined within the near-field environment and will not significantly contribute in a normal scenario to the final impact of the repository. Conversely, anionic fission products such as 129I, 36Cl, 14C, 79 Se, 99Tc, etc., are not expected to sorb significantly and will therefore slowly diffuse out of the near-field environment. They will therefore be the principal measurable radionuclides that could reach the surface environment and contribute to any dose impact of a repository. Key radionuclides in such environments and repositories are thus the mobile anionic radionuclides including 129I, 36Cl, 14C, 79Se and 99Tc. In oxidizing conditions such as those encountered in the Yucca Mountain site, the preceding conclusions are modified by the occurrence of oxidized actinide forms with greater solubility and reduced sorption behaviour. This explains why some actinides such as 237Np may diffuse out of the near-field environment and contribute to the final impact of such a repository.

14.12

Conclusion

The repository near field plays the most significant role in the overall confinement performance of any repository. First of all, radionuclides are embedded in waste forms that are tailored (excepted for spent fuel) to have strong confinement properties consistent with the radiotoxicity of the radionuclides. During the first few thousand years the waste forms will

© Woodhead Publishing Limited, 2010

Post-containment performance of geological repository systems

479

evolve without any interaction with the environment, only driven by their intrinsic radioactive decay and disequilibrium. Subsequently, after the corrosion of the container, the waste forms will be altered by the groundwater and subsequently release their radionuclides in the near-field environment. However, the waste forms are highly durable and are altered very slowly, with lifetimes extending over hundreds of thousands of years for high-level waste and tens of thousands of years for intermediate-level waste. The particularity of spent nuclear fuel is that part of the radionuclide inventory is released immediately when water enters the canister (the instant release fraction), which contributes to increasing the long-term impact of a spent fuel repository by comparison to nuclear waste glass. Once released in the near field, radionuclides become involved in a complex set of geochemical processes that includes aqueous reactions with other aqueous ligands or with particles, interaction on the mineral surfaces and transport towards the geosphere. However, the near field of any repository is selected to be an impermeable low-diffusivity medium (bentonite in an engineered barrier system, compacted clay-rock, etc.). Radionuclide migration is therefore quite limited and does not extend much except for some mobile elements: most of the radionuclides will not migrate further than a few decametres over the long term, therefore ensuring negligible impact. The mobile elements are mainly anions such as 129I, 36Cl, 14C and 79Se, which are not retarded by surface interactions (anion exclusion). In order to understand the near-field evolution, a high-level dedicated R&D approach has been developed over several decades by the scientific community. It relies first of all on an accurate and detailed understanding at the smallest scale of the actual mechanisms governing waste-form evolution and radionuclide migration. This knowledge must then be integrated in multi-scale models that are simplified to derive performance-oriented models for use in performance assessments and safety analyses. The current focus of R&D deals with near-field coupling phenomena, which have been demonstrated to be largely beneficial for the overall repository performance. Although the safety of the repository can already be demonstrated at the current state of knowledge, further R&D will be required to enhance confidence in long-term predictions and ensure continuous improvement of modelling tools to reduce the remaining uncertainties as science progresses.

14.13

References

Adenot F and Richet C (1995), ‘Modelling of the chemical degradation of a cement paste; long term prediction’, MRS 1995 Fall Meeting, Boston, Massachusetts. Adenot F, Richet C and Faucon P (1997), ‘Long-term prediction of concrete durability in radioactive waste management: influence of the pH of the

© Woodhead Publishing Limited, 2010

480

Geological repository systems for safe disposal

aggressive solution’, in International Conference on Engineering Materials, Ottawa, Canada, 8–11 June 1997, Proceedings vol. 2. Advocat T, Crovisier J L, Vernaz E, Ehret G and Charpentier H (1991), ‘Hydrolysis of R7T7 nuclear waste glass in dilute media: mechanisms and rate as a function of pH’, in Scientific Basis for Nuclear Waste Management, 212th edition, vol. XIV, Materials Research Society, Pittsburgh, Pennsylvania, p. 57. Alonso U, Missana T, Patelli A and Rigato V (2007), ‘Bentonite colloid diffusion through the host rock of a deep geological repository’, Physics and Chemistry of the Earth, 32, 469–476. ANDRA (2006), Dossier 2005 Argile, ANDRA Report. Angeli F, Gaillard M, Jollivet P and Charpentier T (2006), ‘Influence of glass composition and alteration solution on leached silicate glass structure: a solidstate NMR investigation’, Geochimica Cosmochimica Acta, 70(10), 2577–2590. Angeli F, Charpentier T, Gaillard M and Jollivet P (2008), ‘Influence of zirconium on the structure of pristine and leached soda lime borosilicates glasses: towards a qualitative approach by 17O MQMAS NMR’, Journal of Non-crystalline Solids, 354, 3713–3722. Arab M, Cailleteau C, Angeli F, Devreux F, Girard L and Spalla O (2008), ‘Aqueous alteration of five-oxide silicate glasses: experimental approach and Monte Carlo modelling’, Journal of Non-crystalline Solids, 354, 155–161. Artinger R, Buckau G, Zeh P, Geraedts K, Vancluysen J, Maes A and Kim J I (2003), ‘Humic colloid mediated transport of tetravalent actinides and technetium’, Radiochimica Acta, (91), 743–750. Atkins M, Damidot D and Glasser F P (1994), ‘Performance of cementitious systems in the repository’, Materials Research Society Symposium Proceedings, 333, 315–326. Avrami M (1939), ‘Kinetics of phase change. I. General theory’, Journal of Chemistry and Physics, 7, 1103–1112. Avrami M (1940), ‘Kinetics of phase change. II. Transformation-time relations for random distribution of nuclei’, Journal of Chemistry and Physics, 8, 212–224. Avrami M (1941), ‘Kinetics of phase change. III. Granulation, phase change and microstructure’, Journal of Chemistry and Physics, 9(1941), 177–184. Baron J and Ollivier J-P (1996), Les Be´tons. Bases et Donne´es pour leur Formulation, Eyrolles. Bary B and Sellier A (2004), ‘Coupled moisture–carbon dioxide–calcium transfer model for carbonation of concrete’, Cement and Concrete Research, 34, 1859– 1872. Bazer-Bachi F, Tevissen E, Descostes M, Grenut B, Meier P, Simonnot M O and Sardin M (2006), ‘Characterization of sulphate sorption on Callovo-Oxfordian argillites by batch, column and through-diffusion experiments’, Physics and Chemistry of the Earth, 32, 552–558. Beaucaire C, Pitsch H, Toulhoat P, Motellier S and Louvat D (2000), ‘Regional fluid characterisation and modelling of water–rock equilibria in the Boom Clay Formation and in the Rupelian aquifer at Mol, Belgium’, Applied Geochemistry, 15(5), 667–686. Beaucaire C, Tertre E, Coreau N, Juery A and Legrand S (2008), ‘A multi-site ion exchange model to predict contaminants sorption in sediments’, Geochimica Cosmochimica Acta, 72, A62.

© Woodhead Publishing Limited, 2010

Post-containment performance of geological repository systems

481

Berger S, Cau Dit Coumes C, Le Bescop P and Damidot D (2009a), ‘Hydration of calcium sulfoaluminate cement by a ZnCl2 solution: investigation at early age’, Cement and Concrete Research, 39(12), 1180–1187. Berger S, Frizon F and Joussot-Dubien C (2009b), ‘Formulation study on caesiumbased and caesium containing geopolymers’, Advances in Applied Ceramics, 108(7), 412–417. Bradbury M H and Baeyens B (1998). ‘A physicochemical characterisation and geochemical modelling approach for determining porewater chemistries in argillaceous rock’, Geochimica Cosmochimica Acta, 62(5), 783–795. Bradbury M and Baeyens B (1999), ‘Modelling the sorption of Zn and Ni on Camontmorillonite’, Geochimica Cosmochimica Acta, 63, 325–336. Bradbury M and Baeyens B (2005), ‘Modelling the sorption of Mn(II), Co(II), Ni(II), Zn(II), Cd(II), Eu(III), Am(III), Sn(IV), Th(IV), Np(V) and U(VI) on montmorillonite: linear free energy relationships and estimates of surface binding constants for some selected heavy metals and actinides’, Geochimica Cosmochimica Acta, 69, 875–892. Bradbury M and Baeyens B (2006), ‘Modelling sorption data for the actinides Am (III), Np(V) and Pa(V) on montmorillonite’, Radiochimica Acta, 94, 619–625. Bradbury M H, Baeyens B, Geckeis H and Rabung T (2005), ‘Sorption of Eu(III)/ Cm(III) on Ca-montmorillonite and Na-illite. Part 2: Surface complexation modeling’, Geochimica Cosmochimica Acta, 69, 5403–5412. Brocskowski M E, Goldik J S, Santos B, Noel J and Shoesmith D (2006), Materials Research Society Symposium Proceedings, 985, 3–14. Brown P, Curti E and Grambow B (2005), Chemical Thermodynamics of Zirconium, Chemical Thermodynamics Series vol. 8, OECD Publications. Bruno J, Duro L and Grive M (2002), ‘The applicability and limitations of thermodynamic geochemical models to simulate trace element behaviour in natural waters. Lessons learned from natural analogue studies’, Chemical Geology, 190, 371–393. Bruno J, Bosbach D, Kulik D and Navrotsky A (2007), Chemical Thermodynamics of Solid Solutions of Interest in Nuclear Waste Management, Chemical Thermodynamics Series vol. 10, OECD Publications. Bunker B C (1994), ‘Molecular mechanisms for corrosion of silica and silicate glasses’, Journal of Non-crystalline Solids, 179, 300–308. Bureau G, Delaye J M, Peuget S and Calas G (2008), ‘Molecular dynamics study of structural changes versus deposited energy dose in a sodium borosilicate glass’, Nuclear Instruments and Methods in Physics Research Section B – Beam Interactions with Materials and Atoms, vol. 266, pp. 2707–2710. Cailleteau C, Angeli F, Devreux F, Gin S, Jestin J, Jollivet P and Spalla O (2008), ‘Insight into silicate – glass corrosion mechanisms’, Nature Materials, 7(12), 978–983. Camaro S, Moulinier D, Fromont M, Naud G, Masson H and Sellier J-Y (1999), ‘Evaluation of bituminized waste reactivity – experimental approach, Radwaste Bituminization’99’, in Proceedings of the International Workshop on the Safety and Performance Evaluation of Bituminization Process for Radioactive Waste, Rez, Prague. Camaro S, Ragetly Q and Riglet-Martial C (2003), ‘Compose´ pie´geur de l’hydroge`ne, proce´de´ de fabrication et utilisation’, Patent 2,859,202.

© Woodhead Publishing Limited, 2010

482

Geological repository systems for safe disposal

Cau-dit-Coumes C, Courtois S, Peysson S, Ambroise J and Pera J (2009), ‘Calcium sulfoaluminate cement blended with OPC: a promising binder to encapsulate low-level radioactive slurries of complex chemistry’, Cement and Concrete Research, 39(9), 740–747. CEA (1983), La Solidification des De´chets par le Bitume, CEA, Paris. Chaix P, Camaro S, Simondi B, Vistoli P P and Blanc V (2001a), ‘Long-term behaviour of bituminised waste: modelling of auto-irradiation and leaching’, Materials Research Society Symposium Proceedings, 663, 9. Chaix P, Moulin V, Poinssot C and Cocaud A (2001b), ‘Long term behaviour of standard compacted waste package: current status and research issues’, Conference GLOBAL ‘01, Paris, September 2001. Chamssedine F, Sauvage T, Peuget S and Martin G (2010), ‘Measurements of the helium diffusion coefficient in R7T7 nuclear glass by using 3He(d, α)1H nuclear reaction analysis technique’, Submitted to Journal of Nuclear Materials. Chen Y and Pearson F J (2008), ‘Modelling plutonium solubility for Yucca Mountain performance assessment’, Radiochimica Acta, 96, 521–526. Choppin G R (1992), ‘The role of natural organics in radionuclide migration in natural aquifer systems’, Radiochimica Acta, 58(9), 113–120. Choppin G R (2006), ‘Actinide speciation in aquatic systems’, Marine Chemistry, 99, 83–92. Christensen H (1998), ‘Calculations simulating spent-fuel leaching experiments’, Nuclear Technology, 124, 165–174. Christensen H and Bjergbakke E (1987), ‘Radiation induced dissolution of UO2’, Materials Research Society Symposium Proceedings, 84, 115–122. Christensen H, Sunder S and Shoesmith D W (1994), ‘Oxidation of nuclear fuel (UO2) by the products of water radiolysis: development of a kinetic model’, Journal of Alloys and Compounds, 213/214, 93–99. Conradt R (2008), ‘Chemical durability of oxide glasses in aqueous solution: a review’, Journal of American Ceramic Society, 91(3) 728–735. Corbel C, Sattonnay G, Guilbert S, Garrido F, Barthe M-F and Je´gou C (2006), ‘Addition versus radiolytic production effects of hydrogen peroxide on aqueous corrosion of UO2’, Journal of Nuclear Materials, 348, 1–17. Dahn R, Scheidegger A M, Manceau A, Schlegel M L, Baeyens B, Bradbury M H and Chateigner D (2003), ‘Structural evidence for the sorption of Ni(II) atoms on the edges of montmorillonite clay minerals: a polarized X-ray absorption fine structure study’, Geochimica Cosmochimica Acta, 67, 1–15. De Bonfils J (2007), ‘Effets d’irradiations sur la structure de verres borosilicate´s – comportement a` long terme des matrices vitreuses de stockage des de´chets nucle´aires’, PhD thesis, University Claude Bernard, Lyon, September 2007. Degueldre C and Bolek M (2009), ‘Modelling colloid association with plutonium: the effect of pH and redox potential’, Applied Geochemistry, 24, 310–318. Dejonghe P (1999), ‘Development and industry applications of bituminization processes, Radwaste Bituminization’99’, in Proceedings of the International Workshop on the Safety and Performance Evaluation of Bituminization Process for Radioactive Waste, Rez, Prague, pp. 11–16. Deladerrie`re N, Delaye J-M, Augereau F, Despaux G and Peuget S (2008),

© Woodhead Publishing Limited, 2010

Post-containment performance of geological repository systems

483

‘Molecular dynamics study of acoustic velocity in silicate glass under irradiation’, Journal of Nuclear Materials, 375, 120. Delaye J-M and Ghaleb D (2005a), ‘Combining two types of molecular dynamics for rapid computation of high-energy displacement cascades. Part 1: description of the CoMoD method’, Physics Review B, 71, 224203. Delaye J-M and Ghaleb D (2005b), ‘Combining two types of molecular dynamics for rapid computation of high-energy displacement cascades. Part 2: application of the CoMoD method to a 70 keV cascade in a simplified nuclear glass’, Physics Review B, 71, 224204. Delaye J-M and Ghaleb D (2006), ‘Damage inhomogeneity in the core region of displacement cascades in simplified nuclear glass’, Journal of Nuclear Materials, 348, 243–255. De Pablo J, Casas I, Gimenez J, Molera M, Rovira M, Duro L and Bruno J (1999), ‘The oxidative dissolution mechanism of uranium dioxide. I. The effect of temperature in hydrogen carbonate medium’, Geochimica Cosmochimica Acta, 63(19/29), 3097–3103. Descostes M, Blin V, Bazer-Bachi F, Meier P, Grenut B, Radwan J, Schlegel M L, Buschaert S, Coelho D and Tevissen E (2008), ‘Diffusion of anionic species in Callovo-Oxfordian argillites and Oxfordian limestones (Meuse/Haute–Marne, France)’, Applied Geochemistry, 23, 655–677. de Windt L, Schneider H, Ferry C, Catalette H, Lagneau V, Poinssot C, Poulesquen A and Je´gou C (2006), ‘Modelling spent nuclear fuel alteration and radionuclide migration in disposal conditions’, Radiochimica Acta, 94, 787– 794. Dewonck S, Descostes M, Blin V, Tevissen E, Radwan J, Poinssot C, Cartalade A, Brisset P and Coelho D (2006), ‘In situ diffusion experiments in CallovoOxfordian mudstone’, Geochimica Cosmochimica Acta, 70, A140. Distinguin M and Lavanchy J M (2007), ‘Determination of hydraulic properties of the Callovo-Oxfordian argillite at the Bure site: synthesis of the results obtained in deep boreholes using several in situ investigation techniques’, Physics and Chemistry of the Earth, 32, 379–392. Doremus R H (1975), ‘Interdiffusion of hydrogen and alkali ions in a glass surface’, Journal of Non-crystalline Solids, 19, 137. Drot R, Roques J and Simoni E (2007), ‘Molecular approach of the uranyl/mineral interfacial phenomena’, C-R Chimie, 10, 1978–1091. Dzombak D A and Morel F M M (1990), Surface Complexation Modelling; Hydrous Ferric Oxides, Wiley-Interscience, New York. Ewing R C, Weber W J and Clinard F W (1995), Progress in Nuclear Energy, 29, 63. Fares T, Peuget S, Chamssedine F, Sauvage T, Haussy J, Deschanels X, Roudil D, Charles B and Je´gou C (2009), ‘Helium diffusion in doped nuclear waste glasses’, in Proceedings of the 12th International Conference on Environmental Remediation and Radioactive Waste Management, ICEM’09, Liverpool, UK, October 2009. Ferrand K, Abdelouas A and Grambow B (2006), ‘Water diffusion in the simulated French nuclear waste SON68 contacting silica rich solutions: experimental and modelling’, Journal of Nuclear Materials, 355, 54–67. Ferry C, Lovera P, Poinssot C and Garcia P (2005), ‘Theoretical approaches of

© Woodhead Publishing Limited, 2010

484

Geological repository systems for safe disposal

diffusion coefficient enhanced by alpha self irradiation in spent nuclear fuel’, Journal of Nuclear Materials, 346, 48–55. Ferry C, Piron J P and Stout R (2006a), ‘Effect of helium accumulation on the spent fuel microstructure, MRS scientific basis for nuclear waste management XXX’, Materials Research Society Symposium Proceedings, 985, 111–116. Ferry C, Poinssot C, Cappelaere C, Desgranges L, Je´gou C, Miserque F, Piron J P, Roudil D and Gras J M (2006b), ‘Specific outcomes of the research on the spent fuel long term evolution in interim dry storage and deep geological disposal’, Journal of Nuclear Materials, 352, 246–253. Ferry C, Piron J P, Poulesquen A and Poinssot C (2007), ‘Radionuclides release from the spent fuel under disposal conditions: re-evaluation of the instant release fraction, XXXIst Scientific Basis for Nuclear Waste Management’, Material Research Society Symposium Proceedings, 1107, 447–454. Frizon F and Cau-dit-Coumes C (2006), ‘Cementation of ILW ion exchange resins: impact of sulfate ions released by radiolysis on hydrated matrix’, Journal of Nuclear Materials, 359, 162–173. Frugier P, Martin C, Ribet I, Advocat T and Gin S (2005), ‘The effect of composition on the leaching of three nuclear waste glasses: R7T7, AVM and VRZ’, Journal of Nuclear Materials, 346, 14–24. Frugier P, Gin S, Lartigue J E and Deloule E (2006), ‘SON68 glass dissolution kinetics at high reaction progress: mechanisms accounting for the residual alteration rate’, Materials Research Society Symposium Proceedings on Scientific Basis for Nuclear Waste Management 2005, Ghent, Belgium, pp. 305–312. Frugier P, Gin S, Minet Y, Chave T, Bonin B, Godon N, Lartigue J-E, Jollivet P, Ayral A, De Windt L and Santarini G (2008), ‘SON68 nuclear glass dissolution kinetics: current state of knowledge and basis of the new GRAAL model’, Journal of Nuclear Materials, 381(1–3), 8–21. Frugier P, Chave T, Gin S and Lartigue J-E (2009), ‘Application of the GRAAL model to leaching experiments with SON68 nuclear glass in initially pure water’, Journal of Nuclear Materials, 392, 552–567. Galle´ C, Peycelon H, Le Bescop P, Be´jaoui S, L’Hostis V, Bouniol P and Richet C (2006), ‘Concrete long-term behaviour in the context of nuclear waste management: experimental and modelling research strategy’, Journal de Physique IV, 136, 25–38. Gamsja¨ger H, Bugajski J, Gajda T, Lemire R and Preis W (2005), ‘Chemical Thermodynamics of Nickel, Chemical Thermodynamics Series vol. 6, OECD Publications. Gauthier-Lafaye F (2002), ‘Two billion year old natural analogs for nuclear waste disposal: the natural nuclear fission reactors in Gabon (Africa)’, C R Physique, 3, 839–849. Geckeis H and Rabung T (2008), ‘Actinide geochemistry: from the molecular level to the real system’, Journal of Cont. Hydrology, 102, 187–195. Geneste G, Bouyer F and Gin S (2006), ‘Hydrogen–sodium interdiffusion in borosilicate glasses investigated from first principles’, Journal of Noncrystalline Solids, 35(28–29), 3147–3152. Gin S (2000), ‘Protective effect of the alteration gel: a key mechanism in the long term behaviour of nuclear waste glass’, in Scientific Basis for Nuclear Waste

© Woodhead Publishing Limited, 2010

Post-containment performance of geological repository systems

485

Management, edited by K Hart and G R Lumpkin, 663rd edition vol. XXIV, Materials Research Society, Sydney, p. 207. Gin S and Mestre J P (2001), SON68 nuclear glass alteration kinetics between pH 7 and pH 11.5, Journal of Nuclear Materials, 295, 83. Glasser F P (1997), ‘Fundamental aspects of cement solidification and stabilisation’, Journal of Hazardous Materials, 52, 151–170. Glasser F P (2002), ‘Characterization of the barrier performance of cements’, Materials Research Society Symposium Proceedings, 713, JJ9.1.1–JJ9.1.12. Gorgeon L (1994), ‘Contribution a` la mode´lisation physics-chime´que de la re´tention de radioe´le´ments a` vie longue par des mate´riaux argileux’, PhD thesis, Pierre et Maria Curie Universite´, Paris. Grambow B (1985), ‘A general rate equation for nuclear waste glass corrosion’, in Scientific Basis for Nuclear Waste Management, edited by C M Jantzen, J A Stone and R C Ewing, 44th edition, vol. VII, Materials Research Society, Pittsburgh, Pennsylvania, p. 15. Grambow B and Strachan D M (1998), ‘A comparison of performance of nuclear waste glass by modelling’, Materials Research Society, 112, 713. Grambow B, Mennecart T, Fattahi M and Blondiaux G (2004), ‘Electrochemical aspects of radiolytically enhanced UO2 dissolution’, Radiochimica Acta, 92, 603–609. Grambow B et al. (2008), ‘RTDC-1 Final Synthesis Report, RTD Component 1: Dissolution and release from the waste matrix’, EU NF-PRO Project FI6WCT-2003-02389. Gras J M, Do Quang R, Masson H, Lie´ven T, Ferry C, Poinssot C, Debes M and Delbecq J M (2007), ‘Perspectives on the closed fuel cycles – implications for high level matrices’, Journal of Nuclear Materials, 362, 383–394. Grenthe I, Fuger J, Konings R, Lemire R, Muller A, Nguyen-Trung C and Wanner H (1992), Chemical Thermodynamics of Uranium, Chemical Thermodynamics Series vol. 1, Elsevier. Guilbert S, Sauvage T, Erramli H, Barthe M-F, Desgardin P, Blondiaux G, Corbel C and Piron J-P (2003), ‘Helium behaviour in UO2 polycrystalline disks’, Journal of Nuclear Materials, 321, 121–128. Guilbert S, Sauvage T, Garcia P, Carlot G, Barthe M-F, Desgardin P, Blondiaux G, Corbel C, Piron J-P and Gras J M (2004), ‘Helium migration in implanted UO2 sintered disks’, Journal of Nuclear Materials, 327, 88–96. Guillaumont R, Fangha¨nel T, Fuger J, Grenthe I, Neck V, Palmer D and Rand M (2003), Update on the Chemical Thermodynamics of Uranium, Neptunium, Plutonium, Americium and Technetium, Chemical Thermodynamics Series vol. 5, OECD Publications. Gurban I, Laaksoharju M, Made B and Ledoux E (2003), ‘Uranium transport around the reactor zone at Bangombe and Okelobondo (Oklo): examples of hydrogeological and geochemical model integration and data evaluation’, Journal of Cont. Hydrology, 61, 247–264. Gwinner B, Sercombe J, Tiffreau C, Simondi-Teisseire B, Felines I and Adenot F (2006), ‘Modelling of bituminized radioactive waste leaching, Part 2: experimental validation’, Journal of Nuclear Materials, 349, 107–118. Hattori T, Saito T, Ishida K, Scheinost A, Tsuneda T, Nagasaki S and Tanaka S (2009), ‘The structure of monomeric and dimeric uranyl adsorption complexes

© Woodhead Publishing Limited, 2010

486

Geological repository systems for safe disposal

on gibbsite: a combined DFT and EXAFS study’, Geochimica Cosmochimica Acta, 73, 5875–5988. Horseman S T, Higgo J J W, Alexander J and Harrington J F (1996), Water, Gas and Solute Movement through Argillaceous Media, Nuclear Energy Agency, OECD Publications, 290 pp. Hu Q, Zavarin M and Rose T (2008), ‘Effect of reducing groundwater on the retardation of redox-sensitive radionuclides’, Geochemical Transactions, 9,12, doi:10.1186/1467-4866-9-12. Hummel W, Anderegg G, Puigdomenech I, Rao L and Tochiyama O (2007), Chemical Thermodynamics of Compounds and Complexes of U, Np, Pu, Am, Tc, Se, Ni and Zr with Selected Organic Ligands, Chemical Thermodynamics Series vol. 9, OECD Publications. Hwang Y, Pigford T H, Chambre P L and Lee W W L (1992), ‘Analyses of masstransport in a nuclear waste repository in salt’, Water Resources Research, 28 (7), 1857–1868. IAEA (1991), ‘Evaluation of spent fuel as a final waste form’, IAEA Technical Report Series 320. IAEA (1993), ‘Bituminization processes to condition radioactive wastes’, IAEA Technical Report Series 352. Jacquier P, Ly J and Beaucaire C (2004), ‘The ion-exchange properties of the Tournemire argillite – I. Study of the H, Na, K, Cs, Ca and Mg behaviour’, Applied Clay Sciences, 26, 163–170. Je´gou C, Muzeau B, Broudic V, Peuget S, Poulesquen A, Roudil D and Corbel C. (2005a), ‘Effect of external gamma irradiation on dissolution of the spent UO2 fuel matrix’, Journal of Nuclear Materials, 341, 62–82. Je´gou C, Muzeau B, Broudic V, Poulesquen A, Roudil D, Jorion F and Corbel C (2005b), ‘Effect of alpha irradiation on UO2 surface reactivity in aqueous media’, Radiochimica Acta, 93, 35–42. Johannesson K H, Stetzenbach K J and Hodge V F (1995), ‘Speciation of the rare earth element neodymium in groundwaters of the Nevada Test Site and Yucca Mountain and implications for actinide solubility’, Applied Geochemistry, 10, 565–572. Johnson W A and Mehl R F (1939), Transactions of American Institute of Mining Engineers, 135, 416. Johnson L H and Shoesmith D W (1988), ‘Spent nuclear fuel’, in Radioactive Wasteform for the Future, edited by W Lutze and R C Ewing, Elsevier Science Publishers B.V., Chapter 11, pp. 635–698. Johnson, L, Garisto N C and Stroess-Gascoyne S (1985), Proceedings of Waste Management, p. 479. Johnson L H, Ferry C, Poinssot C and Lovera P (2005) ‘Spent fuel radionuclide source term model for assessing spent fuel performance in geological disposal. Part I – Assessment of the instant release fraction’, Journal of Nuclear Materials, 346, 56–65. Jollivet P, Angeli F, Cailleteau C, Devreux F, Frugier P and Gin S (2008), ‘Investigation of gel porosity clogging during glass leaching’, Journal of Noncrystalline Solids, 354, 4952–4958. Kamali S, Ge´rard B and Moranville M (2003), ‘Modelling the leaching kinetics of

© Woodhead Publishing Limited, 2010

Post-containment performance of geological repository systems

487

cement-base materials. Influence of materials and environment’, Cement and Concrete Composites, 25(4–5), 451–458. Kersting A, Efurd D, Finnegan D, Rokop D, Smith D and Thompson J (1999), ‘Migration of plutonium in groundwater at the Nevada Test Site’, Nature, 397, 56–59. Kolgomorov A N (1937), Izv. Akad. Nauk. SSSR, Ser. Fiz. 3, 335. Ledieu A, Devreux F, Barboux P, Sicard L and Spalla O (2003), ‘Leaching of borosilicate glasses: I – Experiments’, Journal of Non-crystalline Solids, 1–25. Lemire R, Fuger J, Nitsche H, Potter P, Rand M, Rydberg J, Spahiu K, Sullivan J, Ullman W, Vitorge P and Wanner H (2001), Chemical Thermodynamics of Neptunium and Plutonium, Chemical Thermodynamics Series vol. 4, OECD Publications. Leroy P, Revil A, Altmann S and Tournassat C (2007), ‘Modelling the composition of the pore water in a clay-rock geological formation (Callovo-Oxfordian, France)’, Geochimica Cosmochimica Acta, 71(5), 1087–1097. Limousin G, Gaudet J P, Charlet L, Szenknect S, Barthe`s V and Krimissa M (2007), ‘Sorption isotherms: a review on physical bases, modelling and measurement’, Applied Geochemistry, 22, 249–275. Liu X, Lu X, Wang R, Zhou H and Xu S (2008), ‘Surface complexes of acetate on edge surfaces of 2:1 type phyllosilicate: insights from density functional theory calculation’, Geochimica Cosmochimica Acta, 72(24), 5896–5907. Lutze W and Ewing R (1988), Radioactive Wasteform for the Future, North Holland Publisher, 778 pp., ISBN 978-0444871046. MacQuarrie K M and Mayer K (2005), ‘Reactive transport modelling in fractured rock: a state-of-the-science review’, Earth-Sciences Review, 72, 189–227. Mainguy M, Tognazzi C, Torrenti J-M and Adenot F (2000), ‘Modelling of leaching in pure cement paste and mortar’, Cement and Concrete Research, 30, 83–90. Martin Ph, Ripert M, Carlot G, Parent Ph and Laffon C (2004), ‘A study of molybdenum behaviour in UO2 by X-ray absorption spectroscopy’, Journal of Nuclear Materials, 326, 132–143. Martin G, Garcia P, Labrim H, Sauvage T, Carlot G, Desgardin P, Barthe M-F and Piron J-P (2006), ‘A NRA study of temperature and heavy ion irradiation effects on helium migration in sintered uranium dioxide’, Journal of Nuclear Materials, 357, 198–205. Mathieu G, Dymitrowska M and Bourgeois M (2008), ‘Modelling of radionuclide transport through repository components using finite volume finite element and multidomain methods’, Physics and Chemistry of the Earth, 33, S216–S224. Matzke H J (1997), ‘Alpha self-irradiation of waste glasses: state of the knowledge’, in Proceedings of the Universite´ d’e´te´ CEA-Valrho, pp. 149–166. Mazurek M, Gautschi A, Marschall P, Vigneron G, Lebon P and Delay J (2008), ‘Transferability of geoscientific information from various sources (study sites, underground rock laboratories, natural analogues) to support safety cases for radioactive waste repositories in argillaceous formations’, Physics and Chemistry of the Earth, 33, S95–S105. Meeus A and Luycx P (1999), ‘Conditioning of decontamination solutions in renewed Eurobitum facility at Belgoprocess’, Radwaste Bituminization’99, in Proceedings of the International Workshop on the Safety and Performance. Evaluation of Bituminization Process for Radioactive Waste, Rez, Prague.

© Woodhead Publishing Limited, 2010

488

Geological repository systems for safe disposal

Miller W, Alexander R, Chapman R, McKinley I and Smellie J (1994), Natural Analogue Studies in the Geological Disposal of Radioactive Wastes, Studies in Environmental Science 57, Elsevier. Montarnal P, Mugler C, Colin J, Descostes M, Dimier A and Jacquot A (2007), ‘Presentation and use of a reactive transport code in porous media’, Physics and Chemistry of the Earth, 32, 507–517. Motellier S, Ly J, Gorgeon L, Charles Y, Hainos D, Meier P and Page J (2003), ‘Modelling of the ion-exchange properties and indirect determination of the interstitial water composition of an argillaceous rock. Application to the Callovo-Oxfordian low-water-content formation’, Applied Geochemistry, 18, 1517–1530. Mukhopadhyay S, Sonnenthal E L and Spycher N (2009), ‘Modelling of coupled heat transfer and reactive transport processes in porous media: application to seepage studies at Yucca Mountain, Nevada’, Journal of Porous Media, 12, 725–748. Muzeau B, Je´gou C, Delaunay F, Broudic V, Brevet A, Catalette H, Simoni E and Corbel C (2009), ‘Radiolytic oxidation of UO2 pellets doped with alphaemitters (238/239Pu)’, Journal of Alloys and Compounds, 467, 578–589. Neck V, Altmaier M and Fanghanel T (2007), ‘Solubility of plutonium hydroxides/ hydrous oxides under reducing conditions and in the presence of oxygen’, CR Chimie, 10, 959–977. Neilson R M and Dole L R (1986), ‘Chemical considerations for the immobilization of low-level radioactive waste’, in Radioactive Waste Technology, The American Society of Mechanical Engineers, pp. 317–349. Ojovan M I, Pankov A and Lee W E (2006), ‘The ion exchange phase in corrosion of nuclear waste glasses’, Journal of Nuclear Materials, 358, 57. Olander D (2004), ‘Thermal spike theory of athermal diffusion of fission products due to the alpha decay of actinides in spent fuel (UO2)’, SKB report TR-04-17. Olin A, Nola¨ng B, O¨hman L-O, Osadchii E and Rose´n E (2005), Chemical Thermodynamics of Selenium, Chemical Thermodynamics Series vol. 7, OECD Publications. Orlhac X, Fillet C, Sempere R and Phalippou J (2001), ‘An assessment of isothermal crystallization kinetics in glass by mathematical modelling’, Journal of Noncrystalline Solids, 291, 1–13. Petit J C (1992), ‘Natural analogues for the design and performance assessment of radioactive waste forms: a review’, Journal of Geochemical Exploration, 46, 1– 33. Peuget S, Cachia J-N, Je´gou C, Deschanels X, Roudil D, Broudic V, Delaye J-M and Bart J-M (2006), ‘Irradiation stability of R7T7-type borosilicate glass’, Journal of Nuclear Materials, 354, 1–13. Peuget S, De Bonfils J, Panczer G, De Ligny D, Henry S, Noel Py and Chenet A (2010), ‘Effect of chemical composition on borosilicate glass behaviour under irradiation’, Journal of Non-crystalline solids, Accepted. Peycelon H, Blanc C and Mazoin C (2006), ‘Long-term behaviour of concrete: influence of temperature and cement binders on the degradation (decalcification/hydrolysis) in saturated conditions’, Revue Europe´enne de Ge´nie Civil, 10(9), 1107–1125. Pipon Y, Toulhoat N, Moncoffre N, Raimbault L, Scheidegger A M, Farges F and

© Woodhead Publishing Limited, 2010

Post-containment performance of geological repository systems

489

Carlot G (2007), ‘Thermal diffusion of chlorine in uranium dioxide studied by secondary ion mass spectrometry and X-ray absorption spectroscopy’, Journal of Nuclear Materials, 362, 416–425. Piron J P, Pelletier, M and Pavageau J (2000), ‘Helium behaviour in spent UO2 and MOX fuels’, in IAEA International Seminar on Fission Gas Behaviour in Water Reactor Fuels, Cadarache, France, 26–29 September 2000, paper 21. Poinssot C and Gras J M (2008), ‘Key scientific issues related to the sustainable management of spent nuclear fuels in the back-end of the fuel cycle’, in Materials Research Society Symposium Proceedings on Scientific Basis for Nuclear Waste Management XXXI, vol. 1124, p. 1124–Q02-01. Poinssot C and Toulhoat P (2007), ‘From nanosecond to million years: predicting the long term fate of radionuclides in a deep geological disposal, scientific conference for the 150 years of the French Society of Chemistry’, in Invited Conference, Paris, July 2007. Poinssot C, Baeyens B and Bradbury M. (1999a), ‘Experimental and modelling studies of caesium sorption on illite’, Geochimica Cosmochimica Acta, 63, 3217–3227. Poinssot C, Toulhoat P and Gras J-M (1999b), ‘R&D needs for accounting for the long term evolution of spent nuclear fuel in interim storage and geological disposal: objectives and programs’, in GLOBAL’99: International Conference on Future Nuclear Systems, Jackson Hole, USA, 30 August–2 September 1999. Poinssot C, Jegou C, Toulhoat P, Piron J P and Gras J M (2000), ‘New approach for the RN source terms from spent nuclear fuel in geological disposal conditions’, in Materials Research Society Symposium Proceedings on Scientific Basis for Nuclear Waste Management XXIV, vol. 663. Poinssot C, Toulhoat P, Grouiller J P, Pavageau J, Piron J P, Pelletier M, Dehaudt P, Cappelaere C, Limon R, Desgranges L, Jegou C, Corbel C, Maillard S, Faure´ M H, Cicariello J C and Masson M (2001), ‘Synthesis on the long term behavior of the spent nuclear fuel’, CEA Report R-5958, 2 vols., 616 pp. Poinssot C, Toulhoat P, Gras J M and Vitorge P (2002), ‘Long term evolution of spent nuclear fuel in long term storage or geological disposal. New findings from the French PRECCI R&D program and implications for the definition of the RN source term in geological repository’, Journal of Nuclear Science and Technology, Suppl. 3, 473–476. Poinssot C, Lovera P, Ferry C and Gras J M (2003), ‘Consequences of the anticipated long term evolution of spent nuclear fuel for the assessment of the release rate of radionuclides’, in Materials Research Society Symposium Proceedings on Scientific Basis for Nuclear Waste Management XXVI, vol. 757, pp. 35–42. Poinssot C, Lovera P and Ferry C (2004), ‘Radionuclide release model for spent nuclear fuel in geological disposal for performance assessment’, in Materials Research Society Symposium Proceedings on Scientific Basis for Nuclear Waste Management XXVIII, vol. 824, pp. 71–79. Poinssot C, Ferry C, Lovera P, Je´gou C, and Gras J-M (2005a), ‘Spent fuel radionuclide source term model for assessing spent fuel performance in geological disposal. Part II: Matrix alteration model and global performance’, Journal of Nuclear Materials, 346(2005), 66–77. Poinssot C, Grambow B, Kelm M, Spahiu K, Martinez A, Johnson L, Cera E, de

© Woodhead Publishing Limited, 2010

490

Geological repository systems for safe disposal

Pablo J, Quinones J, Wegen D, Lemmens K and McMenamin T (2005b), ‘Mechanisms governing the release of radionuclides from spent nuclear fuel in geological repository: major outcomes of the European Project SFS’, in Materials Research Society Symposium Proceedings on Scientific Basis for Nuclear Waste Management XXIX, Ghent, Belgium. Poinssot C, Ferry C and Poulesquen A (2006), ‘New trends in the field of spent nuclear fuel radionuclides release in a deep geological repository’, in Materials Research Society Symposium Proceedings on Scientific Basis for Nuclear Waste Management XXX, vol. 985, pp. 111–116. Pointeau V, Deditius A P, Miserque F, Renock D, Becker U, Zhang J, Clavier N, Dacheux N, Poinssot C and Ewing R C (2009), ‘Synthesis and characterization of coffinite’, Journal of Nuclear Materials, 393, 449–458. Pourcelot L and Gauthier-Lafaye F (1999), ‘Hydrothermal and supergene clays of the Oklo natural reactors: conditions of radionuclide release, migration and retention’, Chemistry and Geology, 157, 155–174. Quinones J, Iglesias E, Rodriguez N and Nieto J (2005) ‘Modelling of the spent fuel dissolution rate evolution for repository conditions. Matrix Alteration Model results and sensitivity analyses’, in Materials Research Society Symposium Proceedings on Scientific Basis for Nuclear Waste Management XXIX, vol. 932, pp. 433–440. Quinones J, Iglesias E, Rodriguez N and Nieto J (2009), ‘Influence of the specific surface area on spent nuclear fuel dissolution rates’, in Materials Research Society Symposium Proceedings on Scientific Basis for Nuclear Waste Management XXXII, vol. 1124, pp. 99–105. Rabung T, Pierret M C, Bauer A, Geckeis H, Bradbury M and Baeyens B (2005), ‘Sorption of Eu(III)/Cm(III) on Ca-montmorillonite and Na-illite. Part 1: Batch sorption and time-resolved laser fluorescence spectroscopy experiments’, Geochimica Cosmochimica Acta, 69(23), 5393–5402. Rande M, Fuger J, Grenthe I, Neck V and Rai D (2009), Chemical Thermodynamics of Thorium, Chemical Thermodynamics Series vol. 11, OECD Publications. Rard J, Rand M H, Anderegg G and Wanner H (1999), Chemical Thermodynamics of Technetium, Chemical Thermodynamics Series vol. 3, Elsevier. Redonnet C, Leroy B and Masson H (1999), ‘The bituminization in La Hague Plant. Description of the process and of the fire risk prevention’, Radwaste Bituminization’99, in Proceedings of the International Workshop on the Safety and Performance Evaluation of Bituminization Process for Radioactive Waste, Rez, Prague. Reiller P, Evans N D M and Szabo G (2008), ‘Complexation parameters for the actinides(IV)-humic acid system: a search for consistency and application to laboratory and field observations’, Radiochimica Acta, 96(6), 345–358. Reilly S, Runde W and Neu M (2007), ‘Solubility of plutonium(VI) carbonate in saline solutions’, Geochimica Cosmochimica Acta, 71, 2672–2679. Ribet I and Gin S (2004), ‘Role of neoformed phases on the mechanisms controlling the resumption of SON68 glass alteration in alkaline media’, Journal of Nuclear Materials, 324, 152. Ribet I, Godon N, Gin S, Minet Y, Jollivet P, Frugier P, Vernaz E, Cavedon J M and Petitjean V (2004), ‘The V0–Vr operational model for the long term

© Woodhead Publishing Limited, 2010

Post-containment performance of geological repository systems

491

behaviour of vitrified R7T7 waste packages’, Atalante 2004 Conference Proceedings on Advances for Future Nuclear Fuel Cycles, Nimes, France, p.1. Riglet-Martial C, Adenot F, Camaro S and Blanc V (2005), A New Chemical Process to Prevent the Swelling of the Radioactive Waste Packages in Organic Media, Global, Stukuba, Japan. Ronchi C and Hiernaut J-P (2004), ‘Helium diffusion in uranium and plutonium oxides’, Journal of Nuclear Materials, 325, 1–12. Rosenberg N D, Gdouski G E and Knauss K G (2001), ‘Evaporative chemical evolution of natural waters at Yucca Mountain, Nevada’, Applied Geochemistry, 16, 1231–1240. Rotenberg B, Marry V, Dufreˆche J F, Giffaut E and Turq P (2007a), ‘A multiscale approach to ion diffusion in clays: building a two-state diffusion–reaction scheme from microscopic dynamics’, Journal of Colloid and Interface Science, 309, 289–295. Rotenberg B, Marry V, Dufreˆche J F, Malikova N, Giffaut E and Turq P (2007b), ‘Modelling water and ion diffusion in clays: a multiscale approach’, CR Chimie, 10, 1108–1116. Rotenberg B, Morel J P, Marry V, Turq P and Morel-Desrosiers N (2009), ‘On the driving force of cation exchange in clays: insights from combined microcalorimetry experiments and molecular simulation’, Geochimica Cosmochimica Acta, 73, 4034–4044. Roudil D, Deschanels X, Trocellier P, Je´gou Ch, Peuget S and Bart J-M (2004), ‘Helium thermal diffusion in a uranium dioxide matrix’, Journal of Nuclear Materials, 325, 148–158. Roudil D, Je´gou Ch, Deschanels X, Peuget S, Raepsaet C, Gallien J P and Broudic V (2006), ‘Effects of alpha self-irradiation on actinide-doped spent fuel surrogate matrix’, in Materials Research Society Symposium on Scientific Basis for Nuclear Waste Management XXIX, vol. 932, pp. 529–536. Roudil D, Je´gou Ch, Broudic V, Muzeau B, Peuget S and Deschanels X (2007), ‘Gap and grain boundaries inventories from pressurized water reactor spent fuels’, Journal of Nuclear Materials, 362, 411–415. Roudil D, Bonhoure J, Pik R, Cuney M, Je´gou C and Gauthier-Lafaye F (2008), ‘Diffusion of radiogenic helium in natural uranium oxides’, Journal of Nuclear Materials, 378, 70–78. Runde W, Conradson S, Efurd D, Lu N, VanPelt C and Tait C (2002), ‘Solubility and sorption of redox-sensitive radionuclides (Np, Pu) in J-13 water from the Yucca Mountain site: comparison between experiment and theory’, Applied Geochemistry, 17, 837–853. Sattonnay G, Ardois C, Corbel C, Lucchini J-F, Barthe M-F, Garrido F and Gosset D (2001), ‘Alpha-radiolysis effects on UO2 alteration in water’, Journal of Nuclear Materials, 288, 11–19. Schindler P W and Stumm W (1987), ‘The surface chemistry of oxides, hydroxides and oxide minerals’, in Aquatic Surface Chemistry, edited by W Stumm, WileyInterscience, New York. Schlegel M, Charlet L, Chateigner D and Manceau A (1999a), ‘Sorption of metal ions on clay minerals I. Polarized EXAFS evidence for the adsorption of Co on the edges of hectorite particles’, Journal of Colloid and Interface Science, 215, 140–158.

© Woodhead Publishing Limited, 2010

492

Geological repository systems for safe disposal

Schlegel M, Charlet L and Manceau A (1999b), ‘Sorption of metal ions on clay minerals II. Mechanism of Co sorption on hectorite at high and low ionic strength and impact on the sorbent stability’, Journal of Colloid and Interface Science, 220, 392–405. Schlegel M, Manceau A, Charlet L, Chateigner D and Hazemann J L (2001), ‘Sorption of metal ions on clay minerals. III. Nucleation and epitaxial growth of Zn phyllosilicate on the edges of hectorite’, Geochimica Cosmochimica Acta, 65(22), 4155–4170. Sercombe J, Adenot F, Tiffreau C, Gwinner B and Felines I (2004), ‘Modelling of water uptake and salt leaching in bituminized radioactive waste: comparison between numerical, analytical and experimental results’, Materials Research Society Symposium Proceedings, 807, 377–382. Sercombe J, Gwinner B, Tiffreau C, Simondi-Teisseire B and Adenot F (2006), ‘Modelling of bituminized radioactive waste leaching, Part 1: Constitutive equations’, Journal of Nuclear Materials, 349, 96–106. Shoesmith D W (2000), ‘Fuel corrosion processes under waste disposal conditions’, Journal of Nuclear Materials, 282, 1–31. Shoesmith D W, Kolar M and King F (2003), ‘A mixed-potential model to predict fuel (uranium dioxide) corrosion within a failed nuclear waste container’, Corrosion, 59, 802–816. Silva R, Bidoglio G, Rand M H, Robouch P, Wanner H and Puigdomenech I (1995), Chemical Thermodynamics of Americium, Chemical Thermodynamics Series vol. 2, Elsevier. Spahiu K, Werme L and Eklund U B (2000), Radiochimica Acta, 88, 507–511. Spahiu K, Cui D and Lundstrom M (2004), ‘The fate of radiolytic oxidants during spent fuel leaching in the presence of dissolved near field hydrogen’, Radiochimica Acta, 92, 625–629. Stumm W and Morgan J J (1996), Aquatic Chemistry, Wiley Interscience Publication, 1022 pp. Sunder S, Miller N H and Shoesmith D W (2004), ‘Corrosion of uranium dioxide in hydrogen peroxide solutions’, Corrosion Science, 46, 1095–1111. Taylor H F W (1997), Cement Chemistry, Academic Press. Tertre E, Berger G, Simoni E, Castet E, Giffaut E, Loubet M and Catalette H (2006), ‘Europium retention onto clay minerals from 25 to 150 8C: experimental measurements, spectroscopic features and sorption modelling’, Geochimica Cosmochimica Acta, 70(18), 4563–4578. Tertre E, Beaucaire C, Coreau N and Juery A (2009), ‘Modelling Zn(II) sorption onto clayey sediments using a multi-site ion-exchange model’, Applied Geochemistry, 24, 1852–1861. Tevissen E, Soler J M, Montarnal P, Gautschi A and Van Loon L R (2004), ‘Comparison between in situ and laboratory diffusion studies of HTO and halides in Opalinus Clay from Mont Terri’, Radiochimica Acta, 92, 781–786. Toulhoat P (2002), ‘Confinement and migration of radionuclides in a nuclear waste deep repository’, CR Physique, 3, 975–986. Toulhoat P, Grambow B and Simoni E (2005), ‘Migration ou confinement des radioe´le´ments?’, L’Actualite´ Chimique, 285–286, 41–53. Tournassat C, Ferrage E, Poinsignon C and Charlet L (2003), ‘The titration of clay

© Woodhead Publishing Limited, 2010

Post-containment performance of geological repository systems

493

minerals II. Structure-based model and implications for clay reactivity’, Journal of Colloids Interface Science, 273, 234–246. Tournassat C, Neaman A, Villieras F, Bosbach D and Charlet L (2004), ‘Nanomorphology of montmorillonite particles: estimation of the clay edge sorption site density by low-pressure gas adsorption and AFM observations’, American Mining, 88, 1989–1995. Tournassat C, Gaucher E, Fattahi M and Grambow B (2007), ‘On the mobility and potential retention of iodine in the Callovian-Oxfordian formation’, Physics and Chemistry of the Earth, 32, 539–551. Vernaz E (2002), ‘Role of neoformed phases on the mechanisms controlling the resumption of SON68 glass alteration in alkaline media’, CR Physique, 3, 813. Vitorge P and Poinssot C (2005), ‘Comparaison des comportements chimiques du neptunium et du plutonium dans des eaux naturelles’, L’Actualite´ Chimique, 285–286, 54–59. Voigt W, Brendler V, Marsh K, Rarey R, Wanner H, Gaune-Escard M, Cloke P, Vercouter Th, Bastrakov E and Hagemann S (2007), ‘Quality assurance in thermodynamic databases for performance assessment studies in waste disposal’, Pure Applied Chemistry, 79, 883–894. Weber W J and Roberts F P (1983), ‘A review of radiation effects in solid nuclear waste forms’, Nuclear Technology, 60, 178–198. Weber W J, Ewing R C, Angell C A, Arnold G W, Cormack A N, Delaye J M, Griscom D L, Hobbs L W, Navrotsky A, Price D L, Stoneham A M and Weinberg M C (1997), ‘Radiation effects in glasses used for immobilization of high-level waste and plutonium disposition’, Journal of Materials Research, 12, 1946–1978. Wersin P, Soler J M, Van Loon L, Eikenberg J, Baeyens B, Grolimund D, Gimmi T and Dewonck S (2008), ‘Diffusion of HTO, Br, I, Cs+, Sr-85(2+) and Co60(2+) in a clay formation: results and modelling from an in situ experiment in Opalinus Clay’, Applied Geochemistry, 23, 678–691. Wren J C, Shoesmith D W and Sunder S (2005), ‘Corrosion behavior of uranium dioxide in alpha radiolytically decomposed water’, Journal of the Electrochemical Society, 152, B470–B481. Wronkiewicz D, Watkins C, Baughman A and Miller F (2001), in Materials Research Society Symposium Proceedings, vol. 713, p. 625–632. Zhang K, Wu Y S and Houseworth J E (2006), ‘Sensitivity analysis of hydrological parameters in modelling flow and transport in the unsaturated zone of Yucca Mountain, Nevada, USA’, Hydrogeology Journal, 14, 1599–1619.

© Woodhead Publishing Limited, 2010

15 Safety assessment for deep geological disposal of high-level radioactive waste in geological repository systems P . N . S W I F T , Sandia National Laboratories, USA

Abstract: Safety assessments estimate the long-term performance of geologic repositories for radioactive waste using quantitative models. This paper reviews regulatory standards, the iterative relationship between site characterization and safety assessment, selection of scenarios for analysis and the development of computational models and their linkage into a system analysis. Uncertainty must be acknowledged, and can be accounted for using both conservative deterministic and probabilistic approaches. In addition to generating performance estimates for comparison to regulatory standards, safety assessments can also guide research and model development, evaluate design alternatives, enhance the scientific understanding of the system and contribute to public acceptance. Key words: safety assessment, performance assessment, geologic repositories, spent nuclear fuel, high-level radioactive waste.

15.1

Introduction

Discussions of deep geological disposal of high-level radioactive waste invariably come to the question ‘Is it safe?’, or perhaps to variants of the question such as ‘How safe is it?’ or ‘How can we be sure it is safe?’ These questions are central to considerations of geological disposal regardless of the perspective from which they are proposed: industries and agencies responsible for the management or disposal of wastes, regulators charged with protecting public health and safety, and both opponents of and advocates for specific disposal options all want credible answers to these questions. Ensuring the safety of both workers and the general public while 497 © Woodhead Publishing Limited, 2010

498

Geological repository systems for safe disposal

geological repositories are operating also requires careful attention, but public concerns have tended to focus on the long-term safety of repositories after they have been decommissioned. This emphasis on long-term risks may in part be a function of familiarity; operational safety can be observed and monitored directly, and preclosure risk assessments for repositories are therefore analogous to those undertaken for nuclear power plants or other industrial facilities. Ensuring the safety of repositories over the long geological time period during which significant radioactivity may remain has generally been perceived, however, to be a more difficult problem. How can we make meaningful estimates of the hazards associated with a geological repository for radioactive waste over thousands or hundreds of thousands of years? As is the case for many complex problems in science and engineering, computer modeling provides at least a partial answer. There is general consensus among the nations that manage long-lived radioactive waste that meaningful estimates of long-term performance of repositories are both possible and necessary to support the decision-making process. The importance of modeling based on thorough characterization of the disposal site and accepted principles of physics and chemistry was recognized early in the consideration of repositories (e.g. National Research Council, 1978, p. 4), and disposal programs in many nations undertook quantitative modeling efforts, referred to generally (and synonymously) as safety assessments or performance assessments, to support repository safety analyses (e.g. NEA, 1997). Acknowledging these ongoing efforts, the International Atomic Energy Commission (IAEA) advisory standards for the Geological Disposal of Radioactive Waste, developed jointly with the Organisation for Economic Co-operation and Development Nuclear Energy Agency (OECD/NEA), have defined the process for quantifying performance estimates as ‘safety assessment’ (IAEA, 2006, Section 3.41): Safety assessment is the process of systematically analysing the hazards associated with the facility and the ability of the site and the design of the facility to provide for the safety functions and to meet technical requirements. Safety assessment includes quantification of the overall level of performance, analysis of the associated uncertainties and comparison with the relevant design requirements and safety standards. The specification in the IAEA definition that safety assessments must include an analysis of associated uncertainties is consistent with the recognition that the uncertainties inherent in modeling the far future are potentially large, and must be fully acknowledged in any credible analysis. The IAEA definition in this regard parallels approaches taken in many national programs to acknowledging and accounting for uncertainty (NEA,

© Woodhead Publishing Limited, 2010

Safety assessment for deep geological disposal

499

2004a, 2004b). For example, as early as the middle 1980s the regulatory framework in the United States required a performance assessment that: (1) Identifies the processes and events that might affect the disposal system; (2) examines the effects of these processes and events on the performance of the disposal system; and (3) estimates the cumulative releases of radionuclides, considering the associated uncertainties, caused by all significant processes and events. These estimates shall be incorporated into an overall probability distribution of cumulative releases to the extent practicable (US EPA, 1985, 40 CFR 191.12(2)). This definition remains essentially unchanged in the current United States regulatory framework governing geological repositories (US NRC, 2008, 10 CFR 63.2), and has proven to be an important factor in determining the overall approach to probabilistic repository safety assessments in the United States. Although not all national programs have chosen probabilistic approaches to safety assessments, there is widespread agreement on the need for a consideration of uncertainty (NEA, 2004a). This chapter focuses on the implementation of safety assessments for repositories, examining those aspects that are common to all assessments and using specific examples to demonstrate alternative approaches to achieving useful estimates of future performance. Sections of the chapter comment on the goals of safety assessments, steps in the process of conducting an assessment, approaches to acknowledging and interpreting uncertainty and applications of the results of safety assessments. The chapter concludes with observations on possible future trends in safety assessment and a discussion of sources of additional information for interested readers.

15.2

Goals of a safety assessment

Overall, the primary goal of a safety assessment is straightforward: provide an estimate of the long-term safety of the potential facility. At a finer level of detail, the specifics of how this goal is defined become important. How should we measure safety? How safe is safe enough? Who, or what, is to be protected, and for how long? What level of uncertainty is acceptable? To what extent can future generations be protected from willful human disruption of the disposal system? In general, answers to these questions are societal, or perhaps political, decisions for which science informs the answer but does not solely determine it. As described elsewhere in this volume in more detail (see Chapters 20 and 21), safety can be evaluated in terms of the risk of future health effects, in terms of estimates of future radiation doses or in terms of releases of radiation from the disposal system. Protection standards can be established for environmental criteria (e.g. concentrations of radioactive materials in groundwater) or to protect hypothetical future

© Woodhead Publishing Limited, 2010

500

Geological repository systems for safe disposal

members of the general public. Standards could apply for as long as the wastes remain hazardous (potentially many millions of years) or for shorter periods of specified duration, reflecting beliefs about both the merits of regulating the distant future and the usefulness of attempting quantitative modeling over extraordinarily long times. Examples exist internationally of each of these approaches. The Waste Isolation Pilot Plant (WIPP) currently operating in the United States for disposal of transuranic wastes has regulatory limits set on cumulative radionuclide releases for 10 000 years (US EPA, 1985; US DOE, 1996), the proposed Yucca Mountain high-level radioactive waste repository in the United States has regulatory limits set on mean annual radiation doses to a hypothetical future individual for 1 000 000 years (US DOE, 2008; US EPA, 2008; US NRC, 2008) and preliminary safety assessments for potential candidate sites in Sweden address the risk of harmful human health effects for up to 1 000 000 years (SSI, 1998; SKB, 2006a). Other nations set limits on maximum allowable radiation doses or impacts on human health or the environment for an unlimited period of time (e.g. Canada (CNSC, 2004), Switzerland (HSK and KSA, 1993) and others (see NEA, 2006)). With the exception of the cumulative release standard used for the WIPP, the basic question of ‘how safe is safe enough’ is most commonly addressed through risk- or dose-based standards. Internationally, risk-based limits range from probabilities of harmful health effects of 105/year to 106/year and dose-based limits are generally from 0.1 mSv/year to 1 mSv/year (ICRP, 2007, Section 6.1.3; NEA, 2007, Appendix 2), with 0.3 mSv/year being the value recommended by the IAEA for the time before ‘uncertainties become so large that the criteria may no longer serve as a reasonable basis for decision making’ (IAEA, 2006, Section 2.12). Many nations define the time during which quantitative limits apply at 10 000 years or less; the United States defines a 10,000-year standard of 0.15 mSv/year and a separate, and higher, dose standard of 1 mSv/year to apply between 10 000 years and 1 000 000 years (US EPA, 2008). Several nations also set higher standards (up to 1 mSv/year) for scenarios involving human intrusion into the repository (NEA, 2007, Appendix 2). Although each of these regulatory approaches shares a common goal of protecting future humans and the environment until such time as radioactive decay has reduced the hazard posed by the waste, differences in how regulations are structured can have significant influence on how safety assessments are designed, and, for that matter, on how repositories are sited. For example, setting limits on the allowable cumulative release of radionuclides normalized to the total inventory of the repository, as in 40 CFR 191 (US EPA, 1985), can encourage the development of centralized repositories with large inventories by allowing proportionally larger releases from larger facilities, and avoids creating a regulatory incentive for multiple

© Woodhead Publishing Limited, 2010

Safety assessment for deep geological disposal

501

dispersed repositories with smaller inventories that might each individually result in smaller estimated releases. Regulatory approaches that specify a cumulative release also rely on a specification of a fixed time period, thereby encouraging the selection of sites that emphasize isolation of all radioactive materials during that period. Regulatory approaches that set limits on the maximum allowable annual doses or health risks to an individual during a specified time period, as in the United States high-level waste program and in Sweden, encourage isolation during that time period, but they can also encourage the selection of sites that show a lower estimated maximum annual dose due to either a gradual release of radionuclides through time or a dilution of radionuclides in the environment. Regulations that apply a single limit regardless of the initial inventory of the repository unavoidably carry a potential for encouraging the selection of multiple smaller disposal sites. Regulations may also prescribe assumptions to be made in the safety assessment regarding the characteristics of the future humans potentially at risk. Should they be assumed to be like the humans in the region today? Should they be assumed to have a lifestyle such as subsistence farming that might lead to higher risk? Should they be assumed to have sufficient technology to recognize and avoid or mitigate radiation risks? Answers to these questions, and others like them that address unknowable aspects of future human behavior, will define essential boundary conditions for any safety assessment, and in the absence of clear regulatory guidance, reasonable assumptions should be made and clearly stated. Typically, safety assessments tend to rely on cautious assumptions about human culture and knowledge, assuming that the receptor groups will have lifestyles that result in at least as much risk of exposure as those of today, and that future generations will eventually cease to be aware of the dangers posed by the site. Acts of deliberate sabotage or warfare in the far future are generally considered to be outside the scope of what environmental regulations can control, although many safety assessments include some consideration of the consequences of inadvertent human disruption of the site (NEA, 2007, Appendix 2).

15.3

Steps in a typical safety assessment

Once a candidate repository site has been identified and the goals of the safety assessment have been defined, safety assessments typically develop through four steps, shown in Fig. 15.1. Although they are discussed here as if they were sequential activities, in practice, all four activities are undertaken together, and the safety assessment tends to develop and mature through multiple iterations. First, basic information about the potential disposal system must be developed through characterization of the geological setting, the waste and a conceptual design of the repository.

© Woodhead Publishing Limited, 2010

502

Geological repository systems for safe disposal

15.1

Four steps in a typical safety assessment.

Second, scenarios must be developed that are representative of possible future states of the system relevant to the safety assessment. Third, computational models are developed that can simulate the features, events and processes relevant to the chosen scenarios, incorporating uncertainty about the behavior of the system. Fourth, computational models are combined into a system-level analysis that provides estimates of overall performance that supports programmatic and regulatory decisions.

15.3.1 Characterization of the disposal system Characterization of each of the three main components of a disposal system (i.e. the waste form, the geological setting and the repository design) is often seen as three separate lines of research, and responsibility for these tasks is often split among different entities within a single program. The safety assessment needs input from all three equally, however, and overall performance depends on the integration of design with the site geology and the wastes intended for disposal. The waste form itself defines the total radionuclide inventory for the repository and also impacts performance through its physical and chemical characteristics. For example, the heat generated by radioactive decay of spent nuclear fuel can be an important factor in repository design and performance, and repositories for spent fuel may need different design concepts than those primarily intended for vitrified waste from reprocessing activities.

© Woodhead Publishing Limited, 2010

Safety assessment for deep geological disposal

503

15.3.2 Selection of scenarios for analysis It is neither practical nor possible to analyze all possible future states of a repository system, and essentially all safety assessments choose instead to focus their efforts on a relatively small set of scenarios that are broadly representative of important aspects of the range of future conditions. Multiple approaches have been used to develop scenarios with varying degrees of formalism and documentation (NEA, 1999), but all approaches share common goals. Scenarios selected for analysis must be suitable for quantitative analysis with computer models and the set of scenarios must satisfy programmatic or regulatory requirements regarding the comprehensiveness of the assessment. Formal proof that a scenario development process has considered the complete set of all possible future events is impossible and emphasis instead is placed on documenting a sound process that demonstrates comprehensiveness. If any aspect of the future needs to be considered quantitatively in the safety assessment, it must be accounted for in one or more of the scenarios chosen for analysis. If relevant scenarios can be identified that are not accounted for in the existing analyses and that cannot be shown to be insignificant, the set of scenarios should be expanded to include them. One commonly adopted approach to demonstrating the comprehensiveness of the scenario development process begins with the identification of a list of all factors potentially relevant to the long-term performance of the disposal system. Typically, these factors have been categorized as features, events or processes, using common-sense definitions for the terms. For example, the US NRC (2003, Section 3) defines a feature to be ‘an object, structure, or condition that has a potential to affect disposal system performance’, an event to be ‘a natural or human-caused phenomenon that has a potential to affect disposal system performance and that occurs during an interval that is short compared with the period of performance’ and a process to be ‘a natural or human-caused phenomenon that has a potential to affect disposal system performance and that operates during all or a significant part of the period of performance’. In practice, the distinction of whether a specific phenomenon is a feature, an event or a process is secondary to the recognition that, whatever it is, it needs to be evaluated, and features, events and processes are often grouped together with the generic acronym ‘FEPs’. As shown in Fig. 15.2, after potentially relevant ‘FEPs’ are identified and cataloged in the initial comprehensive list, they are then evaluated, or screened, using previously specified criteria such as the probability of occurrence or the significance of the consequences of the FEP, and FEPs that meet the screening criteria are then used to construct the scenarios for analysis. The comprehensiveness of the process is demonstrated by the

© Woodhead Publishing Limited, 2010

504

Geological repository systems for safe disposal

15.2 Using a feature, event and process list to develop scenarios for safety assessments.

comprehensiveness of the initial FEP list and the documentation of the screening evaluations. All potentially relevant FEPs should be mappable to an FEP on the initial list and there should be clear traceability to documentation of either how the FEP has been accounted for in the systemlevel analysis or how its exclusion has been justified. The initial lists of potentially relevant FEPs are unavoidably subjective, in the sense that any individual FEP can be subdivided narrowly or lumped coarsely (e.g. a single FEP could be identified as ‘colloidal transport of radionuclides’ or many tens of separate FEPs could be defined if transport of each radionuclide in each environment is treated separately). In practice, subjective distinctions like this make little difference in the comprehensiveness of the list; a single FEP describing how colloidal transport has been evaluated for each radionuclide of interest is fully equivalent to multiple separate FEPs. In general, FEPs are most useful if they are defined at the broadest level for which a coherent technical discussion can be presented; little value is added by defining thousands of similar FEPs each of which will require separate documentation. FEP lists have been developed using various techniques, including formal elicitation using top-down classification schemes, freely structured brainstorming and reviews of relevant literature. FEP lists have been published by the NEA and multiple repository programs (NEA, 2000, and references cited therein; SKB, 2006b; SNL, 2008a, 2008b). The use of an FEP list to demonstrate comprehensiveness of the analysis is an expected part of the regulatory process in the United States (US NRC, 2003, Section 2.2.1.2.1.2) and the FEP analyses have been used in regulatory applications for both the WIPP (US DOE, 1996) and the proposed Yucca Mountain repository (US DOE, 2008).

© Woodhead Publishing Limited, 2010

Safety assessment for deep geological disposal

505

Screening criteria for FEPs are most useful if they are explicitly documented and agreed upon in advance by regulators and stakeholders. In the United States, screening criteria are provided directly in the US EPA regulations: FEPs may be excluded from a performance assessment if they can be shown to have an annual probability of occurrence less than one chance in 100 000 000 or if it can be shown that results of the performance assessment would not change significantly if the FEP were omitted (US EPA, 2008, 40 CFR 197.36(a)(1)). If formal FEP screening criteria are not specified by regulations, FEP analyses may still provide valuable justification for the choices made in constructing the system-level models, and documentation of the FEP process is an important step in building confidence in the safety assessment. Additional confidence in the completeness of the treatment of FEPs that are included in the system-level modeling can be gained through systematic evaluation of possible interactions. Both influence diagrams and interaction matrices have been used to confirm that safety assessment models provide a comprehensive treatment of plausible interactions among FEPs (see, for example, SKI, 1996, and Pers et al., 1999). Analyses such as these help identify coupled processes that may have been overlooked in initial screening. Grouping of FEPs retained for analysis into scenarios can logically be done by mapping FEPs to the safety functions of the associated components (both natural and engineered barriers) (e.g. ONDRAF/ NIRAS, 2001, Section 11.2). This approach allows identification of those FEPs that are potentially important to the performance of the system and the evaluation of their impacts, either quantitatively or qualitatively through separate analyses (e.g. ANDRA, 2005a, Section 6.1.5.5).

15.3.3 Developing computational models for relevant processes Once preliminary scenarios have been selected for analysis, computational models are developed to simulate the major physical and chemical processes relevant to the behavior of the system. At a minimum, safety assessments require models for the behavior of engineered barriers in the geological setting, including degradation mechanisms, release of radionuclides from the waste form and engineered barriers and the transport of radionuclides away from the repository to the human environment. Full development of a safety assessment is likely to require detailed models for groundwater flow in the region, corrosion and degradation of engineered materials in a changing chemical environment, dissolution and mobilization of waste forms and contaminant transport both as dissolved and colloidal species. Depending on programmatic and regulatory requirements, models may also be needed

© Woodhead Publishing Limited, 2010

506

Geological repository systems for safe disposal

for human exposure pathways and estimation of radiation doses. Models may need to consider boundary conditions that change through time, e.g. because of long-term climate change, and they should take into account coupled interactions such as those between hydrology, chemistry and the heat generated by the waste. Sufficient data must be collected to support the development and parameterization of these models, creating a need for ongoing iterations between model development and site characterization work. As discussed later in Section 15.4, uncertainty in both the choice of models and the parameter values used in them should be acknowledged and characterized. In many programs, development of these process models, and the information needed to provide values for their input parameters, is the largest part of the work needed to support the safety assessment. Some programs have found it advantageous to construct underground research laboratories dedicated in part to developing and testing process models specific to the performance of their potential repository concept (see Chapter 4 of this book).

15.3.4 The system-level analysis Process models are used to build a system-level model capable of estimating the overall performance measures of interest. Depending on the complexity of the processes and the available computational resources, the system-level model may be built by directly coupling process models, by linking process models by transferring inputs and outputs as response surfaces or look-up tables, or by developing and coupling simplified abstractions of the process models that capture only the most important processes. In practice, systemlevel models are typically constructed using a hybrid approach that involves multiple techniques. Because safety assessments develop over periods of several years and add different component models at different stages of development, the final products may reasonably be expected to be composites with varying levels of complexity. The performance assessment model used for the 1996 WIPP Compliance Certification Application (US DOE, 1996) is illustrated schematically in Figs 15.3 and 15.4 as a representative example of a relatively simple systemlevel model that relied on direct coupling of computer codes, transfer of information among models via response surfaces, and the use of simplified abstractions for computational efficiency. Although specific details of the implementation of the 1996 WIPP performance assessment are beyond the scope of this chapter, concepts introduced in the summary below are broadly applicable to many safety assessment system models. As described in detail in Section 6.4 of US DOE, 1996 (see also Helton and Marietta, 2000), computational models were developed for the major

© Woodhead Publishing Limited, 2010

Safety assessment for deep geological disposal

507

15.3 Schematic view of the system model for the 1996 Waste Isolation Pilot Plant performance assessment (from US DOE, 1996, Figure 6-26, p. 6-176).

15.4 Linkage of the major codes in the 1996 Waste Isolation Pilot Plant performance assessment (from US DOE, 1996, Figure 6-25, p. 6-175).

© Woodhead Publishing Limited, 2010

508

Geological repository systems for safe disposal

processes of interest in each of the components of the disposal system. As shown schematically in Fig. 15.3, the BRAGFLO code simulated two-phase fluid flow (gas and brine) in the repository and surrounding host rock. Three codes, FMT, PANEL and NUTS, were used to estimate radionuclide concentrations in brine within the repository and in near-field rocks. One code, SANTOS, was used to characterize creep deformation in the salt surrounding the excavated areas. Two codes, CUTTINGS_S and BRAGFLO_DBR, were used to estimate the release of radionuclides to the land surface following a hypothetical drilling intrusion event and two additional codes, SECO-FL2D and SECOTP2D, were used to model groundwater flow and contaminant transport away from the intruding borehole in the Culebra dolomite aquifer overlying the repository. One code, GRASP-INV, was used to develop multiple geostatistical realizations of possible transmissivity fields in the Culebra aquifer. Figure 15.4 shows how these codes were linked computationally to construct a system model. The major flow and transport codes (shown in the shaded box in Fig. 15.4) were linked directly into a single simulation tool that drew input parameters from a common database. Transfer of information among these codes was fully automated. Two codes, the detailed process model NUTS and the simplified abstraction PANEL, were available in the performance assessment model for estimating radionuclide concentrations, and for most purposes PANEL was used for computational efficiency. Direct couplings of rock creep processes (using SANTOS) and geochemical processes (using FMT) were not included in the performance assessment, in part because of the complexity of the couplings and in part because preliminary analyses indicated that sufficient resolution could be obtained by running these models independently as standalone models and then providing their output to the relevant performance assessment components as response surfaces. Similarly, the GRASP-INV code was run independently of the performance assessment to generate multiple realizations of transmissivity fields that were then sampled for use in the system-level analysis. Output from the linked component models was then used as input to a post-processing code, CCDF_GF, that compiled individual consequence results for individual scenarios into the probabilistic outcomes summed over all scenarios that were required for comparison to regulatory standards. The 1996 WIPP performance assessment was constructed to allow a Monte Carlo approach to the treatment of uncertainty (see Section 15.4), facilitating the relatively rapid calculation of large numbers of realizations. As such, the system-level model represented a compromise between the level of resolution and realism possible through the use of increasingly detailed process models and the computational efficiency needed to run hundreds or thousands of realizations in a reasonable period of time. This compromise

© Woodhead Publishing Limited, 2010

Safety assessment for deep geological disposal

509

appears, in varying forms, in all safety assessments. Computational power has increased through time, but so too has the complexity possible in process models, and, as discussed further in Section 15.6, it appears that safety assessments will always need to rely on system models that are a judicious simplification of the full understanding developed at the level of individual processes. The 2008 Yucca Mountain Total System Performance Assessment (US DOE, 2008, Section 2; SNL, 2008c) relied heavily on the use of response surfaces developed by detailed process models for input to a simplified linkage of abstractions, but even so, the overall system model for the proposed Yucca Mountain repository is significantly more complex than the WIPP model shown in Fig. 15.4.

15.3.5 The iterative nature of safety assessments Early iterations of safety assessments provide critical information to decision makers regarding the viability of the site and facility design, informing interim decisions regarding the future direction of the program. Preliminary safety assessments can help guide research and design activities by confirming the features of the disposal system that make the greatest contributions to performance and by identifying the uncertainties associated with those features that have the largest impact on confidence in the performance estimates. Sensitivity and uncertainty analyses, in particular, can be powerful tools for focusing research activities on the most important uncertainties, as discussed in the following section. As site characterization and repository design activities progress and safety assessments mature, long-term performance estimates become increasingly suitable for direct comparison of programmatic goals and regulatory standards. Details of the regulatory process differ from nation to nation, but in all programs the safety assessment ultimately becomes a key component of the safety case, informing the final decision to operate the repository. As such, final iterations of the safety assessment must meet high standards for both technical excellence and thorough documentation, and must be suitable for review in a legal and regulatory environment.

15.4

Acknowledging uncertainty in a safety assessment

Uncertainty unavoidably enters into safety assessments, and any other large modeling study, from multiple sources. Sources of uncertainty can be categorized in various ways; what is important, from the perspective of the utility and credibility of the safety assessment is that uncertainty from all sources be openly acknowledged and accounted for in the analysis. Chapman and McCombie (2003) usefully describe four primary types of uncertainty: system uncertainty associated with incomplete characterization

© Woodhead Publishing Limited, 2010

510

Geological repository systems for safe disposal

of the disposal system; scenario uncertainty associated with the comprehensiveness of the scenarios chosen for analysis; model uncertainty associated with the choice of conceptual and computational models used to represent the behavior of the system; and parameter uncertainty associated with incomplete knowledge of the specific parameter values used to characterize material properties used as inputs to models. As described in detail in Chapter 17 of this volume, uncertainties can also usefully be categorized consistent with their intrinsic properties and treatment in the models. Thus, it can be helpful in designing a safety assessment to distinguish between aleatory uncertainties, which in general derive from uncertainty about the occurrence of future events, and epistemic uncertainties, which derive from incomplete knowledge about the physical properties of the system. Aleatory uncertainties can be thought of as irreducible; e.g. no amount of additional research will provide a definitive answer about whether or not an earthquake will occur on a given date in the far future. Epistemic uncertainties, on the other hand, can be thought of as reducible. In principle, a more complete characterization of rock properties could be collected at any given site through additional drilling and testing programs. In practice, however, significant amounts of epistemic uncertainty will remain even in a mature safety assessment. It is simply not practical to reduce epistemic uncertainties indefinitely, nor is it necessary if a safety assessment that takes uncertainty into account shows performance to be acceptable.

15.4.1 Using probability and deterministic approaches to account for uncertainty Options for accounting for uncertainty in safety assessments broadly fall into two types: safety assessments may choose to use a probabilistic uncertainty analysis to display a range of outcomes consistent with uncertainty in model inputs or they may choose to bound uncertainty with conservative choices for selected inputs. Probabilistic uncertainty analyses have numerous advantages: they provide sensitivity analysis results that can help guide research programs, they provide decision makers with an unbiased estimate of a measure of the central tendency of estimates of future performance (i.e., mean or median), as well as a display of uncertainty about that measure, and, importantly, they help guard against unintended non-conservatisms that can enter into a system analysis when choices believed to be conservative at a subsystem level result in unforeseen consequences at the system level as models are coupled together. It can be difficult to make confident a priori assertions that certain conditions will always result in poorer performance and greater radionuclide releases; e.g. higher rates of water flow through a disposal vault may cause greater mobilization and transport of radionuclides,

© Woodhead Publishing Limited, 2010

Safety assessment for deep geological disposal

511

but they may also result in a cooler environment with less aggressive water chemistry and could, under some circumstances, prolong the life expectancy of engineered materials. Using an uncertainty analysis in which a range of flow rates are considered allows analysts, and ultimately decision makers, to confirm the relative importance of specific uncertainties. Some regulators, most notably the US EPA (2008) and US NRC (2008), specifically call for probabilistic uncertainty analyses in the safety assessment that supports licensing, emphasizing that the safety assessment should focus on a reasonable expectation of what may actually occur. Deterministic bounding approaches to uncertainty also offer advantages, both in allowing significant simplifications in the analysis and in increasing confidence and public acceptance regarding the safety of the proposed system, and many regulatory programs encourage the use of conservatism (e.g. NEA, 2004b, 2007). Bounding approaches have drawbacks, however, in that they may obscure understanding of how the system is believed to function, they greatly complicate the design of sensitivity analyses that can help guide the program and they have the potential to decrease public acceptance of repositories by causing performance to appear worse, and perhaps much worse, than it will actually be. These observations notwithstanding, conservatism has a role even in fully probabilistic assessments, allowing the implementer to develop simplified screening justifications for potentially complex FEPs or to forgo potentially costly research or modeling programs that would have little or no impact on performance. (Consider, for example, the assumption made in the 1996 WIPP performance assessment (US DOE 1996) that radionuclides will transport without retardation in interbeds within the host salt formation. Even without accounting for retardation processes, these units did not provide a significant release pathway and there was no need for an experimental program to support a reactive transport model.) In other cases, conservatism may allow the implementer to simplify the treatment of difficult or intractable technical issues in a way that satisfies both the public and the regulator that safety has not been compromised. As the authors of the NEA’s Post-Closure Safety Case for Geological Repositories (NEA, 2004a) note, ‘Conservatism is inevitable, and greatly to be preferred to optimism, but should be used and managed judiciously.’

15.4.2 Design of probabilistic safety assessments Safety assessments have relied on two basic approaches to incorporating uncertainty. In one approach, an analysis can be designed using multiple deterministic scenarios, each representing a different possible state of the system (e.g. two separate scenarios might consider the same system with either high- or low-permeability release pathways), and uncertainty can be

© Woodhead Publishing Limited, 2010

512

Geological repository systems for safe disposal

quantified through the probabilities, or weights, assigned to each scenario. Consequences can be calculated for each scenario and, if desired, the weighted sum of consequences can then be presented as an estimate of mean performance accounting for uncertainty. Alternatively, safety assessments can rely on Monte Carlo simulation-based techniques, in which values for uncertain parameters are sampled from predetermined distribution functions (appropriately accounting for possible correlations among parameter values) and multiple deterministic simulations are conducted using different sets of sampled values. Each simulation represents a different possible realization of the future state of the system, conditional on the chosen models and the specific set of sampled input values. For sampling schemes that do not introduce bias into the weighting, each realization provides an equally likely outcome of the model. The mean of the outcomes provides an estimate of the mean performance of the system and the full population of outcomes provides a measure of uncertainty associated with uncertainty in the model inputs. Figure 15.5 provides a simplistic representation of the choices available between scenario and Monte Carlo approaches. At one endpoint of a continuum of options, a fully probabilistic analysis could be constructed using a large number of scenarios, with all relevant uncertainty incorporated into the scenario probabilities. At the other extreme, an analysis could be envisioned in which the future is reduced into a single scenario, with a probability of occurring equal to one, and all uncertainty is incorporated through a single large Monte Carlo sampling. Examples exist of safety assessments using approaches close to either endpoint: e.g. the Electric Power Research Institute (EPRI) of the United States designed its analysis of the proposed Yucca Mountain repository incorporating all uncertainty through scenarios (Apted and Ross, 2005; EPRI, 2005), and preliminary iterations of the US DOE’s Yucca Mountain performance assessment relied on a Monte Carlo sampling for a single scenario for their million-year analyses (US DOE, 1998, 2002). Most probabilistic assessments have chosen approaches that correspond to intermediate positions on the continuum shown schematically in Fig. 15.5, generally defining a relatively small number of scenarios based on the occurrence (or non-occurrence) of major events and incorporating uncertainties associated with the ongoing processes that describe the behavior of the system through separate Monte Carlo analyses for each scenario. An advantage of this approach is that it allows a full sampling of uncertainty for consequences associated with rare events, which otherwise might occur only in a small number of realizations of a single-scenario analysis. Extremely rare events, such as volcanic disruptions, might require millions of Monte Carlo realizations before a sufficient number of random samplings created a stable mean outcome and a useful uncertainty analysis.

© Woodhead Publishing Limited, 2010

Safety assessment for deep geological disposal

513

15.5 Scenarios and Monte Carlo simulation: two endpoint approaches to uncertainty.

15.5

Applications of safety assessment

The primary application of safety assessments can be described broadly as informing decision makers regarding the long-term safety of a proposed disposal system, either with respect to direct comparisons to applicable regulatory standards or in the less formal context of supporting programmatic decisions regarding the viability of the site and conceptual design. Multiple ancillary applications arise during the multi-year development of a repository program, reinforcing the value of undertaking safety assessment modeling.

15.5.1 Direct comparison to regulatory standards Direct comparison of performance estimates to regulatory standards is a fundamental application of safety assessments for those programs for which regulations prescribe compliance with quantitative limits. The presentation of model results must be consistent with applicable regulatory requirements; e.g. for regulations that set a total allowable dose limit over all scenarios, performance measures for each scenario must be summed into an appropriate total dose history. For regulations that prescribe separate standards for performance following unexpected disruptions, results must be displayed separately for separate scenarios. As noted above, the way in which quantitative standards are defined can be a major factor in the design of the safety assessment.

15.5.2 Providing guidance to research and model development Because the selection and evaluation of geological repositories takes years or even decades, there are multiple opportunities for safety assessments to help guide programs for field testing, experimental research and model

© Woodhead Publishing Limited, 2010

514

Geological repository systems for safe disposal

development. Systematic identification and screening of potentially relevant FEPs early in a program’s history can identify information needs that might otherwise be overlooked. Uncertainty and sensitivity analysis results from early iterations of system-level modeling can provide quantitative confirmation of the uncertainties that contribute most to uncertainty in overall performance. Equally importantly, both FEP screening and early sensitivity analyses can identify technical areas where existing knowledge is sufficient to support decision-making, even though uncertainty remains. If potentially relevant FEPs can be shown to have no impact on performance with bounding analyses, or if uncertainty about parameters that characterize FEPs that are included in the system model can be shown to have acceptably small impacts on overall uncertainty, limited resources can be focused elsewhere. Caution is appropriate, however, when using safety assessments to guide scientific programs. Justifications for excluding FEPs from the system model should be revaluated in subsequent iterations to confirm that new information has not changed earlier conclusions, and project researchers should continue to challenge the basis for each decision. Results of systemlevel uncertainty and sensitivity analyses should be interpreted with the recognition that conclusions are conditional on the models and parameter values used to characterize uncertainty in the component subsystems. Model results can only reveal sensitivity to uncertainties that were acknowledged and included in the model inputs, and changes to either the models or their inputs may change conclusions about the relative importance of different processes and parameters. Characterizations of uncertainty in models and parameter values should be evaluated and confirmed with each iteration of the safety assessment to ensure that they are consistent with current understanding.

15.5.3 Evaluating design alternatives Safety assessments can be used to evaluate and compare alternative designs for the engineered components of a disposal system, including the waste form. Early in a repository program, valuable insights may come from comparisons of multiple alternatives at a conceptual level: e.g. what types of canisters will perform best in the estimated range of disposal environments? As design concepts and the safety assessment mature, comparisons of design alternatives can become increasingly specific; e.g. safety assessments can evaluate how increasing the thickness of the canister wall will change estimates of long-term performance. The resulting information may be valuable for decision makers considering operational safety and cost as well as long-term performance, helping programs achieve design goals that must meet multiple competing criteria.

© Woodhead Publishing Limited, 2010

Safety assessment for deep geological disposal

515

As discussed in Section 15.4.1, probabilistic approaches to the treatment of uncertainty have advantages over deterministic bounding approaches in this regard, because they are less likely to introduce conservatisms that may mask the significance of design changes. If overall performance is dominated by conservative assumptions about one aspect of the system, potentially meaningful changes in the performance of other components may be overlooked. Similarly, if no credit is taken in a safety assessment for the long-term performance of an engineered component, the analysis will be unable to distinguish between alternatives that may provide benefit. Analysts should be aware of the possible impacts of conservatisms and should consider providing comparisons of design alternatives at both subsystem and system levels to provide insight into the extent to which capabilities of components may be masked by other aspects of the safety assessment. In some cases the design alternatives under consideration may represent very different disposal concepts, such as comparisons of the performance of a given site for disposal of either spent nuclear fuel or vitrified waste from reprocessing operations. Comparisons among alternatives that fundamentally change the concept or mission of the repository can be useful, but analysts should be careful to verify that component models remain appropriate for the altered components and conditions. Assumptions that change the extent to which results are truly comparable (e.g. changes in the inventory emplaced in the repository) should be clearly stated whenever results are presented. Assumptions that transfer costs and risks to locations away from the repository, such as reprocessing of spent fuel, or that result in additional societal benefits, such as the production of additional energy, should be noted, but analysis of their impacts is outside the scope of what is generally included in a repository safety assessment. Safety assessments may also be called upon to provide probabilistic evaluations of repository performance for a range or envelope of possible designs using uncertainty analysis techniques, rather than direct comparisons of alternatives, but the limitations of the available approaches should be understood. Uncertainties associated with choices about design alternatives differ from uncertainties about future behavior of the system in that design uncertainties will no longer exist after the repository has been constructed, operated and decommissioned. Design choices should not, therefore, be generally included in sampling for Monte Carlo uncertainty analyses, and probabilistic safety assessments should not be used as the basis for formal a priori arguments that all designs within a specified envelope will result in a mean performance estimate that is in compliance with regulatory criteria. Sampling over ranges of values for a variety of design parameters (or for material properties associated with alternative designs) carries the implicit assumption that any of the sampled values could be chosen for

© Woodhead Publishing Limited, 2010

516

Geological repository systems for safe disposal

construction and operation of the repository. Rigorous evaluations of performance for each combination of design alternatives would require completing separate samplings of all other uncertainties for each combination of design parameters, which is likely to result in an unreasonably complex analysis. In practice, safety assessments can address the desire to evaluate an envelope of possible design choices by focusing on a single preferred design for the primary analysis and providing a limited set of supplemental analyses on representative alternative designs that are chosen qualitatively to represent reasonable bounds for the desired envelope. The resulting set of analyses can then provide a strong qualitative basis for arguing before repository construction begins that there is a wide range of final design specifications that could result in acceptable performance. A final safety assessment based on the final design, once it is known, could then confirm the earlier conclusion.

15.5.4 Enhancing scientific understanding of the behavior of the system In meeting their primary goal of evaluating long-term performance, safety assessments fulfill a corollary function that should be acknowledged to be of equal and perhaps greater importance: safety assessments enhance the scientific understanding of the behavior of the system and thereby become a key component of the technical basis for confidence in the safety of a proposed repository. Safety assessments are not simply descriptive models of what is already known. Rather, they become research tools for expanding knowledge about how coupled processes interact in the complex natural and engineered environments of the repository. Results of safety assessments should be thoroughly analyzed to verify that they are consistent with basic understanding of the underlying processes and apparent inconsistencies should be resolved. When approached as an iterative process that emphasizes interpretation, analysis and explanation, safety assessment is a scientific discipline in its own right, and one that has the potential to provide new insights that go beyond the sum of its parts. Done well, successive iterations of safety assessments build confidence among the technical peer community, both within a repository program and externally, that project scientists and engineers have a sound understanding of how the proposed repository could function in a range of uncertain future conditions.

15.6

Future trends

There is little doubt that scientific understanding of physical processes relevant to repository performance will continue to advance, and improve-

© Woodhead Publishing Limited, 2010

Safety assessment for deep geological disposal

517

ments in computational models and hardware will allow increasingly more realistic simulations at both the subsystem or process and system levels. Safety assessments of the future will be capable of doing much more than those of today, without question. There is also no doubt, however, that significant uncertainties will remain about how best to characterize future events and their interactions with complex natural and engineered systems, and safety assessments will therefore continue to need to acknowledge and account for those uncertainties. As process-level models improve and increase in complexity, couplings between models will also improve, options for propagating uncertainty through a system-level model will multiply and the overall analytical power of the safety assessment will increase. New insights into complex interactions among system components will become possible and improvements in computational design will allow more efficient analysis and testing of results. As a result, confidence is likely to increase among analysts and reviewers in the peer community that the safety assessment provides an appropriate and reasonable representation of the uncertainty associated with the possible future states of the system. The increasing complexity of future safety assessments will be accompanied by increasingly complex documentation, however, highlighting competing goals of clarity and realism. Both the public and the scientific peer community reasonably expect a safety assessment to be informed by the best available understanding. Confidence comes in part from the knowledge that the assessment has been as comprehensive and realistic as reasonably possible. Confidence also comes from clear and straightforward statements of how the system works, what matters and what does not matter, and why. Comprehensive and complete documentation of complex system-level models, regardless of how well prepared it may be, may not meet expectations of clarity for all audiences, and managers of large safety assessments are likely to find themselves being asked to simplify the system model to focus only on the most important processes. This is a reasonable request, and simplified safety assessment models have clear advantages. They run more rapidly, allowing quick turnaround of analyses considering alternative input assumptions (e.g. more or less favorable assumptions about important parameters or alternative designs for engineered components), and documentation can be far less cumbersome. Simplifications must be made with care, however, because once a process or parameter is removed from a model its importance can no longer be evaluated, and some insights may be lost from consideration. Choices between realism and simplicity, and between complexity and clarity, are likely to remain open questions in safety assessments of the future because there are sound rationales underlying both goals. Individual repository programs will find their own paths, consistent with their own national needs, and resulting safety assessments will be likely to range from

© Woodhead Publishing Limited, 2010

518

Geological repository systems for safe disposal

large analyses emphasizing full realism, while accepting the burden of complexity, to simpler analyses that focus only on the major processes. Regardless of the approaches taken, all safety assessments must address the challenge of building confidence with the technical peer community, with regulators and with the public.

15.7

Sources of further information and advice

Publications of the NEA, as indicated in the references for this paper, provide one of the best sources of summary information about repository programs internationally, and typically provide excellent citations to primary sources. Chapman and McCombie (2003) provide a good overview of all aspects of radioactive waste disposal, including a clear discussion of the essential elements of safety assessments. Definitive information on regulatory requirements in each nation must come from the individual regulatory agencies; citations are given in this paper for representative examples. The definitive sources for information on safety assessments are in all cases the technical reports prepared by the organizations that performed the assessment. The examples discussed in this report, including the US WIPP Compliance Certification Application (US DOE, 1996), the SR-CAN safety assessment in Sweden (SKB, 2006a), the Dossier 2005 in France (ANDRA, 2005a, 2005b), US EPRI analyses (EPRI, 2005; Apted and Ross, 2005) and the US Yucca Mountain License Application (US DOE, 2008) are available on the internet, and although each represents a daunting volume of information, there is no simple substitute for reviewing the original documentation.

15.8

Acknowledgments

Abe van Luik, Mark Tynan and Geoff Freeze provided constructive reviews of earlier versions of this manuscript. The author’s understanding of safety assessment has benefited from the experience of countless others who have worked in the field, including the named reviewers and numerous colleagues at Sandia National Laboratories and elsewhere in the peer community. The summary presented here reflects the work of many; any errors or misconceptions, however, are the responsibility of the author. The author is an employee of Sandia Corporation, the management and operating contractor for Sandia National Laboratories operating under Contract DEAC04-94AL85000 with the US Department of Energy, and the publication is identified by Sandia National Laboratories as SAND2009-0068P. The United States Government retains and the publisher, by accepting the article for publication, acknowledges that the United States Government retains a non-exclusive, paid-up, irrevocable, world-wide license to publish or

© Woodhead Publishing Limited, 2010

Safety assessment for deep geological disposal

519

reproduce the published form of this manuscript, or allow others to do so, for United States Government purposes. The statements expressed in this paper are those of the author and do not necessarily reflect the views or policies of the United States Department of Energy or Sandia National Laboratories.

15.9

References

ANDRA (Agence Nationale pour la Gestion des De´chets Radioactifs) (2005a), Dossier 2005: Argile. Tome (English translation: Safety Evaluation of a Geological Repository, but original documentation written in French remains ultimately the reference documentation). ANDRA (Agence Nationale pour la Gestion des De´chets Radioactifs) (2005b), Dossier 2005: Argile. Tome (English translation: Evaluation of the Feasibility of a Geological Repository in an Argillaceous Formation, but original documentation written in French remains ultimately the reference documentation). Apted M and Ross A M (2005), ‘Program on technology innovation: evaluation of a spent fuel repository at Yucca Mountain, Nevada, 2005 Progress Report’, EPRI Report 1010074, Electric Power Research Institute, Palo Alto, California. Chapman N and McCombie C (2003), Principles and Standards for the Disposal of Long-Lived Radioactive Wastes, Waste Management Series vol. 3, Elsevier. CNSC (Canadian Nuclear Safety Commission) (2004), ‘Managing radioactive waste: regulatory policy, P-290’, Canadian Nuclear Safety Commission, Ottawa, Canada. EPRI (Electric Power and Research Institute) (2005), ‘EPRI Yucca Mountain Total System Performance Assessment Code (IMARC) Version 8’, EPRI Report 10118143, Electric Power Research Institute, Palo Alto, California. Helton J C and Marietta M G (eds) (2000), ‘The 1996 performance assessment for the Waste Isolation Pilot Plant’, Special Issue, Reliability Engineering and System Safety, 69(1–3), 1–451. HSK and KSA (Swiss Federal Nuclear Safety Inspectorate and Federal Commission for the Safety of Nuclear Installations) (1993), ‘Schutzziele fu¨r die Endlagerung radioactiver Abfa¨lle, Richtline fu¨r schweizerische Kernanlagen (Protection objectives for the disposal of radioactive wastes), HSK-R-21, Villigen, Switzerland. IAEA (International Atomic Energy Agency) (2006), ‘Geological disposal of radioactive waste’, Safety Requirements WS-R-4, International Atomic Energy Agency, Vienna, Austria (Jointly sponsored by the International Atomic Energy Agency and the Organisation for Economic Co-operation and Development Nuclear Energy Agency). ICRP (International Commission on Radiological Protection) (2007), ‘The 2007 Recommendations of the International Commission on Radiological Protection’, in Annals of the ICRP, ICRP Publication 103, edited by J. Valentin, Elsevier.

© Woodhead Publishing Limited, 2010

520

Geological repository systems for safe disposal

National Research Council (1978), Geological Criteria for Repositories for HighLevel Radioactive Wastes, National Academy of Sciences, Washington, DC. NEA (Nuclear Energy Agency) (1997), Lessons Learnt from Ten Performance Assessment Studies, Organisation for Economic Co-operation and Development Nuclear Energy Agency, Paris, France. NEA (Nuclear Energy Agency) (1999), Scenario Development Methods and Practice: An Evaluation Based on the NEA Workshop on Scenario Development, Madrid, Spain, May 1999, Organisation for Economic Co-operation and Development Nuclear Energy Agency, Paris, France. NEA (Nuclear Energy Agency) (2000), Features, Events, and Processes (FEPs) for Geologic Disposal of Radioactive Waste: An International Database, Organisation for Economic Co-operation and Development Nuclear Energy Agency, Paris, France. NEA (Nuclear Energy Agency) (2004a), Post-closure Safety Case for Geological Repositories: Nature and Purpose, NEA 3679, Organisation for Economic Cooperation and Development Nuclear Energy Agency, Paris, France. NEA (Nuclear Energy Agency) (2004b), Management of Uncertainty in Safety Cases and the Role of Risk, Workshop Proceedings, Stockholm, Sweden, 2–4 February 2004, NEA 5302, Organisation for Economic Co-operation and Development Nuclear Energy Agency, Paris, France. NEA (Nuclear Energy Agency) (2006), Consideration of Timescales in Post-Closure Safety of Geological Disposal of Radioactive Waste, NEA/RWM/IGSC(2006)3, Organisation for Economic Co-operation and Development Nuclear Energy Agency, Paris, France. NEA (Nuclear Energy Agency) (2007), Regulating the Long-term Safety of Geological Disposal, NEA 6182, Organisation for Economic Co-operation and Development Nuclear Energy Agency, Paris, France. ONDRAF/NIRAS (Belgian Agency for Radioactive Waste and Enriched Fissile Materials) (2001), ‘SAFIR 2: Safety Assessment and Feasibility Interim Report 2’ NIROND 2001-06E. Pers K, Skagius K, So¨dergren S, Wiborgh M, Hedin Al, More´n L, Sellin P, Stro¨m A, Pusch R and Bruno J (1999), ‘SR 97 – identification and structuring of processes’, Technical Report TR-99-20, Svensk Ka¨mbra¨nslehantering AB (Swedish Nuclear Fuel and Waste Management Company), Stockholm, Sweden. SKB (Svensk Ka¨mbra¨nslehantering AB (Swedish Nuclear Fuel and Waste Management Company)) (2006a), ‘Long-term safety for KBS-3 repositories at Forsmark and Laxemar – a first evaluation, Technical Report TR-06-09. SKB (Svensk Ka¨mbra¨nslehantering AB (Swedish Nuclear Fuel and Waste Management Company)) (2006b), ‘FEP report for the safety assessment SRCAN’, Technical Report TR-06-20. SKI (Statens Ka¨rnkraftinspektion (Swedish Nuclear Power Inspectorate)) (1996), ‘SKI site-94: deep repository performance assessment project, Volume 1’, SKI Report 96:36. SNL (Sandia National Laboratories) (2008a), ‘Features, events, and processes for the total system performance assessment: methods’, ANL-WIS-MD-000026 REV 00, US Department of Energy Office of Civilian Radioactive Waste Management, Las Vegas, Nevada.

© Woodhead Publishing Limited, 2010

Safety assessment for deep geological disposal

521

SNL (Sandia National Laboratories) (2008b), ‘Features, events, and processes for the total system performance assessment: analyses, ANL-WIS-MD-000027 REV 00, US Department of Energy Office of Civilian Radioactive Waste Management, Las Vegas, Nevada. SNL (Sandia National Laboratories) (2008c), ‘Total system performance assessment model/analysis for the license application’, MDL-WIS-PA-000005 REV 00 AD01, US Department of Energy Office of Civilian Radioactive Waste Management, Las Vegas, Nevada. SSI (Statens Stra˚lskyddsinstitut (Swedish Radiation Protection Institute)) (1998), ‘The Swedish Radiation Protection Institute’s Regulations on the protection of human health and the environment in connection with the final management of spent nuclear fuel and nuclear waste’, SSI FS 1998:1, Unofficial English translation available at http://www.stralsakerhetsmyndigheten.se/Global/ Publikationer/F%c3%b6rfattning/Str%c3%a5lskydd/1998/ssifs-1998-1e.pdf. US DOE (United States Department of Energy) (1996), ‘Title 40 CFR 191 Compliance Certification Application for the Waste Isolation Pilot Plant’, DOE/CAO-1996-2184, United States Department of Energy Carlsbad Area Office, Carlsbad, New Mexico. US DOE (United States Department of Energy) (1998), ‘Viability Assessment of a repository at Yucca Mountain: total system performance assessment’, DOE/ RW-0508 vol. 3, United States Department of Energy Office of Civilian Radioactive Waste Management, Washington, DC. US DOE (United States Department of Energy) (2002), ‘Yucca Mountain site suitability evaluation’, DOE/RW-0549, United States Department of Energy Office of Civilian Radioactive Waste Management, Washington, DC. US DOE (United States Department of Energy) (2008), ‘Yucca Mountain Repository License Application’, DOE/RW-0573, Rev. 0. US EPA (United States Environmental Protection Agency) (1985), ‘Title 40 Code of Federal Regulations Part 191, Environmental Standards for the Management and Disposal of Spent Nuclear Fuel, High-Level and Transuranic Radioactive Wastes; Final Rule’, Federal Register, 50(182), 38066–38089. US EPA (United States Environmental Protection Agency) (2008), ‘Title 40 Code of Federal Regulations Part 197, Public Health and Environmental Radiation Protection Standards for Yucca Mountain’, Federal Register, 73, 61256. US NRC (United States Nuclear Regulatory Commission) (2003), ‘Yucca Mountain Review Plan, Final Report’, NUREG-1804 Revision 2, United States Nuclear Regulatory Commission Office of Nuclear Material Safety and Safeguards, Washington, DC. US NRC (United States Nuclear Regulatory Commission) (2008), ‘10 Code of Federal Regulations Part 63: Disposal of High-Level Radioactive Wastes in a Geologic Repository at Yucca Mountain, Nevada’, http://www.access.gpo. gov/nara/cfr/waisidx_08/10cfr63_08.html.

© Woodhead Publishing Limited, 2010

16 Safety assessment for near-surface disposal of low- and intermediate-level radioactive waste M . W . K O Z A K , INTERA Inc., USA

Abstract: This chapter discusses approaches to the post-closure safety assessment of near-surface facilities for the disposal of low- and intermediate-level wastes. Specifically, it focuses on approaches that have been developed in many countries over many years directed at establishing confidence in post-closure safety assessments. Compared with geological disposal, near-surface disposal demonstrates a greater diversity of disposal system designs, a wider variety in hydrogeological settings, waste that is more heterogeneous in nature and more difficult to characterize than typical waste streams intended for geological disposal. Many near-surface disposal facilities have been in operation far longer than geological disposals. These differences have led to diverse philosophy and technical approaches in near-surface and geological disposal safety assessments. Key words: safety assessment, near-surface disposal, low-level waste, scenario analysis, uncertainty analysis, safety case.

16.1

Introduction

Approaches to the post-closure safety assessment of near-surface facilities for the disposal of low-and intermediate-level wastes are presented in this section. The discussion is limited to the safety assessment needed for the post-closure period, although it should be recognized that safety assessments are needed for the pre-operational and operational stages of waste management as well. The focus is on approaches that have been developed over many years in many countries directed at establishing confidence in post-closure safety assessments. These approaches should be considered comparable to, and complementary with, the approaches to safety assessment for geological disposal facilities discussed in Chapter 15. However, when compared with geological disposal, for near-surface disposal facilities there is a greater diversity of disposal system designs, a 522 © Woodhead Publishing Limited, 2010

Safety assessment for near-surface disposal

523

wider variety in hydrogeological settings, the waste is more heterogeneous in nature and the waste is more difficult to characterize than typical waste streams intended for geological disposal. Furthermore, many near-surface disposal facilities have been in operation for many decades, in contrast to geological disposal facilities, almost all of which remain in early preoperational stages of development. These differences have led to differences in both philosophy and technical approaches between near-surface and geological disposal safety assessments.

16.2

Definition and performance measures for nearsurface disposal of low- and intermediate-level radioactive waste

Safety assessment (or ‘performance assessment’ in the terminology of NCRP, 2005) has been defined in the context of near-surface disposal as: [Safety] assessment is an iterative process involving site-specific, prospective modeling evaluations of the postclosure time phase of nearsurface disposal systems for low-level waste with two primary objectives: . to determine whether reasonable assurance of compliance with quantitative performance objectives can be demonstrated; and . to identify critical data, facility design, and model development needs for defensible and cost-effective licensing decisions and to develop and maintain operating limits (i.e. waste acceptance criteria). NCRP (2005) emphasizes the focus of safety assessment on making practical decisions throughout the lifecycle of the disposal facility and discusses the need to communicate clearly uncertainties and approaches in the context of regulatory decision making. The iterative nature of safety assessment is also emphasized by NCRP (2005), identifying the safety assessment as a central component in progressively updating understanding, data collection and disposal facility design. Uncertainties in this framework are understood to be uncertainties of importance to a licensing decision rather than uncertainties in the projected performance of the disposal facility. Strategies used in radioactive waste management to limit potential exposures to members of the public or inadvertent human intruders include specific features of the engineered barrier design and the geological setting for the disposal facility. As described by ICRP (2000a), the function of the engineered barriers is to provide an initial period of retention, which is followed by a later period in which dilution and dispersion are the mechanisms by which safety is assured. The engineered barrier system may serve a role in protection of the inadvertent intruder, but its primary function is generally considered to be to provide the initial period of

© Woodhead Publishing Limited, 2010

524

Geological repository systems for safe disposal

retention and to provide for gradual releases in later times. These gradual releases, spread out in time, allow the surrounding environment time to respond by diluting and dispersing the contaminants to innocuous levels. Similarly, the geological setting may function either to retain the contaminants (in the case of low flow through a deep unsaturated layer below the facility) or to provide a highly dispersive environment (high flow rates in humid conditions). These opposite functions of the geological environment are the reason that acceptable waste disposal facilities are found in many different climatological and geological settings. National regulatory authorities are responsible for establishing performance measures for disposal facilities in the post-closure period. While some exceptions exist, most regulatory authorities establish performance measures for near-surface disposal facilities that are consistent with ICRP recommendations (ICRP, 2000a). For near-surface disposal facilities, performance measures are generally established for two distinct situations: . .

protection of inadvertent human intruders and protection of the public.

Protection of inadvertent intruders may be accomplished through one of several management strategies. The combination of strategies is intended to ensure that adequate protection of the inadvertent intruder is achieved. These strategies are: . . . .

depth of disposal, institutional controls, control of waste concentrations and intruder barriers.

The use of a combination of these strategies is used to minimize the likelihood of an intrusion event occurring or to minimize the consequences of the intrusion event should it occur. The approach to achieving a disposal system that is passively safe for an intruder is qualitatively to apply arguments for depth of disposal, institutional controls and intruder barriers to demonstrate that intruders will be precluded from activities on the site for an extended period of time, usually taken to be 100–300 years in national regulations. At the end of that period, an intruder calculation is done with a goal to establishing waste concentrations that will limit the exposure of the intruder and the public if an intrusion event occurs. In national regulations, the dose rates considered to be acceptable under these conditions are in the range 10–50 mSv/year, comparable to the dose rates established by ICRP (2000b) for use in intervention situations. The use of institutional controls in the protection of potential intruders should not be confused with the actual intended duration of institutional controls. There is often an intention to maintain institutional ownership of

© Woodhead Publishing Limited, 2010

Safety assessment for near-surface disposal

525

the site indefinitely, which would preclude inadvertent intrusion from occurring at any time. Instead, the period of institutional control used in safety assessments is an assumed time at which institutional controls may be lost regardless of present intentions. Protection of the public is maintained by protecting individuals living beyond the area under institutional control. This is accomplished by establishing dose constraints applicable to individuals, which are applied in a prospective manner for times in the future. Internationally, dose constraints on the order of 0.1–0.3 mSv/year are recommended (ICRP, 2000a) and most national regulatory authorities specify values in this range. These dose constraints are typically intended to apply for thousands of years, to ensure that the disposal system remains safe at all times (US NRC, 2000). Consequently, in this regard safety assessments for near-surface disposal facilities resemble those for geological disposal facilities. Even in waste categorized as ‘short-lived’ waste, sufficient quantities of long-lived radionuclides are usually present to make the long-term portion of the safety assessment important.

16.3

Key issues and development of safety assessment for near-surface disposal of low- and intermediatelevel radioactive waste

The ICRP conceptual approach to waste management applies to all types of waste and all types of disposal facilities. Both near-surface and geological disposal facilities are designed for an initial period in which the wastes are retained in the repository for decay of short-lived radionuclides, followed by a period in which residual long-lived radionuclides are released to the environment, to produce acceptably low concentrations that are protective of human health and the environment. Controlling the depth of disposal has long been a key strategy for evaluating the concentrations of waste appropriate for a given technology, to distinguish between wastes appropriate for near-surface disposal and waste that must be more isolated. The US Nuclear Regulatory Commission, in its development of its regulation for near-surface disposal (Title 10 of the Code of Federal Regulations, Part 61) examined a number of alternative ways in which an inadvertent human intruder might disrupt a waste trench (Oztunali and Roles, 1986). This approach led to the development of the US waste classification system, which limits the concentrations of waste that are appropriate for disposal in the near surface. The underlying concept is that the number of activities that could result in an inadvertent human intrusion decreases quickly with depth and that therefore the likelihood of an intrusion event decreases with depth. This concept was made explicit in

© Woodhead Publishing Limited, 2010

526

Geological repository systems for safe disposal

Table 16.1

The IAEA waste classification system (IAEA, 1994)

Waste type

Characteristics

Exempt waste

Activity levels at or below clearance levels Low- and intermediate-level Activity levels above waste (LILW) clearance levels and thermal power below about 2 kW/m3 Short-lived LILW waste Restricted long-lived radionuclide concentrations; limitation of long-lived alpha emitting radionuclides to 4000 Bq/g in individual waste packages and to an overall average of 400 Bq/g per waste package) Long-lived LILW Long-lived concentrations exceeding those specified for short-lived LILW High-level waste Thermal power above about 2 kW/m3 and long-lived radionuclide concentrations exceeding disposal facility limitations for short-lived waste

Disposal options No radiological restrictions

Near-surface or geological disposal

Geological disposal

Geological disposal

international guidance by the Nuclear Energy Agency of the OECD (NEA, 1987), who introduced the concept of the ‘normal residential intrusion zone’, which represented the depth of a foundation of a residential home. This zone was stated nominally to be about 3 meters deep, but which could vary according to site-specific considerations. This approach was intended to account, to a certain extent, for the effect introduced by differing depths for excavating foundations in different locations. In addition, since only residential construction was considered, this concept was used to justify omitting some combinations of intrusion scenarios. For instance, it was argued that agricultural lands would only be contaminated by residential construction; deeper construction activities (e.g. road construction) would not be associated with adjacent agricultural land. IAEA (1994) established a waste classification system to assist in identifying disposal options appropriate for each class of waste. This system is shown in Table 16.1. This system is not a performance-based system, in which safety assessment is used to establish waste concentrations appropriate for near-surface disposal. Instead, the system establishes acceptable waste concentrations and thermal power based on generic considerations, regardless of the facility design or potential consequences to an intruder or to members of the public. It should therefore be used with caution, since a

© Woodhead Publishing Limited, 2010

Safety assessment for near-surface disposal

527

real disposal facility may differ in its performance with regard to long-term activity in the waste. More recently, IAEA (2003a) conducted a performance-based analysis that examined a broad variety of scenarios that might result in inadvertent intrusion and included several scenarios that might result in major intrusion events into waste as deep as 9 m. This more detailed performance-based analysis was conducted only as an illustration of how such an approach could be performed and does not constitute international guidance. It is often considered that for waste deeper than the normal residential intrusion zone, the only potential for intrusion would be a drilling event. As the disposal technology gets progressively deeper, it is considered that (as a general rule) the likelihood of drilling intrusion tends to decrease, as boreholes become more costly and less likely to be drilled at great depths. This rule of thumb is not universal; at some sites with deep geological resources, there may not be a decrease in the likelihood of drilling at greater depth. However, for many disposal facilities, the primary motivation for drilling would be for water, and simple economics and practicality puts limits on the well depths in such situations. There is no international consensus on the effect depth should have on the regulatory treatment of inadvertent human intrusion. To address this lack of consensus, IAEA has presented concepts on categorizing highly active wastes by source intensity and half-life, with a view to identifying appropriate depth to disposal (IAEA, 2005). This conceptual approach is depicted in Figure 16.1, in which it is proposed that very highly active, longlived sources may only be appropriate in geological disposal facilities, whereas shorter-lived or less intense sources may be appropriate for nearsurface disposal. However, it is notable that IAEA did not recommend

16.1 Proposed IAEA approach for identifying waste disposal depth based on source strength and radionuclide half-life (IAEA, 2005).

© Woodhead Publishing Limited, 2010

528

Geological repository systems for safe disposal

whether to conduct consequence analyses of potential intrusion events for these various categories.

16.4

Safety assessment methodology for near-surface disposal of low- and intermediate-level radioactive waste

IAEA (2004) described features of a safety assessment methodology that has since been used extensively internationally, shown in Fig. 16.2. The methodology represents international experiences with safety assessment that had been developed over many years (e.g. Kozak et al., 1990; Seitz et al., 1992; Kozak, 1994; IAEA, 1999; US NRC, 2000). The methodology is represented as a sequence of steps in an iterative process. The intent of the iterative process is to integrate data collection,

16.2

A safety assessment methodology (IAEA, 2004).

© Woodhead Publishing Limited, 2010

Safety assessment for near-surface disposal

529

modeling, estimation, interpretation and decision elements into a coherent assessment. These activities are organized as a sequence of assessment context, system description, scenario development and justification, model development and justification, consequence analysis, followed by a sequence of interpretation, decision and data collection activities. An assessment context is intended to set the scope and content of the analysis. Any assessment can only be interpreted within the context in which it is intended, and clarity of purpose and context assist in ensuring clarity in the interpretation of the assessment. The assessment context must therefore specify the regulatory and decision goals, time frames of concern, bias of the analyst and availability of information. As the safety assessment can be used for various purposes throughout the facility lifecycle, it is clear that the assessment context will evolve as well. The system description, which can be considered to be part of the assessment context in some situations, consists of a clear description of the disposal system under study, within the framework of the assessment context. This description may include both data and interpretations of data that are used in the subsequent evaluations. The goal of this stage of the assessment should be a clear conceptual description of the function of the disposal facility and its interrelated parts. There are competing goals of clarity and comprehensiveness in the system descriptions, making this a challenging component of the safety assessment documentation. The safety assessment should address the performance of the disposal system under future anticipated conditions, including events associated with the normal evolution of the facility and less probable events and processes. This is often achieved through the formulation and analysis of a set of scenarios describing alternative future evolution of the disposal facility. The scenario development step in the safety assessment represents the establishment of the boundaries of future conditions of concern in the assessment. It is impractical to think that the scenarios will be fully comprehensive, since the number of scenarios that would be required would be prohibitive. However, the credibility of the safety assessment is dependent on the scenarios representing a credible set of conditions that represent key concerns about the future of the disposal facility. As a result, formal methods and procedures have been developed to permit the development and justification of scenarios. The primary intent of the procedures is to make the assumptions and thought processes that go into the assessment more easily traced and understood by other assessors. Safety assessment must be understood to be more than a purely technical exercise; the need to communicate results of the safety assessment clearly is paramount, and communication of the basis for scenarios is among the most important elements of that communication. Justification of scenarios is frequently a major focus of the regulatory review of a safety assessment.

© Woodhead Publishing Limited, 2010

530

Geological repository systems for safe disposal

The next step in the safety assessment process is to formulate and implement models. Models include conceptual models, their mathematical implementation and usually the numerical implementation of the mathematics. This step can be viewed as taking the concepts developed in the scenario generation step to a more detailed and practical level. As with scenarios, the modeling approaches are a key focus of outside reviewers, and emphasis should be placed on clear communication of the ideas and approaches used in the assessment. Consequence analyses are carried out to produce outputs commensurate with the assessment context. Consequently, the result of this step is a set of numerical values that can be compared with safety criteria. However, owing to the nature of safety assessment analyses, a simple direct comparison is generally insufficient to provide confidence in a regulatory decision. Instead, an additional step is necessary for interpretation of the results in the context of the assessment context, consideration of the degree to which uncertainties have been addressed and propagated through the assessment. At the end of the interpretation step, an evaluation of adequacy of the safety assessment can be made. If the assessment demonstrates compliance with regulatory criteria, and if data and models are deemed to be sufficient to support the decision, then the conclusion of this step should be to accept the safety assessment with a positive conclusion. If, on the other hand, the safety assessment is deemed to be inadequate, this suggests that another iteration of the assessment may be warranted. The new iteration is conducted by answering several sequential questions. First, why is the analysis inadequate? It may be that data are inadequate, that models seem insufficiently supported, that the inventory or activity concentrations in the vault are too large or that some scenarios or models have been omitted from the analysis because of insufficient budget or time. The next question to be answered is whether it is cost effective to resolve these issues. Furthermore, this step must set priorities about which questions are most important to answer first. There may be some situations where the answer to this question may lead to a complete rejection of the disposal concept; i.e. it may be more cost effective to start over at a different location than to try to resolve a particularly difficult issue. More often, the final step in the revision is to collect data, revise models, scenarios, assumptions, designs or parameters appropriately, and to revise the analysis. The safety assessment is then conducted again using the revisions, to determine their effect on the overall system safety. The process generally evolves from simple, conservatively biased modeling and data-estimation techniques to more complex and rigorous approaches as subsequent iterations are conducted. The intent at any point of the assessment is to represent the level of uncertainty that exists about

© Woodhead Publishing Limited, 2010

Safety assessment for near-surface disposal

531

physical phenomena and data at the site. This approach has been shown to be an appropriate way to develop cost-effective decisions about disposal facilities.

16.5

Application of safety assessment for near-surface disposal of low- and intermediate-level radioactive waste

16.5.1 Elements of scenario analysis Scenarios are intended to portray alternative future states of the system, with a view to incorporating uncertainties in future conditions at the facility into the safety assessment. A variety of approaches has been used to develop scenarios. In older literature on near-surface disposal, scenarios were generally implemented simply based on the professional judgment of the safety assessor (e.g. Oztunali and Roles, 1986; Kennedy and Peloquin, 1988). As requirements have grown for improved transparency and traceability for scenarios, improved scenario-generation approaches have been implemented based on the use of features, events and processes (FEPs). The FEP-based scenario approaches were initially developed in the context of geological disposal systems, and have been revised and extended for application in near-surface disposal facilities. The use of FEPs is primarily an approach for improving the defensibility and traceability of a safety assessment analysis. The FEP process can be used either for scenario development, in which the FEPs are regarded as fundamental building blocks from which scenarios are constructed, or in an audit mode, in which scenarios developed using past practice and expert judgment are audited to ensure completeness. In general, the process consists of four steps: (1) identifying a comprehensive list of features, events and processes, (2) screening the comprehensive list to a manageable number, (3) describing the relationships between the features, events and processes and (4) arranging them into calculational cases, or scenarios, for the safety assessment. Each of these four steps is conducted either for developing scenarios from FEPs or in justifying scenarios from FEPs. Differences between published scenario development approaches represent differences between methods for one of these steps, or different ordering of the steps. For instance, the original scenario development procedure developed by Cranwell et al. (1990) only calls for screening the full scenarios, whereas later scenario development approaches emphasize screening at the FEP level, or screening both FEPs and full scenarios (e.g. Skagius et al., 1995; OCRWM, 2008). Despite the differences in approaches and ordering, the concepts of these four steps are the same for all scenario development procedures.

© Woodhead Publishing Limited, 2010

532

Geological repository systems for safe disposal

Considerable international effort has been expended to develop comprehensive FEP lists for geological disposal systems. There is only one approach that has been used for this step: collection and elicitation of expert opinion. NEA (2000) represents a recent compilation of FEPs appropriate for geological disposal. While less attention has been paid to the use of FEPs for near-surface disposal, the IAEA (2004) has published a comprehensive FEP list for near-surface conditions, based on the NEA (2000) FEPs database but modified to accommodate near-surface conditions. Techniques for screening FEPs or scenarios from further consideration can be categorized as (1) based on probability, (2) based on consequence or (3) based on expert judgment. In practice, expert judgment permeates all aspects of any screening procedure. Probabilistic approaches involve the development of fault trees to represent key aspects of the scenario, and the branches of the fault tree are assigned probability values that may be propagated to produce an overall probability of occurrence of the scenario. Chapman et al. (1995) described an alternative method for screening FEPs, based on differentiating FEPs acting within a ‘process system’ and those that act externally to generate scenarios. Broadly speaking, this approach produces a similar structure to the probabilistic approaches, but the level of detail included in the process system FEPs increases significantly. In addition, there was no attempt to develop probabilities associated with the FEPs and scenarios. Instead, the approach of Chapman et al. (1995) relies more heavily on FEP justification using expert judgment, which is not assigned a numerical value. The approach of Chapman et al. (1995) has often been found to be appropriate for assessments with a regulatory endpoint of peak dose or peak risk in time. In this approach, scenarios are viewed as illustrative examples of future behavior. There is no intent to be either comprehensive or mutually exclusive, since there is no intent to apply probability theory to the scenarios. FEPs are divided into external FEPs, which act on the system, and internal FEPs, which are internal to the system. Internal FEPs are either explicitly or implicitly represented in the conceptual models used in safety assessment, whereas external FEPs act essentially as boundary conditions. In addition, a subset of external FEPs are chosen to be ‘scenario-generating FEPs’. These FEPs are chosen to be the ones that best illustrate alternative evolutions of the site that are of greatest concern. They are chosen entirely using judgment, with a view to establishing a reasonably complete set of conditions. An initial step in applying the FEPs to a near-surface disposal facility often consists of an initial screening of the FEP list based on the assessment context and system description. The intent of this screening process is to eliminate four categories of FEPs from further consideration. These categories are:

© Woodhead Publishing Limited, 2010

Safety assessment for near-surface disposal . .

.

.

533

FEPs that are clearly not relevant for the current assessment. There should be no possible argument about the exclusion of these FEPs. FEPs that are not relevant because of the chosen assessment context. These FEPs might potentially be important in the future, if other assessment contexts are applied to the disposal system. An example of this type of FEP might be population risk, if the assessment context focuses on individual risk. FEPs that are not considered to be important. The lack of importance may be the result of the type of disposal system considered or because other FEPs have been judged to be more important for overall system performance. Inclusion in this set of FEPs is more judgmental than the first two exclusion principles, and the attention of reviewers is often focused on this set. FEPs that are not considered because there is no information about them, and for which it is unreasonable to expect information to be available in the foreseeable future. Inclusion in this set of FEPs is the most judgmental of these four categories, and the attention of reviewers is often focused on this set.

The next step in the process is to link the screened FEPs into a coherent structure capable of being analyzed. Effectively, this step involves developing a process system model and identifying the external FEPs acting on it. Formal approaches have been proposed for this step, which are intended to address increasing requirements on justification and traceability. Approaches that have been described in the literature for this step include (1) lists and tables, (2) influence diagrams (Skagius and Wiborgh, 1994) and (3) the interaction matrix approach (Skagius et al., 1995). Long-term, post-closure scenarios are inevitably stylized illustrations of the future rather than predictive sets of conditions. Thus the scenarios should be selected to bracket a sufficiently wide range of possible events, pathways and parameters for the purpose of the assessment. In order to estimate dose, consideration needs to be given to the environmental system into which future releases of radionuclides could occur, and to the future behavior of humans. Peak impacts from releases of radionuclides from a repository might not occur for many hundreds or even thousands of years after disposal. Over such time scales, it is clearly unrealistic to forecast human habits and behavior. Changes may also arise from the natural evolution of the biosphere. In addition, profound environmental changes may arise as the result of future (and inherently unpredictable) human actions. To alleviate these conceptual difficulties, ICRP (2000a) has recommended that assumptions about human behavior should reflect current behavior of present-day people in the vicinity, projected into the future.

© Woodhead Publishing Limited, 2010

534

Geological repository systems for safe disposal

There are significant difficulties in specifying the calculations that need to be undertaken to evaluate radiation dose in the long-term future. A series of assumptions must be made as the basis for such an assessment. Some of these assumptions will necessarily be arbitrary to some extent (e.g. future human behavior and its effect on the biosphere). However, these assumptions must be consistent with the aim of providing a robust, yet reasonable level of assurance regarding the acceptability of possible future releases and their impacts on humans and the environment.

16.5.2 Elements of consequence analysis The quantitative analysis of consequences for a given scenario is performed using mathematical models, which are derived from conceptual models. A conceptual model is a qualitative description of the behavior of a system of concern; the intention is for the conceptual model to contain a set of assumptions that clearly relate to the way the system is believed to function and how that function should be represented within the assessment context. Assumptions may be necessary regarding the geometry and dimensionality of the system, initial and boundary conditions, time dependence and the nature of the relevant physical and chemical processes. A description of the scope of the model is necessary in order to record the assumptions under which it has been developed and the situations to which it applies. Since safety assessments are generally applied in situations in which they must withstand scrutiny by outside (usually skeptical) parties (e.g. regulatory review), there is a strong need for clarity of expression of the conceptual models, to communicate clearly the understanding of the system function to other parties. For the scenarios that are to be quantitatively assessed, it is important that the conceptual model is amenable to mathematical representation. The conceptual model should be described in enough detail to allow appropriate mathematical models to be developed to describe the behavior of the system and its components. This description should be sufficient to provide an estimate of the performance of the system over time, at a level of detail that is appropriate to the assessment context and stage of iteration of the assessment. Documentation of the conceptual model is a crucial component of safety assessment documentation, since it is often the focus of a substantial part of the regulatory scrutiny of the safety assessment. Issues arise because a good geological setting, combined with a robust engineering design, can cause peak individual doses to be delayed for very long periods of time. It should be recognized that doses arising in the far distant future are a representative characteristic of a good site and repository design. In carrying out analyses to peak dose, the regulatory structure should be established in a way that recognizes this fact and does

© Woodhead Publishing Limited, 2010

Safety assessment for near-surface disposal

535

not make compliance assessments for a good site and design more onerous than for a less robust system. There should be a clear understanding that results in the far distant future. Stylized approaches may be invoked to establish the processes and exposure pathways to be considered in the scenario. As processes become increasingly uncertain in the distant future, there has been consideration that model criteria should reflect the increasing uncertainty, and be regarded as increasingly stylized and illustrative (NEA, 2002, 2004a; IAEA, 2003b). Stylized approaches are important in differentiating between scientific uncertainty, which is large and growing at all times, and uncertainty in the regulatory decision, which is manageable at all times. These two concepts need to be kept clearly differentiated. The use of stylized approaches is an important tool in translating between the two. Hence, in situations in which scientific uncertainty is large and growing, it may be appropriate to use stylized approaches to manage those uncertainties. It is important that application of a stylized approach should encompass reasonable bounds on system behavior, so that the uncertainties are managed, not ignored.

16.5.3 Uncertainties and decisions Uncertainty is an intrinsic part of decisions about the prospective behavior of a complex system over long periods of time. There is no question that the uncertainties exist and that the decision maker must account for them. However, questions exist about how to treat these uncertainties in a way that lends itself to defensible decisions regarding compliance. The principal challenge of safety assessment is to translate the results of an uncertain calculation, which inherently does not generally provide clear-cut answers, to the needs of the decision maker, who must ultimately make a binary (yes or no) decision as to whether the system meets the performance objectives. Uncertainties are dealt with in safety assessments by application of two principles: multiple lines of reasoning and conservative bias. Multiple lines of reasoning refers to a technique for considering the consequences of alternative sets of behavior, alternative parameter sets and alternative physical processes, when uncertainty about them exists (Kozak, 1997). Conservative bias refers to a tendency to choose lines of reasoning that produce higher calculated doses. The use of conservative bias means that any errors made by the analyst will tend to be on the side of safety. However, it must also be recognized that the use of extreme or unreasonable levels of conservatism can lead to unreasonable decisions. Consequently, in practice there needs to be an appropriate balance between conservatism and rigor in the analyses (Kozak, 1994). Justification of this balance is a key issue that needs to be addressed in the safety assessment. It is crucial to recognize the unusual nature of safety assessments when

© Woodhead Publishing Limited, 2010

536

Geological repository systems for safe disposal

considering uncertainty. The goal of the safety assessment is to make a regulatory decision based on the available information; as a result, the analyst need not be concerned with representing the uncertainty in the ‘correct’ numerical answer to the analysis. The ‘correct’ answer is unknown, unknowable and largely irrelevant. Instead, the analyst should be concerned with identifying the important assumptions and parameter values that, when changed, can change the decision. This distinction between numerical uncertainty and identification of factors important to the decision is central to understanding the need and the approaches used in addressing uncertainties in safety assessment. NCRP (2005), in identifying and emphasizing these distinctions, chose to introduce the term ‘importance analysis’ to apply to approaches used in safety assessments. Uncertainties of importance in safety assessments have been categorized as being of two kinds: uncertainties due to intrinsic variability in the system and uncertainties due to lack of knowledge. Intrinsic variability may be attributed to a property of the system based on repeated measurements of the property, or may be associated with variability in time or space. Examples of this kind of uncertainty in safety assessment might be a time series of rainfall measurements or repeated measurements of the sorption capacity of a single soil or engineering material. This type of uncertainty is referred to as aleatory uncertainty, stochastic uncertainty or as type A uncertainty (IAEA, 1989). By contrast, uncertainty due to lack of knowledge is founded on incomplete characterization, understanding or measurement of the system. Examples of this type of uncertainty are the projection of behavior into an uncertain and immeasurable future, incomplete inventory information and incomplete understanding of processes governing release and transport phenomena. In addition, intrinsic spatial variability is often included in this category, since the uncertainty results from an incomplete characterization of the material or process at below some scale of concern, or at a scale that cannot practically be measured. The type of uncertainty resulting from lack of knowledge is known as epistemic uncertainty, knowledge uncertainty, subjective uncertainty or type B uncertainty. NCRP (2005) emphasized that epistemic uncertainties dominate safety assessments, to the extent that aleatory uncertainties may be considered to be entirely subsumed by them. The reasons for this dominance of epistemic uncertainties are: . .

Safety assessments are characterized by sparse data sets, which do not lend themselves to full characterization of variabilities. Even when sufficient data exist to characterize the variability, there exist additional uncertainties about whether the data can be used to represent the large spatial domains and time scales needed for safety assessment.

© Woodhead Publishing Limited, 2010

Safety assessment for near-surface disposal .

537

Uncertainties in the future behavior of the disposal system are intrinsically linked to our lack of knowledge of the future, which is substantially greater than variability in current conditions.

Practitioners of safety assessments have been divided into those using deterministic methods, probabilistic methods or a combination of the two (NCRP, 2005). (Several researchers have proposed the use of nonprobabilistic methods, such as fuzzy set theory, but these approaches have not been used in practical safety assessments to date.) Deterministic methods are those in which a single scenario, model and parameter set are used to calculate a single value of the outcome. Treatment of uncertainty is carried out by conducting numerous deterministic analyses and choosing one of them to be the basis for the decision. Probabilistic methods functionally consist of assigning probabilities to scenarios, and in some cases models, and probability density functions for parameters. Conceptually, probabilistic analyses consist only of a series of deterministic analyses with the relationships between the deterministic results expressed using probability theory. A general structure of the way uncertainties can be represented in safety assessment is shown in Fig. 16.3. Particular conceptualizations of the future of the site are embodied in the scenarios. Within each scenario, it is possible to postulate alternative conceptual models. For each model, it is possible to postulate alternative vectors of parameters within the models. The differences between deterministic and probabilistic approaches represent competing ways to choose among the possible analyses that the analyst

16.3 Uncertainty analysis for safety assessment (Kozak et al., 1993).

© Woodhead Publishing Limited, 2010

538

Geological repository systems for safe disposal

might perform and competing ways to represent the output information. Probabilistic analyses attempt to span many of the possible combinations of scenario, model and parameter vector. These approaches compete with single parameter variation approaches and deterministic analyses in current safety assessments. Examination of the literature shows that there is not currently unanimity of opinion about the relative strengths of probabilistic and deterministic approaches for treating uncertainties in safety assessments. Regardless of the approach used, the net effect of the two approaches is similar. If all uncertainties are addressed, the result would be alternative possible doses depending on our conceptualization of the site, the mathematical models we use to represent the conceptualization, the parameters used in the mathematical model and the conceptualization of the evolution of the site in time. The primary difference between the approaches is the manner in which the results of the individual deterministic analyses are integrated and interpreted for the decision-making process. Each approach for implementing uncertainty analyses has both strengths and weaknesses. The deterministic approach has the advantage of simplicity of implementation and transparency of interpretation: one conducts a relatively small set of deterministic analyses and interprets the model behavior based on those analyses. In addition, communicating a singlevalued outcome to a non-specialist audience is easier than communicating the more complex approaches. The deterministic approach also has some drawbacks. First, single parameter variations will not necessarily determine extremes of the output variable, since the extremes of the output of safety assessment models may be caused by unusual combinations of input parameters, and in general there is no way to determine the response of the system to variations in input parameters prior to the analysis. Consequently, troublesome combinations of parameters can potentially be overlooked using this approach. Second, this approach does not lend itself to development of a clear decision rule. At the end of the analysis, the decision maker is left with a set of undifferentiated results; if some of those results violate the performance objective, there are no clear numerical criteria for determining whether the facility complies, only judgmental ones. In probabilistic approaches, scenarios, models and parameter vectors are assigned probabilities and the set of analyses are combined probabilistically to produce a distribution of outcomes. The most common probabilistic approach for treating parameter uncertainty is Monte Carlo analysis. The primary advantage to conducting Monte Carlo analysis is that it provides model results from a large number of input parameter sets. Therefore, the output uncertainty is acknowledged and there is an explicit method for identifying whether the output uncertainty resulting from input parameter uncertainty has been bounded. Clearly, one has more confidence that the

© Woodhead Publishing Limited, 2010

Safety assessment for near-surface disposal

539

output has been bounded with increasing numbers of samples. Surprisingly to some, Monte Carlo analysis has similar drawbacks to deterministic analyses when used for safety assessment. As with deterministic analyses, Monte Carlo analysis does not lend itself to a clear decision rule in the context of safety assessment. The analyst is faced with a large number of realizations and must use judgment to identify the case (or probability value) that represents the decision point. The output distribution does not provide a statistically rigorous estimate of performance of the site. Instead, it provides quantitative information about the relative likelihood of various output values of the model. This information is developed through judgments about the input parameters, leading to qualitative estimates of the shape of the input distributions. Consequently, the shape of the output distribution (and hence the probability level) is a representation of judgmental considerations, rather than fact. Because this issue is not widely appreciated, the use of probabilistic analyses can lead to the mistaken belief that the results are rigorous and that all sources of uncertainty have been adequately accounted for and rigorously propagated. Furthermore, probabilistic analyses can be difficult to interpret and are difficult to communicate to non-expert audiences. NCRP (2005) concluded that both deterministic and probabilistic approaches can be useful tools in the conduct of safety assessments. The most defensible approach is dependent on the specific features of the safety assessment under examination. The process and clarity with which uncertainties are treated are keys to developing a credible safety assessment. The analyst must be able to present a credible argument that the uncertainties have been adequately addressed in a manner that makes the resulting decision credible.

16.6

Future trends

16.6.1 Information management In the past, safety assessments were viewed as one-time technical calculations used to get regulatory approval for a disposal facility. More recently, as discussed in this chapter, there is a recognition that safety assessments are integral and key parts of the management of a waste disposal facility and that they progressively evolve throughout the facility lifecycle. This means that information management approaches are necessary to treat information generated in the past, to support decisions today and to treat past and present information so it is useful in the future. Management processes are necessary to establish the qualification of past data in a regulatory setting. The qualification of data should constitute a set of procedures that permit traceability and transparency of data and their

© Woodhead Publishing Limited, 2010

540

Geological repository systems for safe disposal

interpretation, when such data are to be used in regulatory decisions. Data used in a safety case may be derived from one of several origins: . . . .

data collected within the project subject to the management system, data collected as part of a research program that are not part of the management system, data collected historically, which predate the existence of the management system, and literature information that reflects general knowledge, understanding or measurements not necessarily specifically associated with the project under consideration.

The management system should establish clear processes for qualifying each of these types of information. For example, to qualify historical data, it may be necessary to establish management processes for review of the original data to ensure it is correct and traceable. The management system also needs to accommodate record keeping over the duration of the project lifecycle. Since repository programs have particularly long lifecycles, and since information collected throughout the lifecycle of the repository will be needed for decisions taken late in the lifecycle, there is a particularly stringent requirement on the management system to provide long-lasting traceability and transparency of information. The pressures on this management system are large, particularly given the rapid evolution of computer technologies. Information stored in electronic format may be inaccessible after only a few years, as computer technologies improve and change. Ensuring the traceability of such information over decades, so that it is accessible to future decision makers, is a significant challenge.

16.6.2 Safety case The literature has shown an increasing awareness that developing confidence in the safety of a disposal facility involves more than simply meeting the requirements of safety assessment. This awareness has led to increasing use of the term ‘safety case’ (NEA, 2004b) to refer to this broader concept. NEA (2004b) defines safety case as follows: ‘A safety case is the synthesis of evidence, analyses and arguments that quantify and substantiate a claim that the repository will be safe after closure and beyond the time when active control of the facility can be relied on.’ Elements of the safety case described previously (e.g. NEA, 2004b) have been limited in scope to the technical, and to a lesser extent the managerial, elements that contribute to confidence in the safety of a disposal system. However, IAEA has clearly identified a number of other elements that contribute to confidence in the disposal system (IAEA, 2000, 2008). These

© Woodhead Publishing Limited, 2010

Safety assessment for near-surface disposal

16.4

541

Elements supporting the safety case.

additional elements may be broadly categorized as legal, managerial and financial elements of the safety case. The safety case is therefore seen to be underpinned by four main elements, as shown in Fig. 16.4. Confidence in each of these main elements is necessary to generate regulatory and societal confidence in the overall system for disposal of waste. For many nearsurface disposal facilities, which are in operation or in the post-closure phases of the facility lifecycle, confidence in the non-technical elements of the safety case often outweigh the technical element. The safety case is shown as a table in Fig. 16.4, since a loss of integrity of any one of the elements supporting it causes the safety case to become unstable, just as a table does if one leg is missing. For instance, IAEA (2000) stipulates that ‘The regulatory body shall be provided with adequate authority and power, and it shall be ensured that it has adequate staffing and financial resources to discharge its assigned responsibilities.’ It is clear that the presence of a strong, independent and competent regulator has an enormously positive effect on societal confidence in regulatory decisions. However, in some countries the regulatory authority may not be wholly independent, or may be severely underfunded and understaffed, to the extent that they cannot fulfill their regulatory function. This situation represents a breakdown of the legal requirements for safe management of waste established in IAEA (2000). In this situation, the operating authority may have little trouble in getting a license to operate a disposal facility, but the lack of regulatory function would be likely to erode public confidence, and such a waste management system is ultimately unstable. Similarly, legislation must ‘set out the responsibilities and obligations in respect of financial provision for radioactive waste management and decommissioning’, and ‘set out the arrangements for provision of financial security in respect of any liabilities’. (IAEA, 2000). In the context of waste management systems, which may operate over several generations and require monitoring over additional generations, these stipulations require

© Woodhead Publishing Limited, 2010

542

Geological repository systems for safe disposal

the establishment of a system of financial surety, in which the current operators of the facility establish a trust fund for future generations to use to fund the closure and maintenance of the disposal facility. This financial surety must be established in law and must be firmly guarded against diversion of the resources to unrelated activities. The establishment of such financial systems has a strong positive effect on building trust and confidence in the overall safety case for a waste management system, as it ensures that potential liabilities in the future will be manageable. IAEA (2008) has addressed necessary components supporting the managerial contribution to the safety case. These contributions include the provision of quality assurance and quality control, continuity of record keeping and maintenance of staffing and equipment over the facility lifecycle. The safety case is therefore seen to be broader than the (primarily technical) aspects that have been described in the literature. However, these broader aspects are well understood and well established in IAEA requirements and in the regulatory and legal structures of many countries. All four elements of the safety case are necessary to support a stable and safe waste management program and to ensure regulatory and public acceptance.

16.6.3 Software advances As computer technologies have advanced, improvements in safety assessment modeling have become available. Numerical approaches that were once limited to supercomputers can now be conducted routinely on personal computers. These trends have led to two changes in safety assessment modeling: . .

Models have become more complex. Models have become more flexible.

The trend toward modeling complexity in safety assessments is clear, as demonstrated in a comparison of two successive IAEA programs on safety assessment (IAEA, 1995, 2004). In the earlier program (IAEA, 1995), calculational approaches were founded on analytical, semi-analytical and simple numerical methods. The conceptual models implemented in the program were similarly simple, and often the availability of solution methods required analysts to adapt their conceptual models, rather than adapting their solution methods to reflect accurately the conceptual model. By contrast, in the later program, models included multidimensional solutions, time-varying parameters and other complexities that had previously been impossible to analyze using extant safety assessment software. Therefore, there has been a significant improvement in the ability

© Woodhead Publishing Limited, 2010

Safety assessment for near-surface disposal

543

of safety assessors to match their models to a conceptual model, which has led to a tremendous improvement in the credibility of analyses. This improvement in computational capability has also had a negative effect. The ability to include ever more complex analyses in a safety assessment framework has meant that some assessors have increased the complexity of their analysis to the point that it becomes difficult to communicate. Safety assessment intrinsically requires clear communication of ideas, approaches, methods and data from the developer to a skeptical audience. As the complexity of modeling has increased, so has the trend toward opacity of safety assessments. Modern safety assessors should keep clearly in mind that the technical merit of their analyses is only of value if they can make a convincing case to their regulator and other stakeholders that the assessment is credible. Improved computer technologies have also allowed safety assessment models to become more flexible. Older codes, commonly written in FORTRAN, were written to implement a single conceptual model, and changes to the model required modifications to the code itself. Now, many safety assessments are carried out using integrated compartment-based modeling software packages. These software packages often have objectoriented graphical user interfaces, which allow the analyst to modify the equations to be solved with little effort and without altering the underlying code. This type of software allows rapid deployment of alternative conceptual models, clear visualization of relationships between components of the safety assessment and improved capabilities to conduct ‘what if’ analyses. The primary drawback to such software is the potential for breakdown of the management of traceability of analyses. The very flexibility of these tools means that there is a greater risk of quality assurance failures than when using hardwired, single conceptual model codes. It is anticipated that software and hardware developments will continue to influence the complexity and flexibility of safety assessments. As they do, there will continue to be a primary need for safety assessors to be able to communicate their results and to convince others that the approaches are credible. The pressures on clear communication are expected to increase in the future as technical methods in safety assessments continue to evolve.

16.7 .

Sources of further information and advice

NCRP (2005) provides a comprehensive overview of safety assessment for near-surface waste disposal systems. In addition to discussions of the basic approaches, specific recommendations are made for modeling various elements of the disposal system.

© Woodhead Publishing Limited, 2010

544 .

.

Geological repository systems for safe disposal US NRC (2000) provides extensive discussions of many key issues associated with conducting safety assessments for near-surface disposal facilities, with specific attention to approaches that are acceptable from a regulatory perspective. IAEA (2004) describes the results of a research program intended to explore key issues in safety assessments, and has formed the basis for a substantial international consensus on appropriate approaches to safety assessment.

16.8

References

Chapman N A, Andersson J, Robinson P, Skagius K, Wene C, Wiborgh M and Wingefors S (1995), ‘Systems analysis, scenario construction, and consequence analysis definition for SITE-94’, SKI Report 95:26, Swedish Nuclear Power Inspectorate. Cranwell R M, Guzowski R V, Campbell J E and Ortiz N R (1990), ‘Risk methodology for geologic disposal of radioactive waste: scenario selection procedure’, NUREG/CR-1667, SAND80-1429, Sandia National Laboratories. IAEA (1989). ‘Evaluating the reliability of predictions made using environmental transfer models’, IAEA Safety Series 100, International Atomic Energy Agency, Vienna, Austria. IAEA (1994), ‘Siting of near surface disposal facilities’, IAEA Safety Series 111-G3.1, International Atomic Energy Agency, Vienna, Austria. IAEA, (1995), ‘Safety assessment of near-surface radioactive waste disposal facilities: model intercomparison using simple hypothetical data (Test Case 1)’, First Report of NSARS, IAEA-TECDOC-846, International Atomic Energy Agency, Vienna, Austria. IAEA (1999), ‘Safety assessment for near surface disposal of radioactive waste, Safety Guide’, Safety Standards Series WS-G-1.1, International Atomic Energy Agency, Vienna, Austria. IAEA (2000), ‘Legal and governmental infrastructure for nuclear, radiation, radioactive waste and transport safety’, Safety Requirements Safety Standards Series GS-R-1, International Atomic Energy Agency, Vienna, Austria. IAEA (2003a), ‘Derivation of activity limits for the disposal of radioactive waste to near-surface facilities’, IAEA TECDOC Series 1380, International Atomic Energy Agency, Vienna, Austria. IAEA (2003b), ‘Safety indicators for the safety assessment of radioactive waste disposal’, IAEA-TECDOC-1372, International Atomic Energy Agency, Vienna, Austria. IAEA (2004), ‘Safety assessment methodologies for near surface disposal facilities’, IAEA-ISAM-1, International Atomic Energy Agency, Vienna, Austria. IAEA (2005), ‘Disposal options for disused radioactive sources’, TRS-436, International Atomic Energy Agency, Vienna, Austria. IAEA (2008), The management system for the disposal of radioactive waste’, IAEA Safety Standards Series GS-G-3.4, International Atomic Energy Agency, Vienna, Austria.

© Woodhead Publishing Limited, 2010

Safety assessment for near-surface disposal

545

ICRP (2000a), ‘Radiation protection recommendations as applied to the disposal of long-lived solid radioactive waste’, ICRP Publication 81, International Commission on Radiological Protection. ICRP (2000b), Protection of the Public in Situations of Prolonged Radiation Exposure. The Application of the Commission’s System of Radiological Protection to Controllable Radiation Exposure Due to Natural Sources and Long-Lived Radioactive Residues, Elsevier Science Ltd, Oxford. Kennedy Jr W E and Peloquin R A (1988), ‘Intruder scenarios for site-specific lowlevel radioactive waste classification’ DOE/LLW-71T, US Department of Energy, Idaho Operations Office, Idaho Falls, Idaho. Kozak M W (1994), ‘Decision analysis for low-level radioactive waste disposal facilities’, Radioactive Waste Management and Environment Restoration, 18, 209–223. Kozak M W, (1997) ‘Sensitivity, uncertainty, and importance analyses’, in Proceedings of the 18th Annual Department of Energy Low-Level Radioactive Waste Management Conference, 20–22 May 1997, Salt Lake City, Utah. Kozak M W, Chu M S Y and Mattingly P A (1990), ‘A performance assessment methodology for low-level waste facilities,’ NUREG/CR-5532, US Nuclear Regulatory Commission. Kozak M W, Olague N E, Rao R R and McCord J T, (1993), ‘Evaluation of a performance assessment methodology for low-level radioactive waste, Volume 1: Evaluation of modeling approaches’, NUREG/CR-5927 Vol. 1, US Nuclear Regulatory Commission. NCRP (2005), ‘Performance assessment of low-level waste disposal facilities’, NCRP Report 152, National Council on Radiation Protection and Measurements (NCRP), Bethesda, Maryland. NEA (1987), ‘Shallow land disposal of radioactive waste: reference levels for acceptance of long-lived radionuclides’, Nuclear Energy Agency, Organisation for Economic Co-operation and Development, Paris. NEA (2000), ‘Features, events and processes (FEPs) for geologic disposal of radioactive waste: an international database’, Nuclear Energy Agency, Organisation for Economic Co-operation and Development, Paris. NEA (2002), ‘Nuclear Energy Agency Radioactive Waste Management Committee Integration Group for the Safety Case (IGSC)’, in Workshop on Handling of Time Scales Assessing Post-Closure Safety, NEA/RWM/IGSC(2002)6. NEA (2004a), The handling of timescales in assessing post-closure safety’, Nuclear Energy Agency. NEA (2004b), ‘Post-closure safety case for geological repositories, nature and purpose’, Nuclear Energy Agency OECD/NEA, Paris. OCRWM (2008), ‘Total system performance assessment model/analysis for the License Application’, MDL-WIS-PA-000005 Rev. 0 January 2008, Office of Civilian Radioactive Waste Management, Las Vegas, Nevada. Oztunali O I and Roles G W (1986), ‘Update of Part 61, Impacts Analysis Methodology’, NUREG/CR-4370, US Nuclear Regulatory Commission, Washington, DC. Seitz R R, Garcia R S, Kostelnik K M and Starmer R J (1992), ‘Performance Assessment Handbook for Low-Level Radioactive Waste Disposal Facilities’, Report DOE/LLW-135, US Department of Energy.

© Woodhead Publishing Limited, 2010

546

Geological repository systems for safe disposal

Skagius K and Wiborgh M (1994), ‘Testing of influence diagrams as a tool for scenario development by application on the SFL 3-5 repository concept’, SKB Technical Report 94-47, Swedish Nuclear Fuel and Waste Management Company. Skagius K, Strom A and Wiborgh M (1995), ‘The use of interaction matrices for identification, structuring, and ranking of FEPs in a repository system’, SKB Technical Report 95-22, Swedish Nuclear Fuel and Waste Management Company. US NRC (2000), ‘A performance assessment methodology for low-level radioactive waste disposal facilities, Recommendations of NRC’s Performance Assessment Working Group’, NUREG-1573, US Nuclear Regulatory Commission, Washington, DC.

© Woodhead Publishing Limited, 2010

17 Treatment of uncertainty in performance assessments for the geological disposal of radioactive waste J . C . H E L T O N , Arizona State University, USA; C . J . S A L L A B E R R Y , Sandia National Laboratories, USA

Abstract: The treatment of uncertainty in performance assessments for the geological disposal of radioactive waste is discussed and illustrated. The following topics are considered: (1) the conceptual design and structure of a performance assessment including the separation of aleatory and epistemic uncertainty, (2) the numerical propagation of uncertainty, (3) the computational design of a performance assessment, and (4) sampling-based methods for sensitivity analysis. The presented concepts and techniques are illustrated with results from the 2008 performance assessment for the proposed repository for high-level radioactive waste at Yucca Mountain, Nevada. Key words: aleatory uncertainty, epistemic uncertainty, performance assessment, radioactive waste disposal, sensitivity analysis, uncertainty analysis.

17.1

Introduction

A performance assessment (PA) for a radioactive waste disposal facility, or in general any type of engineered facility, is an analysis intended to answer three questions about the facility (i.e. Q1, Q2 and Q3) and one question about the analysis itself (i.e. Q4): Q1, ‘What can happen?’; Q2, ‘How likely is it to happen?’; Q3, ‘What are the consequences if it does happen?’; and Q4, ‘How much confidence exists in the answers to the first three questions?’ Two types of uncertainty are inherent in the answers to the preceding questions. Questions Q1 and Q2 relate to uncertainty with respect to future events (e.g. seismic events, igneous events, . . . ) at the facility under consideration whose occurrence, within the limits of our ability to predict 547 © Woodhead Publishing Limited, 2010

548

Geological repository systems for safe disposal

the future, is assumed to be random. The descriptor aleatory is usually used for this type of uncertainty. Question Q4 relates to a lack of knowledge with respect to the appropriateness/correctness of the assumptions, models and parameter values that underlie answering questions Q1, Q2 and Q3. The descriptor epistemic is usually used for this type of uncertainty. The nature of aleatory and epistemic uncertainty and the importance of their separation in analyses of complex systems has been discussed by a number of authors (e.g. see References 1 to 7). Answering the four indicated questions leads to an analysis based on three basic mathematical structures or entities: EN1, a probability space characterizing aleatory uncertainty; EN2, a function that predicts the physical behavior of the facility under consideration; and EN3, a probability space characterizing epistemic uncertainty.8,9 The probability space corresponding to EN1 characterizes aleatory uncertainty and provides the basis for answering questions Q1 and Q2. In practice, the function corresponding to EN2 is one or more very complex numerical models and provides the basis for answering question Q3. The probability space corresponding to EN3 characterizes epistemic uncertainty and provides the basis for answering question Q4. The nature of the basic analysis components EN1, EN2 and EN3 is elaborated on in Section 17.2 and illustrated with results from the 2008 PA for the proposed Yucca Mountain (YM) repository for high-level radioactive waste.10 However, this presentation is not intended to be specific to the 2008 YM PA; rather, the 2008 YM PA is used as a convenient and relevant source of examples for general concepts. In YM project documentation, the descriptor total system performance assessment (TSPA) is typically used instead of the simpler and more widely used descriptor performance assessment (PA). Closely associated with the characterization of epistemic uncertainty provided by the probability space corresponding to EN3 and the answering of question Q4 are the concepts of uncertainty analysis and sensitivity analysis, where uncertainty analysis designates the determination of the epistemic uncertainty in analysis results that derives from epistemic uncertainty in analysis inputs, and sensitivity analysis designates the determination of the contribution of the epistemic uncertainty in individual analysis inputs to the epistemic uncertainty in analysis results. Basically, uncertainty and sensitivity analysis are the means by which EN3 gives rise to the answer to question Q4. A number of approaches to uncertainty and sensitivity analysis exist, including differential analysis, response surface methods, variance decomposition methods, and sampling-based (i.e. Monte Carlo) methods.11–18 Of the preceding methods, it is felt that sampling-based methods provide the most broadly applicable uncertainty and sensitivity analysis procedures for use in PAs for radioactive waste disposal. For this reason, sampling-

© Woodhead Publishing Limited, 2010

Treatment of uncertainty in performance assessments

549

based methods are discussed in Sections 17.3 to 17.5. Specifically, the propagation of aleatory and epistemic uncertainty is discussed in Section 17.3; the importance of appropriate analysis design is discussed in Section 17.4; and sensitivity analysis is discussed in Section 17.5. The presentation then ends with a concluding discussion in Section 17.6.

17.2

Conceptual structure of a performance assessment

The first entity underlying a PA, EN1, corresponds to a probability space (A, A, pA), where A is the set of everything that could occur in the particular universe under consideration (i.e. over some specified time period for the facility under analysis), A is a suitably restricted set of subsets of A for which probability is defined, and pA is the function that defines probability for elements of A (i.e. if S is an element of A, then pA(S) is the probability of S). In the usual terminology of probability theory, A is called the sample space or sometimes the universal set, elements of A are called elementary events, elements of A are called events, pA is called a probability measure, and pA(S) is the probability of the event S. In the terminology of radioactive waste disposal, elements of A are often called futures, elements of A are often called scenarios or scenario classes, and pA(S) is the probability of scenario S. Although the concept of a probability space is important conceptually and convenient notationally, calculations involving a probability space (A, A, pA) are often described with a density function dA(a), where ð ½17:1 pA ðSÞ ¼ dA ðaÞdS S

for S Î A, a Î S and dS corresponding to an increment of volume from S. Then, the expected value, variance, cumulative distribution function (CDF), and complementary cumulative distribution function (CCDF) at time t (year) associated with a real-valued function y = f(t|a) defined on A are defined by ð fðtjaÞdA ðaÞdA; ½17:2 EA ½fðtjaÞ ¼ A

ð VA ½fðtjaÞ ¼

A

ffðtjaÞ  EA ½fðtjaÞg2 dA ðaÞdA;

½17:3

ð pA ½fðtjaÞ4y ¼

A

dy ½fðtjaÞdA ðaÞdA;

© Woodhead Publishing Limited, 2010

½17:4

550

Geological repository systems for safe disposal

and

ð pA ½y4fðtjaÞ ¼

A

dy ½fðtjaÞdA ðaÞdA;

½17:5

respectively, where

dy ½fðtjaÞ ¼

1 0

if fðtjaÞ4y otherwise;

dy ½fðtjaÞ ¼

1

if fðtjaÞ > y

0

otherwise,

and dA represents an increment of volume from A. A CCDF is defined in Equation [17.5] because of the typical usage of CCDFs rather than CDFs to represent uncertainty in risk assessments. In particular, a CCDF answers the question ‘How likely is it to be this bad or worse?’, which is usually the question asked with respect to individual consequences in a risk assessment. However, conversion between CCDFs and CDFs is straightforward as a CDF is simply one minus the corresponding CCDF. In turn, the q quantile value (e.g. q = 0.05, 0.5 ~ median, 0.95) for f(t|a) is the value y such that ð q ¼ pA ½fðtjaÞ4y ¼ dy ½fðtjaÞ dA ðaÞ dA: ½17:6 A

For notational purposes, the q quantile value for f(t|a) can be represented by Qq[f(t|a)]. The variance VA[f(t|a)] provides less information than the CDF and CCDF defined by the probabilities pA[f(t|a) ≤ y] and pA[y < f(t|a)] respectively, and is rarely used in the summary of results obtained in a PA. In PAs for radioactive waste disposal, the probability space (A, A, pA) for aleatory uncertainty is usually defined to characterize the occurrence of potential future events over some time period of interest (e.g. for a time period corresponding to the initial 104 years after repository closure) that could affect the behavior/evolution of the repository and the waste contained in the repository. Specifically, each element a of the sample space A is a vector of the form a ¼ ½a1 ; a2 ; . . . ; an , where the elements of a characterize the properties of one potential sequence of occurrences over the time interval under consideration. For example, the future behavior of the repository might be affected by a single class of disruptive events (e.g. seismic events) whose occurrence is characterized by a Poisson process with a rate constant l (year1). Each individual event is characterized by a time t (year) of occurrence and a vector p of additional properties (e.g. peak ground velocity, number of damaged waste packages, . . . ). Then, for a specified time period [a, b] (e.g. [a, b] = [0, 104 year]), each future a would be a vector of the form a ¼ ½n; t1 ; p1 ; t2 ; p2 ; . . . ; tn ; pn ;

© Woodhead Publishing Limited, 2010

½17:7

Treatment of uncertainty in performance assessments

551

where n is the number of occurrences in the time interval [a, b], a4t1 4t2 4    4tn 4b are the times of the individual occurrences, p1, p2, . . . , pn are vectors indicating the properties of the individual occurrences, and aN = [0] represents the future in which no events occur. In turn, the sample space A would have the form  A ¼ a : a ¼ ½n; t1 ; p1 ; t2 ; p2 ; . . . ; tn ; pn  for (i) n ¼ 0; 1; 2; . . . ; (ii) a4t1 4t2 4     ðiiiÞ pi 2 P for i ¼ 1; 2; . . . ; n ;

4tn 4b; and

½17:8

where P is the set of all possible values for the property vector p and scenario probabilities pA(S) are defined for subsets S of A. For example, pA ðAn Þ ¼ f½lðb  aÞn =n!g exp½lðb  aÞ

½17:9

for An ¼fa: a ¼ ½n; t1 ; p1 ; t2 ; p2 ; . . . ; tn ; pn 

 for a4t1 4t2 4    4tn 4b and pi 2 P for i ¼ 1; 2; . . . ; n :

The sets An are examples of the elements of A. However, much more complex examples exist (e.g. subsets of A defined with restrictions involving elements of P); in such cases, definitions of the corresponding set probabilities can become very complicated. As an example, the representation of aleatory uncertainty in the 2008 PA for the proposed YM repository is summarized in Table 17.1. The second entity underlying a PA, EN2, corresponds to a model, or more realistically a large system of interacting models, that predicts the evolution of the repository and various summary measures associated with this evolution (e.g. groundwater flow, radionuclide release and transport, radiation exposures, . . . ). Notationally, this model can be represented by a function of the form f ðtjaÞ ¼ ½f1 ðtjaÞ; f2 ðtjaÞ; . . . ; fm ðtjaÞ;

½17:10

where t corresponds to time (year), each element fj(t|a) of f(t|a) is a specific calculated result, and a is an element of the sample space A for aleatory uncertainty. In general, the value of f(t|a), and indeed the actual structure of the individual models that are combined to produce f(t|a), will change with changing values for a. As an example, the component models and associated connections that define f(t|a) for nominal conditions (i.e. for aN) in the 2008 PA for the YM repository are summarized in Fig. 17.1. An overview of the processes incorporated into the definition of f(t|a) is given in References 19 to 21, and

© Woodhead Publishing Limited, 2010

552

Geological repository systems for safe disposal

Table 17.1 Representation of aleatory uncertainty in the 2008 YM PA (Section J4.4 of Reference 10) Individual futures: a ¼ ½nEW; nED; nII; nIE; nSG; nSF; aEW ; aED ; aII ; aIE ; aSG ; aSF  where, for a time interval [a, b] (e.g. [0, 104 year] or [0, 106 year]), nEW = number of early waste package (WP) failures, nED = number of early drip shield (DS) failures, nII = number of igneous intrusive (II) events, nIE = number of igneous eruptive (IE) events, nSG = number of seismic ground (SG) motion events, nSF = number of seismic fault (SF) displacement events, aEW = vector defining the nEW early WP failures, aED = vector defining the nED early DS failures, aII = vector defining the nII igneous intrusive events, aIE = vector defining the nIE igneous eruptive events, aSG = vector defining the nSG seismic ground motion events, and aSF = vector defining the nSF seismic fault displacement events. Sample space for aleatory uncertainty: A ¼ fa: a ¼ ½nEW; nED; nII; nIE; nSG; nSF; aEW ; aED ; aII ; aIE ; aSG ; aSF g Example scenario classes: Nominal, AN ¼ fa: a 2 A and nEW ¼ nED ¼ nII ¼ nIE ¼ nSG ¼ nSF ¼ 0g; early WP failure, AEW ¼ fa: a 2 A and nEW51g; early DS failure, AED ¼ fa: a 2 A and nED51g; igneous intrusive, AII ¼ fa: a 2 A and nII51g; igneous eruptive, AIE ¼ fa: a 2 A and nIE51g; seismic ground motion, ASG ¼ fa: a 2 A and nSG51g; seismic fault displacement, ASF ¼ fa: a 2 A and nSF51g; early failure, AE ¼ AEW [ AED ; igneous, AI ¼ AII [ AIE ; seismic; AS ¼ ASG [ ASF Scenario class probabilities: pA(AN) = probability of no disruptions of any kind; pA(AEW) = probability of one or more early WP failures; pA(AED) = probability of one or more early DS failures; pA(AII) = probability of one or more II events; pA(AIE) = probability of one or more IE events; pA(ASG) = probability of one or more SG motion events; pA(ASF) = probability of one or more SF displacement events; pA(AE) = probability of one or more early failures; pA(AI) = probability of one or more igneous events; pA(AS) = probability of one or more seismic events.

a more detailed description is provided in Section 6 of Reference 10 and in the extensive technical reports cited in this reference. The third entity underlying a PA, EN3, corresponds to a probability space (E, E, pE) for epistemic uncertainty. The conceptual properties associated with the probability space (E, E, pE) are the same as indicated in Equations [17.1] to [17.6] for the probability space (A, A, pA) for aleatory uncertainty.

© Woodhead Publishing Limited, 2010

Treatment of uncertainty in performance assessments

553

17.1 Component models and associated connections that define f(t|aN) for nominal conditions in the 2008 PA for the YM repository (Figure 6.1.4-1 of Reference 10).

In general, the elements of the sample space E are vectors of the form e ¼ ½ eA ; eM 

¼ eA1 ; eA2 ; . . . ; eA;nEA ; eM1 ; eM2 ; . . . ; eM;nEM ¼ ½e1 ; e2 ; . . . ; enE ;

½17:11

nE ¼ nEA þ nEM;

where eA ¼ ½e1 ; eA2 ; . . . ; eA;nEA  is a vector of epistemically uncertain quantities used in the characterization of aleatory uncertainty (e.g. a rate

© Woodhead Publishing Limited, 2010

554

Geological repository systems for safe disposal

term that defines a Poisson process), eM ¼ ½eM1 ; eM2 ; . . . ; eM;nEM  is a vector of epistemically uncertain quantities used in the evaluation of f(t|a) (e.g. a distribution coefficient), and the concept of an uncertain quantity is interpreted broadly enough to include designators for possible values for poorly known functions or models. When notationally convenient, the probability space (E, E, pE) can be represented with a density function dE(e). Reference was made to ‘poorly known functions or models’ in the preceding paragraph. This form of epistemic uncertainty is often given the designation model uncertainty22 and involves a situation where there are multiple alternative models for a process and the analysts involved are not sure which is the appropriate model to use in the specific analysis context under consideration. Although there is no clear divide between where parameter uncertainty ends and model uncertainty begins, the designation model uncertainty is usually used in reference to a situation involving a finite number of structurally distinct models for a process. If probability is being used to characterize epistemic uncertainty mathematically, then a probability distribution would be defined to represent the analysts’ degree of belief with respect to which alternative model is the appropriate model to use. With this approach, the ‘model’ would simply be one more epistemically uncertain variable that is sampled (i.e. an integer-valued pointer variable would be sampled, with the different values for this variable designating the use of different models). Most large analyses have a few pointer variables that identify different possible models. Usually, model uncertainty involves a very limited number of alternative models. In this situation, a possibility is to perform a ceteris paribus analysis in which the entire analysis is performed repeatedly with a different model used in each repetition. However, this approach has the potential to be very computationally demanding. In practice, the probability space (E, E, pE) is defined by assigning probability distributions to the individual elements of e. In addition, correlations and other restrictions involving the elements of e may also be specified. The specified distributions serve as mathematical summaries of all available information with respect to where the appropriate values for the elements of e are located and are often developed through expert review processes.23–32 As an example, the 2008 PA for the YM repository considered nE = 392 epistemically uncertain analysis inputs. Examples of these inputs are presented in Table 17.2. Specifically, Table 17.2 contains variables that appear in the example sensitivity analyses presented in Section 17.5. All 392 epistemically uncertain variables considered in the 2008 YM PA are listed and discussed in Tables K3-1, K3-2 and K3-3 of Reference 10. With the introduction of the probability space (E, E, pE) for epistemic uncertainty, the representation for the system model in Equation [17.10]

© Woodhead Publishing Limited, 2010

Treatment of uncertainty in performance assessments

555

Table 17.2 Examples of the nE = 392 epistemically uncertain variables considered in the 2008 YM PA (Tables K3-1, K3-2 and K3-3 of Reference 10) DSNFMASS – scale factor used to characterize uncertainty in radionuclide content of defense spent nuclear fuel (dimensionless). Distribution: triangular. Range: 0.45 to 2.9. Most likely: 0.62. HLWGRNDS – specific surface area of high-level waste glass degradation rind (m2/g). Distribution: uniform. Range: 10 to 38. IGRATE – frequency of intersection of the repository footprint by a volcanic event (year1). Distribution: piecewise uniform. Range: 0 to 7.766107. INFIL – pointer variable for determining infiltration conditions: 10th, 30th, 50th or 90th percentile infiltration scenario (dimensionless). Distribution: discrete. Range: {1, 2, 3, 4}. SCCTHR – stress threshold for stress corrosion cracking (MPa). Distribution: uniform. Range: 315.9 to 368.55. SCCTHRP – residual stress threshold for stress corrosion cracking nucleation of Alloy 22 (as a percentage of yield strength in MPa) (dimensionless). Distribution: uniform. Range: 90 to 105. SZFIPOVO – logarithm of flowing interval porosity in volcanic units (dimensionless). Distribution: Piecewise uniform. Range: 5 to 1. SZGWSPDM – logarithm of the scale factor used to characterize uncertainty in groundwater specific discharge (dimensionless). Distribution: piecewise uniform. Range: 0.951 to 0.951. THERMCON – selector variable for one of three host-rock thermal conductivity scenarios (low, mean and high) (dimensionless). Distribution: discrete. Range: {1, 2, 3}. WDCRCDEN – ratio of stress corrosion cracking area to unit of seismic damaged area for a waste package (dimensionless). Distribution: uniform. Range: 0.00327 to 0.0131. WDGCA22 – temperature-dependent slope term of Alloy 22 general corrosion rate (K). Distribution: truncated normal. Range: 666 to 7731. Mean: 4905. Standard deviation: 1413. WDGCUA22 – variable for selecting distribution for general corrosion rate (low, medium or high) (dimensionless). Distribution: discrete. Range: {1, 2, 3}. WDNSCC – stress corrosion cracking growth rate exponent (repassivation slope) (dimensionless). Distribution: truncated normal. Range: 0.935 to 1.395. Mean: 1.165. Standard deviation: 0.115. WDZOLID – deviation from median yield strength range for outer waste package lid (dimensionless). Distribution: truncated normal. Range: 3 to 3. Mean/median/mode: 0. Standard deviation: 1.

© Woodhead Publishing Limited, 2010

556

Geological repository systems for safe disposal

becomes f ðtja; eM Þ ¼ ½f1 ðtja; eM Þ; f2 ðtja; eM Þ; . . . ; fm ðtja; eM Þ

½17:12

and the representation for the density function associated with the probability space (A, A, pA) for aleatory uncertainty becomes dA(a|eA). In turn, results of the form in Equations [17.1], [17.2], [17.4] and [17.5] become ð ½17:13 pA ðSjeA Þ ¼ dA ðajeA ÞdS; S

ð EA ½fðtja; eM ÞjeA  ¼

A

fðtja; eM ÞdA ðajeA ÞdA;

½17:14

ð pA ½fðtja; eM Þ4yjeA  ¼

A

and

dy ½fðtja; eM ÞdA ðajeA ÞdA;

½17:15

ð pA ½y < fðtja; eM ÞjeA  ¼

A

dy ½fðtja; eM ÞdA ðajeA ÞdA;

½17:16

where f (τ|a,eM) corresponds to one of the functions fj(t|a,eM) contained in f(t|a,eM). Similarly, the q quantile value for f(t|a) defined in Equation [17.6] now depends on e = [eA, eM] and is appropriately represented by QAq[f(t|a, eM)|eA]. As e changes, each of the preceding quantities also changes. In turn, the quantities in Equations [17.13] to [17.16] have distributions that derive from the probability space (E, E, pE) for epistemic uncertainty. Similarly, elements f(t|a, eM) of f(t|a, eM) for fixed values of a also have distributions that derive from the probability space (E, E, pE) for epistemic uncertainty. For example, the expected value, CDF, CCDF and q quantile QEq{EA[f(t|a, eM)|eA]} for EA[f(t|a, eM)|eA] resulting from epistemic uncertainty are defined by ð EE fEA ½fðtja; eM ÞjeA g ¼ EA ½fðtja; eM ÞjeA dE E

ð ð ¼

E

A

fðtja; eM ÞdA ðajeA ÞdA dE;

© Woodhead Publishing Limited, 2010

½17:17

Treatment of uncertainty in performance assessments

557

ð pE fEA ½fðtja; eM ÞjeA 4yg ¼

E

dy fEA ½fðtja; eM ÞjeA gdE ðeÞdE ð

ð ¼

E

dy

A

fðtja; eM ÞdA ðajeA ÞdA dE ðeÞdE; ½17:18

ð pE fEA ½y < fðtja; eM ÞjeA g ¼

E

dy fEA ½fðtja; eM ÞjeA gdE ðeÞdE ð

ð ¼

E

dy

A

fðtja; eM ÞdA ðajeA ÞdA dE ðeÞdE; ½17:19

and the value of y such that q ¼ pE fEA ½fðtja; eM ÞjeA 4yg, respectively. Similarly, the expected value, CDF, CCDF and q quantile QEq[f(t|a, eM)] for f(t|a, eM) resulting from epistemic uncertainty are defined by ð EE ½fðtja; eM Þ ¼ fðtja; eM ÞdEM ðeM ÞdEM; ½17:20 EM

ð pE ½fðtja; eM Þ4y ¼

EM

dy ½fðtja; eM ÞdEM ðeM ÞdEM;

½17:21

ð pE ½y < fðtja; eM Þ ¼

EM

dy ½fðtja; eM ÞdEM ðeM ÞdEM;

½17:22

and the value of y such that q ¼ pE ½fðtja; eM Þ4y, respectively. The preceding results associated with f(t|a, eM) are defined with respect to the probability space (EM, EM, pEM) for eM because eM is the only epistemically uncertain quantity under consideration.

17.3

Propagation of uncertainty

The propagation of aleatory uncertainty is considered first. Direct evaluation of the integrals in Equations [17.13] to [17.16] is usually not possible. As a result, some form of numerical procedure must be used. The two most widely used procedures are simple random sampling and stratified sampling. In simple random sampling, a random sample

½17:23 aj ¼ a1j ; a2j ; . . . ; anj ;j ; j ¼ 1; 2; . . . ; nSA; is generated from A in consistency with the definition of the probability space (A, A, pA). In general, the properties of (A, A, pA), and hence of the

© Woodhead Publishing Limited, 2010

558

Geological repository systems for safe disposal

resultant sample in Equation [17.23], will depend on eA. In turn, the results in Equations [17.13] to [17.16] are approximated by pA ðSjeA Þ%

nSA X

dS ðaj Þ=nSA;

½17:24

j¼1

EA ½fðtja; eM ÞjeA %

nSA X

fðtjaj ; eM Þ=nSA;

½17:25

j¼1

pA ½fðtja; eM Þ4yjeA %

nSA X



dy fðtjaj ; eM Þ =nSA;

½17:26

j¼1

and pA ½y < fðtja; eM ÞjeA %

nSA X



dy fðtjaj ; eM Þ =nSA;

½17:27

j¼1

respectively, with dS(aj) = 1 if aj Î S and dS(aj) = 0 if aj Ï S. In stratified sampling, the sample space A is subdivided into a sequence of subsets Aj, j = 1, 2, . . . , nSA, with the properties that Èj Aj = A and Ai ÇAj = Æ for i ¹ j. Then, the results in Equations [17.13] to [17.16] are approximated by pA ðSjeA Þ%

nSA X

dS ðaj ÞpA ðAj jeA Þ;

½17:28

j¼1

EA ½fðtja; eM ÞjeA %

nSA X

fðtjaj ; eM ÞpA ðAj jeA Þ;

½17:29

j¼1

pA ½fðtja; eM Þ4yjeA %

nSA X



dy fðtjaj ; eM Þ pA ðAj jeA Þ;

½17:30

j¼1

and pA ½y < fðtja; eM ÞjeA %

nSA X



dy fðtjaj ; eM Þ pA ðAj jeA Þ;

½17:31

j¼1

respectively, where aj is a representative element of Aj. Event trees are often used to assess the effects of aleatory uncertainty and, in essence, are simply algorithms for implementing stratified sampling.

© Woodhead Publishing Limited, 2010

Treatment of uncertainty in performance assessments

559

The widely used Kaplan/Garrick ordered triple representation for risk33 corresponds to a summary description of an analysis based on stratified sampling. Specifically, this representation defines risk by the set ðS j ; pSj ; cS j Þ;

j ¼ 1; 2; . . . ; nS;

½17:32

where Sj is a set of similar occurrences, pSj is the probability of Sj, and cSj is a vector of consequences associated with Sj. In the context of the stratified sampling results in Equations [17.28] to [17.31], Sj = Aj, pSj = pA(Aj|eA), and cSj = f(aj|eM). The propagation of epistemic uncertainty is now considered. As for the propagation of aleatory uncertainty, direct evaluation of the integrals that formally define the propagation of epistemic uncertainty is unlikely to be possible in a real analysis. Simple random sampling and stratified sampling are possibilities for the propagation of epistemic uncertainty. However, because of its efficient stratification properties, Latin hypercube sampling is widely used for the propagation of epistemic uncertainty in complex and computationally demanding analyses.34,35 Latin hypercube sampling operates in the following manner to generate a sample of size nSE from the distributions D1, D2, . . . , DnE associated with the elements of e = [e1, e2, . . . , enE], where the distributions D1, D2, . . . , DnE in effect define the probability space (E, E, pE) for epistemic uncertainty. The range of each ej is exhaustively divided into nSE disjoint intervals of equal probability and one value eij is randomly selected from each interval. The nSE values for e1 are randomly paired without replacement with the nSE values for e2 to produce nSE pairs. These pairs are then randomly combined without replacement with the nSE values for e3 to produce nSE triples. This process is continued until a set of nSE nE-tuples

½17:33 ei ¼ ei1 ; ei2 ; . . . ; ei;nE ; i ¼ 1; 2; . . . ; nSE; is obtained, with this set constituting the Latin hypercube sample (LHS). If needed, a restricted pairing technique exists that can be used to induce a specified rank correlation structure in an LHS.36,37 Once the LHS in Equation [17.33] is generated, the results in Equations [17.17] to [17.19] can be approximated by EE fEA ½fðtja; eM ÞjeA g%

nSE X

EA ½fðtja; eMi ÞjeAi =nSE

i¼1

%

" nSE X nSA X i¼1

# fðtjaj ; eMi Þ=nSA =nSE;

j¼1

© Woodhead Publishing Limited, 2010

½17:34

560

Geological repository systems for safe disposal pE fEA ½fðtja; eM ÞjeA 4yg%

nSE X

dy fEA ½fðtja; eMi ÞjeAi g=nSE

i¼1

%

nSE X

dy

" nSA X

i¼1

#

½17:35

fðtjaj ; eMi Þ=nSA =nSE;

j¼1

and pE fy < EA ½fðtja; eM ÞjeA g%

nSE X

dy fEA ½fðtja; eMi ÞjeAi g=nSE

i¼1

%

nSE X

" dy

i¼1

nSA X

#

½17:36

fðtjaj ; eMi Þ=nSA =nSE;

j¼1

respectively, and the q quantile QEq{EA[f(t|a, eM)|eA]} for EA[f(t|a, eM)|eA] resulting from epistemic uncertainty can be obtained from the approximation in Equation [17.35]. Similarly, the results in Equations [17.20] to [17.22] can be approximated by EE ½fðtja; eM Þ%

nSE X

fðtja; eMi =nSE;

½17:37

i¼1

pE ½fðtja; eM Þ4y%

nSE X

dy ½fðtja; eMi Þ=nSE;

½17:38

i¼1

and pE ½y < fðtja; eM Þ%

nSE X

dy ½fðtja; eMi Þ=nSE;

½17:39

i¼1

respectively, and the q quantile QEq[f(t|a, eM)] for f(t|a, eM) resulting from epistemic uncertainty can be obtained from the approximation in Equation [17.38]. Distributional results for other quantities dependent on e and eM are obtained in a similar manner. The propagation of uncertainty is now illustrated with results from the 2008 PA for the YM repository. This analysis used an LHS of the form ei ¼ ½eAi ; eMi ;

i ¼ 1; 2; . . . ; nSE;

½17:40

of size nSE = 300 from nE = 392 epistemically uncertain variables (i.e. from the sample space E for epistemic uncertainty as indicated in conjunction with Equation [17.11] and Table 17.2). A large number of analysis results conditional on specific realizations a of

© Woodhead Publishing Limited, 2010

Treatment of uncertainty in performance assessments

561

17.2 Uncertainty analysis results for DN (τ |aN, eM): (a) DN(τ |aN, eMi) for all (i.e. nSE = 300) LHS elements, and (b) estimated CDF and CCDF for DN(66105 year |aN, em) (Figure J5-1 of Reference 10).

aleatory uncertainty (i.e. elements of the function f(t|a, eM); see Table K3-4 of Reference 10) were analyzed as part of the 2008 YM PA. As a single example, results for dose DN(τ |aN, eM) (mrem/year) to a reasonably maximally exposed individual (RMEI) under nominal conditions (i.e. for the future aN indicated in conjunction with Equation [17.7] involving no disruptions of any kind) are shown in Fig. 17.2. The RMEI is a hypothetical individual specified by the US Nuclear Regulatory Commission (NRC) for use in assessing compliance with post-closure requirements for the YM repository.38 Specifically, Fig. 17.2(a) contains 300 individual curves corresponding to DN(τ |aN, eMi) for each of the 300 LHS elements indicated in Equation [17.40]. The spread in these curves provides a representation of the epistemic uncertainty present in the estimation of DN(τ |aN, eM). The mean and quantile curves (i.e. q = 0.05, 0.5 ~ median, 0.95) in Fig. 17.2(a)

© Woodhead Publishing Limited, 2010

17.3 Uncertainty analysis results for DSG(t|a, eM): (a) CCDFs for DSG(104 year|a, eMi) for all (i.e. nSE = 300) LHS elements, (b) CCDF for EA[DSG(104 year|a, eM)|eA], (c) EA[DSG(t|a, eM)|eA] for all LHS elements, and (d) mean and quantile curves for EA[DSG(t|a, eM)|eA] (Figures J8.3-5 and J8.3-10 of Reference 10).

562 Geological repository systems for safe disposal

© Woodhead Publishing Limited, 2010

Treatment of uncertainty in performance assessments

563

provide additional summaries of the epistemic uncertainty present in the estimation of DN(τ |aN, eM) and are defined as indicated in Equations [17.20] and [17.21] and estimated as indicated in Equations [17.37] and [17.38]. The CDF and CCDF for DN(66105 year |aN, eM) in Fig. 17.2(b) are defined and estimated as indicated in Equations [17.21], [17.22], [17.38], and [17.39] and provide a more detailed summary of the epistemic uncertainty in DN(66105 year |aN, eM) than is provided by selected quantiles (e.g. by q = 0.05, 0.5 and 0.95 at 66105 year). Many additional results of this type are presented in Appendices J and K of Reference 10. Distributions over aleatory uncertainty are now considered. As a single example, dose DSG(t|a, eM) to the RMEI from seismic ground motion events is considered (Fig. 17.3). Although these results constitute a subset of the results considered in the 2008 YM PA, they conceptually correspond to the outcomes of a PA in which seismic ground motion events are the only occurrences under consideration (see Appendix J of Reference 10). Individual CCDFs for DSG(104 year|a, eM) conditional on eMi for each element of the LHS in Equation [17.40] are shown in Fig. 17.3(a). In concept, the CCDFs in Fig. 17.3(a) can be defined and approximated as indicated in Equations [17.22] and [17.27]. However, in this analysis, as is often the case in other analyses, this type of naı¨ ve sampling is very inefficient because of the large number of futures a that result in no consequences of interest (i.e. waste package damage and resultant dose to the RMEI in this example). To avoid this inefficiency, the CCDFs in Fig. 17.3(a) were estimated by only sampling futures that involved waste package damage resulting from seismic ground motion and then implementing a correction for this restricted sampling. Specifically, the CCDFs in Fig. 17.3(a) were estimated from the exceedance probabilities  

pA DSG 104 yearja; eMi > yjeAi nSA  X   ~ SG jei ½17:41 %pA A dy DSG 104 yearjaj ; eMi =nSA; j¼1

~ SG is the subset of ASG that contains only futures that cause where (1) A ~ SG jei Þ for A ~ SG depends in waste package damage, (2) the probability pA ðA part on uncertain waste package properties defined by elements of eM, and ~ SG (see Section J8.3 of (3) the individual futures aj were sampled from A Reference 10 for additional details). Although the vector eA did not contain any quantities that affected the probability of seismic events in the 2008 YM PA, eA is retained in the notation used in Equation [17.41] for conceptual generality. As indicated in Equations [17.14] and [17.25], CCDFs of the form appearing in Fig. 17.3(a) can be reduced to expected values EA[DSG(104

© Woodhead Publishing Limited, 2010

564

Geological repository systems for safe disposal

year|a, eMi)|eAi] over aleatory uncertainty. With use of the approximation procedure indicated in conjunction with Equation [17.41], this reduction becomes   EA DSG 104 yearja; eMi jeAi ~ SG jei Þ %pA ðA

nSA X

  DSG 104 yearjaj ; eMi =nSA

½17:42

j¼1

and produces 300 expected values EA[DSG(104 year|a, eMi)|eAi] that can be summarized with a CCDF (Fig. 17.3(b)). The CCDF in Fig. 17.3(b) provides a representation of the epistemic uncertainty present in the estimation of EA[DSG(104 year|a, eM)|eA] and corresponds to results indicated in Equations [17.19] and [17.36] with the estimation of EA[DSG(104 year|a, eMi)|eAi] in Equation [17.36] modified to be consistent with the estimation procedure in Equation [17.42]. Selected quantiles (i.e. q = 0.05, 0.5, 0.95) are also shown in Fig. 17.3(b). As indicated in conjunction with Equations [17.18] and [17.35], these quantiles are obtained by solving the equation q ¼ pE fEA ½fðtja; eM ÞjeA 4yg for y. However, once the CCDF in Fig. 17.3(b) is constructed, these quantiles can also be obtained by (1) starting at 1 – q on the ordinate, (2) drawing a horizontal line to the CCDF, and then (3) drawing a vertical line to the abscissa to obtain the quantile QEq{EA[f(t|a, eM)|eA]}. The 300 expected doses EA[DSG(104 year|a, eMi)|eAi] summarized in the CCDF in Fig. 17.3(b) correspond to the 300 expected doses above τ = 104 year in Fig. 17.3(c). In concept, sampling-based calculations of the form indicated in Equation [17.42] can be performed at multiple times to obtain the 300 time-dependent expected dose curves in Fig. 17.3(c) (i.e. one curve for each LHS element). The spread of these curves provides a summary of the epistemic uncertainty in EA[DSG(104 year|a, eM) |eA] as a function of time. As a result of specific properties of seismic events in the 2008 YM PA, it was possible to use appropriately designed quadrature procedures to calculate directly the time-dependent expected values in Fig. 17.3(c). Within numerical error, this procedure produced the same expected doses as the sampling-based procedure in Equation [17.42] but was numerically more efficient because it avoided the calculations involved in initially calculating CCDFs at a sequence of times (see Section J8.3 of Reference 10 for details). Thus, although the CCDFs in Fig. 17.3(a) are important for showing the effects of aleatory uncertainty, they did not provide a computationally efficient means of estimating the expected dose curves in Fig. 17.3(c). The determination of mean and quantile results as indicated in Fig. 17.3 (b) can be carried out for a sequence of times. Plotting these results as functions of time produces the mean and quantile curves in Fig. 17.3(d),

© Woodhead Publishing Limited, 2010

Treatment of uncertainty in performance assessments

565

17.4 Mean and quantile curves for pA[DSG(104 year|a, eM) > y|eA] (Figure J8.3-10 of Reference 10).

with these curves providing an overall summary of the epistemic uncertainty in the expected dose EA[DSGt|a, eM)|eA] as a function of time. If desired, the epistemic uncertainty associated with the CCDFs in Fig. 17.3(a) can be summarized with mean and quantile curves in a manner analogous to that shown in Fig. 17.3(d) for expected dose curves (Fig. 17.4). The summary curves in Fig. 17.4 are defined by means and probabilities analogous to those defined by Equations [17.17], [17.18], [17.34] and [17.35] with pA[DSG(104 year|a, eM) > y|eA] replacing EA[f(t|a, eM)|eA]. As indicated in Equations [17.28] to [17.31], stratified sampling also provides a basis for estimating results involving aleatory uncertainty. An example of this approach in the context of radioactive waste disposal is given in Reference 39, and an additional example of this approach in the context of a probabilistic risk assessment for a nuclear power plant is given in Reference 40. Further, many additional examples of the propagation of aleatory and epistemic uncertainty in the context of radioactive waste disposal are given in Appendices J and K of Reference 10.

17.4

Computational design of a performance assessment

To incorporate aleatory and epistemic uncertainty appropriately in a PA for a complex system such as a radioactive waste disposal facility requires a careful computational design. Fundamental to this design is a decomposition of the calculations that allows results of interest to be evaluated and, at

© Woodhead Publishing Limited, 2010

566

Geological repository systems for safe disposal

Table 17.3 Decomposition of expected dose EA[D(τ|a, eM)|eA] into expected incremental doses EA[DC(τ|a, eM)|eA] from individual scenario classes (Section J4.6 of Reference 10) ð Dðtja; eM ÞdA ðajeA ÞdA EA ½Dðtja; eM ÞjeA  ¼ A ) ð ( X DN ðtjaN ; eM Þ þ DC ðtja; eM Þ dA ðajeA ÞdA % A

with



C2MC

Dðtja; eM Þ%DN ðtjaN ; eM Þ þ

P C2MC

DC ðtja; eA Þ

MC ¼ fEW; ED; II; IE; SG; SFg X ð ¼ DN ðtjaN ; eM Þ þ DC ðtja; eM ÞdA ðajeA ÞdA ¼ DN ðtjaN ; eM Þ þ ¼ DN ðtjaN ; eM Þ þ

C2MC

A

C2MC

AC

X ð X

DC ðtja; eM ÞdA ðajeA ÞdA

EA ½DC ðtja; eM ÞjeA Þ

C2MC

where aN corresponds to the single future associated with the nominal scenario class AN in which no disruptions of any kind occur, DN ðtjaN ; eM Þ is the dose to the RMEI that results solely from processes associated with the nominal scenario class, and DC ðtja; eM Þ is the incremental dose to the RMEI that results solely from the effects of the disruptions that result in the future a being an element of the scenario class (i.e. set) AC.

the same time, holds computational costs to an acceptable level. As already indicated, sampling-based methods are likely to play a significant role in the computational implementation of a PA. However, given the size and complexity of a PA for a real system, such methods cannot be implemented without careful planning. Without such planning, appropriate uncertainty and sensitivity results will not be available, appropriate levels of resolution in analysis results will not be present, and the overall computational cost of the analysis could be prohibitive. In general, such designs will depend on the properties of the particular analysis under consideration and also on the resources available to carry out the analysis (i.e. time, money, computer power, . . . ). A core property of such analyses is maintaining a separation of aleatory and epistemic uncertainty. As an example, the computational design used for the 2008 YM PA is briefly described. This design is based on decomposing the calculation of dose to the RMEI and other analysis outcomes of interest on the basis of the scenario classes AN, AEW, AED, AII, AIE, ASG and ASF indicated in Table 17.1. In turn, this permits the decomposition of the calculations in the manner indicated in Table 17.3. The decomposition in Table 17.3 conserves probability but does entail the assumption that there are no synergisms between the effects of the individual scenario classes that significantly affect the value of the expected dose EA[D(t|a, eM)|eA]. Simplifying assumptions of various forms are inevitable in the computational implementation of a large

© Woodhead Publishing Limited, 2010

Treatment of uncertainty in performance assessments

567

Table 17.4 Integration procedures used to obtain expected incremental dose EA[DC(τ|a, eM)|eA] for individual scenario classes in the 2008 YM TSPA (Sections J5 to J8 of Reference 10) Nominal conditions: DN(τ|aN, eM) . Always zero for [0, 26104 year] in YM 2008 TSPA . Combined with seismic ground motion for [0, 106 year] Early WP and DS failures: EA[DWP(t|a, eM)|eA], EA[DDS(t|a, eM)|eA] . Summation of probabilistically weighted results for individual failures Igneous intrusive events: EA[DII(t|a, eM)|eA] . Quadrature procedure Igneous eruptive events: EA[DIE(t|a, eM)|eA] . Combined quadrature/Monte Carlo procedure Seismic ground motion events: EA[DSG(t|a, eM)|eA] . Quadrature procedure for [0, 26104 year] . Monte Carlo procedure for [0, 106 year] Seismic fault displacement events: EA[DSF(t|a, eM)|eA] . Quadrature procedure

analysis; however, such assumptions need to be justified. In this case, the indicated decompositon in Table 17.3 was justified in part on the basis of the probabilities of the individual scenario classes and in part on the basis of the size of the doses to the RMEI associated with the individual scenario classes. The decomposition in Table 17.3 reduces the determination of the expected dose EA[D(t|a, eM)|eA] from the evaluation of an integral of a very complex function D(τ|a, eM) over the entire sample space A for aleatory uncertainty to the integration of less complex functions DC(τ|a, eM) over subsets AC of A. In turn, this allows the design of computational procedures for the evaluation of each expected result EA[DC(t|a, eM)|eA] that are appropriate for the integration of the function DC(τ |a, eM) that underlies the determination of EA[DC(t|a, eM)|eA] (Table 17.4). As indicated in conjunction with Equation [17.40], epistemic uncertainty is propagated in the 2008 YM PA with use of an LHS of size nSE = 300. This results in the calculations indicated in Tables 17.3 and 17.4 being repeated 300 times and produces the estimates EA ½Dðtja; eMi ÞjeAi %DN ðtjaN ; eMi Þ X þ EA ½DC ðtja; eMi ÞjeAi 

½17:43

C2MC

in Fig. 17.5(a) for i = 1, 2, . . . , 300 and 0 ≤ τ ≤ 20 000 year. An important part of the computational design is the use of the same LHS in the evaluation of each of the expected doses EA[DC(τ |a, eMi)|eAi] in Equation [17.43]. Without this consistency, the summation in Equation [17.43] would

© Woodhead Publishing Limited, 2010

568

Geological repository systems for safe disposal

17.5 Expected dose to RMEI: (a) expected dose EA[D(τ |a, eMi)|eAi] for individual LHS elements, and (b) mean and quantile curves for EA[D(τ |a, eM)|eA] (Figure J9.2-1 of Reference 10).

not be a valid approximation to EA[D(τ|a, eMi)|eAi] and the resultant representation of the epistemic uncertainty in EA[D(τ|a, eM)|eA] in Fig. 17.5 (a) would not be possible. The mean and quantile curves for EA[D(τ|a, eM)|eA] in Fig. 17.5(b) are estimated similarly to the mean and quantile curves in Fig. 17.3(d). The mean curve EE{EA[D(τ|a, eM)|eA]} in Fig. 17.5(b) is formally defined in Equation [17.17] and is one of the primary regulatory quantities specified by the NRC for the YM repository. Specifically, EE{EA[D(τ|a, eM)|eA]} is required to be less than 15 mrem/year for the time period [0, 104 year] after repository closure and to be less than 100 mrem/year for the time period [104, 106 year] after repository closure.41 A detailed description of the computational strategy used to separate

© Woodhead Publishing Limited, 2010

Treatment of uncertainty in performance assessments

569

aleatory and epistemic uncertainty in the 2008 YM PA is given in Appendix J of Reference 10. Additional examples of computational strategies used to separate aleatory and epistemic uncertainty in large analyses are given in References 40, 42 and 43.

17.5

Sensitivity analysis

The LHSs indicated in Equations [17.33] and [17.40] result in mappings between epistemically uncertainty analysis inputs and analysis results of interest of the form ½eMi ; fðtja; eMi Þ;

i ¼ 1; 2; . . . ; nSE;

½17:44

and fei ; EA ½fðtja; eMi ÞjeAi g;

i ¼ 1; 2; . . . ; nSE;

½17:45

where (1) f(t|a, eM) is an analysis result at time t (year) conditional on future a and uncertain input values contained in eM, and (2) EA[f(t|a, eM)|eA] is the expected value of f(t|a, eM) over aleatory uncertainty conditional on uncertain analysis inputs contained in e = [eA, eM]. Sampling-based sensitivity analysis involves the exploration of the mappings in Equations [17.44] and [17.45] to determine how the values for individual elements of e = [eA, eM] affect the values of f(t|a, eM) and EA[f(t|a, eM)|eA]. As an example, a sensitivity analysis for results of the form indicated in Equation [17.44] is presented in Fig. 17.6. In this example, f(t|a, eM) = NCSFL(t|aN, eM), where NCSFL(t|aN, eM) is the number of failed commercial spent fuel waste packages at time t under nominal conditions. Specifically, (1) the 300 time-dependent values for NCSFL(t|aN, eM) corresponding to the results in Equation [17.44] are shown in Fig. 17.6(a); (2) partial rank correlation coefficients (PRCCs) between NCSFL(t|aN, eM) and individual elements of eM are shown in Fig. 17.6(b), with PRCCs taking values between 1 and + 1, positive values indicating that variables tend to move up and down together, negative values indicating that variables tend to move in opposite directions, and the absolute values of PRCCs providing a measure of the importance of individual elements of eM; (3) a stepwise rank regression for NCSFL(106 year|aN, eM) is shown in Fig. 17.6(c), with variable importance and effects indicated by the order in which variables are selected in the stepwise process, the incremental changes in R2 values as successive variables are selected in the stepwise process, and the signs and absolute values of the standardized rank regression coefficients (SRRCs) in the regression model; and (4) a scatterplot of the points [WDGCA22i, NCSFL(106 year|aN, eMi)], i = 1, 2, . . . , nSE = 300, is shown in Fig. 17.6(d), with this scatterplot clearly showing the dominant effect of WDGCA22 on

© Woodhead Publishing Limited, 2010

17.6 Uncertainty and sensitivity analysis results for NCSFL(t|aN, eM): (a) NCSFL(t|aN, eMi) for all (i.e. nSE = 300) LHS elements, (b) PRCCs for NCSFL(t|aN, eM), (c) stepwise rank regression for NCSFL(106 year|aN, eM), and (d) scatterplot for [WDGCA22i, NCSFL (106 year|aN, eMi)], i = 1, 2, . . . , nSE = 300 (Figure K2-1 of Reference 10).

570 Geological repository systems for safe disposal

© Woodhead Publishing Limited, 2010

Treatment of uncertainty in performance assessments

571

17.7 Sensitivity analysis results for expectations over aleatory uncertainty: (a) PRCCs for EA[DSG(τ |a, eM)|eA] and (b) PRCCs for EA[D(τ |a, eM)|eA] (Figures K7.7.1-1 and K8.1-1 of Reference 10).

the uncertainty in NCSFL(106 year|aN, eM). Definitions of the elements of eM appearing in Fig. 17.6 are given in Table 17.2. As an additional example, sensitivity analyses for results of the form indicated in Equation [17.45] are presented in Fig. 17.7. In this example, EA[f(t|a, eM)] corresponds to EA[DSG(τ |a, eM)|eA] (i.e. the expected incremental dose due to seismic ground motion) in Fig. 17.7(a) and to EA[D(τ |a, eM)|eA] (i.e. the expected dose due to all futures) in Fig. 17.7(b). In both examples, PRCCs are used to indicate variable importance. The results corresponding to the mapping in Equation [17.45] are shown in Figs 17.3(c) and 17.5(a), respectively. This provides an example where both intermediate results (i.e. EA[DSG(τ |a, eM)|eA]) and final results (i.e. EA[D(τ |a, eM)|eA]) are analyzed. Definitions of the elements of e appearing in Fig. 17.7 are given in Table 17.2. Additional details on sampling-based approaches to sensitivity analysis are

© Woodhead Publishing Limited, 2010

572

Geological repository systems for safe disposal

available in several review articles.11,18,35,44 Further, many additional examples from the 2008 YM PA are presented in Appendix K of Reference 10.

17.6

Concluding discussion

The novelty of simply being able to carry out the computations that underlie a complex analysis is past. Now, penetrating questions about the analysis itself are being asked and must be answered. Rightfully, many of these questions relate to the treatment and representation of uncertainty. Answering questions with respect to the uncertainty present in the results of an analysis begins with a clear conceptual design for the analysis. In most analyses, this design begins with a separation of what is intended to be considered as aleatory uncertainty and what is intended to be considered as epistemic uncertainty. Failure to establish this separation can result in a commingling of aleatory and epistemic uncertainty in a way that makes the interpretation of uncertainty and sensitivity analysis results difficult, if not impossible. For example, treating the occurrence time for a Poisson process as though it is an epistemic uncertainty in a sampling-based uncertainty and sensitivity analysis could completely obscure the effects and importance of epistemically uncertain variables present in models for physical processes. In turn, an appropriate separation of aleatory and epistemic uncertainty leads to an analysis that conceptually consists of three basic mathematical components or entities: (1) a probability space that characterizes aleatory uncertainty, (2) a function that defines analysis results of interest, and (3) a probability space that characterizes epistemic uncertainty. The introduction of these three entities and the integrals that derive from them and define the CDFs and CCDFs that display the effects of aleatory and epistemic uncertainty in analysis results may seem pedantic. However, this introduction provides a high-level view of what is being calculated and a pathway into the more detailed and algorithmic procedures needed to actually calculate results. Unfortunately, many large analyses develop (or, as is sometimes said, accrete) over long periods of time, with the result that it can be very difficult to ascertain exactly what is being calculated. This is particularly true when an analysis involves a mixing of aleatory and epistemic uncertainty and one or more sampling processes from assorted variables. Introduction of a clear conceptual structure and a path from this structure to computational results of interest greatly facilitates the design, implementation, documentation and review of an analysis. Verification and validation are fundamental parts of any analysis on which important decisions are to be based, where (1) verification is the process of determining that a model implementation accurately represents the developers’ conceptual description of the model and the solution to the model and (2) validation is the process of determining the degree to which a

© Woodhead Publishing Limited, 2010

Treatment of uncertainty in performance assessments

573

model is an accurate representation of the real world from the perspective of the intended uses of the model (see Reference 45, p. 3, and References 46 to 51). A clear and conceptually consistent analysis description is an essential component of analysis verification and validation. Further, sensitivity analysis is a powerful tool for checking for analysis errors, and thus is an important component of analysis verification, and model validation is an important contributor to the insights that ultimately lead to the definition of the probability space that characterizes epistemic uncertainty. The distributions that characterize the epistemic uncertainty in individual analysis inputs and thus, in effect, define the probability space for epistemic uncertainty should be neither deliberately pessimistic (i.e. conservative in the sense of leading to overly negative analysis outcomes) nor deliberatively optimistic (i.e. non-conservative in the sense of leading to overly positive analysis outcomes). Rather, the goal in defining these distributions should be to provide an ‘honest’ characterization of the uncertainty in the knowledge base pertaining to these variables. The use of deliberately pessimistic or deliberatively optimistic uncertainty characterizations distorts the results of uncertainty and sensitivity analyses and, as a consequence, can lead to flawed assessments of system behavior. The importance of avoiding deliberately pessimistic analyses has been emphasized by a number of individuals.52–56 If a deliberatively conservative analysis is desired, it should be preceded by an analysis that is neither deliberatively conservative nor deliberatively non-conservative. Without such a precursor analysis, it is difficult meaningfully to assess the potential implications of results obtained in a conservative analysis. It is important to recognize that the probability distributions that characterize epistemic uncertainty are not invariant properties of the quantities under consideration. Rather, these distributions are numerical characterizations of the currently available knowledge with respect to these quantities and how they are used in the particular analysis under consideration. As a result, these distributions will change as the underlying knowledge base changes. Because of this dependence on a current knowledge base, a careful description of the rationale used to specify the distributions used for epistemically uncertain quantities is a very important part of analysis documentation. This presentation has used probability as the mathematical structure for the characterization of aleatory and epistemic uncertainty. Recently, there has been significant interest in alternative mathematical structures for the characterization of epistemic uncertainty such as interval analysis, possibility theory and evidence theory.57–63 The rationale for the use of these alternatives to probability is that they involve less restrictive (i.e. mathematically structured) assumptions than probability and, as a result, provide more realistic characterizations of the available knowledge about a

© Woodhead Publishing Limited, 2010

574

Geological repository systems for safe disposal

quantity. Specifically, the use of probability to characterize epistemic uncertainty has the potential to imply a greater resolution in the available uncertainty information than is actually present. Some of these alternative uncertainty representations may find use in future analyses where there is limited information available for use in the characterization of epistemic uncertainty. However, their implementation has the potential to be very computationally demanding.64 Of these mathematical structures, the authors feel that evidence theory is the structure most likely to find use in future analyses. Stepwise rank regression and partial rank correlation have long been primary techniques used in sampling-based sensitivity analyses (e.g. see References 11, 18 and 44). However, these techniques can fail to provide appropriate insights into system behavior when the relationships between analysis inputs and analysis results are both non-linear and non-monotonic. In such situations, sensitivity analyses based on techniques such as nonparametric regression65,66 and complete variance decomposition67–70 are likely to be productive in future analyses. Several additional techniques that also have the potential to be useful in such situations are listed and described in Reference 18. Quality documentation is an essential part of any meaningful analysis. In turn, a clear conceptual structure is an essential starting point in the development of quality analysis documentation and the explanation of the treatment of uncertainty within the analysis. For a large analysis, it can never be expected that everyone will agree on how the analysis was performed and how uncertainty was treated within the analysis. However, everyone should be able to know exactly what was assumed and done within the analysis.

17.7

Acknowledgments

This work was performed at Sandia National Laboratories (SNL), which is a multiprogram laboratory operated by Sandia Corporation, a Lockheed Martin Company, for the US Department of Energy’s (DOE’s) National Nuclear Security Administration under Contract No. DE-AC0494AL85000. The views expressed in this article are those of the authors and do not necessarily reflect the views or policies of the DOE or SNL. The United States Government retains and the publisher, by accepting this article for publication, acknowledges that the United States Government retains a non-exclusive, paid-up, irrevocable, worldwide license to publish or reproduce the published form of this article, or allow others to do so, for United States Government purposes.

© Woodhead Publishing Limited, 2010

Treatment of uncertainty in performance assessments

17.8 1. 2. 3. 4.

5.

6. 7.

8.

9.

10.

11.

12.

13. 14. 15.

16.

575

References

Parry G W and Winter P W, ‘Characterization and evaluation of uncertainty in probabilistic risk analysis’, Nuclear Safety, 1981, 22(1), 28–42. Apostolakis G, ‘The concept of probability in safety assessments of technological systems’, Science, 1990, 250(4986), 1359–1364. Helton J C, ‘Treatment of uncertainty in performance assessments for complex systems’, Risk Analysis, 1994, 14(4), 483–511. Hoffman F O and Hammonds J S, ‘Propagation of uncertainty in risk assessments: the need to distinguish between uncertainty due to lack of knowledge and uncertainty due to variability’, Risk Analysis, 1994, 14(5), 707– 712. Helton J C and Burmaster D E, Guest Editorial, ‘Treatment of aleatory and epistemic uncertainty in performance assessments for complex systems’, Reliability Engineering and System Safety, 1996, 54(2–3), 91–94. Pate´-Cornell M E, ‘Uncertainties in risk analysis: six levels of treatment’, Reliability Engineering and System Safety, 1996, 54(2–3), 95–111. Helton J C, ‘Uncertainty and sensitivity analysis in the presence of stochastic and subjective uncertainty’, Journal of Statistical Computation and Simulation, 1997, 57(1–4), 3–76. Helton J C, ‘Mathematical and numerical approaches in performance assessment for radioactive waste disposal: dealing with uncertainty’, in Modelling Radioactivity in the Environment, edited by E M Scott, Elsevier Science, New York, 2003, pp. 353–390. Helton J C, Anderson D R, Jow H-N, Marietta M G and Basabilvazo G, ‘Conceptual structure of the 1996 performance assessment for the Waste Isolation Pilot Plant’, Reliability Engineering and System Safety, 2000, 69(1–3), 151–165. SNL (Sandia National Laboratories), ‘Total system performance assessment model/analysis for the license application’, MDL-WIS-PA-000005 Rev 00, AD 01, Las Vegas, Nevada, US Department of Energy Office of Civilian Radioactive Waste Management, 2008. Helton J C, ‘Uncertainty and sensitivity analysis techniques for use in performance assessment for radioactive waste disposal’, Reliability Engineering and System Safety, 1993, 42(2–3), 327–367. Hamby D M, ‘A review of techniques for parameter sensitivity analysis of environmental models’, Environmental Monitoring and Assessment, 1994, 32(2), 135–154. Saltelli A, Chan K and Scott E M (eds), Sensitivity Analysis, John Wiley and Sons, Inc., 2000. Frey H C and Patil S R, ‘Identification and review of sensitivity analysis methods’, Risk Analysis, 2002, 22(3), 553–578. Cacuci D G and Ionescu-Bujor M, ‘A comparative review of sensitivity and uncertainty analysis of large-scale systems – II: Statistical methods’, Nuclear Science and Engineering, 2004, 147(3), 204–217. Ionescu-Bujor M and Cacuci D G, ‘A comparative review of sensitivity and uncertainty analysis of large-scale systems – I: Deterministic methods’, Nuclear Science and Engineering, 2004, 147(3), 189–2003.

© Woodhead Publishing Limited, 2010

576 17. 18.

19.

20.

21.

22.

23. 24. 25.

26.

27.

28.

29. 30.

31.

Geological repository systems for safe disposal Saltelli A, Ratto M, Tarantola S and Campolongo F, ‘Sensitivity analysis for chemical models’, Chemical Reviews, 2005, 105(7), 2811–2828. Helton J C, Johnson J D, Sallaberry C J and Storlie C B, ‘Survey of samplingbased methods for uncertainty and sensitivity analysis’, Reliability Engineering and System Safety, 2006, 91(10–11), 1175–1209. Sevougian S D, Behie A, Bullard B, Chipman V, Gross M B and Stathum W, ‘Yucca Mountain 2008 performance assessment: modeling disruptive events and early failures’, in Proceedings of the 2008 International High-Level Radioactive Waste Management Conference, September 7–11, 2008: American Nuclear Society, 2008, pp. 533–541. Mackinnon R J, Behie A, Chipman V, Chen Y, Lee J, Lee P L, Mattie P D, Mehta S, Mon K, Schreiber J D, Sevougian S D, Stockman C T and Zwahlen E, ‘Yucca Mountain 2008 performance assessment: modeling the engineered barrier system’, in Proceedings of the 2008 International High-Level Radioactive Waste Management Conference, September 7–11, 2008, American Nuclear Society, 2008, pp. 542–548. Mattie P D, Hadgu T, Lester B, Smith A, Wasiolek M and Zwahlen E, ‘Yucca Mountain 2008 performance assessment: modeling the natural system’, in Proceedings of the 2008 International High-Level Radioactive Waste Management Conference, September 7–11, 2008, American Nuclear Society, 2008, pp. 550–558. Mosleh A, Siu N, Smidts C and Liu C, Proceedings of Workshop I in Advanced Topics in Risk and Reliability Analysis, Model Uncertainty: Its Characterization and Quantification, NUREG/CP-0138, US Nuclear Regulatory Commission, Washington, DC, 1994. Hora S C and Iman R L, ‘Expert opinion in risk analysis: the NUREG1150 methodology’, Nuclear Science and Engineering, 1989, 102(4), 323–331. Meyer M A and Booker J M, Eliciting and Analyzing Expert Judgment: A Practical Guide. Academic Press, New York, 1991. Keeney R L and Winterfeldt D V, ‘Eliciting probabilities from experts in complex technical problems’, IEEE Transactions on Engineering Management, 1991, 38(3), 191–201. Thorne M C and Williams M M R, ‘A review of expert judgement techniques with reference to nuclear safety’, Progress in Nuclear Safety, 1992, 27(2–3), 83– 254. Thorne M C, ‘The use of expert opinion in formulating conceptual models of underground disposal systems and the treatment of associated bias’, Reliability Engineering and System Safety, 1993, 42(2–3), 161–180. Budnitz R J, Apostolakis G, Boore D M, Cluff L S, Coppersmith K J, Cornell C A and Morris P A, ‘Use of technical expert panels: applications to probabilistic seismic hazard analysis’, Risk Analysis, 1998, 18(4), 463–469. Ayyub B M, Elicitation of Expert Opinions for Uncertainty and Risks, CRC Press, Boca Raton, Florida, 2001. Cooke R M and Goossens L H J, ‘Expert judgement elicitation for risk assessment of critical infrastructures’, Journal of Risk Research, 2004, 7(6), 643–656. Cooke R, Experts in Uncertainty: Opinion and Subjective Probability in Science, Oxford University Press, Oxford and New York, 1991.

© Woodhead Publishing Limited, 2010

Treatment of uncertainty in performance assessments 32.

33. 34.

35.

36.

37.

38.

39.

40. 41.

42.

43.

44.

45.

46. 47.

577

Garthwaite P H, Kadane J B and O’Hagan A, ‘Statistical methods for eliciting probability distributions’, Journal of the American Statistical Association, 2005, 100(470), 680–700. Kaplan S and Garrick B J, ‘On the quantitative definition of risk’, Risk Analysis, 1981, 1(1), 11–27. McKay M D, Beckman R J and Conover W J, ‘A comparison of three methods for selecting values of input variables in the analysis of output from a computer code’, Technometrics, 1979, 21(2), 239–245. Helton J C and Davis F J, ‘Latin hypercube sampling and the propagation of uncertainty in analyses of complex systems’, Reliability Engineering and System Safety, 2003, 81(1), 23–69. Iman R L and Conover W J, ‘A distribution-free approach to inducing rank correlation among input variables’, Communications in Statistics: Simulation and Computation, 1982, B11(3), 311–334. Iman R L and Davenport J M, ‘Rank correlation plots for use with correlated input variables’, Communications in Statistics: Simulation and Computation, 1982, B11(3), 335–360. Helton J C and Sallaberry C J, ‘Conceptual basis for the definition and calculation of expected dose in performance assessments for the proposed high level radioactive waste repository at Yucca Mountain, Nevada’, Reliability Engineering and System Safety, 2009, 94, 677–698. Helton J C and Iuzzolino H J, ‘Construction of complementary cumulative distribution functions for comparison with the EPA release limits for radioactive waste disposal’, Reliability Engineering and System Safety, 1993, 40(3), 277–293. Helton J C and Breeding R J, ‘Calculation of reactor accident safety goals’, Reliability Engineering and System Safety, 1993, 39(2), 129–158. US NRC (United States Nuclear Regulatory Commission), ‘Disposal of highlevel radioactive wastes in a geologic repository at Yucca Mountain, Nevada’, Code of Federal Regulations 10 Part 63, 2008, http://www.access.gpo.gov/nara/ cfr/waisidx_08/10cfr63_08.html. Helton J C, Johnson J D, Shiver A W and Sprung J L, ‘Uncertainty and sensitivity analysis of early exposure results with the MACCS reactor accident consequence model’, Reliability Engineering and System Safety, 1995, 48(2), 91–127. Helton J C and Marietta M G (guest eds.), ‘Special Issue: The 1996 Performance Assessment for the Waste Isolation Pilot Plant’, Reliability Engineering and System Safety, 2000, 69(1–3), 1–451. Helton J C and Davis F J, ‘Sampling-based methods’, in Sensitivity Analysis, edited by A Saltelli, K Chan and E M Scott, John Wiley & Sons, Inc., New York, 2000, pp. 101–153. American Institute of Aeronautics and Astronautics, AIAA Guide for the Verification and Validation of Computational Fluid Dynamics Simulations, AIAA G-077-1998, American Institute of Aeronautics and Astronautics, Reston, Virginia, 1998. Roache P J, Verification and Validation in Computational Science and Engineering, Hermosa Publishers, Albuquerque, New Mexico, 1998. Oberkampf W L and Trucano T G, ‘Verification and validation in

© Woodhead Publishing Limited, 2010

578

48.

49. 50.

51.

52.

53.

54.

55. 56.

57. 58. 59. 60. 61.

62. 63.

64.

Geological repository systems for safe disposal computational fluid dynamics’, Progress in Aerospace Sciences, 2002, 38(3), 209–272. Oberkampf W L, Trucano T G and Hirsch C, ‘Verification, validation, and predictive capability in computational engineering and physics’, Applied Mechanics Review, 2004, 57(5), 345–384. Roache P J, ‘Building PDE codes to be verifiable and validatable’, Computing in Science and Engineering, 2004, 6(5), 30–38. Babuska I and Oden J T, ‘Verification and validation in computational engineering and science: basic concepts’, Computer Methods in Applied Mechanics and Engineering, 2004, 193(36–38), 4057–4066. Trucano T G, Swiler L P, Igusa T, Oberkampf W L and Pilch M, ‘Calibration, validation, and sensitivity analysis: what’s what’, Reliability Engineering and System Safety, 2006, 91(10–11), 1331–1357. Nichols A L and Zeckhauser R J, ‘The perils of prudence: how conservative risk assessments distort regulation’, Regulatory Toxicology and Pharmacology, 1988, 8, 61–75. Sielken Jr R L, Bretzlaff R S and Stevenson D E, ‘Challenges to default assumptions stimulate comprehensive realism as a new tier in quantitative cancer risk assessment’, Regulatory Toxicology and Pharmacology, 1995, 21, 270–280. Caruso M A, Cheok M C, Cunningham M A, Holahan G M, King T L, Parry G W, Ramey-Smith A M, Rubin M P and Thadani A C, ‘An approach for using risk assessment in risk-informed decisions on plant-specific changes to the licensing basis’, Reliability Engineering and System Safety, 1999, 63, 231– 242. Pate´-Cornell E, ‘Risk and uncertainty analysis in government safety decisions’, Risk Analysis, 2002, 22(3), 633–646. Diaz N J, Realism and conservatism, remarks by Chairman Diaz at the 2003 Nuclear Safety Research Conference, October 20, 2003’, in NRC News, No. S03-023, US Nuclear Regulatory Commission, Washington, DC, 2003. Klir G J and Wierman M J, Uncertainty-Based Information, Physica-Verlag, New York, 1999. Wolkenhauer O, Data Engineering: Fuzzy Mathematics in Systems Theory and Data Analysis, John Wiley & Sons, Inc., New York, 2001. Jaulin L, Kieffer M, Didrit O and Walter E, Applied Interval Analysis, Springer-Verlag, New York, 2001. Ross T J, Fuzzy Logic with Engineering Applications, 2nd edition, John Wiley & Sons, Inc., New York, 2004. Ross T J, Booker J M and Parkinson W J (eds.), Fuzzy Logic and Probability Applications: Bridging the Gap, Society for Industrial and Applied Mathematics, Philadelphia, Pennsylvania, 2002. Klir G J, ‘Generalized information theory: aims, results, and open problems’, Reliability Engineering and System Safety, 2004, 85(1–3), 21–38. Helton J C, Johnson J D and Oberkampf W L, ‘An exploration of alternative approaches to the representation of uncertainty in model predictions’, Reliability Engineering and System Safety, 2004, 85(1–3), 39–71. Helton J C, Johnson J D, Oberkampf W L and Storlie C B, ‘A sampling-based computational strategy for the representation of epistemic uncertainty in

© Woodhead Publishing Limited, 2010

Treatment of uncertainty in performance assessments

65.

66.

67. 68.

69. 70.

579

model predictions with evidence theory’, Computational Methods in Applied Mechanics and Engineering, 2007, 196(37–40), 3980–3998. Storlie C B and Helton J C, ‘Multiple predictor smoothing methods for sensitivity analysis: description of techniques’, Reliability Engineering and System Safety, 2008, 93(1), 28–54. Storlie C B and Helton J C, ‘Multiple predictor smoothing methods for sensitivity analysis: example results’, Reliability Engineering and System Safety, 2008, 93(1), 55–77. Sobol’ I M, ‘Sensitivity estimates for nonlinear mathematical models’, Mathematical Modeling and Computational Experiment, 1993, 1(4), 407–414. Saltelli A, Tarantola S and Chan K P-S, ‘A quantitative model-independent method for global sensitivity analysis of model output’, Technometrics, 1999, 41(1), 39–56. Li G, Rosenthal C and Rabitz H, ‘High-dimensional model representations’, The Journal of Physical Chemistry, 2001, 105(33), 7765–7777. Rabitz H and Alis O F, ‘General foundations of high-dimensional model representations’, Journal of Mathematical Chemistry, 1999, 25(2–3), 197–233.

© Woodhead Publishing Limited, 2010

18 Assessment of expert judgments for safety analyses and performance assessment of geological repository systems K . E . J E N N I , Insight Decisions, USA; A . v a n L U I K US Department of Energy, USA

Abstract: Expert judgments are used throughout safety analyses and performance assessments for geological disposal facilities. The term ‘expert elicitation’ in this area has come to be associated with multi-expert, multiyear studies focused on assessing expert judgments about highly complex, highly uncertain factors such as seismic and volcanic hazards. Less attention has been paid to the many other ways in which expert judgment is used and on the availability of structured approaches to improve both the quantification and the documentation of those expert judgments. Depending on the application, there are less intensive approaches to bringing discipline to the process of assessing, using and documenting expert judgment. Key words: expert elicitation, multiple experts, aggregation of expert judgments, quantifying uncertainty.

18.1

Introduction

Safety analyses and performance assessments for geological disposal systems generally require modeling the projected performance of both engineered and natural components of a planned repository for many thousands of years into the future. Developing credible models of how yetto-be built systems will perform over many generations requires that modelers estimate numerous factors for which there is very limited (or, at early stages, even a complete lack of) relevant data on which to base those estimates. It is necessary to rely on informed, defensible and credible expert judgments to develop the required models and/or model inputs. The term expert judgment can cover a wide variety of judgments. Expert judgment can 580 © Woodhead Publishing Limited, 2010

Assessment of expert judgments

581

be qualitative or quantitative, formally assessed or informal; it includes judgments about anything from model selection, to data applicability, to individual model parameters. The key factor that differentiates ‘expert judgment’ from a ‘guess’ is that it is provided by one or more technical experts about information or data within their area of technical expertise, and it is based on the training and knowledge of those experts (Meyer and Booker, 1991; Kotra et al., 1996). Chapter 15 of this volume describes the four steps of a safety analysis for a geological disposal system as follows: (1) characterize the proposed system, (2) identify representative scenarios for analysis, (3) develop computational models for relevant processes and account for uncertainty in models and parameters, and (4) construct system-level analysis and estimate overall performance metrics (see Fig. 15.1). This becomes an iterative approach, since knowledge concerning the site as well as the size and nature of the proposed system increases and matures as site characterization and design efforts proceed. The role of expert judgment in step 3, and in particular its role in ‘accounting for’ uncertainty in models and model inputs, is widely recognized. Some limit the use or the definition of the expert judgment further, arguing that using ‘historical data’ to develop estimates of the uncertain quantities is always preferable, when it is possible, to using expert judgment to develop those estimates (Hora, 2007). Such arguments fail to recognize that expert judgment is required even in the selection of the appropriate ‘historical data’ to address a particular question. As recognized by the US Nuclear Regulatory Commission in NUREG1563: Branch Technical Position on the Use of Expert Elicitation in the HighLevel Radioactive Waste Program (Kotra et al., 1996), the use of expert judgment in safety assessments is ubiquitous. While much expert judgment is informal, and for some of those informal uses no standard ‘assessment’ approaches exist, we believe it is important to recognize the very large role expert judgments play in safety analyses and performance assessments beyond where they are typically discussed. Expert judgment is involved, in various forms, in virtually every step of the performance assessment and safety analysis processes; approaches and tools for formalizing the assessment of those judgments are described in this chapter.

18.1.1 Expert judgments in safety analyses and performance assessments for geological repositories Expert judgment plays a role in all four safety analysis steps mentioned above, and the importance of expert judgments in steps 1, 2 and 4 is perhaps even greater for being less recognized than in step 3. Because data collection starts during the site characterization phase of a

© Woodhead Publishing Limited, 2010

582

Geological repository systems for safe disposal

safety analysis, expert judgments come into play from the beginning: deciding what data to collect or monitor, what experiments to conduct to generate relevant information, and so forth. All these decisions involve scientists using their expert judgment about what will be most important to understand about the site and the repository system. The fact that expert judgment is required to identify and screen the ‘features, events and processes’ (FEPS) that can affect the performance of a geological repository is fairly well recognized. As described in Chapter 15, considerable effort is made to ensure that the list of potential scenarios is comprehensive and well-defined, and different approaches with different levels of formality have been used in scenario development (NEA, 1999). While care is taken to ensure a comprehensive initial list of scenarios or FEPS, expert judgment comes into play again in ‘screening’ those FEPS. The criteria for screening may be defined by regulation or by the implementing organization, but whether a scenario meets those criteria involves expert judgments. At the stage where screening decisions are made, there is often not yet clear and compelling information allowing each FEP to be characterized in terms of either its probability of occurrence or its impact on performance (typical screening criteria): screening decisions are often made on informal expert judgment with only preliminary information, and, therefore, such initial decisions may be suitable for subsequent review. In the third and fourth safety-evaluation stages, both the development and implementation of computational models and systems-level analysis require use of expert judgment. In 2001, DOE conducted a thorough review of the various sources of uncertainty in the total system performance assessment for the site recommendation decision for the Yucca Mountain repository (BSC, 2001). That report discusses three types of uncertainty, all related to the development and implementation of models and systems-level analyses, each of which requires some form of expert judgment. Conceptual model uncertainty is defined as uncertainty about the fundamental processes being evaluated, which leads to alternative representations (conceptual models) of those processes, all of which may be consistent with available data or equally credible. Where conceptual model uncertainty exists, subject matter experts and/or system modelers typically select one or more conceptual models for implementation and to carry forward in the systems-level analysis. In the 2001 study, DOE found that in most cases where conceptual model uncertainty was recognized, only one conceptual model was carried forward, usually the one believed to be the more conservative among competing models. We believe this is typical of performance assessments for highly complex systems like geological repositories and we note that determining what is truly ‘conservative’ can be very difficult in complex analyses. Such a judgment should be made in

© Woodhead Publishing Limited, 2010

Assessment of expert judgments

583

terms of fundamental performance assessment results, rather than at the level of the individual conceptual models. Representational model uncertainty refers to uncertainties introduced by the translation of a conceptual model into a mathematical representation of that model and associated computer codes. Expert judgments involved in this process are typically not recognized: in this case the modeler is the expert and makes the simplifications believed necessary to represent a subject matter expert’s conceptual model. Collaboration between the subject matter expert and the modeler is a requisite for assuring model credibility. Finally, model parameter uncertainty is the most commonly recognized type of uncertainty and the one for which expert judgment is most recognizably required. Model parameters are often treated probabilistically, with estimates of the range and distribution of values provided by subject matter experts. See, for example, Table 17.2 of this volume listing several individual parameters from DOE’s TSPA model with their associated uncertainty. There is yet a fourth type of uncertainty, sometimes disparagingly labeled ‘the unknown unknowns’; this is uncertainty about the occurrence of future events that have not been conceptualized. It is a type of uncertainty that is often viewed as entirely ‘subjective’ and is typically thought to increase with longer and longer time horizons. It is this type of uncertainty that caused the Yucca Mountain standard, per the recommendation of the National Academy of Sciences (NAS, 1995), to cut off calculations at the time when the geologic stability of the location became uncertain (at about a million years). The difficulty in predicting future technology and human action many generations in the future led the NAS to propose a similar approach for addressing the possibility of an inadvertent intrusion into a repository at some point in the future. In the US, the NAS (1995) suggested this to be a scientifically intractable problem and recommended that the regulators require a stylized calculation of the robustness of a repository to an intrusive event.

18.1.2 Structuring and assessing expert judgments Formal processes for assessing and quantifying expert judgment arose from the field of decision analysis (DA) in the late 1960s and early 1970s. Over the ensuing 40 years, scientific understanding about how people understand (and systematically misunderstand) uncertainties and probabilities evolved, and approaches used to assess expert judgments evolved to accommodate that understanding. In the area of safety assessments, focus on the importance of understanding extreme, and extremely rare, events (such as very large earthquakes and potential volcanic events) on facilities and structures led to increased use of multiple expert opinions or judgments and

© Woodhead Publishing Limited, 2010

584

Geological repository systems for safe disposal

formalized ‘expert elicitation’ to support a robust estimate of the likelihood of such events, and in some cases their range of potential consequences. Again the techniques used to formally assess such judgments evolved to meet this need. The sections below provide a brief discussion of the evolution of techniques for assessing expert judgments that were motivated, at least in part, by those advances in both understanding and decisionmaker’s needs. In Sections 18.2 through 18.5, we provide an overview of approaches for structuring and assessing expert judgments, and how they have evolved over time, with a focus on accurately representing uncertainty about quantities or events of interest. Most of these approaches are illustrated as they can or have been used to assess judgments about models and model parameters, which is where formal assessment approaches are most often used. In Section 18.6 we return to the issue of where and how expert judgments are used throughout the process of safety analyses and performance assessment, and discuss how some of the approaches described next might be used most effectively to add rigor to less formal assessments while minimizing the increased analysis burden such formalized approaches involve.

18.2

Quantifying uncertainties for decision analyses

Decision analysis (DA) is a systematic and analytical approach for improving decision making that began to come of age as a unique and applied field of study and research in the late 1960s (Howard, 1966; Raiffa, 1968). Early work in the field focused on the critical issue of how to make rational decisions in the face of significant uncertainties and technical complexity. Rational decisions demand logical treatment of uncertainty, and a fundamental tenet of DA is the use of probability to quantify uncertainty. To a decision analyst, probability is subjective: it is best defined as a measure of an individual’s state of knowledge about an uncertain event or quantity (see, for example, Chapter 5 of Raiffa, 1968). One of the first problems confronting early decision analysts was how to ‘translate’ expert judgments into probabilities that could be used in a decision tree. Computational limits at the time led to most practical analyses being formulated as decision trees with a limited number of decisions and uncertainties represented, and with each uncertainty represented by a relative small number of discrete outcomes.

18.2.1 Probability encoding This process of using expert judgment to generate the probabilistic inputs for a decision analysis became known as ‘probability encoding’, defined by Spetzler and Stael von Holstein (1975) as ‘the process of extracting and

© Woodhead Publishing Limited, 2010

Assessment of expert judgments

585

quantifying individual judgment about uncertain quantities’ and characterized as ‘one of the major functions required in the performance of decision analysis’. Several approaches for assessing uncertainties were, and continue to be, used. In a direct assessment, the relevant expert is simply asked, ‘directly’, what he/she believes the probability of the uncertain event to be or, alternatively, what event outcome corresponds to a specific probability: What is the probability that the outcome will be less than x? or What value do you expect to see less than 50% of the time? For example, in the performance assessment for the proposed Yucca Mountain repository conducted by the Electric Power Research Institute (EPRI) (Kessler and Vlasity, 2003; EPRI, 2005), one factor included in the model is uncertainty about whether ‘flow focusing’ will occur. In the logic tree illustrating key uncertainties, this is captured simply as the probability of flow focusing: ‘no’ flow focusing is defined as meaning that the percolation flux of water into the repository drifts is everywhere the same, whereas ‘flow focusing’ is defined to mean that the percolation flux at some locations is zero, and at other locations it is significantly higher than the average value. Specifically, in the EPRI model flow focusing is modeled with a factor of four, meaning that 25% of the repository receives four times the average percolation flux and the remaining 75% has zero percolation flux (Kessler and Vlasity, 2003). In a direct assessment of this uncertainty, the expert might simply be asked: What is the probability that flow focusing (with a factor of 4) will occur? Direct assessment may work well if the expert providing the judgment is experienced with probability theory and is highly knowledgeable about and has a well-formulated opinion about the quantity being assessed. It works less well if the expert is not familiar with or comfortable with probability statements, and when he or she does not yet have a well-formulated opinion about the quantity. Direct assessment also performs poorly when the probability of the event is extremely low and when such estimates are required they are typically developed through modeling rather than direct assessment (although such a model and its inputs may themselves be developed through expert assessment). To assist experts in thinking probabilistically, decision analysts may use indirect assessment questions, typically using betting questions or ‘reference lotteries’ to facilitate the probability estimate and inform the expert’s intuition about the meaning of different expressions of probability. In a reference lottery, the analyst describes a common chance event (e.g. the probability a fair coin toss lands ‘heads’) and asks the expert to compare the likelihood of the variable being assessed to the likelihood of this reference lottery. An expert’s judgment (probability) about a specific event can be inferred by the tradeoffs he or she is willing to make between these lotteries

© Woodhead Publishing Limited, 2010

586

Geological repository systems for safe disposal

involving the uncertainty of interest and the ‘reference lottery’ involving uncertainties with known probabilistic outcomes. Using the example described above, in an indirect assessment an expert might be asked to choose between the following ‘lotteries’: Q1. Would you prefer to bet that a single roll of a six-sided die will result in an even number or that the flow focusing factor will be equal to 4? (Given the assessment result indicated in Kessler and Vlasity (2003), a 13.8% chance of a flow focusing factor of 4, the expert should be able to easily answer the question: he or she would prefer to bet on the dice). Q2. Would you prefer to bet that the dice will result in a ‘1’ or that the flow focusing factor will equal 4? (This question should be more difficult for the expert to answer, as he or she is comparing a subjective estimate of about 14% with a reference lottery with a probability of about 17%). Through iterations with various reference lotteries the expert and the assessor arrive at a probability that reflects the expert’s subjective judgment about the quantity or event of interest. Making estimates of uncertain quantities where the range of outcomes is continuous adds a complication to the assessment task, but the same basic approaches (direct and indirect assessment) are typically used. In direct assessment of continuous distributions, the expert may be asked to describe possible outcomes with a probability distribution (e.g. a normal distribution with a mean of x and a standard deviation of y): to assist in this a variety of distributional forms may be shown to the expert. Again, this assessment technique may be applicable when working with experts with detailed knowledge of probability theory, but in our experience even those comfortable with probability may not fully understand the implications of their selected distributions. A careful assessor will explore those implications with the expert rather than just ‘writing down’ his or her first statement of the probability distribution: this iterative assessment process is described more completely in the next section on elicitation protocols. More typically, a direct assessment of continuous distribution will be conducted by direct assessment of specific quantiles of the distribution. For example, if the flow focusing factor described above were to be assessed as a continuous variable, the expert might be asked to estimate flow focusing factors corresponding to various probabilities: e.g. what flow focusing factor is sufficiently high that you think there is only a 10% chance it would be exceeded (to assess the 90th percentile). In assessing quantiles, direct and indirect assessment approaches can be used.

© Woodhead Publishing Limited, 2010

Assessment of expert judgments

18.3

587

Biases in judgments and the development of formal probability elicitation protocols

As DA evolved, so did the interest of cognitive psychologists in studying human judgment and decision making. While decision analysts styled themselves as studying ‘normative’ decision making – the study of how to improve decision making through application of subjective expected utility theory – the psychologists focused on ‘descriptive’ decision making – the study of how individuals (and groups) actually make decisions. Not surprisingly, there are some gaps between what the psychologists found about how people do decide and what the normative experts believed to be how people should decide to maximize their long-term benefit (Bell et al., 1988). One of the goals of practicing decision analysts has always been to narrow this gap and to provide tools for decision-makers to improve their decisions.

18.3.1 Judgmental biases The findings of psychologists about how people make judgments about uncertainties related directly to decision analysis practice. In brief, extensive studies showed that people are not good ‘intuitive statisticians’, expressing uncertainty in terms of probability being very difficult for most people, and there are many ‘heuristics and biases’ that are prevalent in unaided judgment, meaning that people err in predictable ways when asked to estimate probabilities. The list of common heuristics and biases is described extensively elsewhere (Kahneman et al., 1982; Gilovich et al., 2002). A brief summary of the more common biases is provided below. Availability For various reasons, some information is more easily recalled than other information; it is more available to the memory. Information that is more available has an outsized influence on estimates of likelihood: if an incident is easily recalled, people tend to overestimate the likelihood of similar incidents. This heuristic is likely to work quite well when estimating the frequency of events for which the expert has a significant amount of personal experience, where the frequency with which the expert encounters similar events is about the same as the frequency with which such events occur. However, it leads to consistent and predictable errors in other situations: overestimation of the likelihood of events that are easily recalled and underestimation of events that are less easily recalled. A classic example of the availability bias was described in Fischhoff and MacGregor (1982), wherein subjects were shown consistently to overestimate the frequency of

© Woodhead Publishing Limited, 2010

588

Geological repository systems for safe disposal

some causes of death over others. Causes of death that are more ‘sensational’ and more publicized (such as death by firearms or by natural disasters) were overestimated, while causes of death that are more prevalent but receive much less attention (such as death by stroke or heart disease) were underestimated. Anchoring and adjustment A bias often related to the availability bias, anchoring and adjustment (sometimes simply called ‘anchoring’) refers to a tendency to focus on an initial value (the ‘anchor’) and to adjust the estimate from that first value: typically the adjustment is insufficient (Tversky and Kahneman, 1974). The source of the initial value may come from anywhere: it may be an estimate or event that is particularly available, as discussed above, it may be based on the expert’s personal experiences or it may even be inadvertently suggested by the problem formulation or the response scale (Hora et al., 1992). Regardless of the source, different initial values often lead to different ‘final’ estimates, which are relatively closer to the initial values. Lichtenstein et al. (1978) describe a study wherein subjects were asked to estimate the frequency of death by a variety of causes in the United States and were provided with a starting point of the number of people who die each year from traffic accidents. One group was told that about 1000 people die from traffic accidents each year and a second group was told that about 50 000 people die from traffic accidents each year. Estimates of the frequency of death (from all causes) were systematically lower for subjects who received the low anchor value than for those who received the high anchor value. Overconfidence Overconfidence or apparent overconfidence refers to a tendency to give probability estimates that are too close to zero or one, or to give uncertainty distributions that are too narrow, indicating more certainty in the estimates than the expert actually has (Capen, 1976; Lichtenstein et al., 1982). Classic illustrations of the overconfidence bias are created by asking individuals to provide their best estimates, with ranges representing their uncertainty about those estimates, of the answers to general knowledge questions (e.g. the height of Hoover Dam in feet or the population of the state of Ohio as reported in the 2000 US Census). Asked a number of such questions, people will be ‘surprised’ by actual answers outside their confidence bounds much more often than they should be. For example, in Capen’s study, he asked subjects to provide estimates of their 98, 90, 80, 50 and 30% confidence intervals. Regardless of the confidence interval requested, 65 to 70% of the time the actual answers fell outside the assessed confidence intervals.

© Woodhead Publishing Limited, 2010

Assessment of expert judgments

589

Overconfidence tends to increase as the task difficulty increases, and individual differences in overconfidence are difficult to predict (Lichtenstein et al., 1982). Two approaches have been taken to address this issue of overconfidence in judgments. One is to train experts in the issue and actively work to overcome the bias in the expert assessment process by focusing attention on the extreme values first and by encouraging the expert to think through scenarios that might lead to values outside their initial confidence intervals. The second approach (which may be combined with the first) is to use a measure of expert calibration, defined by how accurately their estimates are for a set of ‘seed’ questions where an actual answer can be known, as either an adjustment to or a weight applied to that expert’s assessment. These two approaches will reappear later in this chapter, in Section 18.5, where aggregation of expert judgments is discussed. Representativeness This refers to a tendency to judge the likelihood of an event or outcome based on how similar it is to the characteristics expected from the process or population it represents. People expect the small sample to represent the characteristics of the larger sample, sometimes referred to as a ‘belief in the law of small numbers’ (Tversky and Kahneman, 1971). This bias leads to overreliance on case-specific information considered to be representative of the population or process being assessed at the expense of adequate consideration of more general information that may be equally relevant to the larger population of interest.

18.3.2 Formal elicitation protocols Recognition of the difficulties individuals have in making a clear and honest assessment of their own state of knowledge about any particular uncertain event led decision and risk analysts to develop formal approaches to eliciting expert judgment that includes detailed procedures for recognizing and mitigating against those biases. Analysts at Stanford University and the Stanford Research Institute (SRI) developed the ‘SRI protocol’ for probability elicitation in the early 1970s. This protocol is typically described as consisting of five or six steps, with iteration as necessary (Spetzler and Stael von Holstein, 1975; Merkhofer, 1987; Morgan and Henrion, 1990). These steps underlie almost all formal elicitation processes in some form: 1.

Motivate the assessment task. In this step the analyst (the interviewer, the elicitor) introduces the assessment task and describes the process that will be used to assess the quantities of interest. The discussion should describe the importance of the assessments required, how they

© Woodhead Publishing Limited, 2010

590

2.

3.

Geological repository systems for safe disposal will be used in any further analysis and may also include some discussion of why this expert, in particular, is being asked to provide input to the analysis. Depending on the expert’s level of comfort and experience with probability and probabilistic estimates, the analyst may choose to have a general discussion about the use of probability to quantify uncertainty and the concept of subjective probability as a measure of an individual’s state of knowledge. During this step it is important to explore potential motivational biases that might affect the expert’s judgments. In safety and performance assessments, in particular, experts may be used to providing (or being asked to provide) ‘conservative’ estimates, which might lead them to provide estimates they believe to be conservative rather than providing the most accurate statement of their beliefs about the quantity being assessed (which is the more typical goal of an expert assessment). Structure the assessment task and the variables to be assessed. In this step, the analyst and the expert discuss the uncertain quantity about which information is desired and the analyst assists the expert in identifying one or more clearly defined variables that can be assessed with relative ease. It is important to decompose the variable of interest such that the expert does not have to do any detailed modeling in his or her head before making the necessary estimates: the structuring process should identify any underlying events and uncertainties that affect the quantity of interest so that they can be assessed directly. Clarifying these relationships improves the quality of the overall assessment and may help the expert in identifying any additional information he or she might wish to have before beginning the ‘encoding’ task. Each variable to be assessed should pass the ‘clairvoyance test’: it should be specified in sufficient detail that a clairvoyant could provide a correct value for the quantity in question. Using the example of the flow focusing factor described above, note that the description of what a flow focusing factor of 4 means is fairly detailed: a focusing factor of 4 means that four times the ‘average’ percolation flux will be seen over 25% of the repository footprint and a percolation flux of zero will be seen over the remainder of the footprint. This level of detail and clarity is necessary to ensure a meaningful assessment. During the structuring process, the variables to be assessed should be described in whatever terms are most meaningful to the expert: e.g. this might mean framing the assessment questions in terms of ‘odds’ rather than ‘probability’ if that is how the expert thinks about uncertainties; it means using units that are familiar to the expert rather than to the analyst. Establish the conditions for the assessment. There are two goals to the conditioning step. First is to help the expert to think carefully and consciously about the assessment task and what information he or she

© Woodhead Publishing Limited, 2010

Assessment of expert judgments

4.

5.

591

brings to bear on the questions and, in some cases, what additional information he or she would like to obtain to inform his or her judgments. The second goal is to make the expert aware of the various cognitive biases described above, to identify those that might be in play and to discuss how they may inadvertently affect the expert’s judgments and what can be done to counteract those biases. For example, discussing what the most important factors affecting the expert’s judgment are will identify likely anchors, and the analyst can help the expert expand his or her thinking by asking him or her to create scenarios wherein that information would be less reliable, in the process identifying other information that could be used to inform his or her judgments. Ideally the result of the conditioning step will be a list of all the assumptions and all the information on which the expert’s judgment are ‘conditioned’. Some restructuring may occur during the conditioning step. Encode the expert’s probability estimates. In this step the analyst and the expert work together to express the expert’s judgment about the quantities of interest as accurately as possible in probabilistic terms. Any of the specific encoding techniques described previously may be used here: direct and indirect assessments, use of reference lotteries, and so forth. It is usually recommended that an analyst start by assessing extreme values for the quantity of interest (e.g. values that are at the very high and very low ends of what the expert thinks might occur or be realistic), to help counteract anchoring and overconfidence biases. Similarly, when moving from identifying the extreme values to assessing the expert’s estimate of the probability or frequency of a particular outcome, it is recommended that the analyst start by asking reference questions that are easy to answer – that they pick a reference probability, for example, that is much larger than what the analyst thinks the expert’s subjective probability will be, followed by a probability that is much lower. This helps the expert to become comfortable with the assessment questions and makes it easier for him or her to identify their ‘indifference’ probability. During the encoding task itself, it may be necessary to revisit parts to the structuring and conditioning: the need to make quantitative statements about the likelihood of uncertain quantities tends to sharpen one’s thinking. Verify the assessment. The final step in the expert interview is the process of validating the assessment: of ensuring that the expert both understands and stands by the judgments as encoded. Typically the analyst shows the expert a graphical representation of the assessment results, e.g. a cumulative distribution and/or a probability density function, and explores the major implications with the expert. This should include revisiting and confirming the absolute minimum or

© Woodhead Publishing Limited, 2010

592

6.

Geological repository systems for safe disposal maximum values specified by the expert (probability bounds) and/or some of the more extreme values, such as the 1st and 99th percentiles of the distribution, as well as the central tendency of the distribution and the ‘most likely’ values represented by any modes within the probability distribution. When the expert is comfortable with the representation, it may be useful for the analyst to pose some hypothetical bets based on the distribution to test the expert’s understanding of and confidence in his or her estimates. For example, the expert could be asked if he or she would prefer to bet that the value is greater than [the 90th percentile] or less than [the 10th percentile]. If the encoded distribution accurately reflects the expert’s judgment, he or she should be indifferent between the two bets. Obviously, if inconsistencies are found during the verification step, portions of the prior steps may need to be revisited. As noted above, when working with experts who are familiar with probability and are comfortable providing direct estimates of probability distributions, it is particularly important to verify those estimates through some of these testing-type questions. Determine how to represent the encoded distribution. In the original descriptions of this elicitation protocol, the last step is described as ‘discretize the distribution’, referring to the need to represent the encoded distribution with a discrete number of branches in a decision tree. As computational resources are now cheap and simulation-based models are common, the need to represent any probability distribution by a discrete number of branches has diminished, replaced for simulation models by the need to represent the encoded distribution with a closed functional form, typically by a ‘named’ probability distribution. Assuming sufficient points were encoded to define the shape of the cumulative distribution, any number of commercially available software tools can be used to fit a probability distribution to those points. The encoded distribution is then represented in models by a probability distribution with parameters that were derived to match the assessed cumulative distribution. If desired, the cumulative distribution (or the fitted distribution) can be ‘discretized’ using any of several fairly standard approaches. In discretizing, the range of possible values is divided into intervals, a point is chosen to represent that interval and a probability equal to the probability that the actual value falls in the corresponding interval is assigned to the selected point value (Keefer and Bodily, 1983; Miller and Rice, 1983; Smith, 1993). This allows a continuous distribution to be represented by a few discrete points, appropriately weighted.

© Woodhead Publishing Limited, 2010

Assessment of expert judgments

18.4

593

Assessments with multiple experts

When assessments of the same quantity are conducted with different experts, the results are not necessarily the same. Philosophically, this creates no problem for decision analysts, as subjective probability is often defined as a measure of a particular individual’s state of knowledge. Practically, however, it raises difficult questions. When experts disagree, who is ‘right’ and what input should the analyst use in his models? What inputs should the decision-maker rely upon in making decisions? Should expert judgments be combined in some way and, if so, how should they be combined?

18.4.1 When experts’ assessments differ Experts’ assessments may differ for a number of reasons, and it is rarely clear at the outset why they differ. Two basic philosophies and approaches have emerged regarding how to deal with the issue of differing assessments. One interpretation is that some experts’ assessments are ‘better’ than others, in the context of the elicitation variables or the elicitation process. Just as some weather forecasters are more accurate than others, some expert’s assessments will be more accurate than others. ‘Better’ experts may be more knowledgeable about the variables of interest, less affected by motivational bias, less subject to or better able to counteract the influences of cognitive biases, or just more skilled at expressing uncertainty in terms of probabilities. This philosophy, combined with concerns about the effect of group biases and the information sharing between experts, is embodied in the ‘Classical Approach’ to expert elicitation described by Roger Cooke (1991). In the classical approach, experts are treated as independent samples from a population of experts being used to develop an estimate of variables of interest (the name itself was chosen to highlight the parallels with hypothesis testing in classical statistics). There is little if any interaction between the experts, to maintain independence, and each expert is evaluated in terms of the ‘accuracy’ of their estimates. The final assessment is a weighted combination of the individual expert’s judgments. This differential weighting of experts is described further in Section 18.5. A second approach, most thoroughly developed by the Senior Seismic Hazard Analysis Committee in the late 1990s and published as US Nuclear Regulatory Commission’s NUREG/CR-6372 (SSHAC, 1997), commonly called the SSHAC Approach, is based on a philosophy that disagreements that result from differences in problem formulation and differences in information availability can be minimized through facilitated expert interactions. Any differences that remain are a reflection of genuine uncertainty in the range of scientific knowledge: different experts bring different knowledge bases and different perspectives to the problem, and

© Woodhead Publishing Limited, 2010

594

Geological repository systems for safe disposal

even equally informed, completely unbiased experts may disagree. The SSHAC process combines extensive interactions among experts to facilitate information sharing, with individual assessments designed to minimize the effects of group pressure on the individual experts. Final assessment results are created by weighting all experts equally, under the philosophy that the resulting distribution provides a more comprehensive estimate of the range of scientific knowledge. Detailed assessment guidelines have been developed that are consistent with each of these philosophies. Similarities and differences are described below.

18.4.2 Common elements of assessments with multiple experts Regardless of philosophy, the processes for assessing expert judgment from multiple experts are similar. The steps for conducting such assessment have been described in detail in a variety of sources: we summarize three such descriptions as a way of highlighting the similarities in process as well as some of the key philosophical differences that can be seen in practice. Table 18.1 lists the recommended steps in conducting an ‘expert elicitation’ study as described in: (1) the Branch Technical Position on the Use of Expert Elicitation in the High-Level Radioactive Waste Program (Kotra et al., 1996), hereafter the BTP; (2) the SSHAC Guidelines and (3) the Procedures Guide for Structured Expert Judgment (Cooke and Goossens, 1999), hereafter the Procedures Guide, developed through a long-term joint study by the US Nuclear Regulatory Commission and the Commission for the European Communities. The steps and recommendations of this latter report are derived from the Classical Approach of Cooke (1991). The similarities in process are evident from the table. All the approaches include some version of the following steps: . . . . . . .

Define the objectives of the study and verify appropriateness of/need for expert elicitation. Select the expert panel. Structure the assessment through careful problem description and decomposition; clarify the assessment issues. Conduct pre-elicitation training to familiarize the experts with the assessment tasks and probability encoding concepts. Conduct the elicitation and provide post-elicitation feedback. Aggregate the judgments. Document the assessment and assessment results.

© Woodhead Publishing Limited, 2010

Assessment of expert judgments

595

Table 18.1 Three perspectives on the process of assessing judgments from multiple experts NRC Branch Technical Position (NUREG-1563)

SSHAC Guidelines

European Commission on Nuclear Science and Technology: Procedures Guide for Structured Expert Judgment

1 Definition of objectives 2 Selection of experts 3 Refinement of issues and problem decomposition 4 Assembly and dissemination of basic information 5 Pre-elicitation training 6 Elicitation of judgments 7 Post-elicitation feedback 8 Aggregation of judgments 9 Documentation

1 Identify technical issues 2 Identify and select expert panel 3 Discuss and refine technical issues 4 Train experts 5 Group interactions and individual elicitations 6 Analysis, aggregation and resolution of disagreements 7 Documentation and communication

Preparation for elicitation 1 Define case structure 2 Identify target variables 3 Identify query variables 4 Identify performance variables 5 Identify experts 6 Select experts 7 Define elicitation document 8 Conduct dry run 9 Train experts Elicitation 10 Expert elicitation session Post-elicitation 11 Combine expert assessments 12 Conduct discrepancy and robustness analyses 13 Provide feedback 14 Conduct postprocessing analyses 15 Documentation

18.4.3 Key differences in approaches to assessments with multiple experts As mentioned, the formal expert assessment process embodied in the SSHAC Approach and the Classical Approach differ in their underlying philosophies, which can be seen in how the steps of the process are defined as shown in Table 18.1. The most obvious embodiment of that difference is typically seen in the aggregation approach: SSHAC based approaches generally combine expert assessments using equal weights, and the Classical Approach typically does not (discussed further in Section 18.5 below). However, the difference in underlying philosophies is also seen in several other ways, including the overall assessment process and interactions, and the amount and type of post-elicitation feedback provided to the experts.

© Woodhead Publishing Limited, 2010

596

Geological repository systems for safe disposal

Process and expert interactions In the Classical Approach, as embodied in the Procedures Guide, structuring the assessments and developing a set of assessment questions is the responsibility of the assessment or analysis team, not of the experts. While the analysts may be assisted by subject matter experts in this process, every attempt is made not to use the expert panel members for this process. As can be seen by the level of detail in the ‘preparation’ steps in Table 18.1, a great deal of emphasis is placed on careful and detailed specification of a set of elicitation questions prior to any extensive interactions between the assessment team and the experts. In the more detailed description of what should be included in the elicitation documents, the Procedures Guide includes a description of ‘basic information’ relevant to the assessment, consistent with the statement of step 4 of the BTP, but it is not clear what that ‘basic information’ includes. In the SSHAC Approach, however, responsibility for structuring the assessments and developing assessment questions is shared between the assessment or analysis team and the experts, both as a group and individually. The expert panels typically meet several times prior to any actual elicitations: to structure the assessment question(s), to identify information and models that are available or that they wish to have, and to discuss among the panel members any issues relevant to the assessment that they wish to discuss (Coppersmith et al., 2009a). Although not called out clearly as a separate step in the SSHAC process, as part of this interactive process the assessment team often takes primary responsibility for collecting the relevant information, data and models and ensuring they are available to all expert panel members. A key goal of the expert interactions is to develop a panel of experts who are equally knowledgeable and equally well-informed prior to beginning individual assessments. This level of interaction is required by the SSHAC process, as it is a critical part of ensuring that the group of experts can represent the informed technical community (Hanks et al., 2009). This collaboration between the assessment team and the experts enables a more fluid assessment process and, in particular, allows relatively straightforward incorporation of conceptual model uncertainty in the assessment process, rather than being limited to parameter uncertainty (see Section 18.1.1). For example, in the update to the probabilistic volcanic hazard analysis (PVHA-U) for the proposed Yucca Mountain repository completed by the DOE in 2008 (SNL, 2008), each of eight experts developed their own models of volcanism in the Yucca Mountain region. Numerous alternative conceptual models of the spatial distribution of volcanism, the temporal distribution of volcanism and even the characteristics of an igneous event were proposed and discussed by the experts during several workshops.

© Woodhead Publishing Limited, 2010

Assessment of expert judgments

597

Across all the experts, at least four distinct conceptual models of the location of future events were specified: models based on the concept of zones, defined by geological data and features, two alternative models based on the location of past events and models that combined the location of past events with explicit consideration of geological data. Because each of these alternative conceptual models was included in at least one of the individual expert assessments, all are represented in the final elicitation results. Feedback to experts In the approach described in the Procedures Guide, each expert provides his or her answers to a set of assessment questions individually, without interaction with other experts, guided by the elicitation document. The expert is responsible for doing his or her own research, identifying relevant information and models, etc., without discussion with other experts. Each expert provides answers to exactly the same questions, including assessments related to a number of ‘seed variables’, discussed further in Section 18.5. Experts are given individual feedback on their own performance during the elicitation interviews. However, only after the analysis is complete are they allowed to see the range of judgments from the full panel. Following the SSHAC process, the elicitation is also conducted with each expert individually or with small teams with related expertise, but feedback is provided both to individuals or teams and the panel as a whole in an iterative process. While each individual or team is tasked with developing assessments of the same overarching questions, the specific assessment or ‘elicitation’ questions asked of each may differ slightly based on the model formulation and structure that is most helpful to and relevant to that expert’s conceptual model. No ‘seed variables’ or calibration questions are used; rather the experts are provided information on the various heuristics and biases that may affect their judgment and assessments are structured to minimize the impacts of those biases. Extensive feedback on the implications of the assessments is provided to the experts, both during individual elicitation interviews and in feedback to the entire panel. Feedback occurs both in terms of exploring or ‘verifying’ individual assessments, as described above, and through explanation of the implications of those assessments on the larger issues of interest. Several recent papers that have reviewed a wide range of SSHAC-type studies discuss the critical role of feedback in obtaining comprehensive and accurate assessments (Coppersmith et al., 2009a; Hanks et al., 2009); both of these studies emphasize the need for more feedback during the assessment process.

© Woodhead Publishing Limited, 2010

598

Geological repository systems for safe disposal

18.5

Aggregating assessments from multiple experts

As shown in Table 18.1, all approaches include some type of combination or aggregation of the individual expert’s assessments. It is worth noting that the issue of aggregating expert opinion is not unique to safety analyses, and various other approaches have been explored in depth elsewhere (Clemen and Winkler, 2007). In general, these approaches are described either as behavioral aggregation or mathematical aggregation. Behavioral aggregation simply means that experts are brought together in some manner and their assessments are combined in a manner acceptable to the experts themselves, or to the decision maker. The Delphi Method (Linstone and Turoff, 1975), for example, involves distributing the initial assessments of each individual back to the entire group and asking each to reevaluate and update their assessment based on their review of the assessments of other experts. This process is repeated until some predetermined stopping point is reached: consensus or some number of feedback rounds. Typically an average is taken at the end of the process to represent the aggregated assessments. The Nominal Group Technique (Delbecq and VandeVen, 1971) follows a similar process except that the experts are gathered and, through facilitated discussion, come to agreement on either a consensus distribution or a mutually acceptable mathematical aggregation. At the limit of ‘behavioral aggregation’, Bolado and Badea (2008) describe elicitations conducted by Phillips and colleagues wherein facilitators used group processes to help a group of experts develop a consensus distribution without individual assessments. Far more common in expert elicitations for safety analyses and performance assessments is mathematical aggregation, wherein the individual expert assessments are combined mathematically into a single distribution to be used in analysis. The most common aggregation approach is called a ‘linear opinion pool’ (Clemen and Winkler, 2007), a simple mathematical weighting and summing of individual assessments. A series of more complex mathematical combinations known as Bayesian approaches have been also been proposed and demonstrated, but are not widely used and will not be discussed further here.

18.5.1 Scoring rules and differential weighting of expert assessments As mentioned above, the Classical Approach includes differential weighting of expert assessments: the weights are derived based on an extension of early work evaluating the accuracy of different forecasters. The difficulty of knowing how ‘good’ or how accurate a probabilistic forecaster is at estimating the likelihood of uncertain events was recognized well before the

© Woodhead Publishing Limited, 2010

Assessment of expert judgments

599

development of the elicitation protocols described above (Brier, 1950; Winkler, 1969). ‘Scoring rules’ are mathematical formulas that use an expert’s assessed probability in combination with the occurrence or nonoccurrence of the event in question to create a ‘score’ that provides some measure of the accuracy of the estimator. Scoring rules were first used as a measure of the quality of the forecast in areas where the event occurrence is known after the fact, such as weather forecasts (Brier, 1950). In the case of expert elicitations for safety assessments, it is rarely the case that the ‘actual’ values will be known in any reasonable time frame, if ever. In this context it is particularly difficult to judge the quality of an assessment or of an individual expert’s judgments. Focus instead has been on the development of ‘strictly proper scoring rules’, which are scoring rules designed to motivate the expert to provide his or her most accurate and nonstrategic response (Winkler, 1969; Cooke, 1991). The Classical Approach uses a strictly proper scoring rule to derive weights for each expert’s assessments. To do so, experts are asked to provide their judgments on a set of performance or ‘seed’ or ‘performance’ variables, in addition to providing judgments about the variables of interest for the actual elicitation problem (see step 4 in the Procedure Guide column of Table 18.1). Seed variables are quantities that are relevant to the assessment task, which have ‘known’ or knowable values but which the experts are unlikely to know before the assessment. They are similar in concept to the ‘almanac’ questions used to demonstrate or measure overconfidence, but in this context they are carefully defined to be relevant to the assessment task and are used to help determine how well the expert is calibrated in his or her use of probabilistic judgments, and how informative the judgments are. The mathematics can be complicated and are derived elsewhere (Cooke, 1991); conceptually, calibration is a measure of how frequently the actual values for the performance questions fall within the ranges estimated by the expert. Assume, for example, that an expert is asked 10 seed questions, and for each is asked to provide an estimate of the 5th, 25th, 50th, 75th and 95th percentiles. A well-calibrated expert would end up with one of ten answers outside of the 5th–95th percentile range; five answers that fall within what they specified as the 25th–75th percentile range, etc. Because the seed variables correspond to variables for which an ‘actual’ value is known, an estimate of the expert’s calibration can be made based on his or her calculated calibration on the seed variables. The informativeness of an expert’s assessment can be thought of as a measure of how ‘concentrated’ the assessment is. All other things being equal, experts or assessments that are narrower or more tightly focused around some value are considered to be more informative than those that are more dispersed.

© Woodhead Publishing Limited, 2010

600

Geological repository systems for safe disposal

In the Classical Approach, these calculated values for the calibration and informativeness of each expert are used to weight the different judgments and create an aggregated distribution. In the SSAC Approach, for reasons described previously, expert judgements are combined with equal weights.

18.5.2 Comparing aggregation methods In a recent review of aggregation approaches, Clemen and Winkler (2007) review the performance of different approaches. All the difficulties discussed above in terms of understanding how ‘good’ an individual assessment is apply equally to trying to understand the ‘goodness’ of different aggregation approaches: the best that can be done is to compare assessments (individual and aggregated) about variables for which the true value can later be determined. ‘Better’ aggregation approaches are those that give values closer to the ultimate ‘true’ values. Clemen and Winkler concluded: ‘ . . . mathematical aggregation outperforms intuitive aggregation, and mathematical and behavioral approaches tend to be similar in performance, with mathematical rules having a slight edge.’ Regarding the mathematical aggregation approaches, they also conclude that ‘ . . . simple combination rules (e.g. a simple average) tend to perform quite well.’ More complex mathematical combinations can sometimes perform better than simpler rules, but their performance is less predictable, as well as being more difficult to explain.

18.6

Degrees of rigor and formality in assessing expert judgments

As described previously, expert judgments of many types are used throughout the safety analysis and performance assessment processes. The ways in which these judgments are recognized, made and documented define a continuum, ranging from unrecognized judgments at one end to formal multiple-expert ‘expert elicitation’ studies at the other end. Examples of expert judgments that may not even be recognized or acknowledged as being judgments include: judgments made about what data to collect and what systems to monitor early in the site characterization process (e.g. ground water monitoring in the Yucca Mountain region began as early as the late 1970s: the location of monitoring wells, monitoring frequency, etc., were all decisions requiring expert judgment); judgments made by modelers about how to implement or represent a subject matter expert’s conceptual model in a computer code. Other expert judgments are acknowledged, but made informally and qualitatively, such as in discussions of conceptual model uncertainty. In

© Woodhead Publishing Limited, 2010

Assessment of expert judgments

601

many performance assessments and modeling exercises, subject matter experts will describe alternative conceptual models that are consistent with the available data and scientifically sound. They may describe the implications of those alternatives and the rationale behind the conceptual model or models they ultimately prefer. Continuing down the line, many estimates of model parameters or inputs are based on documented expert judgments, but are not formally assessed. In these cases subject matter experts simply provide the necessary input values based on their own modeling, data and judgment. They may be encouraged to provide ranges for the parameters and to provide the best estimate of their uncertainty about the parameter, but they are not provided with tools or assistance to help ensure their assessments accurately reflect their level of knowledge. Judicious use of the assessment techniques described in Sections 18.2 and 18.3, especially the verification techniques, could significantly increase the quality and the defensibility of this type of expert judgment. At the other end of the continuum are expert judgments developed through formal expert elicitation approaches, such as those used to develop probabilistic seismic hazard analyses (Stepp et al., 2001) and probabilistic volcanic hazard analyses (Perry et al., 1998), and being used widely in European Union studies (Bolado, 2008) and elsewhere. In this case the use of expert judgment is clear, and assessment protocols and tools are fully implemented throughout the study. In many cases (e.g. in most probabilistic hazard analyses), expert elicitation may be used to develop the inputs to models, which are then used to calculate outputs that are relevant to the decision-makers (e.g. a mean hazard curve for a seismic hazard). It is particularly important in such cases to verify not only the specific expert assessments but also the model construct and the calculated results with the expert’s overall judgments. Andrews et al. (2007) provide a specific illustration of the general principle that results from an expert elicitation need to be verified for physical realism over the range of probabilties for which the result is intended to be used. Andrews et al. (2007) criticize the probabilistic seismic hazard analysis for Yucca Mountain (Stepp et al., 2001) for including calculated ground motions that have been judged to be physically impossible. Application of verification techniques to the calculated hazard results may well have highlighted this result and led to reassessments that would have eliminated such results from the final calculated hazard. The recently completed update to the probabilistic volcanic hazard analysis for Yucca Mountain (SNL, 2008) included significant feedback to the experts as part of the verification step, including feedback on the hazard results calculated from their models and input assessments, and recent recommendations of lessons learned from such studies stress the importance of this type of feedback (Hanks et al., 2009).

© Woodhead Publishing Limited, 2010

602

Geological repository systems for safe disposal

There is one giant step in this continuum: there is a significant increase in the quality and the defensibility of the expert judgments obtained when moving from documented but unassessed expert judgment to formal expert assessments. There is, however, a large increase in cost, level of effort and time required to obtain these estimates. The costs and the length of time some of these formal multi-expert elicitations require has led some to argue that the elicitations are being ‘used’, inappropriately, as a substitute for performing additional scientific work. This idea is also reflected in a critique of the SSHAC recommendations by a panel of the US National Academy of Sciences (NAS, 1997), which suggested that the SSHAC document overemphasized the so-called Level IV approach, the full-scale multi-expert studies described above, at the expense of less intensive approaches that preserve the benefits of formal assessment with less cost and effort. The NRC recently sponsored a series of workshops to collect ‘lessons learned’ from SSHAC studies completed over the past ten or more years, and discussions focused on how to reduce the level of effort required for such studies (Hanks et al., 2009). We believe that the intensive approaches described above do have a role: they may be the only way to obtain defensible and comprehensive estimates of intractable uncertainties for higher-risk endeavors such as nuclear installations or high-consequence industrial accident scenarios. We also believe that judicious use of the assessment techniques described in Sections 18.2 and 18.3 could significantly increase the quality and the defensibility of many of the expert judgments required for safety analyses and performance assessments, with only a modest increase in the level of effort required. The assessment approaches given above are described as if they require a great deal of formality and a number of detailed steps, but in practice these assessments of expert judgments can be conducted at varying levels of detail. The increased effort yields definite benefits: increased defensibility of the analyses, increased ownership of the analyses and results by the technical people providing input and more robust estimates and model results. Table 18.2 lists some of the levels or types of expert judgments described above, an example or two of where that type of judgment has been used if available and our recommendations for where and when each type should be used.

18.7

Future trends

Some of what we expect (and hope) to see in the future as it relates to the assessment and use of expert judgments in safety analyses and performance assessments is described in the previous section. We see a growing recognition of the role of expert judgment throughout the safety analysis and performance-assessment processes, and a concomitant need for better

© Woodhead Publishing Limited, 2010

© Woodhead Publishing Limited, 2010

Should rarely be used. Judgments made in this way can be improved significantly with relatively little effort: at a minimum, clarification of the ultimate use of the judgment, the importance of appropriately quantifying uncertainty and a discussion of potential biases that might affect unaided judgments can lead to significantly more robust judgments Should be used for most model inputs for a safety analysis or performance assessment. The degree of rigor and level of detailed interaction between the subject matter expert and the assessor can be established based on the importance of the variable or judgment to the overall assessment of system performance and on the amount of relevant data available to support the assessment. The minimal approach is described above. For more critical judgments, more structure, perhaps including structured discussion and assessments from more than one individual or technical team, may be appropriate

Descriptions of conceptual model uncertainty in some modeling efforts; description of alternative conceptual models and the rationale for the selected conceptual models

A specific model input variable that has been described by a technical expert or expert team encouraged to express uncertainty

Structured probabilistic inputs for model parameters; more frequently seen for variables for which there is little or no data, at early stages in a repository program

Use only for issues of critical importance to system performance, where Probabilistic seismic and volcanic hazard analyses (PSHAs and PVHAs) typically use significant uncertainty or disagreement exists and there is a high need for this formalized approach: extremely limited robust, defensible estimates data and the need to make estimates of critical variables over very long time frames require this use of expert judgment

Non-quantitative assessments: acknowledgment and discussion of the use of expert judgment

Documented judgments, usually including uncertainty, by one or more experts

Assessed judgments, using one or more of the structured approaches discussed in this chapter to quantify uncertainties

Formal expert assessments involving multiple experts, extensive interaction, detailed assessment and feedback, followed by development of an aggregate representation of the relevant expert’s knowledge

Appropriate for describing conceptual model uncertainty

Decisions by modelers about Never. Any time expert judgment is used both the judgment and the basis representational models used to implement for that judgment should at a minimum be documented and tested/ a specific conceptual model confirmed with other knowledgeable experts

Unrecognized or unacknowledged expert judgments

Recommended uses

Examples

Degrees of formality in assessments of expert judgment: current and recommended uses

Type of expert judgment/ assessment

Table 18. 2

604

Geological repository systems for safe disposal

guidance and perhaps new approaches for formalizing the assessments and the documentation of those judgments. This should include more formality in the recording of decisions made throughout the modeling process: decisions about conceptual model selection, about model implementation and about how the inputs required for those models are obtained. When empirical data are used to support model inputs, it is essential that the appropriateness of the data is considered and documented: the ranges of conditions to be modeled as compared to the ranges experimentally evaluated and the rationale for extrapolating the experimental data to provide meaningful input to the model must be clearly understood and documented. Understanding this importance early in the site characterization process will be helpful in ensuring that careful documentation is maintained throughout the process. Those gathering the data will need to explain its limits and the handoff between the ‘data gatherers’ and the ‘data users’ needs to be careful and thorough. We have heard it argued that it often ‘goes without saying’, that this communication and documentation is in place. Experience has shown, however, that it is often not done until there is a regulatory inquiry leading to a mad scramble to document what was decided, and on what basis, in the past. As noted, expert judgments are used informally early in the process in terms of defining site characterization activities and defining and screening scenarios. Screening to ensure that only the most important or consequential scenarios are modeled has the potential to save significant amounts of time and money in the site characterization process, but adequate models are often not available early in the process to support expert assessment of the probabilities and consequences of various scenarios. The increased use of simplified models very early in the safety analysis and performance assessment process would allow for more rigorous assessment of the screening criteria and, perhaps, more succinct yet appropriate performance assessments. Inputs for simplified but realistic (in terms of the treatment of uncertainty) models can be developed initially using primarily expert judgment: uncertainties would be expected to be significant in the early stages, with highly simplified models being replaced by more sophisticated models as the site characterization and model process proceeds. As the number of performance assessments for geological repositories conducted worldwide grows, we see an increased opportunity for the development of these simplified early models. Similarly, a final development that we expect and hope to see will be increased international cooperation and sharing of models and information between countries using similar geological locations and repository designs. If, for example, the Swiss have done a thorough and complete formal expert assessment on some aspect of the behavior of a repository in clay and the French are evaluating a potential repository in similar hard-rock clays, we

© Woodhead Publishing Limited, 2010

Assessment of expert judgments

605

would like to see the latter be able to make maximum use of the assessments that have already been completed. It is likely that future developments will be incremental and not revolutionary. Updates or slight modifications to the methodology result every time it is used in a new application. It is noted, for example, that the Swiss PEGASOS expert elicitation, according to its elicitation team, led to several improvements in methodology as recently as 2009 (Coppersmith et al., 2009b). Publication of post-elicitation lessons learned should be encouraged.

18.8

Sources for further information and advice

For information on decision analysis and probability assessments generally, see Clemen (1996) and Morgan and Henrion (1990). Chapters 6 and 7 of Morgan and Henrion (1990), Chapter 5 of Bolado and Badea (2008) and Hora (2007) all provide additional details and step-by-step guidance for developing specific elicitation questions using the various approaches discussed in Section 18.2.1 above. Each of those authors also discuss some of the heuristics and biases affecting human judgments; for a comprehensive description and discussion of these biases, see Kahneman et al. (1982) and Gilovich et al. (2002). Meyer and Booker (1991) describe several approaches to assessing expert opinions. Cooke (1991) describes the Classical Approach to assessments with multiple experts, and an application of that approach to assessing volcanic hazard can be found in Aspinall and Cooke (1998) and the Procedures Guide described above (Cooke and Goossens, 1999). Coppersmith et al. (2009a) describe what they term ‘formal expert assessment’, based in part on the SSHAC process, along with several applications of that approach, and Hanks et al. (2009) describe a set of ‘lessons learned’ from SSHAC-type studies. Bolado (2008) outlines a procedure for assessing expert judgment entirely consistent with the SSHAC process. In terms of combining expert judgments, Clemen and Winkler (2007) provide a nice overview and review of earlier work from the field of decision analysis.

18.9

Acknowledgments

The authors thank Kevin Coppersmith, Timothy Nieman, Roseanne Perman and Bob Youngs for many interesting and detailed discussions of expert elicitation protocols and philosophy and Peter Swift and John Helton for sharing early drafts of their chapters to enhance integration. The

© Woodhead Publishing Limited, 2010

606

Geological repository systems for safe disposal

opinions and recommendations expressed in this chapter are those of the authors alone.

18.10 References Andrews D J, Hanks T C and Whitney J W (2007) ‘Physical limits on ground motion at Yucca Mountain’, Bulletin of the Seismological Society of America, 97, 1771–1792. Aspinall W and Cooke R M (1998), ‘Expert judgement and the Montserrat volcano eruption’, in Proceedings of the 4th International Conference on Probabilistic Safety Assessment and Management PSAM4, edited by A Mosleh and R A Bari, vol. 3, pp. 2113–2118, 13–18 September 1998, New York City. Bell D E, Raiffa H and Tversky A (1988), ‘Descriptive, normative, and prescriptive interactions in decision making’, in Decision Making: Descriptive, Normative, and Prescriptive Interactions, edited by Bell, Raiffa and Tversky, Cambridge University Press, New York. Bolado R (2008), ‘An expert judgement protocol to assess solubility limit distributions for key chemical elements in a generic Spanish repository in granite’, European Commission, Euratom Research and Training Programme on Nuclear Energy, Available from http://www.ip-pamina.eu/publications/ reports/index.html (accessed 5 November 2009). Bolado R and Badea A (2008), ‘Review of expert judgement methods for assigning PDFs’, European Commission, Euratom Research and Training Programme on Nuclear Energy, Available from http://www.ip-pamina.eu/publications/ reports/index.html (accessed 5 November 2009). Brier G W (1950), ‘Verification of forecasts expressed in terms of probability’, Monthly Weather Review, 78, 1–3. BSC (Bechtel SAIC Company) (2001), ‘Uncertainty analyses and strategy’, Report for the US Department of Energy, Yucca Mountain Site Characterization Office, Report Document SA011481M4. Capen E C (1976), ‘The difficulty of assessing uncertainty’, Journal of Petroleum Technology, 28, 843–850. Clemen R T (1996), Making Hard Decisions: An Introduction to Decision Analysis, 2nd edition, Duxbury Press, Belmont, California. Clemen R T and Winkler R L (1999), ‘Combining probability distributions from experts in risk analysis’, Risk Analysis, 19, 187–204. Clemen R T and Winkler R L (2007), ‘Aggregating probability distributions’, in Advances in Decision Analysis: From Foundations to Applications, edited by W Edwards, R F Miles Jr and D von Winterfeldt, Cambridge University Press, Cambridge. Cooke R M (1991), Experts in Uncertainty: Opinion and Subjective Probability in Science, Oxford University Press, New York. Cooke R M and Goossens L H J (1999), ‘Procedures guide for structured expert judgement’, Report EUR18820, Commission of the European Communities, Brussels–Luxembourg. Coppersmith K J, Jenni K E, Perman R C and Youngs R R (2009a), ‘Formal expert assessment in probabilistic seismic and volcanic hazard analysis’, in Volcanic

© Woodhead Publishing Limited, 2010

Assessment of expert judgments

607

and Tectonic Hazard Assessment for Nuclear Facilities, edited by C B Connor, N A Chapman and L J Connor, Cambridge University Press, New York. Coppersmith K J, Youngs R R and Sprecher C (2009b), ‘Methodology and main results of seismic source characterization for the PEGASOS Project, Switzerland’, Swiss Journal of Geosciences, 102, 91–105. Delbecq A L and VandeVen A H (1971), ‘A group process model for problem identification and program planning’, Journal of Applied Behavioral Science, 7, 466–491. EPRI (Electric Power Research Institute) (2005), ‘EPRI Yucca Mountain total system Performance Assessment Code (IMARC) Version 8, EPRI Report 10118143, Electric Power Research Institute, Palo Alto, California, Available from http://my.epri.com/portal/server.pt?open=512&objID=366&&PageID =224848 &mode=2&in_hi_userid=2&cached=true (accessed 6 November 2009). Fischhoff B and MacGregor D (1982), ‘Subjective confidence in forecasts’, Journal of Forecasting, 1, 155–172. Gilovich T, Griffin D and Kahneman D (2002). Heuristics and Biases: The Psychology of Intuitive Judgment, Cambridge University Press, Cambridge. Hanks T C, Abrahamson N A, Boore D M, Coppersmith K J and Knepprath N E (2009), ‘Implementation of the SSHAC Guidelines for Level 3 and 4 PSHAs – experience gained from actual applications’, US Geological Survey Open-File Report 2009–1093, 66 pp., Available from http://pubs.usgs.gov/of/2009/1093/ (accessed 6 November 2009). Hora S C (2007), ‘Assessing probabilities from experts,’ in Advances in Decision Analysis: From Foundations to Applications, edited by W Edwards, R F Miles Jr and D von Winterfeldt, Cambridge University Press, Cambridge. Hora S C, Hora J A and Dodd N G (1992), ‘Assessment of probability distributions for continuous random variables’, The Journal of Behavioral Decision Making, 6, 133–147. Howard R A (1966), ‘Decision analysis: applied decision theory’, in Proceedings of the Fourth International Conference on Operational Research, pp. 55–71, Wiley– Interscience. Reprinted in Howard R A and Matheson J E (1984), The Principles and Applications of Decision Analysis, Strategic Decisions Group, Menlo Park, California. Kahneman D, Slovic P and Tversky A (eds) (1982), Judgment under Uncertainty: Heuristics and Biases, Cambridge University Press, Cambridge. Keefer D L and Bodily S E (1983), ‘Three-point approximations for continuous random variables’, Management Science, 29, 595–609. Kessler J H and Vlasity J A (2003), ‘EPRI performance assessment results for the Yucca Mountain repository’, in Proceedings of the International High Level Radioactive Waste Management Conference, Las Vegas, Nevada. Kotra J P, Lee M P, Eisenberg N A and DeWispelare A R (1996), ‘Branch technical position on the use of expert elicitation in the high-level radioactive waste program’, NUREG-1563, US Nuclear Regulatory Commission, Washington, DC. Lichtenstein S, Slovic P, Fischhoff B, Layman M and Combs B (1978), ‘Judged frequency of lethal events’, Journal of Experimental Psychology: Human Learning and Memory, 4, 551–578.

© Woodhead Publishing Limited, 2010

608

Geological repository systems for safe disposal

Lichtenstein S, Fischhoff B and Phillips L D (1982), ‘Calibration of probabilities: the state of the art to 1980’, in Judgment under Uncertainty: Heuristics and Biases, edited by D Kahneman, P Slovic and A Tversky, Cambridge University Press, New York. Linstone H A and Turoff M (1975), The Dephi Method: Techniques and Applications, Addison Wesley, Reading, Massachusetts. Merkhofer M W (1987), ‘Quantifying judgmental uncertainty: methodology, experiences, and insights’, IEEE Transactions on Systems, Man, and Cybernetics, SMC-17, 741–752. Meyer M A and Booker J M (1991), Eliciting and Analyzing Expert Judgment: A Practical Guide, Academic Press, San Diego, California. Miller A C and Rice T R (1983), ‘Discrete approximations of probability distributions’, Management Science, 29, 352–362. Morgan G M and Henrion M (1990), Uncertainty: A Guide to Dealing with Uncertainty in Quantitative Risk and Policy Analysis, Cambridge University Press, New York. NAS (National Academy of Sciences) (1995), Technical Bases for Yucca Mountain Standards, National Research Council, National Academy Press, Washington, DC. NAS (National Academy of Sciences) (1997), Review of Recommendations for Probabilistic Seismic Hazard Analysis: Guidance on Uncertainty and Use of Experts Panel on Seismic Hazard Evaluation, National Research Council, National Academy Press, Washington, DC. NEA (Nuclear Energy Agency) (1999), Scenario Development Methods and Practice: An Evaluation Based on the NEA Workshop on Scenario Development, Madrid, Spain, May 1999, Organisation for Economic Co-operation and Development Nuclear Energy Agency, Paris, France. Perry F V, Crowe B M, Valentine G A, and Bowker L M (1998), ‘Volcanism Studies: Final Report for the Yucca Mountain Project’, LA-13478, Los Alamos National Laboratory, Los Alamos, New Mexico. Raiffa H (1968), Decision Analysis: Introductory Lectures on Choices under Uncertainty, Random House, New York. Smith J E (1993), ‘Moment methods for decision analysis’, Management Science, 39, 340–358. SNL (Sandia National Laboratories) (2008), ‘Probabilistic volcanic hazard analysis update (PVHA-U) for Yucca Mountain, Nevada’, TDR-MGR-PO-000001 REV 00, US Department of Energy Office of Civilian Radioactive Waste Management, Las Vegas, Nevada. Spetzler C S and Stael von Holstein C A S (1975). ‘Probability encoding in decision analysis’, Management Science, 22, 340–358. SSHAC (Senior Seismic Hazard Analysis Committee) (1997), ‘Recommendations for probabilistic seismic hazard analysis: guidance on uncertainty and use of experts’, NUREG/CR-6732, US Nuclear Regulatory Commission, Washington, DC. Stepp J C, Wong I, Whitney J, Quittmeyer R, Abrahamson N, Toro G, Youngs R, Coppersmith K, Savy J, Sullivan T and Yucca Mountain PSHA Project Members (2001), ‘Probabilistic seismic hazard analyses for ground motions and fault displacements at Yucca Mountain’, Earthquake Spectra, 17, 113–151.

© Woodhead Publishing Limited, 2010

Assessment of expert judgments

609

Tversky A and Kahneman D (1971), ‘Belief in the law of small numbers’, Psychological Bulletin, 76, 105–110. Tversky A and Kahneman D (1974), ‘Judgment under uncertainty: heuristics and biases’, Science, 185, 1124–1131. Winkler R L (1969), ‘Scoring rules and the evaluation of probability assessors’, Journal of the American Statistical Association, 64, 1073–1078.

© Woodhead Publishing Limited, 2010

19 Application of knowledge management systems for safe geological disposal of radioactive waste H . U M E K I , Japan Atomic Energy Agency, Japan; H . T A K A S E , Quintessa K.K, Japan

Abstract: Information overload caused by the rate at which data can be produced and the ease with which it can be accessed poses a significant challenge to those charged with maintaining an overview of large, complex, multidisciplinary projects. Geological disposal of long-lived radioactive waste is a technical area characterised by a breadth of multidisciplinary knowledge wider than almost any other industry. The particular challenges for radwaste require the development of a system that pushes the current limits of information technology (IT). This chapter illustrates an approach to the problem that focuses on a formal knowledge management system (KMS) and associated knowledge base that utilise state-of-the-art tools, developed in the field of knowledge engineering (KE) and IT. It considers whether advanced KM tools can contribute to the development, review, communication and control of the iterative evolution of safety cases. Key words: knowledge management system (KMS), knowledge engineering (KE), strategic environmental assessment (SEA), research and development R&D, quality management system (QMS), argumentation model (AM), knowledge acquisition design system (KADS), artificial neural network (ANN).

19.1

Introduction

The breathtaking expansion in technology over the last few decades has helped solve many problems – but also created others. The greenhouse effect, destruction of the ozone layer, pollution of land and sea, etc., are dramatic global issues that, rightly, catch attention. However, from a pragmatic point of view, there is an insidious problem that is only gradually being recognised as a major concern – information overload caused by the exponential growth 610 © Woodhead Publishing Limited, 2010

Application of knowledge management systems

611

in the rate at which data can be produced and the ease with which it can be accessed. Sociologically, this factor has been identified as a cause of alienation of the older generation and those with less access to information technology (IT), but it is also a growing headache for those charged with maintaining an overview of large, complex, multidisciplinary projects. There are few areas where this is more evident than geological disposal of long-lived radioactive waste (radwaste), which is a technical area characterised by a breadth of multidisciplinary knowledge that is wider than almost any other industry - covering geology to radiation physics, materials science to microbiology, archaeology to engineering, public communication to advanced IT. It also has an unparalleled depth in time, in terms of project implementation (around 100 years for the siting, construction, operation and final closure of facilities that are being planned today – matched maybe by some mediaeval cathedrals) and the associated safety case (hundreds of thousands or millions of years – much longer than the existence of modern man!). Of course this problem is not unique and pressure in other affected industries (notably aerospace) has led to the growth of the discipline of knowledge management (KM). Nevertheless, the particular challenges for radwaste require that a system be developed that pushes – or exceeds – the limits of the current state-of-the-art of IT. Indeed, the initiatives developed here may well be more widely applicable in the future – when nuclear waste may well be seen to be one of the minor challenges that our global society is facing (McKinley, 2009). In this chapter, by considering global trends and using illustrative examples from Japan, an approach to cutting this ‘Gordian Knot’ is illustrated. This focuses on a formal knowledge management system (KMS) and associated knowledge base that utilise state-of-the-art tools, developed in the field of knowledge engineering (KE) and IT. In particular, the critical question, of whether advanced KM tools can contribute to the development, review, communication and control of the iterative evolution of safety cases, will be considered.

19.2

Knowledge management: definitions and nomenclature

The quote from Francis Bacon – ‘Knowledge is Power’ – has never been more true than at the present time. The major global problem at the beginning of the 21st century may be presented as resulting from unequal access to resources, highlighted by the increasing divergence between the rich and the poor. The fundamental difficulty with addressing this problem may result, to a large extent, from unequal access to knowledge. In fact, the

© Woodhead Publishing Limited, 2010

612

Geological repository systems for safe disposal

critical issue is not actually information access per se, but possession of the tools and intellectual resources that allow knowledge to be usefully harnessed. Here, ‘knowledge’ is used as a global term, which encompasses all of the science and technology (implicitly including social science, economics, medicine, etc.) underpinning a repository project. This can be classified as common knowledge (e.g. that established in component disciplines – geology, chemistry, materials science, civil engineering, etc.), generic waste management knowledge and project-specific knowledge. Knowledge is not static, but evolves with time in line with the general progress in science and technology. In addition, experience is associated directly with individual staff and also accumulates with time. Therefore, an active programme of experience transfer is needed to ensure that this is passed to younger generations before older staff members retire (an acknowledged concern throughout the nuclear industry and the driver for initiatives like the ITC, www.itc-school.org). The term ‘knowledge management’ covers all aspects of the development, integration, quality assurance (QA), communication and maintenance/ archiving of knowledge, including data, information, understanding and experience. It is an active process, which is focused by specific programme or project requirements (which may be developed and structured by a requirements management system). A KMS aims to optimise ‘knowledge management’ in terms of formalised approaches and tools for creating, storing, retrieving and disseminating knowledge. Knowledge management is a term commonly used in many areas of technology but, in general, focus is on conventional approaches to the systematic handling of technical information. Indeed, knowledge management, for projects like deep geological disposal, has tended so far to involve only formalisation of the old system of experts (‘gurus’) in key technical areas supported by electronic access to the technical literature. Such an approach has served for the last decades, but is now seen to be straining at the seams. There are increasingly strong indications of a ‘phase transition’ in knowledge space, which may soon render such an approach impotent and which thus merits urgent attention. As considered below, this is particularly critical in the Japanese programme (JNC, 2005; Kawata et al., 2006; Miyamoto et al., 2006; Umeki, 2007).

19.3

Disposal programme structures and knowledge flows

Geological disposal programmes are unique in terms of their duration, complexity and political sensitivity. For example, even after three decades of

© Woodhead Publishing Limited, 2010

Application of knowledge management systems

613

generic planning (e.g. JNC, 2000), implementation of the first Japanese HLW repository, starting from initial site investigations through to the closure of the filled-up repository, could run until the end of this century (e.g. NUMO, 2004). Even after closure of a repository, post-closure monitoring may continue for an even longer period. In line with other generic national projects initiated in the 1970s and 1980s, there was little formal structuring of the knowledge base and information flows in the early days, with much of the supporting knowledge production carried out in a bottom-up manner, as defined by the technical experts involved. As programmes mature and move towards the sensitive stages of site selection and licensing for construction and operation, there is an accepted need for a transparent process of decision-making based on structured, quality assured and openly accessible knowledge bases. Given their long duration, implementation is generally planned in a stepwise manner, which, in most programmes, involves communication with an increasing range of stakeholders. Recently, there has been increased discussion of the desire to maintain the ability of future generations to contribute to this decisionmaking process (e.g. NEA, 2004a) and assured reversibility of decisions, should this be required (e.g. KASAM and IAEA, 2000; NEA, 2001). A key role of waste management organisations is to structure and construct sufficiently complete knowledge bases in time for project milestones, identifying important gaps in order to focus the associated R&D that generates required new knowledge. In the 1980s, or even in the 1990s, it was possible for top managers in the nuclear waste field to have a reasonably comprehensive overview of all relevant technical work contributing to a repository project. Since then, there has not only been breathtaking growth in basic knowledge but work has become more international and has been opened up to wider scrutiny – with increasing emphasis on non-technical aspects associated with open communication and public acceptance (e.g. Kawata et al., 2006). Except possibly in the smallest and most isolated programmes, the traditional approach is beginning to lead to an obvious loss of overview, synthesis and flexibility. In retrospect, some of the problems experienced by the very technocratic approach to early repository programme development can be attributed to the narrow range of knowledge considered to be significant by key decision-makers (the decide–announce–defend approach, leading to safe concepts that were unacceptable to stakeholders). More recently, the move towards devolving much decision-making to consensus among stakeholders may lead to the opposite situation – compromise concepts that are not straightforward to assess and/or tricky to implement. An important problem here has been termed ‘knowledge asymmetry’, which is reflected most strongly in the consensus that geological disposal is safe by experts in this area, the strong concerns expressed by many professional and

© Woodhead Publishing Limited, 2010

614

Geological repository systems for safe disposal

academic groups who are non-specialist and the common assumption that safe disposal is inherently impossible by politicians, nuclear opponents and the general public (e.g. McKinley, 2009). In summary, the range and quantity of knowledge to be assessed in nuclear waste disposal is expanding explosively while the needs in terms of its synthesis for decision-making and its communication to stakeholders are becoming much more challenging. This problem was recognised already at the end of the 20th century in several programmes and has given rise to focused efforts to manage knowledge more efficiently. However, the rapid expansion of the knowledge base results in a ‘red queen’s race’, where all efforts in conventional approaches aim, at best, to avoid losing ground. To escape from this Catch 22 situation, a paradigm shift is needed in the way in which knowledge management is implemented.

19.4

Identification of critical problems and development of solutions

The amount of diverse information that must be collated, analysed, integrated and summarised to form the basis for a repository project is already so huge (e.g. US DOE for YMP licensing application, http://www. ocrwm.doe.gov/repository/index.shtml) that the effort already invested to its management in longer-running programmes can give rise to inertia and resistance to change. Even if key staff acknowledge that their system is failing or dysfunctional, they are simply too busy to develop alternatives. It is thus clearly a challenge for top management to assess the source of problems and identify optimised solutions. This is particularly the case when such solutions cannot be developed by stepwise modification of existing procedures, but require introduction of novel methods and tools. As an additional complication, the advanced KM approaches that come into consideration are not well established in the radwaste field and need to be demonstrated as not only applicable for the roles identified but also robust and user-friendly in practice. The fundamental dichotomy is that advanced KM technology shows its advantages when implemented in a comprehensive manner – requiring parallel development and implementation of a range of tools and necessitating a commitment from all involved staff – whereas caution and limited resources would favour piecewise introduction of tools that are developed sequentially, allowing experience from a lead team to gradually disperse through the entire organisation. When the alternative is a complete technical breakdown of project integration and control, however, it is necessary eventually to make a leap in the dark. A clear hope in many cases has been that one national programme would

© Woodhead Publishing Limited, 2010

Application of knowledge management systems

615

take the lead and, at least, demonstrate the fundamental feasibility of such an approach and give indications of the costs and benefits associated with the transition. As illustrated in the following section, this has now been done by the Japan Atomic Energy Agency (JAEA), which should open the door for other programmes to follow.

19.5

Japan Atomic Energy Agency (JAEA) knowledge management system (KMS): the basic concept

In Japan, problems associated with the information explosion have been observed, in particular, in R&D areas supporting the high-level radioactive waste (HLW) disposal programme and are highlighted in the ‘H17’ study (JNC, 2005). When the total supporting information base is expressed as bytes of data, it is already in the petabyte range and is expanding exponentially – roughly in line with ‘Moore’s law’ expansion of computing speeds and data storage capacities. Great difficulties were associated with the process of integrating this huge amount of information/data in order to develop the syntheses needed to model geological environments, develop engineering designs for repositories and carry out associated safety assessment. As noted above, identification of the problem is not sufficient – commitment is required from top managers that a major change in operational practice is needed and there needs to be sufficient confidence that new approaches will work efficiently and gain the acceptance of all involved staff. As a first stage, therefore, the basic concept for an advanced KMS was developed. Some key boundary conditions here were that the Japanese siting process for an HLW repository would be based on a volunteering approach and that subsequent stepwise decision-making should involve dialogue with interested stakeholders. Although this involves particular challenges associated with assuring flexibility to tailor the programme to the sites that come forward and to communicate with all relevant parties, it provided a unique window of opportunity to make the KM transition during the hiatus while waiting for volunteers to come forward. Figure 19.1 (Umeki et al., 2008) illustrates the structure and major elements of the system proposed by JAEA. Although only a conceptual outline, it can be seen that the emphasis is on interaction – with two-way flows between the knowledge base and the central guiding knowledge office, the R&D sectors that produce focused new knowledge, the internet, which is the interface to the wider international community, a think tank that attempts to anticipate the inherently unpredictable future and, most importantly, the end users – the justification of the entire exercise!

© Woodhead Publishing Limited, 2010

616

Geological repository systems for safe disposal

19.1 The structure and elements of the knowledge management system (Umeki et al., 2008) (METI: Ministry of Economics, Trade and Industry).

The vast quantity and rate of production of knowledge, its complexity, the need to manipulate it in a rigorous, quality-assured manner and to make it available to a wide range of users make it clear that, wherever possible, the management processes indicated in Fig. 19.1 should be automated, using the most advanced tools available. It is also very important to note here that the JAEA knowledge base is intended to provide technical support to both the implementer and regulator. Given the limitations in resources of experienced staff in this field, there are particular advantages in establishing an agreed common knowledge base that would serve the needs of all stakeholders – including the general public. JAEA was selected to lead this project as an impartial technical organisation with the critical depth of experience needed – ranging from basic research (knowledge production) up to the top-level of use of such knowledge to develop and assess repository sites, designs and associated long-term safety cases (e.g. JNC, 2005). The essence of the concept is that all relevant information will be held in a central electronic database (backed up appropriately). Already, most key documentation is produced in electronic form and associated raw data and all manipulation and modelling of it is available on digital files. Key supporting references from past work can be scanned in and any critical data from earlier work can be digitalised. A great challenge is format standardisation and ensuring that information remains readable, despite evolution of software and hardware in future decades. Such problems are,

© Woodhead Publishing Limited, 2010

Application of knowledge management systems

617

however, common to large-scale projects that are now ongoing in many different fields to produce electronic archives. The archive must be fully searchable by methods allowing contextual analysis of written text and autonomic analysis of data. This is certainly not a trivial job, but requires processing capacities that are minor compared to some of the massive and complex data mining projects being developed, and implemented (e.g. by national security agencies: http://www.fas.org/sgp/crs/intel/RL31798.pdf). In terms of communication, it was recognised that users will include not only technical specialists but also lay stakeholder groups – including politicians and the general public. To facilitate communicating relevant information (combating the problem of ‘knowledge asymmetry’ noted above), the need for appropriate output interfaces was identified, which emphasise simple natural language and presentation of data in the form of autonomically generated visual material, including animations. The proposed structure allows gaps in the knowledge base to be identified and provides guidance for the R&D needed to fill them. The main limitation is that this tends to be constrained by expectations based on the current technical state-of-the-art. For projects running over many decades, the consequences of future advances in science and technology should also be considered. Although there are clearly limitations to the extent to which this can be done, the approach of using focused ‘think tanks’ was selected for further investigation, as a complement to more traditional methods of setting R&D goals and priorities. The development of a conceptual KMS is challenging in itself. Based on the considerations above, however, to be of real use this should not simply be a passive tool to archive and disseminate information. It requires internal analytical facilities to synthesise and integrate material from a diversity of sources, identify trends and inconsistencies and, ideally, even produce feedback to the data producers. In effect, it should replace many of the functions of the network of peer reviewers and expert advisors, who currently carry out such work. When particular emphasis is placed on advanced electronic information management approaches, the fundamental basis appeared feasible in that: . .

.

Already, most key information for repository projects is available electronically and accessible via internet/intranet systems. It is reasonable to expect that this will very soon provide effectively 100% coverage. Increasingly sophisticated content-recognition/cross-referencing systems allow relationships between documents and any form of datasets to be defined in much more detail than traditional document labelling/ keyword approaches. The development of autonomic data mining techniques involving network agents, bots, etc., is currently an area of very rapid progress,

© Woodhead Publishing Limited, 2010

618

.

Geological repository systems for safe disposal which allows much of the information gathering, sorting and compilation processes to be automated. The combination of expert systems with autonomic learning approaches (e.g. based on neural networks) allows, at least in principle, many of the key processes involved in knowledge management – collation, synthesis, review, etc. – to be largely automated.

This concept, together with an outline implementation plan, found wide support within the Japanese waste management programme and was summarised in a number of publications, which also aroused considerable international interest (e.g. Kawata et al., 2006; Umeki, 2007; Umeki et al., 2008). The key identified challenge was, however, actually to demonstrate applicability of novel KE technology that, to date, was established only in other areas or in research studies.

19.6

JAEA KMS: demonstration of application to safety case development

There has been considerable effort made to rationalise procedures associated with repository implementation in order to make the information handling process clearer. Over the last few years, this has included a standardised terminology and a logical flow – in particular for the processes of demonstrating post-closure repository safety, potentially for hundreds of thousands or even millions of years. The set of arguments used to address this task is often labelled ‘the safety case’ (NEA, 2004b; IAEA and NEA, 2006). In Japan, there is a tendency to use a more extended definition of the term ‘safety case’ than is common elsewhere, considering it to be an integration of all arguments that support the licensing requirement that a specific repository is sufficiently safe (e.g. Kawamura and McKinley, 2008). Safety is thus demonstrated during all programme phases – site characterisation, construction, operation, closure/institutional control and post-closure. Emphasis is, however, on ‘sufficient’ safety, recognising not only that complete safety is impossible in any human activity but also that any real project will involve trade-offs between safety during different project phases and also between safety and other global project requirements. In this regard, it is compatible with considerations associated with development of a strategic environmental asessment (SEA), as is now increasingly common for major projects (e.g. http://www.sea-info.net/). The critical question, highlighted in the introduction, is whether advanced KM tools can contribute to the development, review, communication and control of the iterative evolution of safety cases. This has been a focus for JAEA since 2005 and considerable progress has been made in each of these areas. In this section, such progress is overviewed with an emphasis on

© Woodhead Publishing Limited, 2010

Application of knowledge management systems

619

consideration of the extent to which an integrated KMS can be implemented with existing technology. In the following sections more details are provided on the tools investigated and their strengths and weaknesses for particular applications. The starting point for a comprehensive safety case (or SEA) is a definition of the wastes to be disposed of and the project boundary conditions. Here it is important to recognise that such a project will be licensed only in a couple of decades time and will operate until the end of this century – over which time major technical and societal changes are to be expected. Such developments cannot be predicted in detail, but the ‘Think Tank’ approach (Umeki, 2007) has provided a useful tool for focusing consideration of the uncertainties involved and highlighting relevant areas where significant changes are likely. Thus, although the initial focus is on HLW, it is clear that, in the future, there will be more emphasis on other types of waste, including those from future advanced fuel cycles. This requirement has led to initiation of an innovative study of the implications of future nuclear technology developments in terms of waste disposal needs (Miyamoto et al., 2006; Makino et al., 2009a). Once the repository requirements and boundary conditions are defined, a repository concept must be developed. In Japan a rather wide interpretation of the ‘repository concept’ is used, defined (NUMO, 2004) as a conceptual design of all surface and underground repository structures, along with a description of how the repository can be constructed, operated and sealed. This also includes an evaluation of operational and long-term safety and an assessment of environmental impacts and socioeconomic aspects. The concept is dynamic, evolving with the programme as it moves from early generic studies through to siting and, eventually, licensing for construction and operation. Tailoring a repository concept to volunteer sites is a management challenge; this is recognised in the NUMO structured approach (NUMO, 2007), which requires iterative evolution of site characterisation plans, repository designs and associated performance and safety assessments (PA and SA) as the programme develops. NUMO is in the process of developing a requirements management system that will guide this process in a top-down manner. Ideally this would interface to the supporting JAEA KMS but, until NUMO’s system becomes operational, an alternative KE method of providing top-level guidance has been investigated, based on an argumentation model (AM) approach (Fig. 19.2). AMs for overall safety cases and SEAs have been developed using commercial software packages, although a tailored tool for this purpose is under development (see Section 19.8). The advantage of these is that the fundamental arguments that support the safety (or other justification) of the disposal project can be developed in a structured manner, using natural language in a way that makes the overall case easier to follow for non-

© Woodhead Publishing Limited, 2010

19.2 Top levels of an SEA expressed as an argumentation model. The electronic version allows access to further sublevels of detail by clicking on the (+) symbols.

620 Geological repository systems for safe disposal

© Woodhead Publishing Limited, 2010

Application of knowledge management systems

621

specialists – or even members of the general public. The top-level arguments provide context, but allow hierarchical expansion of detail until the level of specialist technical publications is reached. When implemented in an electronic format (internet based), all key supporting documentation can be hyperlinked, to form a powerful and user-friendly resource for technical audiences. For those less technical – or more interested in key arguments – text-based support can be complemented by the use of animations, video and other advanced communication methods (video links, as indicated in Fig. 19.2). This structured approach to access of knowledge facilitates identification of gaps in understanding or in available databases, particularly when arguments and associated counter-arguments are grouped into classes (cf. Section 19.8). To be rigorously implemented, however, it needs to be complemented by a strict quality management system (QMS). Although assurance of quality is accepted as a key goal within the radwaste business, experience in application of conventional QMS approaches within national programmes has been mixed. The ‘ISO 9001’ type of quality assurance focused on processes (http://www.iso.org/iso/iso_catalogue/management_standards/iso_9000_iso_14000.htm) has been shown to provide some benefits, but the assurance of basic data and the process of rigorously deriving system understanding from input information has proven trickier. Here again the fundamental problem is the information explosion, which has caused collapse of the traditional process of technical peer review. The solution being investigated is to focus review only on the ‘kernels’ of knowledge essential to the safety case, with the level of such review evolving along with overall progress towards facility licensing. As the assessment documentation can also be hyperlinked to electronic text (or other linked material) with a simple visual indicator of quality level, this allows the uncertainties associated with the safety case arguments to be easily seen by users and areas where improvements need to be identified. As emphasised above, both the safety case and the repository concept are dynamic entities, evolving with time as work becomes increasingly focused on specific sites and system understanding increases. Tracking key developments in source knowledge can be handled by established change management procedures, but automating the cascading consequences of such changes on the safety arguments requires more active KE tools. For example, instead of presenting model output in reports, the model can be included in an ‘all-in-one’ electronic report, which will automatically rerun the model in response to any changes in the input parameters (see Section 19.9). In the past, much of the dynamic synthesis was carried out by expert teams, based on experience developed over decades. The JAEA KMS has identified preservation of such tacit knowledge and its transfer to younger

© Woodhead Publishing Limited, 2010

622

Geological repository systems for safe disposal

generations as a key task that will utilise a range of existing and novel teaching/training methods (Miyamoto et al., 2006; Makino et al., 2007). However, these need to be complemented by KE techniques, with a focus of tacit knowledge capture and its formalisation within an expert system (ES) approach (see Section 19.9). It is expected that many areas will be suitable for ES support, but development has concentrated on those that involve multi-disciplinary integration of information (in particular geosynthesis of site characterisation data (Semba et al., 2009)). Although development of such ESs involves considerable effort, it is likely to pay dividends immediately after field work in volunteer sites commences due to the very limited reserves of suitably qualified staff in Japan. The knowledge acquisition phase is complementing input from expert teams at JAEA’s field sites with input from foreign experts with relevant site characterisation experience in other national programmes. On the repository design side, it has already been recognised that the basic H12 design, used to demonstrate feasibility of HLW disposal in Japan (JNC, 2000), needs to be modified in order to tailor it to the specific characteristics of volunteer sites (NUMO, 2004). The initial catalogue of design components is, however, rather limited (McKinley and Scourse, 2009) and it is certain that, with a bit of lateral thinking, a wider spectrum of options could be developed. To facilitate this process, advanced KE methodology for innovative problem solving (TRIZ) is being investigated. This will, in turn, require more realistic approaches to quantitative evaluation of performance in order to allow preferred options to be identified. Even with the expanding power of computers, the capacity required to run such realistic models is enormous and hence lateral thinking has led to the assessment of cloud computing methodology as a costeffective solution to this problem. The knowledge base for such a safety case represents the fundamental inventory of information that is needed to meet the requirements of users. As is evident above, the access options allow a departure from traditional structuring of databases and archives, which has focused on grouping information by technical discipline. This can be replaced by flexible structuring on the basis of specific applications, with hyperlinks crossing disciplines as appropriate. In order to maximise applicability, however, a strict ontology has to be developed, which will allow efficient access by smart search engines and utilisation by other software tools (see Section 19.10). Finally, the range of tools and approaches listed above must be integrated and packaged in a form that provides a user-friendly interface for all interested parties. This requirement is rather novel – particularly in terms of the wide range of user technical ability that needs to be considered. A novel web-based system has thus been developed – called CoolRep (the name

© Woodhead Publishing Limited, 2010

Application of knowledge management systems

623

being derived from Coolbiz, a good example of lateral thinking by the Japanese government to reduce peak summer power loads, http://en. wikipedia.org/wiki/Cool_Biz_campaign). Although currently set up as a demonstration (predominantly in Japanese), this is intended to develop into a fully bilingual interface that illustrates application of all KM tools and provides access to the JAEA knowledge base at any level selected by the user. This global interface is complemented by groupware that facilitates the development of the tools and knowledge that will finally be incorporated in CoolRep after appropriate QA (see Section 19.10). Current emphasis is on demonstration of the fundamental feasibility of advanced KE and IT tools. A final decision with regard to full-scale implementation (concept and supporting software) will include assessment of factors such as: . . . . . . . . . . .

ease of initiation and maintenance, assured preservation of information (note that with the timescale of many decades, this is a far from trivial process), security (critically this may be a defined source of information for both license applications and the review of such applications by regulators and other stakeholders), rigorous quality assurance of both data and processes, ability to implement change management automatically, including assessment of consistency/compatibility, autonomic knowledge generation (inference, data mining, etc.), ease of use by a range of users, which includes not only interactive functions that encourage feedback but also the transparency needed to allow the credibility of knowledge sources to be assessed, ability to be interfaced with relevant tools utilised by the user (e.g. electronic requirements management system), flexibility to evolve with time, guaranteed long-term support, established application to similar projects.

Given that the time until a high-level waste repository is licensed for closure is in the order of a century (or more), developing a system with sufficient robustness and longevity will clearly involve some novel challenges. So far, the various components of the demonstration KMS have been evaluated both internally and externally and finally presented at an international workshop in 2008 (http://www.jaea.go.jp/04/tisou/kms/pdf/081111_ws.pdf). Feedback to date has been very positive and reflected to the update of CoolRep demonstration of support of the safety case for geological disposal in Japan (termed H22, http://kms1.jaea.go.jp/CoolRep/).

© Woodhead Publishing Limited, 2010

624

Geological repository systems for safe disposal

19.3 Hierarchy of knowledge creation, synthesis and application for radioactive waste disposal projects.

19.7

Assessment of knowledge engineering and advanced information technology

The knowledge management tasks outlined in the previous section can be structured as shown in Fig. 19.3. Basic knowledge elements are compiled or created through focused R&D in the bottom layer and then synthesised to form a set of goal-oriented knowledge representations at the intermediate level. These include site descriptive models (SDM), repository designs, and assessment models and datasets, together with a framework for total system assessment. This output, in turn, is fed into the top level and used to construct a safety case and SEA, on the basis of a safety strategy. Successful knowledge management within the overall scope depicted in Fig. 19.3 requires optimal organisation and active participation of all project staff, together with application of advanced KE and IT tools. In this context, it should be emphasised that the goal of a KMS is not to automate all knowledge management processes – rather to provide ‘intelligent support’ to the teams involved. In practice, implementation of a KMS involves a cycle in which intelligent systems grow more sophisticated with extended application and the human experts learn how to make the best use of support they provide, resulting in changes to the working environment that provide the basis for a further iteration of tool development. In this section, experience in application of novel KMS tools and methodology within the JAEA project is outlined, focused on their pros and cons for specific tasks. This follows the structure of Fig. 19.3 in a top-down manner.

© Woodhead Publishing Limited, 2010

19.4 Support of the safety case by formal development of arguments and structured utilisation of the knowledge base (Osawa et al., 2009a).

Application of knowledge management systems

© Woodhead Publishing Limited, 2010

625

626

Geological repository systems for safe disposal

19.8

Constructing and visualising safety case arguments for geological disposal of radioactive waste

As discussed in Section 19.6, the safety case as defined in Japan (or an SEA) can be considered as the top-level constraint on all work carried out within a geological disposal project. Knowledge can be classified in terms of its role in the safety case and prioritised in terms of its impact on overall safety case argumentation (Fig. 19.4 (Osawa et al., 2009a)). Use of an argumentation model (AM) to represent key components of a safety case has certain advantages: .

.

.

.

The starting point of the argumentation model is a clear claim in natural language; this could be, at the top level, ‘the safety case justifies proceeding with the repository project at a particular programme milestone’ or, at a lower level, that ‘a particular system component contributes to the safety case’. This puts any lower level input clearly into context. The initial claim must be supported by arguments, which can be usefully classified into different types. At present 18 different classes of argumentation scheme are used: these range from consideration of hard laws of science or defined exclusion criteria to softer assessments of common understanding and requirements for public acceptance. Some, or all, of the arguments may give rise to counter-arguments. The advantage of classifying the arguments is that particular classes of supporting argument lead naturally to counters. For example, in the case of ‘argumentation based on analogy’, some of the critical questions that might be considered would be the possible extent of any sampling bias or the potential existence of analogue counter-examples. These counter-arguments then lead to further supporting arguments, generally going into further technical detail, with links wherever appropriate to the supporting knowledge base. This is continued until the case is considered to be sufficient – which inevitably involves expert judgement. In practice, especially early in a programme, it may not be possible to develop a complete argument – some aspects must be covered by assumptions or result in open questions. These can be highlighted and used to focus and set priorities for associated R&D.

Inevitably, argumentation models become increasingly complex as they go into finer technical detail and rapidly lead to cross-linking between different subsystems. To manage the network of argumentation, software tools are essential and a number of existing packages have been examined. Although these were used to establish fundamental practicality, it was seen that customisation would be beneficial for this particular application.

© Woodhead Publishing Limited, 2010

Application of knowledge management systems

627

In the area of legal argumentation, which has some similarities to a safety case, a number of tools to support construction of argumentation models have been developed and applied extensively (Reed and Norman, 2004; Walton, 2005). Such experience forms the basis for development of a tailored argumentation model tool for radioactive waste disposal safety cases, called Scarab (Osawa et al., 2009a). Scarab (supporting tool for constructing argumentation models with an associated knowledge-base) supports argument development using a simple point-and-click interface. To begin, a ‘threat’, i.e. a negative argument that is highlighted by the system due to lack of valid defence against it, is identified by the user as the starting point of an AM. The user then selects an argumentation scheme to form this defence and fills in the necessary entries specified in a template specific to the argumentation scheme selected. Critical questions associated with this argumentation scheme can then be reviewed and activated whenever necessary. If any of the new counter-arguments generate further threats, the procedure iterates further. The AM thus evolves in a dialectic manner, highlighting potential open questions and gaps in the associated knowledge base. Additional functions of Scarab include: (a) Storing existing AMs in a case base and allowing users to key-word search cases similar to the one at hand. (b) Recording all the revisions made to each argumentation model, with comments explaining the reason for changes. (c) Link with groupware (see Section 19.10), which provides a collaborative internet working environment. (d) Supporting discovery of new counter-arguments by using ‘deep’ knowledge of the safety case structure. The last function is achieved by linking with another tailored tool, called KNetwork 2 (Osawa et al., 2009a). KNetwork 2 visualises the knowledge network as a multi-layered structure in three dimensions so that a user can see the network from various angles by rotating it. Its search engine runs through a subnetwork associated with a ‘target’ argument to find a potential flaw at any level of the ‘deeper’ knowledge behind it. The knowledge network contains basic (‘deep’) knowledge relating to the safety case, split into five layers corresponding to argumentation, safety functions, processes and features, models and data respectively. Knowledge elements, represented as nodes in the network, are linked to each other either in a positive form (e.g. ‘supports’, ‘is consistent with’, ‘describes’) or a negative form (e.g. ‘contradicts’, ‘competes with’, ‘is not consistent with’). Within this network, there is a subnetwork associated with any specific argument (or counterargument) in which all the nodes are linked only in the positive form. The search engine finds such a subnetwork associated with a target argument

© Woodhead Publishing Limited, 2010

628

Geological repository systems for safe disposal

and then identifies any external nodes that are linked to it with negative relationships. These are potential new counter-arguments. To date, top-level AMs have been developed for both examples of complete safety cases and an outline SEA. At a lower level, examples have included assessment of the use of low alkali cement, requirements for the next generation of post-closure PA models and structuring of a site-specific geosynthesis. These are illustrated within CoolRep. Feedback has been very positive, especially from those involved in the process of AM production. It was seen that the networks provide clear linkage of specialist work to particular applications, while the flexibility and efficiency of electronic hyperlinks to supporting documentation was considered very useful to users (at all levels). Less positively, it is noticeable that the AMs rapidly become very complex, with considerable duplication, as they are extended to fine technical details. Hyperlinking sub-AMs would, however, appear to be a potential solution to this problem. Although the concept for an applicable QMS has been developed, it has not yet been implemented as it will involve defining QA levels for linkages (applications) in addition to the quality assessment of the basic knowledge units. It should also be noted that the AMs illustrated in CoolRep use commercial ‘Mindmap’ software as a free downloadable reader is available: such sophistication is not yet implemented for Scarab.

19.9

Compiling, synthesising and organising knowledge

Knowledge required for the top strategic level in Fig. 19.3 is created and synthesised at the bottom and intermediate levels. The starting point is identification and organisation of the knowledge that already exists: as indicated in previous sections, this is a major challenge in itself due to the vast amounts of material available. To solve this problem, use of smart search engines is being investigated. In order to use this approach, it is critical to establish a clear vocabulary or taxonomy in order to allow structured contextual searches that go beyond simple key words. Initial efforts have thus concentrated on a definition of clear ontology and examination of approaches to ‘ontology cleaning’ that have been used in other areas. In this particular area, it is practical to proceed in a stepwise manner – concentrating on some specific topics that are focuses for current R&D. So far three contrasting topics have been examined: prediction of monogenetic volcanism, use of low pH concrete and application of chemical thermodynamic databases in a performance assessment. Apart from providing contrasting information to test supporting tools, such detailed workshops have helped to motivate specialist staff and build up their commitment to the new KMS. In defining specifications for the intelligent search engine, it has to be

© Woodhead Publishing Limited, 2010

Application of knowledge management systems

629

19.5 Knowledge base compilation and search engine functions.

considered that the tool has to be able to access not only intranets and the internet but also archive all relevant material in a secure project-specific database (Fig. 19.5) (e.g. Umeki et al., 2009). This material will include a range of different types of data and information which must be quality assured and synthesised into a useful form – involving procedures that incorporate both explicit and tacit knowledge. In order to automate these procedures, the use of expert systems is being examined.

19.9.1

Expert system development based on explicit knowledge

Shallow knowledge is defined as a characteristic of a problem domain and an expert system (ES) based on it is applicable only in some very specific situations. On the other hand, deep knowledge concerns fundamental laws and can thus be used to solve a wider range of problems. Rule-based and case-based systems are important classes of ES, which handle shallow and deep knowledge in distinctive manners. The basic components of a rule-based system are a knowledge-base containing shallow knowledge represented using an IF (situation) THEN (action) format and an inference engine that enables the ES to make deductions using these rules and apply them to a particular problem (Osawa et al., 2009b). The inference engine applies a set of rules in one of two ways, namely forward or backward chaining. Forward chaining starts with a set of facts (data) and works forward to the conclusions or goals of the system. In backward chaining, the system starts with a hypothesis, which is proven (or disproven) by checking the rules within the domain. Forward chaining is applicable to planning, prediction, trouble-shooting and so on, while

© Woodhead Publishing Limited, 2010

630

Geological repository systems for safe disposal

Table 19.1 Examples of the expert systems that have been developed and applied in the ISIS project Forward chaining

Backward chaining

ES for planning borehole drilling ES for identifying redox pairs ES to support the selection of tracer for controlling the measured Eh values drilling fluid ES for quality evaluation of data on groundwater chemistry

backward chaining is applicable to diagnosis, interpretation, explanation, etc. A number of rule-based systems utilising either forward or backward chaining have been developed and applied in the Information Synthesis and Interpretation System (ISIS) project, a component of the overall KMS focused on the Japanese site characterisation programme (Semba et al., 2009). Table 19.1 summarises some ES examples that have been developed and applied in the ISIS project (Makino et al., 2009).

19.9.2

Intelligent systems based on tacit knowledge

In the field of knowledge management, the concept of tacit knowledge refers to a knowledge that is only known by individuals and that is difficult to communicate to the rest of an organisation. Knowledge acquisition is the process of acquiring knowledge from human experts, mainly through interviews; it is regarded as the first stage of planning knowledge management and designing associated intelligent systems. The knowledge acquisition design system (KADS) and its more recent variant, ‘common KADS’, is the most used methodology within Europe for the development of intelligent systems. The KADS approach includes the following knowledge acquisition activities: . . .

elicitation: eliciting the relevant knowledge, analysis: interpreting the knowledge in a specific context, formalisation: formalising knowledge components or flows so that they can be used in a computer tool.

Past experience with the use of KADS to develop input for an ES in a number of different technical areas has been rather mixed. A commonly encountered problem is that IT specialists have been charged with knowledge acquisition and have struggled to determine the relevance and completeness of input provided by specialists. To avoid this, in the JAEA project a team of experienced generalists in the waste management field have modified the conventional KADS approach, so that output from the

© Woodhead Publishing Limited, 2010

Application of knowledge management systems

631

interviews can be represented directly in a form of AM. This not only facilitates the interviews with specialists but allows completeness of the argumentation chain to be readily assessed. Such output is also a useful guide for the formalisation of the knowledge flow by an ES tool (Makino et al., 2009b; Osawa et al., 2009b). An alternative approach is to use machine learning algorithms to identify patterns embedded in the judgements based on tacit knowledge. Investigations have focused on artificial neural networks (ANNs), in which self-organization determines the connection strengths between a large number of artificial neurons. Manta (machine learning by a neural network for tacit knowledge acquisition) is a generic ANN tool that has been applied to a range of problems appearing at the bottom layer in Fig. 19.3. ANNs tend to excel at problems where a large number of repetitive or similar operations are required; an example currently studied is classification of fractures along boreholes captured by borehole TV.

19.9.3

Integration of multiple approaches

For real problems it is not usually practical to separate out automated and manual tasks or those requiring explicit or tacit knowledge. A multi-faceted hybrid approach is needed, which can be handled by the use of ‘blackboard architecture’ (Kendall and Green, 2007). Here the total work block is broken into modules, which can be handled by any appropriate human or electronic ‘expert’, but provide information in an agreed format so that other modules can use it. To facilitate this, working memory is subdivided into regions of a ‘blackboard’, which is the interface for all communication between modules. In the ISIS project this blackboard approach has been investigated as an option for integrating all the data that flows in during a site characterisation programme, assessing trends via various ES modules and identifying ‘surprises’ or deviations from the expectations based on the characterisation plan and SDM (Semba et al., 2009). The system contains a case-base of alarms, anomalies and inconsistencies that have been experienced in past site characterisation programmes and also allows feedback to be obtained from relevant expert groups. This allows the busy field coordinators, not only to be warned of potential problems but also to provide them with possible causes and past responses to similar situations, together with resultant outcomes. Another approach to system integration is Aladdin (all-in-one advanced document integration system), a development environment for selfcontained e-documents equipped with tools, databases and other resources. An example of an included application for PA is called Pairs (performance assessment all-in-one report system) (Makino et al., 2009b). Figure 19.6

© Woodhead Publishing Limited, 2010

632

Geological repository systems for safe disposal

19.6

Concept of Pairs (Umeki et al., 2009).

shows the basic idea of Pairs, which is based on the observation that PA teams repeat similar sequences of the calculations many times, e.g. as a result of new site data making minor changes in the engineered barrier system (EBS) geometry. Pairs is an application consisting of a report represented as an HTML document, blackboard architecture that provides a shared working memory for the sequence of tasks and a control shell that coordinates tasks to be done by a variety of smart tools. An intelligent user interface facilitates editing of the input dataset, concealing the huge input data matrix, most of which is not relevant to specific changes. This system also allows connection via groupware (see the next section) to allow parallel operation by remote members of a PA team and to expedite discussion among them.

19.10 Facilitation of communication, multidisciplinary collaboration and efficient use of resources A key aim of the JAEA KMS is to facilitate communication, both between specialists in different areas at a technical level and also between specialists, generalists and other stakeholders – in particular the general public. CoolRep is a major interface for such communication (Umeki et al., 2009), which is being continually expanded to facilitate dialogue with users. This is complemented by focused workshops on topics such as the KMS itself, QA, the safety case, etc. In the past such workshops have been documented in a traditional manner and records made available on the JAEA website (e.g. http://www.jaea.go.jp/04/tisou/kms/pdf/081111_ws.pdf). In the future,

© Woodhead Publishing Limited, 2010

Application of knowledge management systems

633

however, key workshops will be recorded on video, which will be made available also via CoolRep. In many cases implementation of KMS is not just adding some new tools to an existing working style, but making radical changes in the overall working environment. For this, applications of groupware (also referred to as collaborative software) are being tested. This provides: .

.

.

electronic conferencing/communication tools that facilitate the sharing of information, such as: ○ internet forums: a virtual discussion platform to facilitate and manage online text messages, ○ videoconferencing, ○ document/application sharing: users can access a shared document or application from their PCs simultaneously, ○ live video streaming from labs, URLs, workshops, etc. Collaborative management tools that facilitate and manage group activities, including: ○ an ontology editor that ensures all the members use the same terminology consistent with the computer transactions, ○ a common knowledge-base, ○ project management systems that schedule, track and chart the steps in a project as it progresses. A gateway to all the KE/IT tools discussed in the previous section.

Groupware could also be expanded to provide ‘computer-supported collaborative learning’, an option that will be considered in the future. Finally, from a pragmatic point of view, it has been observed that the transition towards more active linking of computer-based work in different locations also provides opportunities to improve the use of resources. This is currently being investigated for the specific case of ‘cloud’ or ‘grid’ computing, which allows the use of spare capacity on an internal network of PCs to be used to run power-intensive programmes (the best-known example of this is seti@home, http://setiathome.ssl.berkeley.edu/).

19.11 Future trends The content of this chapter can possibly be best summarised by the ‘big messages’ presented in the introduction to the JAEA KMS review workshop (http://www.jaea.go.jp/04/tisou/kms/pdf/081111_ws.pdf): 1.

Based on the volume of information to be handled, it is no longer a question of whether advanced KMS will be introduced into radwaste management programmes or not – only whether such systems can be

© Woodhead Publishing Limited, 2010

634

Geological repository systems for safe disposal

developed and implemented before total collapse of conventional approaches! 2. The Japanese decision to move rapidly into advanced KM was driven by the boundary conditions of the national programme. In particular, imminent loss of many experienced staff as they retire leaves only a small window to capture tacit knowledge (a common problem throughout the nuclear industry). 3. The problem identified is so fundamental that modification of conventional approaches offers no chance of solving it – a complete paradigm shift is required, which emphasises structured processes and use of advanced IT and KE methodology. 4. The developments envisaged are ambitious but appear feasible based on current technology. The individual components needed all exist – they have just never been combined and implemented in the manner planned. 5. The most important resource of any radwaste organisation is its expert manpower. Specialists can be trained using existing structures, but generalists must be cultured, using a combination of tailored training and focused ‘on the job’ transfer of experience from mentors. Project planning must take this critical goal into account! 6. Any KMS development project must have clear applications, deliverables and milestones. The safety case provides a top-level focus for all of the work that is carried out in national programmes (note that this includes both technical and sociopolitical issues as defined in Japan) and allows knowledge creation requirements to be prioritized. 7. Quality management must be implemented in a focused and effective manner that encourages all those involved to adopt a quality culture. KM tools should minimise the effort of QA while making benefits clear to users. 8. Long-term planning can only be carried out with an appreciation of the inherent volatility of project boundary conditions (both technical and sociopolitical): exercises such as the Think Tank can introduce such perspectives to specialists and younger staff. 9. Geosynthesis and next-generation PA are selected as a topic for testing of KMS technology, not because they are easier but because they encompass all the challenges of application to integrated safety case development within more restricted topical areas. 10. JAEA ‘puts its money where its mouth is’ in the status report (H22), which not only describes the key KMS concepts required for nextgeneration safety cases but also incorporates these in the document itself (CoolRep). The case study has shown how the initial work by JAEA has shown the

© Woodhead Publishing Limited, 2010

Application of knowledge management systems

635

basic applicability of advanced KMS technology and indicated how this will develop further in the Japanese programme. As, however, many of the challenges are generic, it is hoped that the initial interest shown by other national waste management organisations will develop into international collaboration, which will provide more resources to accelerate the production, testing and implementation of a fully integrated KMS that can serve, at least, until the next stage of repository licensing.

19.12 Sources of further information and advice Since advanced KMS development for radioactive waste management is a recent initiative, there is little specific documentation available at the moment. However, there is an extensive literature on argumentation models, knowledge management, development of KMS, etc., in other sectors. Following is a brief list of references particularly relevant to points covered in this chapter.

19.12.1 . . .

Kendal and Green (2007) Schreiber et al., (2000) Stefanuk and Kaijiri (2004)

19.12.2 . . . . . .

Argumentation

Toulmin (1958) Walton (2005) Van Eemeren and Grootendorst (2004) Kirschner et al. (2003) Reed and Norman (2004) Makau and Marty (2001)

19.12.3 . . . . . .

General knowledge management and knowledge engineering

Expert system and reasoning

Magnani (2001) Magnani and Li (2007) Magnani (2006) Jones (2008) Hu¨llermeier (2007) Li and Du (2008)

© Woodhead Publishing Limited, 2010

636

Geological repository systems for safe disposal

19.12.4 . .

Groupware and search engines

Churchill et al., (2002) Spink and Zimmer (2008)

19.13 Acknowledgements The authors wish to thank Dr I. G. McKinley for support in manuscript preparation and many colleagues at JAEA for their invaluable contribution to development of the concepts outlined herein. Also special thanks to the national and international participants in the JAEA KMS review and QA workshops.

19.14 References Churchill F, Snowdon D N and Munro A J (eds) (2002), Collaborative Virtual Environments; Digital Places and Spaces for Interaction, Springer. Hu¨llermeier E (2007), Case-Based Approximate Reasoning, Springer, Netherlands. IAEA and NEA (2006), Geological Disposal of Radioactive Waste: Safety Requirements, IAEA Safety Standards Series WS-R-4, International Atomic Energy Agency, Vienna, Austria. JNC (2000), ‘H12: Project to establish the scientific and technical basis for HLW disposal in Japan, Project Overview Report, Second Progress Report on research and development for the geological disposal of HLW in Japan’, JNC TN1410 2000-001, Japan Nuclear Cycle Development Institute (now Japan Atomic Energy Agency). JNC (2005), ‘H17: Development and management of the technical knowledge base for the geological disposal of HLW – Knowledge Management Report’, JNC TN1400 2005-022, Japan Nuclear Cycle Development Institute, Tokai, Japan. Jones M T (2008), Artificial Intelligence: A Systems Approach, Infinity Science Press, Hingham. KASAM and IAEA (2000), ‘Retrievability of high level waste and spent nuclear fuel’, in Proceedings of the International Seminar, 24–27 October 1999, Swedish National Council for Nuclear Waste and International Atomic Energy Agency. Kawamura H and McKinley I G (2008), ‘Integrated safety case development for deep geological repositories’, in Proceedings of the International Conference on Underground Disposal Unit Design and Emplacement Processes for a Deep Geological Repository, Prague, 16–18 June 2008, ESDRED. Kawata T, Umeki H and McKinley I G (2006), ‘Knowledge management: the Emperor’s new clothes?’, in Proceedings of the 11th International High-Level Radioactive Waste Management Conference 2006, Las Vegas, Nevada, 30 April–4 May 2006, pp. 1236–1243. Kendal S L and Green M (2007), An Introduction to Knowledge Engineering, Springer-Verlag, London.

© Woodhead Publishing Limited, 2010

Application of knowledge management systems

637

Kirschner P A, Buckingham Shum S J and Carr C S (2003), Visualizing Argumentation, Springer-Verlag, London. Li D and Du Y (2008), Artificial Intelligence with Uncertainty, Chapman & Hall/ CRC, Boca Raton, Florida. McKinley I G (2009), ‘21st century challenges and opportunities for nuclear waste management’, in Proceedings of INC’09. McKinley I G and Scourse E (2009), ‘Materials aspects of advanced repository concepts for higher toxicity waste’, in Proceedings of MRS09. Magnani L (2001), Abduction, Reason, and Science: Processes of Discovery and Explanation, Kluwer Academic/Plenum Publishers, New York. Magnani L (2006), Model Based Reasoning in Science and Engineering, College Publications, London. Magnani L and Li P (2007), Model-Based Reasoning in Science, Technology, and Medicine, Springer-Verlag, Berlin. Makau J M and Marty D L (2001), Cooperative Argumentation: A Model for Deliberative Community, Waveland Press, Long Grove. Makino H, Osawa H, Nakano K, Naito M, Umeki H and McKinley I G (2007), ‘Concept and design of the JAEA KMS for geological disposal of HLW’, in Proceedings of Global 2007. Makino H, Umeki H, Hioki K and McKinley I G (2009a), ‘The challenge of development of a holistic waste management approach to support the nuclear renaissance’, in Proceedings of the WM2009 Conference, 1–5 March 2009, Phoenix, Arizona. Makino H, Hioki K, Umeki H, Yang H, Takase H and McKinley I G (2009b), ‘Practical applications of the KMS: 1. Total system performance assessment,’ in Proceedings of the 12th International Conference on Environmental Remediation and Radioactive Waste Management, ICEM’09, Liverpool, UK, 11–15 October 2009. Miyamoto Y, Umeki H, Ohsawa H, Naito M, Nakano K, Makino H, Shimizu K and Seo T (2006), ‘Key R&D activities supporting disposal of radioactive waste: responding to the challenges of the 21st century’, Nuclear Engineering and Technology, Korean Nuclear Society, 38(6), 505–534. NEA (2001), Reversibility and Retrievability in Geologic Disposal of Radioactive Waste: Reflections at the International Level, OECD Nuclear Energy Agency, Paris, France, ISBN 92-64-02077-2. NEA (2004a), Stepwise Approach to Decision Making for Long-term Radioactive Waste Management: Experience, Issues and Guiding Principles, OECD Nuclear Energy Agency, Paris, France, ISBN 92-64-02077-2. NEA (2004b), Post-closure Safety Case for Geological Repositories – Nature and Purpose, OECD Nuclear Energy Agency, Paris, France, ISBN 92-64-02075-6. NUMO (2004), ‘Repository concepts for volunteer siting environments,’ NUMOTR-04-03. NUMO (2007), ‘The NUMO structured approach to HLW disposal in Japan: staged project implementation at volunteer sites utilising a requirements management system’, NUMO-TR-07-02, Nuclear Waste Management Organization of Japan. Osawa H, Hioki K, Umeki H, Takase H and McKinley I G (2009a), ‘Use of the safety case to focus KMS applications’, in Proceedings of the 12th International

© Woodhead Publishing Limited, 2010

638

Geological repository systems for safe disposal

Conference on Environmental Remediation and Radioactive Waste Management, ICEM’09, Liverpool, UK, 11–15 October 2009. Osawa H, Umeki H, Ota K, Sawada A, Takeuchi S, Semba T, Takase H and McKinley I G (2009b), ‘A structured approach for stepwise development of next-generation technology for integrated site characterization of deep geological repositories’, in Proceedings of the WN2009 Conference, 1–5 March 2009, Phoenix, Arizona. Reed C and Norman T J (eds). (2004), Argumentation Machines: New Frontiers in Argument and Computation, Kluwer Academic Publishers, ISBN: 978-1-40201811-4. Schreiber G, Akkermans H, Anjewierden A, Hoog R, Shadbolt N, Velde W V and Weilinga B (2000), Knowledge Engineering and Management: The Common KADS Methodology, The MIT Press, London. Semba T, Osawa H, Hioki K, Tachibana S, Takase H and McKinley I G (2009), ‘Practical application of the KMS: 2, site characterisation’, in Proceedings of the 12th International Conference on Environmental Remediation and Radioactive Waste Management, ICEM’09, Liverpool, UK, 11–15 October 2009. Spink A and Zimmer M (2008), Web Search: Multidisciplinary Perspectives, Springer-Verlag, Berlin. Stefanuk V and Kaijiri K (2004), ‘Knowledge-based software engineering’, in Proceedings of the Sixth Joint Conference on Knowledge-Based Software Engineering, IOS Press, Amsterdam. Toulmin S E (1958), The Uses of Argument, Cambridge University Press, New York. Umeki H (2007), ‘A challenge for computing in the 21st century: radwaste knowledge management’, in Proceedings of the Joint International Topical Meeting on Mathematics and Computation and Supercomputing in Nuclear Applications (M&C + SNA 2007), Monterey, California, 15–19 April 2007, on CD-ROM, American Nuclear Society, LaGrange Park, Illinois. Umeki H, Osawa H, Naito M, Nakano K, Makino H and McKinley I G (2008), ‘Knowledge management: the cornerstone of a 21st century safety case’, in Proceedings of the OECD-NEA/EC/IAEA International Symposium on Safety Cases for the Deep Disposal of Radioactive Waste: Where Do We Stand?, Paris La De´fense, 23–25 January 2007. Umeki H, Hioki K, Takase H and McKinley I G (2009), ‘Overview of the JAEA knowledge management supporting implementation and regulation of geological disposal in Japan’, in Proceedings of the 12th International Conference on Environmental Remediation and Radioactive Waste Management, ICEM’09, Liverpool, UK, 11–15 October 2009. Van Eemeren F H and Gootendorst R (2004), A Systematic Theory of Argumentation, Cambridge University Press, Cambridge. Walton D (2005), Argumentation Methods for Artificial Intelligence in Law, Springer, Hidelberg, ISBN: 978-3-540-25187-3.

© Woodhead Publishing Limited, 2010

20 Radiation protection principles and development of standards for geological repository systems M . J E N S E N , Swedish Radiation Safety Authority, Sweden

Abstract: This article describes the development of the Swedish standard for geological disposal and some of the rationale behind its formulation. The choice between dose and risk in a health-based standard is described and arguments are given against a too strict and exclusive reliance on risk and probability. The article discusses time scales and cut-off times for a standard, and the additional concepts of optimization and the use of best available techniques. Finally, a recently suggested standard is mentioned, based primarily on best available technology (BAT), using dose or risk as complementary safety indicators. Key words: Swedish standard, geological disposal, nuclear waste, risk, BAT.

20.1

Introduction

This chapter aims to describe the development of the Swedish post-closure regulations for geological disposal, promulgated by the former Swedish Radiation Protection Authority, SSI, and to comment on the regulations by explaining the value, as well as the need for interpretations of radiation protection principles when applied to geological disposal. The issue is treated primarily as a scientific matter, but the reader is cautioned that a comparison between different national standards must also take societal factors into account. Standards are decided within the national legal framework, which may influence both form and content. For example, dose or risk may refer to individuals or groups with different characteristics in different standards. 641 © Woodhead Publishing Limited, 2010

642

Geological repository systems for safe disposal

Table 20.1

SSI and SKI major regulatory activities

SKI

SSI

Safety of nuclear installations Waste fund appropriation Non-proliferation

Radiation safety for workers and patients Radiation protection for the public

Although the term ‘standard’ is not used in the Swedish legislative framework for waste disposal as in the US, the regulations (SSI, 1998) represent the Swedish post-closure geological disposal standard for all radioprotection purposes, and the term is used in the following to designate health-based regulatory requirements for post-closure safety of geological disposal. The term ‘geological repository’ covers a range of disposal categories, from medium depth disposal facilities to repositories for longlived, high-level radioactive waste such as spent nuclear fuel. The focus in this chapter is on the latter. SSI has explained some of the rationale behind its standard (SSI, 2000) and in a few places in the general guidelines on the application of the standard (SSI, 2005), in the following called the guidelines. The author is responsible for other explanatory comments, some of which are based on contributions in the literature made after the publication of those documents. The chapter is concerned with radiation protection. It does not address other aspects of the safety assessment or its methodology, or the regulations of the Swedish Nuclear Power Inspectorate, SKI. In Table 20.1, the term ‘safety’ is primarily connected to barrier integrity. Before the merger of the two authorities the radiation health consequences related to the potential loss of a barrier’s function were considered a radiation protection issue. The fact that barrier integrity in the final analysis is a matter of radiation health was an unwavering subject of discussions between the authorities. The two Swedish authorities SSI and SKI ceased to exist on 1 July 2008 when they merged to create the Swedish Radiation Safety Authority. They are referred to here by their former names, SSI and SKI.

20.2

Understanding safety of geological disposal

There is no simple and quantitative definition of safety, a fact sometimes underlined by the saying ‘safety is a warm and fuzzy feeling’. The concept is therefore open to interpretation, both political and technical, which can be seen in the Swedish example from the 1970s, where a law from 1977 required that the nuclear industry must present a ‘completely’ or ‘absolutely’ safe disposal method. The requirements were considered technical and political in proportions

© Woodhead Publishing Limited, 2010

Radiation protection principles and development of standards Table 20.2

643

Steps in the development of the Swedish standard

1976 The government declaration requires ‘completely safe’ disposal of nuclear waste. 1977 The ‘Stipulation’ law. A parliament committee preparatory document suggests that ‘completely safe’ should not be interpreted in a ‘draconian’ way. 1984 The Nuclear Activities Act. Safety issues are decided by authority decision. No numerical standard for post-closure safety exists. 1986 Every third year SKB´s work and plans are submitted for review. SSI and SKI comment on SKB’s work and formulate expectations of SKB´s safety reporting. 1998 SSI’s standard is promulgated. It contains a risk limit, requirements on protection of the environment, optimization and the use of best available technology (BAT).

that shifted with time, and the final decision (regarding interpretation of completely safe disposal) was taken in the outcome of the nuclear referendum 1980 (Sundqvist, 1991). Table 20.2 describes some steps in the development of the Swedish standard. The heated debate about the Stipulation Act’s requirement ‘completely safe’ showed how expert judgements and political initiatives were mixed in a way that was incomprehensible both for those directly involved and for those who followed the debate from the outside. With the benefit of hindsight we can see how much the historical scenario underlined the value of standards. Little international guidance was available at this time. Circles within the International Commission on Radiological Protection (ICRP) were not concerned with the issue or took the view that geological disposal was a matter mainly of political acceptance. The first publication on disposal appeared later (ICRP, 1985), but the more complex issues dealt with in the debate were not addressed until 20 years later, in Publication 81 (ICRP, 2000). The ICRP Publication 81 updates and clarifies some material from publication (ICRP, 1985), and the new general recommendations (ICRP, 1990), on the two concepts potential and prolonged exposure. It also comments on scenarios involving earthquakes and glaciation, and the treatment of the biosphere. The US Energy Policy Act mentioned below appeared 15 years later.

20.2.1 The US National Academy of Sciences study In 1995, the US National Academy of Sciences (NAS) published their findings from a project funded by the US Environmental Protection Agency, EPA, related particularly to a repository in Yucca Mountain, initiated by a congressional decision in the Energy Policy Act of 1992 (NAS, 1995). It

© Woodhead Publishing Limited, 2010

644

Geological repository systems for safe disposal

turned out to be a very successful political–scientific collaboration. The NAS answered several questions posted by the US Congress, one of them by accepting health-based standards, with requirements on dose and risk, and recommended risk as the preferred regulatory quantity. The present form of the standard was promulgated by the EPA (2008). ‘Dose’ and ‘risk’ in standards are understood in the following to be dose or risk rates, e.g. dose or risk per year. Dose is defined in ICRP’s publications (ICRP, 2007a, 2007b), and risk is discussed further in the next sections. The NAS stated (NAS, 1995, p.125) that ‘imposing subsystem performance requirements may lead to a suboptimal repository design’ and such requirements are not included in the standard. In Sweden such requirements also occurred in the debate of 1978, in the absence of formal regulations. Note that predetermined subsystem regulatory requirements are different from requirements such as ‘safety functions’ (a term used by the Swedish Nuclear Fuel and Waste Management Company, SKB), introduced by the operator for purposes of design and analysis. Regulating with subsystem performance criteria goes far beyond the use of an operator’s analytical tool. It introduces an absolute and restraining control on the operator’s process without full knowledge of consequences for the total system performance. It denies the operator the option to compensate for a weakness in one barrier function by strength in another. Also, such requirements have a built-in risk of infringing on freedom for the operator to choose optimal alternatives (the objection stated in the NAS report). ICRP accepts the application of its system of radiation protection to geological (and other types of radioactive waste) disposal in its Publication 81 (ICRP, 2000), and recommends a constraint of no more than about 0.3 mSv per year for members of the public. This recommendation is generally accepted and can be found in the safety requirements of the International Atomic Energy Agency (IAEA, 2006). Today, regulators in many countries accept ICRP’s use of dose as a figure of merit for repository safety, although with somewhat different views and interpretations of the role of this requirement in the compliance demonstration. A guidance document in the UK (NRPB, 1992), intended for intermediate-level radioactive waste disposal, also made use of a risk for compliance. In Sweden, SSI’s regulation SSI FS 98:1 (SSI, 1998), produced a quantitative criterion in the form of a risk per year criterion. We conclude that dose or risk plays at least some role in most standards.

20.2.2 The impact of the NAS report in Sweden The NAS report’s recommendations were in line with radiation protection experts’ views in Sweden, recommending steering principles related to the

© Woodhead Publishing Limited, 2010

Radiation protection principles and development of standards

645

surface environment effects rather than barrier – i.e. subsystems – performance (Jensen et al., 1995). An impact also occurred in connection with the balance between SSI and SKI on the influence on the licensing process. The form of the standard – called ‘health based’ in the Energy Policy Law – suggested in itself that the regulations’ ultimate goal was radiation protection. This could be interpreted as support for SSI’s work towards a waste disposal post-closure standard, formulated in radiation protection terms. Such a standard was promulgated by SSI (1998). A formal incongruence was often felt between SKI, with a preparatory mandate to propose governmental decisions in most nuclear license matters, and SSI, armed with a quantitative radiation protection standard, a situation that came to an end on 1 July 2008 with the merger of the two authorities.

20.2.3 A Swedish radiation protection standard The Swedish standard (SSI, 1998) was promulgated under the name ‘Regulations on the Protection of Human Health and the Environment in Connection with the Final Management of Spent Nuclear Fuel and Nuclear Waste’. SSI’s criteria were: . . . .

The individual risk per year of harmful effects is constrained to 106. The required use of optimization and best available technique (BAT). Protection of the natural environment, i.e. non-human species, is required. Some non-quantitative requirements are included relating to reporting of human intrusion scenarios.

There is no cut-off time period specified, but reporting is required for two periods: . .

For the first thousand years following closure of a repository, a quantitative analysis is expected. For a period after the first thousand years following closure of a repository, ‘the assessment of the repository’s protective capability shall be based on various possible sequences for the development of the repository’s properties, its environment and the biosphere’.

The term scenario is defined in the guidelines as ‘a description of the development of the repository given an initial state and specified conditions in the environment and their development’.

© Woodhead Publishing Limited, 2010

646

Geological repository systems for safe disposal

20.2.4 Protection of non-human species Protection of non-human species has a place in some post-closure standards. The willingness of ICRP to take on work in the area testifies to the relevance of the subject as a radiation protection discipline (ICRP, 2003). Whether or not protection of non-human species is addressed in a postclosure standard for waste disposal depends very much on the national legislation. Protective legislation of some kind exists in most countries, but not necessarily connection with waste disposal. It is contained in the Swedish standard, but internationally this provision is more an exception than a rule. A number of approaches have been made towards assessment of impact on non-human species from ionizing radiation, e.g. in the US, Canada and the UK (Larsson, 2008). ICRP’s new general recommendations (ICRP, 2007b) acknowledge the need for a separate protection system for nonhuman species. An approach has been made in recent EC-funded projects from 2004 to 2007 (EC, 2004) centred on a reference organism approach to environmental assessment. The initiative can be found via the link www. erica-project.org. The Swedish standard includes requirements of both biodiversity and protection of population. In a total system performance assessment, a so-called TSPA, concentration of radionuclides in the environment will usually be an intermediate result before the individual dose (to humans) is assessed. This result could also be used to assess other environmental effects. It is therefore possible to envisage that, in the future, a common standard for both humans and nonhuman species could be based on an environmental concentration of contaminants. Such a standard endpoint would require that the regulator took some responsibility for the last links in the calculation chain, i.e. the assessment of dose or effect from a known, and suitably defined, environmental dose concentration. In the following, risk is understood to designate risk to humans.

20.3

Dose and/or risk in geological repository systems

20.3.1 Definition of risk The author uses risk, in parallel with probability, like the NAS’s risk definition from the glossary of the NAS (1995, p.195). Risk is thus the probability of adverse health effects for an individual, attributable to exposure from radioactive substances from a repository. It is made up of two components, (1) the probability of receiving a dose and (2) the probability of adverse health effects from that dose.

© Woodhead Publishing Limited, 2010

Radiation protection principles and development of standards

647

Table 20.3 Detriment-adjusted nominal risk coefficients for stochastic effects after exposure to radiation at a low dose rate, whole population, in 102 Sv1 (from ICRP, 2007b, Table 1) Year

ICRP Publication Number

Total

1990 2007

60 103

7.3 5.7

20.3.2 The Swedish risk standard In the development of the Swedish standard, the discussion within the SSI initially evolved around a dose limit. The dose, understood to be the expected or expectation value of dose, was converted to a risk limit using ICRP’s nominal risk coefficients. The decision in favour of risk was influenced by the use of risk limitation in other societal regulatory activities, in connection with environmental protection and occupational health. The NAS report brought up an argument against the use of a dose in a standard, namely that the best estimate of the conversion factor might change over time. In the time between the NAS report of 1995 and the expected license applications for a high-level waste disposal site in Sweden (and the US) only one such revision by ICRP has taken place. The change of nominal risk per unit dose from ICRP (1990) to ICRP (2007b) is around 20%, from 7.3 to 5.76102 Sv1 (see Table 20.3). The NAS statement is correct but the changes involved are so far very modest in relation to the uncertainties one expects to find in a safety assessment. The conclusion is that, if ‘the probability of getting a dose’ is taken into account in the standard, as the NAS recommends, then dose- and risk-based standards are very similar, except for the obvious difference in wording. Still, dose versus risk in the formulation of a post-closure standard is a constant source of consternation for laymen and experts alike. This is perhaps because none of the two types of standards escapes the problems of the unknowable future. The problems are similar but they appear in different forms: .

.

A regulator applying a dose standard must take into account the fact that low probability events leading to high doses can be imagined, and the regulator must reflect this fact in some way, or to quote the NAS (1995): ‘the standard should include consideration of the probabilistic aspect of future exposures’. A regulator applying a risk standard has access to a tool that takes probabilities into account, both for risk per unit dose and for scenarios. The problem is that the probabilities the regulator needs, and has vowed

© Woodhead Publishing Limited, 2010

648

Geological repository systems for safe disposal to use, may not always be available. The risk tool may therefore promise too much.

The next sections give a background to this and explain how the regulator in Sweden has dealt with the problem.

20.3.3 The Swedish standard’s dose or risk limit The SSI (2000) explains how the numerical value in SSI’s criteria was derived from a number of steps of limitations, where (in dose terms): 1

mSv per year is ICRP’s recommended limit for individual members of the public from all sources. 0.1 mSv per year is seen as a planning target for a single activity. This could also be accepted for the total risk burden from several repositories and therefore close to ICRP’s concept ‘risk limit’, a limit from many sources (of repositories). 0.01 mSv per year is a target for the impact of a single repository. This is called a source-related constraint in ICRP’s terminology. A repository is a result of an activity that may continue for a long time, and it cannot be ruled out that individuals in the future may receive dose contributions from more than one repository, hence the last step from 0.1 to 0.01 mSv per year. The difference of 10 between the latter two does not imply 10 repositories at the same site but is somewhat more complicatedly related to the spatiotemporal distribution of the exposure, as discussed below. SSI’s guidelines introduce a dose span to use for the most exposed group: One way of defining the most exposed group is to include the individuals that receive a risk in the interval from the highest risk down to a tenth of this risk. If a larger number of individuals can be considered to be included in such a group, the arithmetic average of individual risks in the group can be used for demonstrating compliance with the criterion for individual risk in the regulations. One example of such an exposure situation is a release of radioactive substances into a large lake that can be used as a source of drinking water and for fishing. If the exposed group only consists of a few individuals, the criterion of the regulations for individual risk can be considered as being complied with if the highest calculated individual risk does not exceed 105 per year. An example of a situation of this kind might be if consumption of drinking water from a drilled well is the dominant exposure path. In such a calculation example, the choice of individuals with the highest risk load should be justified by information about the spread in calculated individual risks with respect to assumed living habits and places of stay.

© Woodhead Publishing Limited, 2010

Radiation protection principles and development of standards

649

Observe that the criterion 106 per year is not changed in the second example. It is applied to a group where only a small number of people are affected by the highest risk, 105 per year. The risk corresponds to a dose of about 0.1 mSv per year (for a scenario probability = 1). The two cases are needed, since a person exposed to risk from several regional repositories could be the reasonably maximally exposed individual (RMEI) of both, if the releases in both cases were to a large lake. This would be less likely in the second case. In both cases a representative person’s risk (taken as the mean of a larger group) would be below 106 per year. The planned repository in Sweden represents waste from only some tens of years´ production of nuclear energy. If repositories were built over hundreds (or thousands) of years, individuals in the public could be exposed to a type of release that could not be stopped in the way that an operating installation can. The assumption about 10 repositories should not be taken too literally; it simply represents an intention to cover a situation with a steady state risk load, i.e. to let the standard in an explicit way allow for sustainable development, or ‘development that meets the needs of the present without compromising the ability of future generations to meet their own needs’ (UN, 1987).

20.4

Probability and risk in geological repository systems

In this section, the author mainly wants to explain that (1) safety assessment is a study of a theoretical concept, a model, rather than predicting the repository’s real future, and that (2) we cannot expect any formal proof of safety based on an exact risk estimate. In our effort to do the best under these difficult circumstances, we need both to examine them in detail and to take initiatives to improve the situation. We address the ‘hardware’ of radiation protection, dose and risk to be discussed faithfully to the ICRP concepts and recommendations, in a first attempt. Intuitively, it is obvious that no strict or mathematical proofs can be given of safety. Protection principles must be applied to exposure situations taking place a hundred thousand or a million years into the future. The exposure appears in an unknown biosphere that may be changed by distant-future-man, a creature of unknown habits and aspirations, whose dose or risk still must be addressed. It is therefore often said that uncertainties forbid a strict regulation. However, the existence of large uncertainties alone does not constitute a reason to abandon traditional concepts. Uncertainties – and in some cases large uncertainties – occur in all areas of radiation protection. The SSI (2000) explains that ‘In many cases it is not possible to calculate an ‘‘exact’’ risk.’ Although the SSI did not in its

© Woodhead Publishing Limited, 2010

650

Geological repository systems for safe disposal

regulation and regulatory guidance intend to rely on dose or risk in a strict and exclusive way, it is valuable to maintain a bookkeeping of the reasons, some of which are offered below, for abandoning their strict use as they arise.

20.4.1 The regulator and operator included in a societal contract Objections and difficulties such as those introduced below could be raised for any industrial activity. In the end, i.e. in its compliance assessment, the regulator must inevitably incorporate a measure of subjective judgement. Since no ‘mathematical’ proofs of safety can be deducted from the operator’s application, or from the regulatory requirements, an important question for the regulatory body is whether it has done its best in its promulgated standard. There is no doubt – in spite of the arguments presented below against the strict and exclusive use of risk – that if a certain scenario had constituted an obvious concern by the regulator in this process, and was found to have consequences in excess of the risk limit, that the scenario alone would constitute grounds for rejection by the regulator, and subsequently (in the author’s opinion) by the government. (In Sweden, most nuclear licensing decisions are made by the government, not by the Radiation Safety Authority.) Risk is therefore still a viable concept in post-closure waste disposal regulation. OECD’s Nuclear Energy Agency (NEA, 2004) mentions that ‘doses and risks calculated on the basis of stylized approaches should be interpreted as illustrations based on agreed sets of assumptions for particular scenarios and not as actual measures of future health detriments and risks’, underlining both a limitation of dose and risk results, and the assessment activity as a study of models based on assumptions that have been agreed. In a compliance demonstration a similar agreement must exist between the regulator and operator as a result of the licensing dialogue.

20.4.2 The alpha and omega of risk Firstly, the focus will be the beginning, the ‘alpha’ of risk, the scenario ‘entrance’ probability, to use a German term (Eintrittswarscheinlichkeit), and below the last component is discussed, the ‘omega’ of risk, i.e. who should be considered at risk. Most of the issues in between are left to the performance assessment experts, and described elsewhere in this book.

© Woodhead Publishing Limited, 2010

Radiation protection principles and development of standards

651

20.4.3 Objections to the use of probability A general philosophical problem with probability, and therefore also with risk, is the objection that has to do with its objective existence. Bruno de Finetti has used the motto ‘probability does not exist’ in his famous work Theory of Probability (de Finetti, 1975). Probability does exist in the sense that a coherent system of probabilities can, and has been (Kolmogorov, 1956), defined by a number of axioms. The problem is whether there is a formal connection of probability to our world, i.e. whether ‘there is a property of the physical world that is in direct correspondence to the theoretical concept of probability’ (Dawid, 2004). Of the two components of risk mentioned above (scenario probability leading to a dose and probability of adverse health effects from the dose), the reservations presented are most relevant for the scenario probability. The second component of risk, the probability of adverse health effects from a dose as given by the ICRP, is perhaps less controversial since ICRP’s system of risk estimates is mainly used as a model, i.e. not necessarily a description of absolute truth but a formalism deemed appropriate for radiation protection purposes. The objection against a strict and exclusive use of risk is thus that disposal regulations must ultimately refer to the development of the repository, i.e. real objects, not models. Models, with attached probabilities, are not reality, not even celebrated models such as quantum mechanics. Michael Audi (1973) explains that ‘To say that the results/from the uncertainty relations/ are inherent in the nature of measurement is to lend our present theories a timeless quantity which, considering the nature of theories, they could not possess.’ Actually, some real life facts like radioactive decay do seem to display some of the properties Dawid (2004) is referring to. Regulators also accept without dispute the ‘scenario’ of exponential decrease in activity with time, but it is an admission that covers only a modest part of the safety assessment for long-lived waste, and it is not directly related to scenario probabilities (unless a single decaying nucleus were to trigger a special sequence of events).

20.4.4 The clairvoyant test Another potential problem with risk is related to the issue that a regulator must assess the performance of a repository in the distant future. The problem can be analysed using the so-called clairvoyant test, implying that a flawless assessment of compliance could be made if a regulator had access to information from the distant future. Let us assume that we were informed by a clairvoyant that the repository suffered no releases at all through all times up to the end of the compliance period. One could then still argue that

© Woodhead Publishing Limited, 2010

652

Geological repository systems for safe disposal

it might have been nearly affected by a disruptive event, e.g. an earthquake that luckily did not occur. In this case perhaps the risk was still too high. You could therefore argue that even the power of a clairvoyant would not be enough to assess compliance. The safe return from a journey does not tell us anything about the risks involved (leaving aside de Finetti’s concern, assuming they exist at all). The author has not come across a practical situation, or even a discussion within the so-called performance assessment community, where the definition of probability was considered in safety assessment for repositories. The theme is more explicit in the discussion of reactor safety studies, e.g. by Apostolakis (2004). It seems unreasonable that there should be a major difference in risk perception between the two disciplines. Many problems in risk assessment for waste disposal are probably connected to the fundamentals of the probability concept. They are just not explicit in the discussion.

20.4.5 Dilution and intergenerational distribution of risk One other issue related to risk is worthwhile to mention, intergenerational risk distribution, sometimes called risk dilution. The concept is used to describe a situation where we assume that a dose is received by an individual in the future, covering many generations. The problem arises because we are not used to applying risk to a group made up of individuals from a series of generations, especially not for hundred thousand or a million years into the future. Suppose an earthquake disturbing the repository would lead to a considerable risk to someone if it occurred, but the yearly risk of the earthquake happening was one in a million. If the assessment period is a million years, someone has a more than even chance of getting a high dose, but we cannot know who, during what time. Does the risk limit apply to ‘the unlucky’ or to the average person in any generation? The two different cases have been attributed by Baltes and Ro¨hlig (2004) to the perspective of different actors, ‘the culprit’ versus ‘the victim’. A proponent of the first type of interpretation may complain that risk is ‘diluted’ in the second. The ambiguousness of risk is not limited to drastic events. A similar phenomenon also appears in the interpretation of results from the technique of probabilistic assessment of a repository’s performance. In the assessment, a series of calculations of the repository’s development are made, each using data parameters, randomly sampled from a given distribution, a so-called Monte Carlo calculation. Mohanty and Codell (2004) point out that two alternatives are possible for a presentation of probabilistic assessment, ‘(1) the ensemble of doses from each Monte Carlo realization at the time of the peak risk (related to the peak-of-the-mean) and (2) the ensemble of peak

© Woodhead Publishing Limited, 2010

Radiation protection principles and development of standards

653

doses calculated from each realization within 10,000 years’. The mean of the doses in case (2) is called mean-of-the-peaks. Averaging risk over generations is not a fundamental problem in the same league as de Finetti’s concern. It should be recognized, however, and its treatment in the application of a standard may have to be agreed between the operator and the regulator. It is mentioned in the general guidelines to the Swedish standard (SSI, 2005, Appendix 1).

20.5

Assessment of probability for scenarios

We have seen how the fundamental or philosophical critique of probability urges us to look at probabilities as perhaps non-objective, i.e. subjective, implying a potential problem (without any final solution offered by the author, unfortunately). However, it is quite obvious that at least the assessment of probability, and therefore risk, is a subjective activity (even if one could argue that the probability still existed objectively). We note that a scenario probability P can always be conservatively taken as 1, avoiding any concern about probability. However, such an approach will be of little value to the operator in the face of unlikely and disruptive scenarios with high dose consequences. Assessment of probabilities can therefore not be avoided completely.

20.5.1 Completeness In a summation of risk contribution from a number of foreseen scenarios, one must consider the independence of scenarios to avoid an overestimate of risk or dose, although overestimation would not be a problem as long as the risk or dose is below the regulatory limit. The main obstacle to a ‘mathematical’ proof of safety lies instead in the proof of completeness, i.e. in the assurance that all possible scenarios have been covered. It is related to the impossible question ‘What is the probability that I forgot something important?’ The issue of completeness is a philosophical problem related to so-called paradigm shifts. That said, we must acknowledge the advanced approach to scenario development, standard in present safety assessments, presenting all considered combinations of the features, events and processes (FEPs) related to the repository, to be available for judgement and comments in a systematic way. The main argument presented here is that we never can expect to know the exact risk. The conclusion on probability is summed up by the common position of the German Reactor Safety and Radiation Protection Safety Committees (SSK, 2002): ‘The difficult communicability of the risk

© Woodhead Publishing Limited, 2010

654

Geological repository systems for safe disposal

approach and difficulties in quantifying the entrance probability still limits the use of the risk criterion.’

20.5.2 The omega of risk In the assessment of the future biosphere and man’s behaviour, ICRP recommends (ICRP, 2000) that we use a reference from today’s regional behaviour and biospheres. The Commission’s recommendations regarding the definition of the person(s) at risk has changed from Publication 43 (ICRP, 1984) mentioning the use of a hypothetical person in particular for exposures in the distant future, superseded by Publication 81 (ICRP, 2000) suggesting the use of the critical group, superseded again by Publication 101 (ICRP, 2007a), where the use of a representative (hypothetical) person again is suggested, using probabilistic methods to describe the characteristics of that person. It is an additional indication of the elusiveness of the critical group concept that it was the only item on which the National Academy’s team could not come to a unanimous agreement (NAS, 1995, Appendix E). The concept of the representative person is open to interpretations in other areas as well. Let us assume that exposure from a repository in a distant future is strongly dependent on geographical location, and then pose the question: ‘As the locations of homes are scattered over the surface near the repository, what is the probability P that someone lived right on the spot of maximum exposure – or should we assume that this always is the case, i.e. that P = 1?’ As mentioned earlier, the basis for the Swedish standard has been the principle of sustainability, implying that the repository’s existence should not interfere with future societies’ needs of land use. It is therefore expected that the repository must be able to coexist with several types of ecosystems. This view also points to the use of probability P = 1 above, indicating that the whole surface of land is available for future societies. This implies in turn that the idea of prediction (of a particular individual’s fate in a future society) is abandoned, and therefore also the uncertainty disappears, in a certain sense. The uncertainties are replaced with decisions about a reference or model future(s). Obviously, the uncertainty remains as to what extent the analysed future biosphere conditions cover the real future, but the decision taken on the reference conditions moves it out of the regulatory regime.

20.6

Time scales in geological repository systems

20.6.1 Human versus geological time As pointed out above, an assessment covering time scales of 0.1 to 1 million years will have to consider even very low probability disruptive events. In

© Woodhead Publishing Limited, 2010

Radiation protection principles and development of standards

655

this connection it should be pointed out that there is not complete international consensus between the ICRP ‘health-related’ and the ‘geological-related’ safety assessment community, at least regarding time and money spent on long-term assessments. According to ICRP (2000), it is important to consider ‘quantitative estimates of dose or risk on the order of 1000 to 10,000 years’. The publication goes on: ‘This approach focuses on that period when the calculation of doses most directly relates to health detriment and also recognises the possibility that over longer time frames the risks associated with cataclysmic geologic changes such as glaciation and tectonic movements may obscure risks associated with the waste disposal system.’ A considerable part of SKB’s recent safety assessment deals with both glaciation and tectonic movements. This substantial body of work is also regulator-driven, keeping in mind that the guidelines requires that ‘For a repository for spent nuclear fuel, or other long-lived nuclear waste, the risk analysis should at least include approximately one hundred thousand years or the period for a glaciation cycle to illustrate reasonably predictable external strains on the repository.’ However, results from long-term assessments can also be valuable for assessing repository safety during shorter periods. The system’s analysed response to what ICRP calls cataclysmic changes could also be used in a different way, to check that the operator has a full understanding of the system and to formulate hypotheses on possible failure mechanisms that need to be analysed. For the same reason, a general description of a repository’s development during long periods also has a general value, independent of the length of the regulatory compliance period. The other issue raised by ICRP is more philosophical. According to ICRP (2000), doses within 10 000 years are more closely directed to health detriment (than doses in the distant future). One can concur with ICRP in the sense that time scales used in a societal industrial project related to health detriment, or rather its prevention by safety measures, seldom go beyond hundreds of years. Similar concerns were behind the requirement in the Swedish standard for a special treatment of the first 1000 years after closure. We recall also NEA (2004 ) pointing out that calculated dose or risk is not always ‘actual measures of future health detriments and risks’. So-called ‘human action’ covers both conscious actions and inadvertent intrusion at the site. Inadvertent human intrusion or actions near the site has been a subject of considerable study (Jensen, 1993; NEA, 1995). It is related to archives and markers that may contribute to our knowledge of the disposal site in the future, in turn related to the more down-to-earth questions of record management systems (RMS), and keeping such systems operative. In connection with deferred decommission of nuclear installations, this points to a time horizon of perhaps 50 years (IAEA, 2008).

© Woodhead Publishing Limited, 2010

656

Geological repository systems for safe disposal

Another potential (conscious) human action is the possibility of reuse of the waste or other types of retrieval, obvious at least in the case of spent nuclear fuel. For a complete description of a repository at time t, one must consider the probability P(t) that the waste has been removed or the repository otherwise disturbed by human action; 1  P(t) therefore amounts to the sum of the scenario probabilities for the undisturbed repository. Scenario probabilities are conditional by their definition (see above) and human action scenarios emphasize this aspect even more. The Swedish standard does not mention a cut-off time, but the guidelines mention that quantitative treatment is expected during 100 000 years. Sometimes, increased time periods of safety demonstration are suggested to bring about higher pressure on the operator or license proponent. The author wishes to caution against extremely long, higher than a million years, assessment periods as a mean of improving safety in general. Both weaknesses and advantages are likely to come out in the study of the repository system over shorter periods.

20.6.2 Justification for a cut-off time Reasons can be established for justification of a regulatory cut-off time, after which no formal safety assessment is required, along the following lines. The distant future of society is unknowable, in a cosmic time scale of a billion years and in some decades of time below. This fact belongs to a series of what can be called eternal problems. These problems were at hand at the time the general decisions were taken about nuclear activities that eventually produced the waste. To quote ICRP (2000): ‘Waste management and disposal operations are an integral part of the practice generating the waste. It is wrong to regard them as a free-standing practice, needing its own justification.’ It is accordingly logical to see the eternal problems, none of which has changed, as part and parcel of the original decisions. The goal of regulation is to achieve a decided activity’s appropriate implementation, not to stand in the way of the original policy decision to carry out the activity. We can therefore sum up by noting that such regulation (of extremely long time periods) – unless it was completely toothless – would run the risk of preventing the regulated activity, by requiring forbiddingly difficult proofs of the operator.

© Woodhead Publishing Limited, 2010

Radiation protection principles and development of standards

20.7

657

Optimization and best available technology (BAT) in geological repository systems

SSI (1998) has required that optimization is performed and consideration has been taken to the best available technique, BAT, in the planning and construction of a repository. These tools are valuable additions to the risk standard, the value and limitation of which was discussed above. Optimization has been a cornerstone in ICRP’s work, particularly since the recommendations of ICRP Publication 26 (ICRP, 1977). It is assumed that the reader is acquainted with the general concepts of justification, optimization and dose limits, and their role in ICRP’s protection system. They are applied here to a disposal system.

20.7.1 Justification The principle of justification relates to the acceptance of the primary activity, in the case of nuclear waste the decision to introduce nuclear power. Justification decisions will also influence the boundary conditions for the required optimization process, or the application of the BAT principle, e.g. in a process to choose the best site in terms of potential exposure from future releases. Site selection is an important part in the optimization process or in the application of BAT, in Sweden, but all possible sites are not available. In Sweden, the legal system lets the local authorities have a considerable influence over the siting process.

20.7.2 Optimization Optimization is very much a developing concept regarding its definition and perception. When the optimization principle (used in the following to denote optimization of protection) was explained in Publication 26 (ICRP, 1977), it was accompanied by a mathematical expression containing dose reduction for a collective and the cost of the dose reduction. Later formulations by ICRP of the optimization principle have been broadened considerably to the more general formulation of reducing doses, i.e. to answer the general question ‘Have I done all I can to reduce doses?’ This formulation puts optimization closer to the definition of best available technology (BAT), discussed in the next section. The NAS comments on optimization in no less than eight sentences on the same page (NAS, 1995, p.125). It first notes that the concept has been applied over 30 years (without commenting on the concept’s major change from ICRP Publication 26 to ICRP Publication 60). Cost and benefits are mentioned, suggesting that the team had the concept from ICRP Publication 26 (1977) in mind. It concludes that ‘there is no scientific

© Woodhead Publishing Limited, 2010

658

Geological repository systems for safe disposal

basis for including the ALARA principle/ICRP’s definition of optimization/ into the EPA standard’. The report then offers that it is sound engineering practice to consider whether dose and risk can be reduced in a cost-effective way, and the last sentence on the page reminds us that ‘subsystem requirements might result in suboptimal repository design’. It is therefore clear that the NAS is no stranger to optimal design. The problem is perhaps rather the inclusion of a non-quantitative or ‘soft’ concept into a legal document within the strict American legal tradition, related to ‘the difficulties in demonstrating technical or legal compliance’ (NAS, 1995, p.125).

20.7.3 BAT, or best available technology BAT, or best available technology, is often associated with the OSPAR Convention. The aim of the Convention is to protect the marine environment of the North-East Atlantic. OSPAR’s activities started in 1972 with the Oslo Convention against dumping. It was broadened to cover land-based sources and the offshore industry by the Paris Convention of 1974. These two conventions were unified, updated and extended by the 1992 OSPAR Convention. The OSPAR Convention requires Contracting Parties to apply best available technology (BAT). Just as for optimization, BAT is accompanied by a short explanation or extension, formulated for BAT as ‘not entailing excessive cost’, generating the acronym BATNEEC. ICRP has gradually come to accept the principle of BAT, and in its Publication 101 (ICRP, 2007a) it expresses support for the concept.

20.7.4 Optimization versus BAT for waste disposal In 2005, SSI introduced the following definitions of optimization and applying BAT in the guidelines related to the Swedish standard: . .

Optimization: taking measures to limit doses and risk using risk calculations as a tool. Applying BAT: taking measures to limit dose and risk by activities that may prevent, limit or delay releases from the repository’s barriers.

Following SSI’s guidance document (SSI, 2005), BAT is applied through such measures as scientific approaches, applying good engineering and quality assurance practices, etc. Such components are important, but their proper application cannot be converted directly to dose reductions. In the guidance document optimization and the use of BAT are thus given separate definitions, but they are both relatively close to the new ICRP formulation of optimization ‘have I done everything I can to limit doses’, suggesting that optimization is a ‘state of mind’. To the author’s knowledge,

© Woodhead Publishing Limited, 2010

Radiation protection principles and development of standards

659

no other regulatory body has introduced such a distinction between the two concepts. However, the main idea in applying the concepts is to ensure that all possible alternatives have been looked at throughout the repository development, whether or not they can be presented with a calculated risk, and that the best has been chosen, given economic and societal constraints.

20.7.5 Arguments against and for optimization and the use of BAT in a standard Arguments can be raised against the use of optimization and the use of BAT in regulations. The regulatory alternative is the easily understood concept ‘good enough’. Why ask for more? The problem is the definition of what is ‘good enough’. Hypothetically, if the repository’s exact risk were known or if it could be bounded, i.e. shown to be less than a regulatory limit, with near certainty, then the risk criterion would be enough. A near-surface repository with a content of nuclides with short half-lives comes close to such a situation, if the waste is protected by only very modest barriers, so that the near future consequences could be reliably calculated. If the dose rate was found to be lower than, say, 10 mSv per year, corresponding to a yearly risk of about 106, the level of protection would perhaps be ‘good enough’. In the case of a geological repository of spent nuclear fuel, the situation is different. The absence of knowledge about an exact risk, concomitant with a strong societal expectation that the best should be done, has led SSI to emphasize the requirements of BAT and optimization. It is worth noting that the use of BAT is also a question about terminology. A development towards state-of-the-art methods for safety assessment has been going on, for example, in the OECD/NEA working groups, related to safety assessment, for decades without the use of the term BAT, but many of the activities would probably meet the criteria explained (SSI, 2005) regarding their work on methodology.

20.8

Future trends

In the presentation of the Swedish standard, risk has been presented as the canonical principle accompanied by BAT and optimization. It is possible to reverse the two. In a Symposium in 2007, Klaus-Ju¨rgen Ro¨hlig presented a paper titled ‘Focus on isolation and containment rather than on potential hazard: an approach to regulatory compliance for the post-closure phase’ (Baltes et al., 2007). The first sentence in the Symposium’s extended summary for the paper is very illuminating: ‘This presentation described a regulatory approach focused on isolation and containment of waste rather

© Woodhead Publishing Limited, 2010

660

Geological repository systems for safe disposal

than on potential hazard and dose.’ In the paper, the repository’s containment performance was judged by the use of six so-called safety indicators, of which dose was one. The concept is very close to a BAT standard, with an attached selection of supporting concepts, dose being one. The work presented above may have implications for a future German standard, but in Sweden and other countries with already established standards it is unlikely that significant change in the standards would come about without compelling reasons. There is one development – with potential for creating change and reasons for revisiting the standards – in the European Union at least. A number of European regulators have joined to define a common view of the expectation of a repository safety case and its regulatory review (Besnus et al., 2007). In the European Union, nuclear safety regulations by directives from the European Commission are a disputed subject, but there is a possibility of future common arrangements that potentially could influence national regulations. Both the activities of the European regulators’ group and the European Commission’s own activities may develop in the future, and in this process it is natural that all new ideas are taken into account.

20.9

References

Apostolakis G E (2004), ‘Risk in technical and scientific studies: general introduction to uncertainty management and the concept of risk’, in Proceedings from an OECD/NEA Workshop, Management of Uncertainty in Safety Cases and the Role of Risk, 2–4 February 2004, Stockholm, Sweden. Audi M (1973), The Interpretation of Quantum Mechanics, The University of Chicago Press, Chicago, Illinois. Baltes B, Becker A, Kindt A and Ro¨hlig K-J (2007), ‘Focus on isolation and containment rather than on potential hazard: an approach to regulatory compliance for the post-closure phase’, in Proceedings of an OECD/NEA Symposium, 23–25 January 2007, Paris, France. Baltes B and Ro¨hlig K-J (2004), ‘Development of safety criteria in Germany: aim, process and experiences’, in Proceedings from an OECD/NEA Workshop, Management of Uncertainty in Safety Cases and the Role of Risk, 2–4 February 2004, Stockholm, Sweden. Besnus F, Vigfusson J, Smith R, Nys V, Bruno G, Metcalf V, Ruı´ z Lo´pez C, Ruokola E, Jensen M, Ro¨hlig K-J and Bodenez P (2007), ‘European pilot study on the regulatory review of the safety case for geological disposal of radioactive waste’, in Proceedings of an OECD/NEA Symposium, 23–25 January 2007, Paris, France. Dawid A P (2004), ‘Probability, causality and the empirical world: A Bayes–de Finetti–Popper–Borel Synthesis’, Statistical Science, 19(1). de Finetti B (1975), Theory of Probability, vols 1 and 2, John Wiley & Sons, Inc., New York.

© Woodhead Publishing Limited, 2010

Radiation protection principles and development of standards

661

EC (2004), Environmental Risks from Ionising Contaminants: Assessment and Management, EC Contract FI6R-CT-2004-508847. EPA (2008), ‘40 CFR Part 197, Public Health and Environmental Radiation Protection Standards for Yucca Mountain, Nevada; Final Rule’, Federal Register, 73(200). IAEA (2006), ‘Geological disposal for radioactive waste’, IAEA Safety Requirements WS-R-4, IAEA, Vienna, Austria. IAEA (2008), ‘Long term preservation of information for decommissioning projects’, Technical Report Series 467, IAEA, Vienna, Austria. ICRP (1977), Recommendations of the International Commission on Radiological Protection, Publication 26, Pergamon Press, Oxford. ICRP (1984), Principles of Monitoring for the Radiation Protection of the Population, Publication 43, Pergamon Press, Oxford. ICRP (1985), Radiation Protection Principles for Disposal of Solid Radioactive Waste, Publication 46, Pergamon Press, Oxford. ICRP (1990), Recommendation of the International Committee on Radiological Protection, Publication 60, Elsevier. ICRP (2000), Radiation Protection Recommendations as Applied to the Disposal of Long-Lived Solid Radioactive Waste, Publication 81, Elsevier. ICRP (2003), A Framework for Assessing the Impact of Ionising Radiation on Nonhuman Species, Publication 91, Elsevier. ICRP (2007a), Assessing Dose of the Representative Person for the Purpose of Radiation Protection of the Public and the Optimisation of Radiological Protection, Publication 101, Elsevier. ICRP (2007b), The 2007 Recommendations of the International Commission on Radiological Protection, Publication 103, Elsevier. Jensen M (ed.) (1993), Conservation and Retrieval of Information: Elements of a Strategy to Inform Future Societies about Nuclear Waste Repositories (Nordiske Seminar- og Arbejdsraporter 1993:596), The Nordic Council of Ministers, Copenhagen. Jensen M, Nolin J and Sundqvist G (1995), ‘Environmental protection or safety review: a cross-scientific analysis of the steering principles in Swedish management of spent nuclear fuel’, SSI-report 95-10 (in Swedish). Kolmogorov, A N (1956), Foundations of the Theory of Probability, Chelsea Publishing Company, New York. Larsson, C-M (2008), ‘An overview of the ERICA integrated approach to the assessment and management of environmental risks from ionising contaminants’, Journal of Environmental Radioactivity, 99(9). Mohanty S and Codell R B (2004) ‘Ramifications of risk measures in implementing quantitative performance assessment for the proposed radioactive waste repository at Yucca Mountain, Nevada, USA’, Risk Analysis, 24(3). NAS (1995), Technical Bases for Yucca Mountain Standards, National Research Council, National Academy Press, Washington, DC. NEA (1995), ‘Future human action at radioactive disposal sites’, A Report from the Working Group on Assessment of Future Human Action at Radioactive Disposal Sites, OECD/NEA, Paris, France. NEA (2004), Post-closure Safety Case for Geological Repositories, OECD/NEA, Paris, France.

© Woodhead Publishing Limited, 2010

662

Geological repository systems for safe disposal

NRPB (1992), ‘Radiological protection objectives for the land-based disposal of solid radioactive wastes’, Documents of the NRPB, 3(3). SSI (1998), The Swedish Radiation Protection Institute’s Regulations on the Protection of Human Health and the Environment in Connection with the Final Management of Spent Nuclear Fuel and Nuclear Waste, SSI FS 1998:1, SSI, Stockholm, Sweden (unofficial translation). SSI (2000), The Swedish Radiation Protection Institute´s Regulations Concerning the Final Management of Spent Nuclear Fuel and Nuclear Waste with Background and Comments, SSI Report 2000:18, SSI, Stockholm, Sweden. SSI (2005), The Swedish Radiation Protection Authority’s Guidelines on the Application of the Regulations (SSI FS 1998:1) Concerning Protection of Human Health and the Environment in Connection with the Final Management of Spent Nuclear Fuel and Nuclear Waste, SSI FS 2005:5, SSI, Stockholm, Sweden (unofficial translation). SSK (2002), Common Position by the German Reactor Safety and Radiation Protection Safety Committees, Regarding the Environment Ministry Questions to Update the Repository Safety Criteria, The German Radiation Commission 182nd Meeting, 4–6 December 2002 (in German). Sundqvist G (1991), Democracy and Expertise: The Waste Problem of Nuclear Power in Connection with the Fuelling of Ringhals 3, VEST, Magazine for Science Studies 4, Gothenburg, Sweden (in Swedish). UN (1987), The Brundtland Report, Our Common Future, United Nations.

© Woodhead Publishing Limited, 2010

21 Development of risk-informed, performancebased regulations for geological repository systems T . M c C A R T I N a n d J . K O T R A , US Nuclear Regulatory Commission, USA; G . W I T T M E Y E R , Center for Nuclear Waste Regulatory Analyses, USA

Abstract: The US Nuclear Regulatory Commission (NRC) developed regulations for a potential high-level radioactive waste (HLW) repository at Yucca Mountain, Nevada, using risk-informed, performance-based principles. NRC used the risk-informed, performance-based approach to focus its regulations for a potential repository at Yucca Mountain on those aspects of repository performance important to protecting public health and safety and the environment. This chapter discusses how risk-informed, performance-based methodologies were used to improve the efficiency and effectiveness of regulations for the (1) requirements for a multiple barrier approach, (2) characteristics of the biosphere and (3) consideration of features, events and processes in the performance assessment. Key words: risk-informed regulation, geological disposal, performance assessment, high-level radioactive waste.

21.1

Introduction

The US Nuclear Regulatory Commission (NRC) developed regulations for a potential high-level radioactive waste (HLW) repository at Yucca Mountain, Nevada, using risk-informed, performance-based principles. ‘Stated succinctly, risk-informed, performance-based regulation is an approach in which risk insights, engineering analysis and judgment, and performance history are used to (1) focus attention on the most important activities, (2) establish objective criteria based upon risk insights for evaluating performance, (3) develop measurable or calculable parameters 663 © Woodhead Publishing Limited, 2010

664

Geological repository systems for safe disposal

for monitoring system and licensee performance, and (4) focus on the results as the primary basis for regulatory decision making’ (NRC, 2001a, p. 55732). By contrast, earlier NRC regulations for geological disposal of HLW contained certain prescriptive requirements (e.g. subsystem criteria). These criteria, issued in the early 1980s, failed to gain widespread acceptance in the technical community. They were ‘criticized as overly prescriptive, lacking in both a strong technical basis and a clear technical nexus to the overall performance objective (i.e., the EPA [U.S. Environmental Protection Agency] standards), and unclear in their wording’ (NRC, 1999, p. 8649). The emergence of performance assessment technology in the 1980s and 1990s has led to capabilities for evaluating physical and geological processes relevant to HLW disposal in support of regulatory methods. These, in turn, have resulted in tools for quantifying the overall performance of geological disposal and its corresponding uncertainties. NRC used the risk-informed, performance-based approach to focus its regulations for a potential repository at Yucca Mountain on those aspects of repository performance important to protecting public health and safety and the environment. For example, the NRC used its performance assessment capability to develop and risk-inform regulatory methods in important areas such as (1) requirements for a multiple barrier approach, (2) characteristics of the biosphere and (3) consideration of features, events and processes in the performance assessment.

21.2

Regulatory principles and methodologies for safe geological disposal

A long-standing principle of geological disposal has been a reliance on multiple barriers to limit the release and transport of radionuclides. Engineered barriers (such as the waste packages and waste forms) should complement and work with the geological or natural barriers (such as sorption properties of geological layers or constrained movement of water) so that safety is not solely dependent on a single barrier or phenomenon. NRC’s generic regulations for geological disposal, issued in 1983, stated this principle as the ‘engineered barrier system works to control the release of radioactive material to the geologic setting and the geologic setting works to control the release radioactive material to the accessible environment’ (NRC, 1983, p. 28223). The international community continues to embrace the multiple barrier approach. The International Atomic Energy Agency (IAEA) published a Safety Requirements document in 2006 setting forward requirements for multiple safety functions. The requirements document states: The natural and engineered barriers shall be selected and designed so as

© Woodhead Publishing Limited, 2010

Development of risk-informed performance-based regulations

665

to ensure that post-closure safety is provided by means of multiple safety functions. That is, safety shall be provided by means of multiple barriers whose performance is achieved by diverse physical and chemical processes. The overall performance of the geological disposal system shall not be unduly dependent on a single barrier or function (IAEA, 2006, p. 19). Although the requirement for engineered and geological barriers appear in both NRC’s 1983 generic regulations (hereafter referred to as Part 60) and its 2001 site-specific regulations for Yucca Mountain, the approach for demonstrating compliance has evolved significantly. The development of NRC regulations for geological disposal in 1983 represented a unique application of the defense-in-depth philosophy to a first-of-a-kind facility. While waste is being emplaced and before a geological repository is closed, its operation is amenable to regulatory oversight and control in much the same manner as any other NRC licensed facility. Application of defense-indepth principles for regulation of repository performance, for long time periods after closure, however, must account for the difference between a geological repository and an operating facility. Whereas an operating facility may rely on active safety systems and the potential for active control and intervention, a closed repository is essentially a passive system. Assessment of the long-term safety of a closed repository, therefore, is best evaluated by consideration of the relative likelihood of threats to its integrity and performance. Although it is fairly easy to identify multiple, diverse barriers that comprise the engineered and geological systems, the performance of any of these systems and their respective subsystems cannot be considered either truly independent or totally redundant. The Nuclear Waste Policy Act of 1982, as amended, directed the NRC to develop technical requirements and criteria for high-level waste repositories that provide for a system of multiple barriers. Although the law requires a system of multiple barriers, the issue of how the performance of those barriers should be assessed was a major issue throughout the development, promulgation and implementation of the Part 60 regulations. During the late 1970s and early 1980s, when the NRC was first developing repository criteria, quantitative techniques for assessing repository performance were in their infancy. This lack of experience with, and confidence in, the quantitative methods for addressing uncertainty was a key factor in selecting an approach for implementing the multiple barrier requirement. Then it was believed that compensation for uncertainties in assessing the overall performance of a repository could only be achieved by introducing conservatism. Intentional addition of conservatism, either by making the measure of performance unduly stringent or by using worst-case, bounding assumptions in the evaluation, were argued to be impractical from a

© Woodhead Publishing Limited, 2010

666

Geological repository systems for safe disposal

regulatory point of view. Thus, for Part 60, the NRC elected to prescribe minimum performance standards for each major system (as they were envisioned at the time) as well as to require the overall system to comply with the EPA standards. The standards for each major system, which became known as ‘subsystem requirements’, were: (1) the length of time the waste package remains intact (300 to 1000 years), (2) the rate of subsequent releases from the engineered system (release rate of 105 per year) and (3) the pre-emplacement ground-water travel time to the accessible environment (1000 years). Much of the bases for the NRC staff’s recommendations of the numerical values for the subsystem criteria were generic judgments of what was thought to be feasible. For example, the release rate criterion of 105 per year was recommended based primarily on staff consideration of typical leach rates of a range of synthetic glass waste forms that were then under consideration by the US Department of Energy (DOE). Despite NRC explanations of its concerns with assessing overall performance and the basis for the values selected, the Part 60 subsystem criteria were widely criticized for many years after their promulgation. The subsystem criteria were criticized as overly prescriptive, lacking sound technical basis and unclear in their wording. Besides the DOE and representatives of the nuclear industry, critics included the National Academy of Sciences Board on Radioactive Waste Management, former NRC Commissioner James R. Curtiss and the US Nuclear Waste Technical Review Board. In a 1983 report (NAS, 1983), the Waste Isolation Systems Panel of the NAS Board on Radioactive Waste Management found that NRC’s proposed numerical criteria ‘ . . . are of questionable importance to long-term safety and are proposed without technically valid basis’. The panel took particular exception to the release rate criterion, stating that ‘ . . . it is both unnecessary and unrealistic for the NRC staff to prescribe a single fractional dissolution rate that must apply to essentially all of the 1000-year inventory of radionuclides . . . is not only unnecessary but it is also likely impossible to achieve.’ With criticism of the Part 60 subsystem criteria as a backdrop, the NAS Committee on Technical Bases for Yucca Mountain issued a report that found, among other things, that (1) the physical and geologic process relevant to a Yucca Mountain repository ‘ . . . are sufficiently quantifiable and the related uncertainties sufficiently boundable that the performance [of a repository] can be assessed over timeframes during which the geological system is relatively stable or varies in a boundable manner’ (NAS, 1995, p. 9) and (2) ‘ . . . because it is the performance of the total system in light of the risk-based standard that is crucial, imposing subsystem performance requirements might result in suboptimal repository design’ (NAS, 1995, p. 13). After the NAS issued its recommendations for Yucca Mountain standards, in 1995, the NRC began development of regulations for Yucca

© Woodhead Publishing Limited, 2010

Development of risk-informed performance-based regulations

667

Mountain. As part of its rule-making activities, the NRC noted (NRC, 1999, p. 8649) that more than 15 years of performance assessment experience provided significantly greater confidence in the technical ability to assess overall repository performance and to address and quantify the corresponding uncertainty than existed when Part 60 was promulgated (see, for example, Bonano et al., 1989; Electric Power Research Institute, 1990, 1992, 1996; NRC, 1992, 1995, 1997; Sandia National Laboratories, 1992; Pacific Northwest Laboratory, 1993; DOE, 1995). As a result, the NRC found no need for added quantitative requirements beyond compliance with the overall performance goal (i.e. a dose limit). Also, NRC noted that quantitative criteria for repository subsystems would needlessly limit the flexibility of the repository developer in achieving compliance with the overall performance goal. Instead, NRC adopted an approach to regulation that provides flexibility to the developer to select and defend the barriers it decides to rely on for demonstrating the safety of the repository. This approach resulted, in large measure, from a greater confidence, relative to 1983, in the efficacy of contemporary techniques to evaluate the long-term performance of a geological repository. NRC’s site-specific regulations for Yucca Mountain in Part 63 impose criteria for the performance assessment that must be used for demonstrating compliance with the overall safety standard (i.e. dose limit). To ensure the necessary confidence that the repository will perform as intended, the required performance assessment must: 1. 2.

3. 4.

5. 6.

Consider features, events and processes (FEPs), including disruptive processes and events. Consider events that have at least a 108 per year chance of occurring (unless the timing and magnitude of the dose would not be significantly affected by the omission). Account for uncertainties and variabilities in parameter values. Consider alternative models of features and processes that are consistent with the available data and scientific understanding and evaluate the effects that alternative conceptual models have on performance of the repository. Provide a technical basis for either inclusion or exclusion of specific FEPs in the performance assessment. Provide a technical basis for models used in the performance assessment such as comparisons made with outputs of detailed process-level models and or empirical observations (e.g. laboratory testing, field investigations and natural analogs).

A performance assessment meeting these criteria can be used appropriately and confidently to evaluate the behavior of the repository. The fundamental component of a risk-informed, performance-based approach applied to

© Woodhead Publishing Limited, 2010

668

Geological repository systems for safe disposal

geological disposal is a high-quality performance assessment and associated analyses of performance of the repository. The Part 63 approach for implementing the multiple barrier approach for Yucca Mountain is tied to the performance assessment used to demonstrate compliance with the overall performance standard. NRC explained when it finalized its regulations for Yucca Mountain (NRC, 2001a, p. 55759) that it was ‘confident that evidence for the resilience, or lack of resilience, of a multiple barrier system will be found by examining a comprehensive and properly documented performance assessment of the behavior of the overall repository system’. Thus, NRC’s regulations in Part 63 do not require DOE to perform additional analyses to demonstrate that the repository is comprised of multiple barriers; rather DOE must describe the capabilities of the barriers it relies on, consistent with the performance assessment used to demonstrate overall performance. Such an approach allows ‘DOE to use its available resources effectively to achieve the safest repository without unnecessary constraints imposed by separate, additional subsystem performance requirements’ (NRC, 2001a, p. 55758). The multiple barrier requirement in Part 63 provides valuable input for NRC’s risk-informed, performance-based review of a high-level waste repository. The Yucca Mountain Review Plan (NRC, 2003, p. 2.2-1 to 2.2-2) describes how the staff will use the description of the capabilities of the barriers important to waste isolation to risk-inform the review. NRC expects DOE to provide a technical basis for each barrier commensurate with its significance to overall repository performance. An effective and efficient regulatory review would also seek to focus the review on those aspects, including the uncertainties, of the barriers that were most important to performance. As with the NRC’s previous approach to multiple barriers (i.e. quantitative criteria for repository subsystems), this current risk-informed, performance-based approach to multiple barriers has met with controversy. NRC’s current approach was also challenged in court because it no longer provided for quantitative criteria for repository subsystems. On a July 2004, the United States Court of Appeals for the District of Columbia Circuit upheld NRC’s regulations regarding its approach for the requirements for multiple barriers (i.e. description of each barriers’ capability consistent with the performance assessment used to estimate overall performance).

21.3

Development and application of methodologies

As described above, the risk-informed, performance-based approach is highly dependent on the ability of performance assessment analyses to provide the necessary risk insights. These insights emerge from intermediate performance assessment results such as the waste package lifetime, release rates from the engineered barrier system or transport times in the geosphere.

© Woodhead Publishing Limited, 2010

Development of risk-informed performance-based regulations

669

An important strength of the risk-informed approach is evaluating and understanding the significance of uncertainties associated with the barriers important to waste isolation by allowing the regulator, developer and other interested parties to focus on those aspects most significant to the safety of the facility. For example, if a particular aspect of the repository system were highly uncertain but the uncertainty had no significant impact on overall compliance (i.e. meeting the dose limit), it would be neither necessary nor productive to focus a lot of attention to reduce uncertainty in this area. Conversely, if a particular aspect of the repository system had a significant influence on whether the repository met the dose limit, the uncertainty in the applicant’s understanding of this particular aspect of repository performance may need to be considered in great detail. Although there have been significant advances in performance assessment methods, there are dimensions of the performance of a geologic repository over very long time periods (e.g. thousands of years) that do not lend themselves to scientific inquiry and collection of information to reduce uncertainty. For example, aspects of the biosphere associated with human behavior (e.g. diet, land use) become very uncertain in the distant future and are subject to a wide range of speculation that cannot be reduced through further data collection. Additionally, certain FEPs such as future climate change are uncertain and introduce a range of issues and uncertainties that may present specific challenges to performance assessment (e.g. range and variability of climate). Implementation of a risk-informed, performancebased approach for regulating geological disposal can use the strength of performance assessment methods to assist determining appropriate approaches for addressing potential limitations in performance analyses (e.g. human behavior and FEPs subject to unique uncertainties such as climate change).

21.3.1 Characteristics of the biosphere When the NRC proposed regulations for Yucca Mountain it was acknowledged, for the purposes of analyzing performance of a waste disposal facility, that demonstration of compliance with an individual dose limit over thousands of years would require the use of assumptions about the characteristics of the individuals at risk and the characteristics of the biosphere in which they lived (NRC, 1999, p. 8645). Although unbounded speculation regarding lifestyles of individuals and characteristics of the biosphere do not provide useful information to performance analyses, there is often a variety of information available regarding lifestyles and characteristics of the biosphere useful for conducting performance assessments and assisting implementation of regulatory approaches. In particular, the NRC conducted a number of analyses in support of its Yucca Mountain

© Woodhead Publishing Limited, 2010

670

Geological repository systems for safe disposal

rulemaking with respect to the lifestyles of individuals and characteristics of the biosphere (NRC, 2001b). It is generally understood that the ground-water pathway is the most likely pathway for radiological exposures at Yucca Mountain; thus the NRC undertook a number of analyses related to the use of ground water applicable to the Yucca Mountain region. Because the Yucca Mountain area is an arid region where wells are the primary source of water, a portion of these analyses focused on the factors that might have influenced the location and behavior of the current population in the greater Yucca Mountain area, such as current sources of water supply for existing communities (e.g. water well properties and water use patterns) and the relative costs of drilling and pumping freshwater from shallow and deep wells. The US Geological Survey performed one of the first comprehensive surveys of ground-water availability (as well as ground-water use) in the Yucca Mountain area in 1971 (Thordarson and Robinson, 1971). To assist government officials in assessing damage claims to water wells and springs possibly resulting from nuclear explosions, the US Geological Survey reviewed published and unpublished geological and ground-water data within a 161-kilometer radius of the Nevada Test Site. The census covered 81 367 square kilometers in portions of six counties in southern California and Nevada. The types of information collected for the census included aquifer type, water-table depth, yield and end use. This census identified 6032 water wells (and 754 springs and seeps). Most of the wells (98%) were located in alluvial fan deposits. In addition to being the most extensive surficial geology in the region, alluvial basins are usually the most accessible topographically (occupying physiographic low terrain). Almost 54% of the wells were reported to be less than 60 meters deep; almost 84% were reported at depths less than 150 meters deep. Of the wells identified, 60% were reported to yield at least 380 liters/minute. In terms of end use, most of the wells in the study area were reported to be used for domestic purposes (56%), followed by irrigation (17%), municipal and commercial (6% each), and stock supply (2%). (Nine percent of the wells in the 1971 study area were reportedly no longer in use.) The greatest depths to water in wells currently being pumped to supply agricultural or domestic needs in the immediate Yucca Mountain region occur in the community of Amargosa Valley where depths to water range from 100 to 120 meters (approximately 20 kilometers from the proposed Yucca Mountain repository). The depth to water increases monotonically from approximately 100 to more than 300 meters along a trajectory extending from the community of Amargosa Valley in the south and terminating in the north approximately 10 kilometers from the proposed Yucca Mountain repository. Most of the farming occurs approximately 30

© Woodhead Publishing Limited, 2010

Development of risk-informed performance-based regulations Table 21.1

671

Costs for developing certain wells in southern Nevada

Aquifer type

Well yield in gallons per minute (liters per minute)

Capital cost per well (1982 US dollars)

Alluvial

300–2500 [1135–9462]

$85 000–$500 000

Carbonate

Less than 5000 (18 925)

$2 000 000

to 40 kilometers from the proposed repository where the depth to water is less than 50 meters. Another factor related to the depth to water is the costs of obtaining the water. The costs of producing (e.g. extracting) the water could provide an additional criterion for distinguishing between potential receptor groups in the Yucca Mountain area. When a water table occurs at great depth, extraction and operating costs could be prohibitive for certain scenarios, thereby creating disincentives to agricultural, residential or even commercial development at particular locations within the site. Because of its accessibility (e.g. depth), there is an economic argument that the costs of drilling, developing and pumping ground water from the (deep) carbonate ground-water resource within (below) the Yucca Mountain site are higher in comparison with the costs of using (shallow) alluvial ground-water resources found approximately 30 kilometers from Yucca Mountain. For example, pumping costs will tend to increase because of the additional electricity needed to lift the water from the greater depths, even when the price of electricity drops or remains constant (Schefter, 1984). Previously, the State Engineers Office (State of Nevada, 1982, p. 11) made estimates with regard to the development (and production) of new ground-water resources in southern Nevada, which tend to support this argument (see Table 21.1). Because the climate is arid to semi-arid, agriculture is restricted to areas where the depth to water is shallow enough to make irrigation economically feasible (i.e. Amargosa Desert, Pahrump and Oasis Valleys). Agriculture also tends to be confined to areas where the relief is gentle enough to make center-pivot or furrow irrigation methods technically feasible and where the soil texture is suitable, such as near the distal margin of alluvial fan deposits. Much of the land in the vicinity of Yucca Mountain is controlled by the US government, which has precluded potential development. Although the withdrawal of this land from private development may have affected landuse patterns, the great depth to ground water and limitations to soil irrigability, throughout the controlled area, suggests that farming would not be economically feasible with current technology. Based on current

© Woodhead Publishing Limited, 2010

672

Geological repository systems for safe disposal

technology, agricultural activity will most likely continue to be limited to the Amargosa Desert vicinity. The type of information discussed above can be used to limit speculation and provide a reasonable basis to help support selection of characteristics of potential receptor groups (e.g. location, diet of locally produced food). Recognizing that in areas where surface water is extremely limited the source of water will be obtained from water wells. Water, a basic need for human habitation, also typically needs to be obtained economically for communities to develop. Thus, the depth to water provides certain constraints regarding where people may live and thereby helps define where potentially contaminated water may be withdrawn. Most importantly, the use of well water is an exposure pathway that must be considered. Specification of aspects of the exposure scenario in the standards and regulations for Yucca Mountain are consistent with this information regarding the Yucca Mountain site and region. For example, a water well is assumed to be located where the depth to water is approximately 100 meters below the surface, and an exposed individual (a reasonably, maximally exposed individual, or RMEI, is the term used in the EPA standards for the assumed individual used for radiation exposure calculations in the performance assessment) is assumed to drink 2 liters of well water per day.

21.3.2 Features, events and processes Implementation of a risk-informed, performance-based approach includes the determination of what features, events and processes (FEPs) should be included in the quantitative analysis of performance. FEPs include both those attributes of the geological system that limit release of and exposure to radionuclides (e.g. low water flow, waste characteristics, sorption characteristics of geological media, water chemistry) and those attributes that may disrupt or potentially degrade the performance of the geological system (e.g. seismic events, colloids, enhanced corrosion, climate change). The evaluation of FEPs is especially challenging when the period of performance assessments is for extremely long time periods (e.g. thousands of years and longer). Recently, the EPA published its standards for Yucca Mountain covering the time period beyond 10 000 years out to one million years (EPA, 2008). A fundamental aspect of the EPA standard is the specified approach for limiting undue speculation on future behavior of the site by constraining the features, events and processes that need to be considered in the performance assessment. EPA’s standards limit the assessment of specific features, events and processes in the period after 10 000 years to effects on the repository system that are most relevant (i.e. ignoring lesser or secondary effects that may add to speculation and uncertainties but would not be expected to have a significant effect on the peak dose over a 1 million

© Woodhead Publishing Limited, 2010

Development of risk-informed performance-based regulations

673

year period) (EPA, 2008, p. 61283). For example, DOE’s performance assessment may (1) limit the analysis of seismic activity to the effects caused by damage to the drifts and the waste package, and changes in the elevation of the water table; (2) limit analysis of igneous activity to effects on the waste package that result in the release of radionuclides to the atmosphere or ground water; (3) require DOE to include general corrosion in its analysis of engineered barrier performance; and (4) limit the effect of climate variation to those resulting from increased water flowing to the repository (NRC was given the task of specifying the values to be used to represent climate change). There is a wide range of uncertainties associated with estimating climate changes over the next one million years. Precipitation and temperature are the most readily identified parameters, associated with climate, that directly influence the amount of water, or deep percolation, flowing to the repository horizon. It is the rate of deep percolation, however, that directly influences repository performance; therefore the NRC elected to specify the deep percolation rate to represent the effect of future climate in performance assessments after 10 000 years. To assist the specification of the deep percolation rates the NRC used performance assessment calculations to understand the effect of some of the uncertainties in climate change to determine appropriate methods for implementing climate change in performance assessments. One consideration was how the transition of climate might affect dose estimates (i.e. the variations from dry to wetter conditions that are expected to occur a number of times over a one-millionyear period). A simple test case was performed to examine potential differences in dose when assuming a constant value for deep percolation versus a cyclical variation as a representation of climate change. Figure 21.1 presents the average deep percolation rate used in the two different representations and Fig. 21.2 presents the average doses from the ground-water pathway when waste packages are assumed to be breached at 20 000 years. The timevarying climate test case clearly exhibits a variation consistent with the variation in the deep percolation rate, but the differences between the two dose curves are small given the differences in deep percolation rates between the two different representations for climate change. There may be significant uncertainty regarding the timing of climate variations (e.g. timing and length of dry and wet periods), but these calculations strongly indicate that assuming a constant deep percolation rate instead of a varying rate would not have a large effect on the overall dose. Here a performance assessment capability was used to help understand and support the specification of the deep percolation rate to ensure that the regulatory approach focuses on the significant aspects of performance (i.e. the magnitude of a deep percolation rate rather than timing of when that rate

© Woodhead Publishing Limited, 2010

674

Geological repository systems for safe disposal

21.1 Average deep percolation rates representing constant climate (dashed line) and a time-varying climate (solid line).

21.2 Average dose based on deep percolation rates representing constant climate (dashed line) and a time-varying climate (solid line).

occurs or over what time period). Thus, NRC’s regulations specify the range to be used for the deep percolation rate of 10 to 100 millimeters per year as a

© Woodhead Publishing Limited, 2010

Development of risk-informed performance-based regulations

675

means to limit speculation and focus on those aspects of climate change most significant to performance.

21.4

Future trends

NRC has finalized its risk-informed, performance-based regulations for a proposed repository at Yucca Mountain as an approach to conduct an efficient and effective safety review. The development of performance assessment methods over the past few decades has provided a capability quantitatively to evaluate repository performance, making implementation of the risk-informed, performance-based approach possible. The performance assessment methods are used as a means to evaluate overall performance and also assist the development of regulations and compliance calculations (e.g. multiple barrier requirements, consideration of climate change). Additionally, the use of site-specific information can be used to focus regulatory criteria (biosphere characteristics). Looking forward on risk-informed, performance-based regulations, it must be recognized that the efficiency and effectiveness of implementation of regulatory reviews will, in part, depend on the extent that unnecessary speculation can be constrained. Generic regulations can be difficult to constrain due to the lack of site-specific information. Risk-informed, performance-based regulations can only be efficient and effective if there are appropriate constraints that allow regulatory reviews to focus on the most important safety attributes of the repository system. The NRC has stated that the performance assessment is intended ‘to provide a reasonable test of the safety of the repository by modeling through computer simulations a large number of ‘‘alternative futures,’’ incorporating the features, events, and processes required by the rule to be included in the assessment to determine if there is a reasonable expectation that the disposal system will meet regulatory standards’ (NRC, 2009, p. 10815). Regardless of whether the regulation is site-specific or generic, the regulations must provide appropriate criteria to limit unwarranted speculation and promote a ‘reasonable’ test of repository performance.

21.5

Disclaimer

This chapter was prepared, in part, by employees of the United States Nuclear Regulatory Commission in their own time apart from their regular duties. NRC has neither approved nor disapproved its technical content. The views expressed are those of the authors. They do not represent the views of the NRC and do not represent a final judgment or determination of the matters addressed or of the acceptability of a license application for a geological repository at Yucca Mountain.

© Woodhead Publishing Limited, 2010

676

Geological repository systems for safe disposal

21.6

References

Bonano E J et al. (1989), ‘Demonstration of a performance assessment methodology for high-level waste disposal in Basalt Formation’, NUREG/CR-4759, US Nuclear Regulatory Commission, Washington, DC. DOE (1995), ‘Total system performance assessment – 1995: an evaluation of the potential Yucca Mountain Repository’, Prepared by TRW Environmental Safety Systems, Inc., for Department of Energy Office of Civilian Radioactive Waste Management, BOOOOOOOO-01717-2200-00136, Rev. 01. Electric Power Research Institute (1990), ‘Demonstration of a risk-based approach to high-level waste repository evaluation’, EPRI NP-7057, Palo Alto, California. Electric Power Research Institute (1992), ‘Demonstration of a risk-based approach to high-level waste repository evaluation: Phase 2’, EPRI TR-100384, Palo Alto, California. Electric Power Research Institute (1996), ‘Yucca Mountain total system performance assessment, Phase 3’, EPRI TR-107191, Palo Alto, California. EPA (2008), ‘Public Health and Environmental Radiation Protection Standards for Yucca Mountain, Nevada; Final Rule; 40 CFR 197’, Federal Register, 73 (200), 61256–61289. IAEA (2006), ‘Geological disposal of radioactive waste’, Safety Requirements WSR-4. NAS (1983), ‘A study of the isolation system for geologic disposal of radioactive wastes’, Waste Isolation Systems Panel, Board on Radioactive Waste Management, National Academy Press. NAS (1995), ‘Technical basis for Yucca Mountain Standards’, Committee on Technical Basis for Yucca Mountain Standards, National Academy Press. NRC (1983), ‘Disposal of high-level radioactive wastes in geologic repositories: technical criteria; Final Rule; 10 CFR 60’, Federal Register, 48 (120), 28194– 28224. NRC (1992), ‘Initial demonstration of the NRC’s capability to conduct a performance assessment for a high-level waste repository’, NUREG-1327. NRC (1995), ‘NRC iterative performance assessment Phase 2 – Development of capabilities for review of a performance assessment for a high-level waste repository’, NUREG-1464. NRC (1997), ‘NRC high-level radioactive waste program Annual Progress Report: Fiscal Year 1996’, Prepared for NRC by the Center for Nuclear Waste Regulatory Analyses, NUREG/CR-6513. NRC (1999), ‘Disposal of high-level radioactive wastes in a proposed geologic repository at Yucca Mountain, Nevada; Proposed Rule; 10 CFR 63’, Federal Register, 64 (34), 8640–8679. NRC (2001a), ‘Disposal of high-level radioactive wastes in a proposed geologic repository at Yucca Mountain, Nevada; Final Rule; 10 CFR 63’, Federal Register, 66(213), 55732–55816. NRC (2001b), ‘Preliminary performance-based analyses relevant to dose-based performance measures for a proposed geologic repository at Yucca Mountain’, NUREG-1538.

© Woodhead Publishing Limited, 2010

Development of risk-informed performance-based regulations

677

NRC (2003), ‘Yucca Mountain Review Plan; Final Report’, NUREG-1804, Revision 2. NRC (2009), ‘Implementation of a Dose Standard after 10,000 years; Final Rule; 10 CFR 63’, Federal Register, 74(48), 10811–10830. Pacific Northwest Laboratory (1993), ‘Preliminary total-system analysis of a potential high-level nuclear waste repository at Yucca Mountain’, PNL-8444. Sandia National Laboratories (1992), ‘TSPA 1991: an initial total-system performance assessment for Yucca Mountain’, SAND91-2795. Schefter J E (1984), ‘Declining ground water levels and increasing pumping costs: Floyd County, Texas – A Case Study’, in National Water Summary 1984 – Hydrologic Events, Selected Water Quality Trends, and Ground Water Resources, US Geological Survey, Water Supply Paper 2275. State of Nevada (1982), ‘Water for Southern Nevada’, Carson City, Division of Water Planning/Department of Conservation and Natural Resources, Water Supply Report 2 (Prepared by URS Company and Converse Ward Davis Dixon, with contributions by the Las Vegas Valley Water District). Thordarson W and Robinson B P (1971), ‘Wells and springs in California and Nevada within 100 miles of the point 37 deg. 15 min. N., 116 deg. 25 min. W, on Nevada Test Site’, US Geological Survey 474-85.

© Woodhead Publishing Limited, 2010

22 Environmental monitoring programs and public engagement for siting and operation of geological repository systems: experience at the Waste Isolation Pilot Plant (WIPP) J . C O N C A a n d T . K I R C H N E R , New Mexico State University, USA

Abstract: The Waste Isolation Pilot Plant (WIPP) near Carlsbad, New Mexico is a successful example of an operating deep geological nuclear repository. That success results from several factors, including an optimal geologic and physiographic setting, a strong scientific basis, early regional community support, frequent interactions among stakeholders at all stages of the process, long-term commitment from the upper management of the US Department of Energy (DOE) over several administrations, strong New Mexico State involvement and oversight, and constant environmental monitoring from before nuclear waste was first emplaced in the WIPP underground (in 1999) to the present, ten years after waste disposal operations began. All of these factors should be an essential part of the siting and operations of any nuclear waste repository. Key words: environmental monitoring, radionuclides, plutonium, bedded massive salt, aerosols, deep geological nuclear waste disposal, repository, whole body counting, Waste Isolation Pilot Plant (WIPP), transuranic waste, CEMRC

22.1

Introduction

The success of any high-profile, routinely misperceived project such as a nuclear facility is strongly tied to the degree of public participation and understanding. The Waste Isolation Pilot Project (WIPP) is a deep geological transuranic defense waste disposal site located 2130 ft (655 m) below the surface in a particularly massive portion of the Permian-age Salado Salt Formation in the arid Delaware Basin of southeastern New Mexico and west Texas (Fig. 22.1). The WIPP is situated between Carlsbad 678 © Woodhead Publishing Limited, 2010

Environmental programs and public engagement

679

and Hobbs, two towns each with less than 30 000 people and each about 30 miles (48 km) from the WIPP. The WIPP is an example where public engagement has constantly been at a high level. Constructed and ready for operation in the early 1980s, the WIPP began accepting nuclear waste on 26 March 1999. The WIPP is licensed and permitted to accept only transuranic waste (> 100 nCi/g but <23 Curie/liter) generated from defense operations. As of 2009, the WIPP has disposed of over 65 000 m3 of transuranic waste in over 120 000 containers (55-gallon drums, standard waste boxes and 10drum overpacks; see www.wipp.energy.gov), equivalent to about 350 000 55-gallon drums. While most of the waste contains primarily alpha-emitters like Pu, which do not require significant shielding, a fraction of the waste contains radionuclides that emit the more penetrating gamma radiation. When containers have surface dose rates greater than 200 mrem/h, (2 mSv/h), they must be remotely handled with shielded casks and delivery vehicles. These higher dose rates mostly result from gamma emissions during the decay of isotopes such as 137Cs and 90Sr/90Y that came from reprocessing defense spent fuel. From the standpoint of addressing operational and environmental risk, as well as public concerns, the WIPP has had broad and detailed site characterization and extensive human health and environmental monitoring. In addition to the regulatory compliance monitoring required by the repository licensee, and also conducted by the State of New Mexico and previous entities, the local community demanded a sophisticated environmental monitoring program carried out by an independent academic institution that would conduct a science-based monitoring and research program, not monitoring only for compliance. The Carlsbad Environmental Monitoring and Research Center (CEMRC), in the Institute for Energy and the Environment in the College of Engineering at New Mexico State University, resulted from these negotiations (http://www.cemcr.org). Its construction and operations were funded by a grant from the US Department of Energy (DOE) that allowed complete public transparency and academic freedom. The public was directly involved in the monitoring program and helped determine the environmental media and priority chosen for the monitoring program. Unlike most environmental programs, which only monitor down to compliance or action levels, the mission of CEMRC was to monitor to below background levels, indeed to push the minimum detection limits farther than is normally done in order to observe what would happen in the environment and what effect WIPP operations would have on everything in the environment. As a result some approaches needed to be completely rethought, such as increasing counting times on alpha spectroscopy to over 5000 minutes in order routinely to achieve the femtoCurie level for Pu and Am, or adopting a 12-detector array for in vivo bioassay in order to detect

© Woodhead Publishing Limited, 2010

680

Geological repository systems for safe disposal

© Woodhead Publishing Limited, 2010

22.1 (a) The Waste Isolation Pilot Plant (WIPP), the only operating deep geological nuclear waste repository, is located 700 meters (2150 ft) below the surface excavated in the massive salt of the Salado Formation, in eight panels with seven rooms each, and has been operating successfully since 1999. (b) Contact handled transuranic nuclear waste (> 100 nCi/g of waste with surface exposures less than 200 mrem/h) being transported to the WIPP site in New Mexico in TRUPac II containers. (c) Remote handled nuclear waste (up to 23 Ci/liter, with surface exposures greater than 200 mrem/h), some of it from reprocessing, being transported to the WIPP site in a 72B cask. (d) Over 10 000 nuclear waste drums and standard waste boxes filling 1 of 56 rooms to be filled at the WIPP over a 20-year period; 24 rooms have been filled as of January 2009. Note the higher activity remote handled waste plunged into boreholes in the wall to the left and plugged.

Environmental programs and public engagement

© Woodhead Publishing Limited, 2010

681

682

Geological repository systems for safe disposal

Table 22.1

Site selection criteria for salt repositories

Criteria

Desired attribute

Medium Thickness Depth Tectonic activity Economic resources Population density Dedicated land Climate Location

Bed of rock salt as pure as possible At least 60 m Greater than 300 m and less than 1000 m Minimal Minimal Low No land conflicts Arid At least 1.6 km from the nearest borehole At least 5 km from any groundwater activity Flat

Topography

Pu in the lungs using its 17.5 keV gamma ray. These changes necessitated debates as to the meaning of such low limits, their challenges in the light of traditional procedures and equipment, and what were the statistical ramifications of such a monitoring philosophy. The following discussion looks at the historic context of the selection and construction of the WIPP nuclear repository, and reviews the initial rationale, funding, set-up and subsequent decisions made in the development of the final environmental monitoring program, ending with examples of monitoring results for air and people in the vicinity of the WIPP, the two most important issues for the public. As with air and people, levels of activity in drinking waters, soils, sediments and surface waters have never exceeded historic levels, and are discussed in detail in previous reports that can be found at www.cemrc.org.

22.2

History of salt and site selection of the Waste Isolation Pilot Plant (WIPP)

22.2.1 Site selection The National Academy of Sciences (NAS) and National Research Council (NRC) in 1957 recommended the strategy of using deep-geological repositories as the optimal disposal mechanism for radioactive waste (NAS–NRC, 1957) and stated that the salt deposits were the most promising method for the disposal of high-level wastes and other radioactive wastes with long-lived radionuclides generated from both civilian and defenserelated programs. One result of the NAS recommendations was the development of site selection criteria for salt repositories (Table 22.1), resulting from research at the Oak Redge National Laboratory (ORNL, 1973). These criteria were instrumental in the site selection process for the

© Woodhead Publishing Limited, 2010

Environmental programs and public engagement

683

WIPP (DOE, 1996a, 1997b) and are still relevant for discussion of any radioactive waste repositories in salt. Located in southeastern New Mexico in the northern portion of the Delaware Basin, the WIPP falls within a 25.75-km2 tract of federal land (a 464 mile square = 16 miles2). The WIPP repository is located 655 meters (m), or about 2130 ft, below the surface in a bedded salt formation of Permian age known as the Salado Formation. Near the WIPP, the salt beds dip eastward with a slope of 1 degree. The region is semi-arid with less than 30 cm (12 in) of precipitation annually. There are no perennial streams within 20 km of the WIPP. Within the Salado Formation, ground-water circulation is negligible. Water-bearing units within the Rustler Formation that overlie the Salado Formation produce only small amounts of brine. The underground workings at the WIPP remain dry. If drilling intersects interstitial water or fluid inclusions, they will drain until dry, but no water from beyond the local grains intersected will follow. The WIPP was authorized in 1979 by Congress to demonstrate the safe disposal of radioactive waste generated from weapons production, research and development in the US Defense Program. The present authorization is for the disposal of 175 500 cubic meters of defense-related TRU wastes (Public Law 102-579, 1992, as amended by Public Law 104-201, 1996) with an original operational life of 35 years. The Atomic Energy Commission (AEC) then conducted research at the Oak Ridge National Laboratory (ORNL) from 1957 to 1961 on disposal of radioactive waste in salt. The US Geological Survey (USGS) reported on domestic salt deposits suitable for radioactive waste (radwaste) disposal in 1962 (Fig. 22.2). One of the key areas identified by the USGS was the Delaware Basin, an area within the larger Permian Basin, which spans southeastern New Mexico and large areas in Kansas, Texas and Oklahoma. The ORNL work expanded to include large-scale field programs simulating waste using electric heaters placed in existing salt mines at Lyons, Kansas, called the Project Salt Vault, from 1963 to 1967 (Holdoway, 1972). This work resulted in the 1970 AEC selection of the Lyons site as a possible location for a radioactive waste salt repository. This site location became highly political in 1972 after there were indications of extensive drilling for oil and solution mining of salt in the area around the site. In 1973, the AEC, USGS and ORNL chose an area in the Permian Basin in southeastern New Mexico as best meeting the site selection criteria (Wentz, 1999). A location about 30 miles (48 km) east of Carlsbad, New Mexico, was chosen for exploratory and field investigations. Detailed site characterization and engineering design programs were initiated and continued for several years (Sandia Laboratories, 1978). In 1980, DOE considered this site satisfactory to proceed with subsurface investigations. Two shafts were mined to about 655 m with drifts mined about 1.6 km

© Woodhead Publishing Limited, 2010

Locations of salt deposits in the United States.

Geological repository systems for safe disposal

22.2

684

© Woodhead Publishing Limited, 2010

Environmental programs and public engagement

685

north/south and 1.4 km east/west, developing the potential waste emplacement footprint. In 1983, DOE decided that the site was suitable and construction commenced in late 1983 (Sandia National Laboratories, 1991).

22.2.2 Site characterization and geology The WIPP is located in a relatively flat, sparsely inhabited plateau in the area known as Los Medan˜os (the dunes). The area falls within southeastern New Mexico, in the Pecos Valley Section of the Great Plains Physiographic Province, which has terrain varying from rugged canyons to plains and lowlands. Near the WIPP, a stable surface layer of sand is underlain by the Mescalero Caliche. This caliche, a layer enriched in calcium carbonate material ranging in age from 510 000 to 410 000 years, indicates fairly longterm surface stability. The subset of the Permian Basin containing the WIPP is called the Delaware Basin, a structural basin ranging through southeastern New Mexico and western Texas, containing a thick sequence of sandstones, shales, carbonates and evaporates that became a distinct feature by about 280 million years ago. The Permian sediments of the Delaware basin in this area are divided into four series, the youngest of which is the Ochoan (Fig. 22.3). The Ochoan series is divided into four formations: the Castile (the oldest), the Salado, the Rustler and the Dewey Lake beds (the youngest). The Castile Formation directly underlies the Salado, 244 m below the repository (Fig. 22.3). The Castile Formation contains three relatively thick anhydrite/carbonate units and two thick halite units, and is approximately 430 to 460 m thick. Underneath the Castile Formation is the Bell Canyon Formation, uppermost of the Guadalupian Series (Fig. 22.3), and is comprised of layered sandstones, shales, siltstones and limestones. Approximately 300 m or more thick, it is a target of hydrocarbon exploration in the area. Directly overlying the Salado is the Rustler Formation, ranging from 91 to 107 m thick and comprised of a lower unnamed member, the Culebra Dolomite, the Tamarisk, the Magenta Dolomite and the Forty-niner. The Culebra and Magenta Dolomites are gypsum-bearing dolomites containing numerous cavities (vugs), fractures and silty zones. The other three members contain various amounts of anhydrite, siltstone, claystone and halite. The Dewey Lake Formation overlies the Rustler Formation and consists of about 30 to 170 m of reddish-brown siltstones and claystones with lesser amounts of sandstone. The Santa Rosa Formation overlies the Dewey Lake Formation and is characterized by light reddish-brown sandstones and conglomerates. The Santa Rosa Formation is thin to absent at the WIPP site. Finally, the Pleistocene age Gatun˜a Formation overlies the Santa Rosa and is characterized by a wide range of lithologies from coarse conglomerates to gypsum-bearing claystones.

© Woodhead Publishing Limited, 2010

686

Geological repository systems for safe disposal

22.3

Geologic column of the area around the WIPP repository.

The thick Salado Formation is predominantly halite (NaCl) and is laterally extensive. The Salado contains some distinctive and laterally continuous layers composed mostly of anhydrite (partially hydrated calcium sulphate) and polyhalite (a potassium–magnesium–calcium chloride mineral). These layers are so continuous that geologists use them as marker beds (MB) and number them to designate vertical position within the Salado Formation. The WIPP repository is located between MB 139 and MB 138. DOE selected the Salado Formation for the WIPP repository because it met or exceeded all of the site selection criteria of Table 22.1: .

The Salado has had a simple geologic history and is in a tectonically quiet area.

© Woodhead Publishing Limited, 2010

Environmental programs and public engagement .

. . .

.

687

The Salado halite has extremely low permeability to fluid flow, with hydraulic conductivities ≤1014 m/s (Beauheim and Roberts, 2002) and diffusion coefficients ≤1015 m2/s (Conca et al., 1993), essentially preventing flow into and out of the repository for several hundred million years. The Salado is regionally widespread and extensive, sufficiently deep with little potential for dissolution. The Salado includes continuous halite beds with a simple structure and composition, exhibiting self-healing rock mechanical properties (creep closure). The Salado is sufficiently close to the surface that mining is easy but sufficiently deep that the creep closure property is optimized at about 150 bar geostatic pressure, e.g. the Salado cannot maintain open and connected fractures or pores, but the closure occurs slowly enough that operations can be executed on a reasonable time-scale of years. The Salado is free of mobile groundwater, as compared to existing mines and other potential repository sites.

This creep closure property is rare and, unlike fractured hard rock (volcanics, limestones or granitics), provides a perfect enclosure for any waste, especially nuclear, because the rock cannot sustain a fracture or any other pathway for water or contaminants to get in or out, making the formation molecularly tight. In fact, many, if not most, fluid inclusions in the Salado Salt have intact Permian macrobiomolecules such as DNA, bacterial husks and cellulosics from degraded cell walls of halophilic bacteria that have been preserved for the 230 million years since they were crystallographically trapped by the precipitating salt in the evaporating Permian Sea (Griffith et al., 2008). There are no better testaments to the extremely long-term performance of this host rock and to the absence of geological processes that could breach the repository in the future. Because of the seasonal nature of the rainfall, surface drainage is intermittent in the region around the WIPP, and over 90% of the annual precipitation evapotranspires. The Pecos River, 20 km southwest of the WIPP, is the primary drainage in the region but a natural divide lies between the river and the WIPP such that the Pecos drainage system does not affect the site.

22.2.3 Construction of the WIPP salt repository Waste panels in the underground consist of seven rooms each (Fig. 22.1). Each room is about 100 m long, 10 m wide and 4 m high. Rooms are separated by 30 m thick pillars. The initial design is for eight waste panels separated from each other and the drifts by about 60 m thick pillars. Four

© Woodhead Publishing Limited, 2010

688

Geological repository systems for safe disposal

vertical shafts connect the surface to the underground: the waste shaft, the salt handling shaft, the air intake shaft and the exhaust shaft. The first three shafts have hoists capable of moving personnel, equipment and materials between the surface and the repository, although the air intake shaft is rarely used. In addition to the geological and hydrological barriers inherent to this site, when complete the WIPP will include four types of engineered barriers: shaft seals, panel closures, backfill material around the waste and borehole plugs. Borehole plugs and shaft seals limit migration of liquid and gases in the WIPP shafts and boreholes. Panel closures limit the intercommunication of brine and gases between panels. MgO was selected as the backfill material for the WIPP because it acts as a pH buffer in the range that minimizes radionuclide solubility, sorbs constituents such as water and CO2 (and Pu although no performance credit is taken) and reduces void space as it sets up as sorrel cement when contacted by the Mg-rich brine at that depth should any enter the repository. The MgO is emplaced as dry granular material in plastic bags placed on top of each waste stack, designed to break and spread the MgO around the waste as the ceiling (or back) of the room closes down on the waste. Shaft seals, borehole plugs and panel closures use durable cements and materials that possess low permeability and optimal mechanical and chemical properties under WIPP subsurface conditions.

22.2.4 Between construction and operation EPA promulgated its Environmental Radiation Protection Standards for Management and Disposal of Spent Nuclear Fuel, High-Level and Transuranic Radioactive Wastes (Federal Register, 1985) in September 1985, subsequently codified in 40 CFR 191. In 1987, the US Court of Appeals for the First District (Boston) decided a legal challenge to the EPA standards by the Natural Resources Defense Council and others, vacating and remanding to EPA for reconsideration of Subpart B of 40 CFR Part 191. This resulted in there being no repository standards applicable to the WIPP. However, the Second Modification to the Consultation and Cooperation Agreement (State of New Mexico–Department of Energy, 1987) required DOE to continue its performance assessment planning as though EPA’s 1985 repository disposal standards remained applicable. DOE notified the US Department of the Interior (DOI) in 1991 that the WIPP was ready to begin the Test Phase and the DOI authorized the DOE to transport and emplace wastes (Federal Register, 1991). The New Mexico Attorney General filed a lawsuit to stop the planned shipment (Civil Action No. 91-2527 filed in US District Court for the District of Columbia) and the US District Court issued Orders directing the DOE to cease all activities relating to the WIPP Test Phase, imposing a permanent injunction against

© Woodhead Publishing Limited, 2010

Environmental programs and public engagement

689

any transport or disposal of TRU waste at the WIPP. Test Phase activities went ahead using radioactive waste in laboratories rather than in the WIPP underground. The Land Withdrawal Act of 1992 (Public Law 102-579) established requirements for the initial receipt and permanent disposal of transuranic waste at the WIPP, specifying statutory, regulatory and other requirements. The Act designated the EPA as the primary independent regulator of the WIPP with authority over its long-term performance as a nuclear waste repository. The EPA issued Environmental Radiation Protection Standards for the Management and Disposal of Spent Nuclear Fuel, High-Level and Transuranic Radioactive Wastes in 1993 in 40 CFR Part 191 (Federal Register, 1993a, 1993b). Subsequently in 1996, the EPA issued a Final Rule (40 CFR 194) establishing criteria for use in certifying whether the WIPP complies with the standards set forth in 40 CFR Part 191 (Federal Register, 1996). These regulations provided the final set of standards and criteria for use in determining the suitability of the WIPP as a nuclear waste disposal facility. The EPA will continue to regulate the WIPP until it closes (Public Law 102-579, 1992, as amended by Public Law 104-201, 1996) and will conduct recertification every five years until closure to determine compliance. The DOE is required to disclose changes in conditions or activities that could potentially affect repository performance or result in radiological releases, and the EPA can conduct inspections of the WIPP and other WIPP-related facilities around the nation to verify compliance. The EPA issued its final certification decision in 1998 under the regulations of 40 CFR Parts 191 and 194 after the DOE’s assertion that the WIPP would comply with the EPA’s regulations and would be safe to contain TRU waste for 10 000 years (Federal Register, 1998). Following this certification, the DOE notified Congress that the WIPP was ready, and also petitioned the US District Court to lift the 1992 injunction. The WIPP received the first shipment of TRU waste on 26 March 1999. It is such meticulous and persistent work by all parties that led to the WIPP becoming operational in 1999 and to the overall success of the project since that time.

22.3

History and current status of CEMRC

CEMRC was conceived by a grassroots coalition recognizing the need for a high-quality, independent health and environmental assessment of the WIPP. Many individuals and organizations supported the formation of CEMRC including the citizens of Carlsbad and the surrounding region; the city and county government; local businesses, the New Mexico Congressional Delegation; New Mexico State University; the Carlsbad Department of Development; the New Mexico Radioactive and Hazardous Materials Committee; Westinghouse Electric Corporation; and the US

© Woodhead Publishing Limited, 2010

690

Geological repository systems for safe disposal

Department of Energy (DOE). CEMRC was initially funded for $27 million over a seven-year period (1991–1998). Subsequently, the grant was increased to about $33 million to support operations of the program until 2003. In 2004, CEMRC diversified beyond monitoring to also collaborate and provide radiological services to various customers, and obtained base funding of about $3 million per year for operations into the foreseeable future. The primary goal of CEMRC is to develop and implement an independent health and environmental monitoring program in the vicinity of the WIPP and make the results easily accessible to the public and all interested parties through posting of results on its website, publications and annual reports. CEMRC was also envisioned to function as a nucleus of environmental and radiological research supported through grants and contracts and to provide advanced training and educational opportunities. CEMRC is a 26 000 ft2 radiochemistry facility that includes environmental and general radiochemistry laboratories, a special plutonium– uranium research laboratory, an in vivo bioassay facility, mobile laboratories, computing operations, offices and classrooms (Fig. 22.4). Instrumentation includes over 100 alpha spectrometers, germanium gamma-ray spectrometers, well-geometry (germanium) gamma-ray spectrometers, gas proportional counters, liquid scintillation counters, inductively coupled plasma mass and optical spectrometers, UV–Vis spectrophotometers, ion chromatography, gas chromotographic mass spectrometers, an Nd:YAG master oscillator power oscillator (MOPO) laser system combining photoacoustic, fluorescence and breakdown spectroscopy, an X-ray diffractometer with Go¨bel mirrors and Rietveld refinements, and an unsaturated/saturated flow apparatus (ASTM D6527), along with numerous hand-held surveying and monitoring equipment. The laboratories also contain HEPA-filtered glove boxes and fume hoods, including hoods able to handle HF and perchloric acid. The bioassay and radiochemistry detection limits are unusual and generally lower than any other comparable facility, able to monitor lung and whole body constituents down to 6 keV and routinely monitoring to the femtoCurie levels for Pu and Am in environmental samples. CEMRC follows a strict Quality Assurance Program congruent with the DOE WIPP Quality Assurance Program, is regularly audited and participates in external testing programs such as the Medical and Physical Examination Program (MAPEP), the NIST Radiochemistry Intercomparison Program (NRIP) and the Oak Ridge bioassay interlaboratory comparison program. CEMRC is involved in a wide variety of outreach activities ranging from presentations for local public school students to hosting groups of visiting foreign scientists. Constituents and properties measured by the monitoring program include, but are not limited to, gross alpha/beta, 7Be, 212Bi, 213Bi, 214Bi, 144 Ce, 249Cf, 60Co, 134Cs, 137Cs, 152Eu, 154Eu, 40K, 233Pa, 234mPa, 212Pb,

© Woodhead Publishing Limited, 2010

Environmental programs and public engagement

691

22.4 (a) The CEMRC whole body counter, used for in vivo bioassay of humans, includes an 8-foot cube counting chamber constructed of 10inch-thick pre-WWII cast iron with a full graded-z shield, fabricated using a specially selected foundry. The chamber is equipped with a lung and whole body counting system composed of hyperpure Ge crystal detectors arranged in a 12 detector array, specifically to capture the 17.5 keV Pu peak. (b) Hi-Vol ambient air particulate collectors. (c) Field sampling for surface water and sediments in nearby reservoirs and lakes. (d) Field sampling for soil profiles of target radionuclides and inorganic constituents. (e) GC-MS analysis of WIPP underground ambient air and closed room air samples.

© Woodhead Publishing Limited, 2010

692

Geological repository systems for safe disposal

214

Pb, 106Rh, 125Sb, 208Tl, 228Ac, 234U, 235U, 238U, 230Th, 232Th, 228Th, Am, 238Pu, 239,240Pu, various volatile organic compounds (VOCs) and many inorganic constituents normally analyzed in air, soil and water, particularly constituents of interest to the Resource Conservation and Recovery Act (RCRA). The in vivo bioassay (whole body counting) program at CEMRC participates in the Department of Energy’s In Vivo Laboratory Accreditation Program (DOELAP) via the WIPP, and is currently accredited to perform the following direct bioassays: transuranium elements via L X-ray in lungs, 241Am, 234Th, 235U, fission and activation products in lungs including 54Mn, 58Co, 60Co and 144Ce, and fission and activation products in the total body including 134Cs and 137Cs (and 57Co, 88 Y and 133Ba). The distinct research areas at CEMRC are organized to reflect function and scientific disciplines: Internal Dosimetry (including in vivo bioassay), Radiochemistry, Field Programs, Environmental Chemistry (including inorganic and organic chemistry), Infomatics (simulation modeling and data base management) and Administration. CEMRC research areas are conducted under the philosophy of full academic freedom and independence from direct external control of research activities and outcomes. CEMRC’s monitoring program was initiated prior to the receipt of waste at the WIPP site. Each major research area was thus able to describe and quantify conditions prior to any possible releases from the repository. This baseline characterization focused on documenting the spatial and temporal patterns of physical processes, existing contaminant sources and levels, and population parameters of interest. After the arrival of the first waste shipment on 26 March 1999 CEMRC began the long-term monitoring that would accompany the operational phase of the WIPP. This operational monitoring relies on the results of the baseline characterization to identify key parameters to be monitored, laboratory methods to be employed and the most effective and efficient technologies, spatial scales and frequencies for data collection. For each separate research area, the following sequence of activities were initiated at the outset: 241

1.

2.

3.

Review published scientific literature, technical reports and any available unpublished data and consult with scientific and technical experts and other interested groups. Prepare written documentation of sources of information, status of knowledge, and any significant knowledge gaps. Develop conceptual study design and objectives, and identify the processes and patterns of interest and information to be generated by the study. Design and conduct pilot studies necessary to evaluate equipment and

© Woodhead Publishing Limited, 2010

Environmental programs and public engagement

4. 5. 6.

7.

8.

693

methods, and collect preliminary data on variability for use in development of sampling plans. Develop and implement detailed plan of sampling and analysis. Periodically analyze data generated by the study to evaluate effectiveness in terms of program objectives and modify study as appropriate. Conduct comparative analyses of different study components to produce integrative interpretations and models of patterns and processes. Prepare periodic written and graphical summaries, analyses and interpretations of data for inclusion in reports, publications and presentations. Archive data and documentation in a database to facilitate access by the public and scientific community.

The public, state and local government were engaged from the outset with many public meetings and small focus groups, as well as discipline-specific workshops and invited technical presentations, held to address individual issues and develop a widely acceptable monitoring program for the region.

22.4

Survey of factors related to contaminant exposure and perceptions of environmental risks in the region around the Waste Isolation Pilot Plant (WIPP)

In addition to people working at the WIPP, people who live and work close to the facility in Eddy and Lea counties are at risk of potential exposure for any releases of contaminants that could occur at the WIPP. It was critical to survey the regional population to determine what is important to them in terms of health and the environment so that an appropriate monitoring program could be emplaced that would address their concerns. It was also essential to obtain baselines concerning aspects of human health, the environment and citizen concerns, which change with time. There are two population centers in Eddy County within a 30 mile (48 km) radius of the WIPP. Carlsbad, 26 miles (42 km) west of the facility, is a primarily non-Hispanic white, comparatively affluent community. LovingMalaga, 18 miles (29 km) southwest of the facility, is primarily an Hispanic white, rural community. The city of Hobbs is located 40 miles (64 km) northeast of the WIPP site in Lea County. Carlsbad and Hobbs are similar in size and many characteristics. Based on distance from the WIPP, Hobbs is less likely to be affected by any release of contaminants to the WIPP and, therefore, represents a possible reference population for purposes of future comparisons with populations nearer to the WIPP. In 1995–1996, CEMRC collaborated with investigators from the

© Woodhead Publishing Limited, 2010

694

Geological repository systems for safe disposal

Epidemiology and Cancer Control Program, at the University of New Mexico Heath Sciences Center to conduct a survey of residents within selected communities in the region of the WIPP (CEMRC, 1996). The questionnaire was administered by telephone to 778 households containing 944 individuals in Carlsbad and Loving-Malaga and to 314 households containing 598 individuals in Hobbs. The survey participants were part of a random sample of published residential telephone numbers. To ensure study accuracy, interviews were conducted on weekdays and weekends at a variety of times, using computer software designed for surveys, and a verification study was performed to assess study accuracy and acceptability to survey participants. Questions in the survey were designed to elicit data on population characteristic and lifestyles that are useful for estimating the health risks associated with potential releases from the WIPP, to document extant contaminant exposure sources in the population that are not associated with the WIPP and to evaluate the level and nature of concerns about preserving the environment and protecting health. In Carlsbad and Loving-Malaga, 66% of the households contacted agreed to participate and completed an interview. In Hobbs, 69.8% of the households contacted agreed to participate and completed an interview. The average size of households surveyed was 2.5 people in Carlsbad, 2.9 people in Loving-Malaga and 2.8 people in Hobbs. The proportion of households with children was 33% in Carlsbad, 50% in Loving-Malaga and 43% in Hobbs. The proportion of Hispanic households was 10.5% in Carlsbad, 36.1% in Loving-Malaga and 17.7% in Hobbs. Most respondents in all communities lived in single-family detached dwellings, with a majority of the remainder residing in mobile homes. Non-mobile homes in Carlsbad and Loving-Malaga were usually of wood frame construction and cement slab foundations, whereas brick or cement-block construction was more common in homes in Hobbs. Indoor Rn has not been identified as a problem in southeast New Mexico, less than 15% of all households reported that they had tested their homes. Few people could recall the Rn test results. One pathway of exposure is through the ingestion of contaminated food and water. The 1990 US Census information indicated that most homes in Carlsbad and Loving-Malaga used water from private wells. In addition, approximately 7% of the surveyed households in Carlsbad and LovingMalaga, and 9.4% of surveyed households in Hobbs, used bottled water for drinking and cooking. Over 75% of the surveyed households in all three communities reported that they never ate home-raised dairy foods (eggs and milk) or meats (beef and poultry), although slightly more households in Loving-Malaga reported consumption of these foods. Home and locally grown fruits and vegetables were eaten by almost half the households in all three communities when these foods are in season. In Carlsbad and LovingMalaga, 39% of households reported eating deer and 32% reported eating

© Woodhead Publishing Limited, 2010

Environmental programs and public engagement

695

self-caught fish at least once per year. Rates were slightly lower in Hobbs, with 28% for deer and 24% for fish. Some adults in all communities reported current or past employment in occupations in which they may have been exposed to radioactive or chemically hazardous materials. In Carlsbad and Loving-Malaga, such occupations include work in underground uranium mines (where they may have been exposed to high levels of radon), nuclear fuel production, nuclear weapons production, radioactive waste handling and hazardous waste handling. In Hobbs, respondents reported work in oil and gas exploration and production, medical and dental occupations, nuclear fuel production, radioactive waste handling and hazardous waste handling. Most adults in all communities regarded the environment in their communities to be unpolluted, with air quality generally regarded to be better than that of soil and water. In general, a larger proportion considered the environmental quality of their community favorably. The air was judged to be ‘very clean’ by 44.4% in Carlsbad versus 25.4% in Hobbs, the water was judged to be ‘very clean’ by 38.4% in Carlsbad versus 23.6% in Hobbs and the soil was judged to be ‘very clean’ by 28.8% of those in Carlsbad versus 19.3% of those in Hobbs. Adults surveyed in all communities expressed a high level of concern for protection of the environment and of human health, with 70% of adults ‘concerned’ or ‘very concerned’ about pollution in the environment. High levels of concern were expressed about pesticides, contamination, chemical contaminants in food and adverse heath effects, including birth defects and cancer. Over half of the adults interviewed in all communities (61.9% in Carlsbad and 56.2% in Hobbs) indicated that they would be willing to give samples of blood or urine to support monitoring of pollution in the environment. In general, response patterns for Carlsbad and Hobbs were similar for the parameters surveyed, indicating that using future data from Hobbs as a reference population is feasible. The survey results demonstrated that members of the public in this region have a high level of concern about preserving the environment and protecting health. That the public expressed an overall willingness to give samples of blood and urine suggested that good public participation could be expected for bioassay programs to monitor for the appearance of contaminants potentially arising from releases at the WIPP. The main public interests were with determining the levels of contaminants in people and air, with concerns about drinking water second, and soil and other media a distant third (CEMRC, 1996). On the one hand it is interesting to note that aerosol release to outside air is the only likely vector for exposing the public to contamination from the repository during the operational phase, suggesting that the public was reasonably aware of the main issues on some intuitive level. On the other hand, modeling studies

© Woodhead Publishing Limited, 2010

696

Geological repository systems for safe disposal

conducted by Sandia National Laboratories (DOE, 1997a) show that contamination of ground water accessible by the public is extremely unlikely, so that the public’s interest in monitoring ground water is probably related to a general understanding of exposure pathways rather than a specific knowledge about dominant pathways of exposure from potential WIPP releases.

22.5

Internal dosimetry and whole body monitoring of area citizens

The internal dosimetry program (ID) conducts analyses and consultations for the study and management of radiation exposure on both citizens with no possibility of occupational dose and workers at the WIPP and other facilities with occupational dose. ID is the most public aspect of CEMRC operations. In the Lie Down and Be Counted (LDBC) Program, citizens within a 100-mile radius of the WIPP can simply come into CEMRC for a whole body count. Analyses include collection of information on work and residence history, past and current radiation exposure, bioassays to measure the presence of radionuclides within body tissues (in vivo) or body fluids and excretions (in vitro), and calculation of dose associated with observed body burdens. Consultations include interpretation of bioassay results, with the citizen or worker being monitored immediately after counting. The program has met the requirements and recommendations of the DOE Implementation Guide for Internal Dosimetry Programs (10 CRF 835) and the American National Standards Institute Performance Criteria for Radiobioassay (N13.30) and continues to meet the most current criteria for radiobioassay. CEMRC is also involved in the DOE Laboratory Accreditation Program for internal dosimetry and radiobioassay. The internal dosimetry program is provided as an outreach service to the public in support of education about naturally occurring radiation and CEMRC’s environmental studies, and to provide assessment of potential exposure to radioactive contaminants of concern. The program also is available to provide support to the WIPP and other regional radiological facilities such as the National Enrichment Facility by conducting bioassays for radiation workers. Full-spectrum dosimetry services are available to evaluate internal radiation exposure to radiation workers and members of the public in the case of a radiological accident or event. CEMRC’s in vivo bioassay facility includes a counting chamber measuring 8 feet68 feet68 feet, constructed of 10-inch-thick cast iron with a full graded-z shield (Fig. 22.4(a)). The cast iron composing the chamber was produced for industrial use prior to 1945 and the chamber was fabricated using a specially selected foundry, resulting in very low

© Woodhead Publishing Limited, 2010

Environmental programs and public engagement

697

background radiation from anthropogenic and naturally occurring constituents in the iron. The counting chamber is equipped with a lung and whole body counting system composed of two hyperpure Ge crystal detector arrays. The lung detector array is suspended above the subject and includes four Ge crystals. The whole body detector array is located below the counting bed and includes four Ge crystals each performing at 80% relative efficiency. Lung and whole body counts are conducted simultaneously with the subject positioned on a large bed inside the counting chamber. A dedicated computer serves the integrated electronic system for acquisition and storage of radiation spectra generated by the detectors. Resolution of the system is in the order of 450, 750 and 2100 eV at photon energies of 5.9, 122 and 1332 keV respectively. Minimum detection activities for Pu and Am in lungs are expected to be 20 and 0.1 nCi respectively and for Cs and Co in the whole body are 0.42 and 0.35 nCi respectively. The total count time for the subject is 30 minutes. Prior to undergoing the radiation count, subjects view a videotape that explains the procedure, and further explanation is provided by internal dosimetry staff who are open to all questions on any subject.

22.5.1 The Lie Down and Be Counted Program This program is provided as an outreach service to the public and to support education about naturally occurring and man-made radioactivity present in people who live in the vicinity of the WIPP. The data collected prior to the opening of the WIPP facility (26 March 1999) serve as a baseline for comparisons with periodic follow-up measurements that are slated to continue throughout the 35-year operational phase of the WIPP. Participating in the LDBC consists of a lung and whole body count every two years. Volunteers are recruited through presentations to local community groups and businesses. The entire measurement process takes approximately one hour. A detailed description of the measurement protocol, analysis and instrument detection limits is provided in the CEMRC 1998 report (CEMRC, 1998). In addition, the status of the project and annual results are available on the CEMRC website at www.cemrc.org. As of 1 January 2009, 1116 individuals from the public (citizens not receiving occupational doses) have participated in the LDBC project, using the in vivo protocol. At the time the WIPP opened, 366 of these individuals had been measured, constituting the pre-operational baseline to which subsequent results are compared; 877 counts performed after the opening of the WIPP are considered to be a part of the operational monitoring phase of the WIPP. Recounts began in July 1999 and 638 recount bioassays had been performed through 1 January 2009, while 239 new volunteers have participated in the program since 1 October 2002. In addition, while not

© Woodhead Publishing Limited, 2010

698

Geological repository systems for safe disposal

Table 22.2 Demographic characteristics of the ‘Lie Down and Be Counted’ population sample through to 31 December 2006 Characteristic

2006 sample group (margin of error)*

Census, 2000{

Gender

Male Female

49.9% (46.4 to 53.4%) 48.2% 50.1% (46.6 to 53.6%) 51.8 %

Ethnicity

Hispanic 15.2% (12.7 to 17.7%) 36.7% Non-hispanic 83.7% (81.2 to 86.3%) 63.3%

Age 60 or older

26.9% (24.2 to 29.5%) 24.5%

Currently or previously classified as a radiation worker

7.4% (5.8 to 8.9%)

Consumption of wild game within 3 months prior to count

21.7% (19.2 to 24.2%) NA

Medical treatment other than X-rays using radionuclides

7.6% (6.0 to 9.2%)

NA

European travel within 2 years prior to the count

5.6% (4.2 to 6.9%)

NA

Current smoker

13.5% (11.4 to 15.5%) NA

NA{

*

The margin of error represents the 95% confidence interval of the observed proportion; under complete replication of this experiment, one would expect the confidence interval to include the true population proportion 95% of the time if the sample was representative of the true population. { http://quickfacts.census.gov, United States Department of Commerce, Economics and Statistics Administration, Bureau of the Census. { NA = not available.

part of the LDBC Program, CEMRC has also counted over 3000 radtrained workers in the region from various contractors and related industries. Demographic characteristics (Table 22.2) of the current LDBC cohort are generally consistent with those reported in the 2000 Census for citizens living in Carlsbad. The statistics reported for the bioassay program assume that the individuals participating are a random sample of the population. Given that the bioassay program relies on voluntary participation, randomness of the sample cannot be assured and sampling appears to be biased by ethnicity. The largest deviation between the LDBC cohort and the 2000 Census is undersampling of Hispanics. However, it is important to note that if the presence of a radionuclide is dependent on a subclass of interest (gender, ethnicity, etc.), valid population estimates can still be made by correcting for the proportion of under- or oversampling for the particular

© Woodhead Publishing Limited, 2010

Environmental programs and public engagement Table 22.3 overleaf)

699

Minimum detectable activities 2007–2008 calibration (table continues Radionuclides deposited in the lungs

Radionuclide

CWT = 1.6 Energy MDA (keV) (nCi)

CWT = 2.22 MDA (nCi)

CWT CWT = 3.01 = 3.33 MDA MDA (nCi) (nCi)

AM-241 CE-144 CF-252 CM-244 EU-155 NP-237 Pu-238 Pu-239 Pu-240 Pu-242 Ra-226 Th-232 via Pb-212 Th-232 Th-232 via Th-228 U-233 U-235 Nat U via Th-234

59.50 133.50 19.20 18.10 105.30 86.50 17.10 17.10 17.10 17.10 186.10 238.60 59.00 84.30 440.30 185.70 63.30

0.21 0.55 33.57 32.80 0.33 0.58 37.02 92.12 36.19 43.66 1.94 0.18 42.46 6.01 0.72 0.12 1.92

0.29 0.70 76.77 83.42 0.44 0.78 106.40 264.73 104.00 125.46 2.37 0.22 57.54 8.07 0.89 0.15 2.59

0.16 0.47 18.63 16.37 0.26 0.44 16.27 40.48 15.90 19.18 1.75 0.15 32.50 4.57 0.61 0.11 1.45

CWT = 4.18 MDA (nCi)

CWT = 5.10 MDA (nCi)

CWT = 6.0 MDA (nCi)

0.32 0.45 0.64 0.90 0.78 1.01 1.34 1.76 107.19 261.50 686.54 1761.58 121.22 330.83 982.02 2831.38 0.49 0.66 0.91 1.25 0.88 1.22 1.73 2.42 162.23 502.57 1706.71 5650.94 403.64 1250.42 4246.37 14059.75 158.57 491.24 1668.22 5523.47 191.29 592.60 2012.45 6663.24 2.57 3.20 4.04 5.08 0.24 0.31 0.40 0.51 65.08 90.16 128.42 181.56 9.15 12.66 17.97 25.32 0.97 1.21 1.55 1.96 0.16 0.20 0.25 0.31 2.94 4.08 5.81 8.23

subclass. Baseline monitoring includes only the initial count of individuals made prior to 26 March 1999. Operational monitoring includes the counting of new individuals and the recounting of previously measured participants. Based on the data obtained thus far, there is no evidence of an increase in the frequency of detection of internally deposited radionuclides for citizens living within the vicinity of the WIPP since the WIPP began receipt of radioactive waste. As discussed in detail in the CEMRC 1998 report and elsewhere (CEMRC, 1998; Webb and Kirchner, 2000), the criterion, LC, is used to evaluate whether a result exceeds background; the use of this criterion will result in a statistically inherent 5% false–positive error rate per pairwise comparison (5% of all measurements will be determined to be positive when there is no measurable activity present in the person). The radionuclides being investigated and their minimum detectable activities are listed in Table 22.3. For the baseline measurements (N = 366), the percentage of results greater than LC were consistent with a 5% random false–positive error rate, at the 95% confidence level (1 to 9%), for all radionuclides except 232Th via the decay of 212Pb, 235U / 226Ra, 60Co, 137Cs, 40K, 54Mn, 232Th via the decay of 228Ac (Table 22.3). As discussed in detail in the 1998 report, five of these (232Th via 212Pb, 60Co, 40K, 54Mn (228Ac interference) and 232Th (via 228Ac))

© Woodhead Publishing Limited, 2010

700

Geological repository systems for safe disposal

Table 22.3

continued Radionuclides deposited in the whole body

Radionuclide

Energy (keV)

MDA (nCi)

Ba-133 Ba-140 Ce-141 Co-58 Co-60 Cr-51 Cs-134 Cs-137 Eu-152 Eu-154 Eu-155 Fe-59 I-131 I-133 Ir-192 Mn-54 Ru-103 Ru-106 Sb-125 Th-232 via Ac-228 Y-88 Zn-65 Zr-95

356 537 145 811 1333 320 604 662 344 1275 105 1099 365 530 317 835 497 622 428 911 898 1116 757

0.75 1.46 1.67 0.34 0.35 4.28 0.32 0.42 1.51 0.90 4.06 0.64 0.45 0.41 0.56 0.43 0.36 3.16 1.30 1.16 0.37 1.06 0.56

are part of the shield-room background and positive detection is expected at low frequency. 40K is a naturally occurring isotope of an essential biological element, so detection in all individuals is expected. 137Cs and 235U/226Ra are not components of the shield-room background and were observed at frequencies greater than the 95% confidence interval for the false positive error rate (discussed below in more detail). For the operational monitoring counts (Table 22.4, N = 690), the percentage of results greater than LC were consistent with baseline at a 95% confidence level (margin of error), except for 60Co and 232Th (via 228 Ac). For these radionuclides, the percentage of results greater than LC decreased relative to the baseline. This would be expected for 60Co, since the radionuclide has a relatively short half-life (5.2 years), and the content within the shield has decreased via decay by approximately 75% since the baseline phase of monitoring. The differences in 232Th (via 228Ac) results between the baseline and operational monitoring phase were also observed in 2001 and 2002 and are likely to be due to the replacement of aluminum (which tends to contain Th and U) in some of the detector cryostat components with those manufactured from low radiation background steel. 40 K results have been positive for all participants since operations began,

© Woodhead Publishing Limited, 2010

Table 22.4

‘Lie Down and Be Counted’ results through to 1 January 2008

Radionuclide

Operational monitoring In vivo Baseline counts counts (margin of error) count type (margin of error)* (data prior to (27 March 1999– 27 March 1999) N = 366{ 1 January 2008) N = 690 % of results ≥ LC{

241

Am Lung Ce Lung 252 Cf Lung 244 Cm Lung 155 Eu Lung 237 Np Lung 210 Pb Lung Plutonium isotope Lung 232 Th via 212Pb} Lung 232 Th Lung 232 Th via 228Th Lung 233 U Lung 235 U/226Ra Lung Natural uranium Lung via 234Th 133 Ba Whole 140 Ba Whole 141 Ce Whole 58 Co Whole d 60 Co Whole 51 Cr Whole 134 Cs Whole 137 Cs Whole 152 Eu Whole 154 Eu Whole 155 Eu Whole 59 Fe Whole 131 I Whole 133 I Whole 193 Ir Whole 40 K Whole d 54 Mn Whole 103 Ru Whole 106 Ru Whole 125 Sb Whole 232 Th via 228Ac Whole 88 Y Whole 95 Zr Whole 144

5.2 4.6 4.1 5.7 7.1 3.6 4.4 5.7 34.2 4.9 4.1 5.7 10.7 5.2 body 3.6 body 5.2 body 3.6 body 4.4 body 54.6 body 5.7 body 1.6 body 28.4 body 7.4 body 3.8 body 3.8 body 3.8 body 5.2 body 3.3 body 4.1 body 100.0 body 12.3 body 2.2 body 4.4 body 5.2 body 34.7 body 7.7 body 6.6

% of results ≥ LC

(4.0 to 6.4) (3.5 to 5.7) (3.1 to 5.1) (4.5 to 7.0) (5.8 to 8.4) (2.6 to 4.5) (3.3 to 5.4) (4.5 to 7.0) (31.7 to 36.6) (3.8 to 6.0) (3.1 to 5.1) (4.5 to 7.0) (9.0 to 12.3) (4.0 to 6.4) (2.6 to 4.5) (4.0 to 6.4) (2.6 to 4.5) (3.3 to 5.4) (52.0 to 57.2) (4.5 to 7.0) (1.0 to 2.3) (26.1 to 30.8) (6.0 to 8.7) (2.8 to 4.8) (2.8 to 4.8) (2.8 to 4.8) (4.0 to 6.4) (2.3 to 4.2) (3.1 to 5.1) (100.0 to 100.0) (10.6 to 14.0) (1.4 to 3.0) (3.3 to 5.4) (4.0 to 6.4) (32.2 to 37.2) (6.3 to 9.0) (5.3 to 7.9)

3.9 3.5 5.8 4.7 4.9 4.0 5.9 5.6 33.8 5.3 5.0 9.5 11.3 6.3 2.9 3.9 4.7 2.4 28.3 4.0 2.5 21.4 6.3 2.6 3.6 6.0 4.0 3.9 4.1 100.0 11.8 1.5 3.8 3.6 25.7 6.3 4.0

(3.2 to 4.6) (2.8 to 4.2) (5.0 to 6.7) (3.9 to 5.5) (4.1 to 5.7) (3.3 to 4.8) (5.0 to 6.7) (4.7 to 6.4) (32.0 to 35.5) (4.5 to 6.1) (4.2 to 5.8) (8.4 to 10.5) (10.1 to 12.4) (5.4 to 7.2) (2.3 to 3.6) (3.2 to 4.6) (3.9 to 5.5) (1.8 to 2.9) (26.6 to 30.0) (3.3 to 4.8) (1.9 to 3.1) (19.9 to 23.0) (5.4 to 7.2) (2.0 to 3.2) (2.9 to 4.3) (5.1 to 6.9) (3.3 to 4.8) (3.2 to 4.6) (3.3 to 4.8) (100.0 to 100.0) (10.6 to 13.0) (1.1 to 2.0) (3.1 to 4.5) (2.9 to 4.3) (24.1 to 27.3) (5.4 to 7.2) (3.3 to 4.8)

* The margin of error represents the 95% confidence interval of the observed percentage; under replication of this experiment, one would expect 95% of the confidence intervals to include the true population if the sample was representative of the true population. { N = number of individuals. Baseline counts include only the initial counts during this baseline period. { To determine whether or not activity has been detected in a particular person, the parameter LC is used; the LC represents the 95th percentile of a null distribution that results from the differences of repeated, pair-wise background measurements; an individual result is assumed to be statistically greater than background if it is greater than LC. } These radionuclides are present in the shield background, so they are expected to be detected periodically.

© Woodhead Publishing Limited, 2010

702

Geological repository systems for safe disposal

and range from 792 to 5558 Bq per person with an overall mean (± SE) of 2526 (± 26) Bq per person. Such results are expected since K is an essential biological element contained primarily in muscle, and a theoretical constant fraction of all naturally occurring K is the radioactive isotope 40K. The mean 40K value for males (± SE) was 3104 (± 30) Bq per person, which was significantly greater (p < 0.0001) than that of females, which was 1900 (± 22) Bq per person. This result was expected since, in general, males tend to have larger body sizes and greater muscle content than females. Detectable 137Cs is present in 23 ± 3% (95% confidence level, baseline and operational monitoring counts) of citizens living in the Carlsbad area. These results are consistent with findings previously reported in CEMRC reports and elsewhere (Webb and Kirchner, 2000). Detectable 137Cs body burdens ranged from 4.9 to 77.5 Bq per person with an overall mean (± SE) of 10.6 (± 0.6) Bq per person. The mean 137Cs body burden for males (± SE) was 11.5 (± 0.8) Bq per person, which was significantly greater (p = 0.002) than that of females, which was 8.7 (± 0.3) Bq per person. As previously reported (Webb and Kirchner, 2000) the presence of 137Cs was independent of ethnicity, age, radiation work history, consumption of wild game, nuclear medical treatments and European travel. However, the occurrence of detectable 137Cs was associated with gender where males had a higher prevalence of 137Cs relative to females. Furthermore, the presence of 137Cs was associated with smoking. Smokers had a higher prevalence of detectable 137 Cs (29.7%) as compared to non-smokers (24.1%). It is likely that the association with gender is related to the tendency for larger muscle mass in males than in females, as supported by the 40K results. The association of 137 Cs with smoking could be related to the presence of fallout 137Cs in tobacco, decreased pulmonary clearing capability in smokers or other as yet unidentified factors. These results, particularly the absence of detectable levels of plutonium, suggest that there have been no significant releases from the WIPP. As reported in previous CEMRC reports, the percentage of results greater than LC for 235U/226Ra (11%) are significantly higher than the distribution-free confidence interval for a 5% random false–positive error rate. These data are not nearly as compelling as those for 137Cs, but the large sample size of the current cohort tends to support the observed pattern. Although 235U and 226Ra cannot be differentiated via gamma spectroscopy, it is likely that the signal is the result of 226Ra because the natural abundance of 226Ra is much greater than that of 235U.

22.6

Air monitoring of geological repository systems

The CEMRC ambient aerosol monitoring studies focus on both man-made and naturally occurring radionuclides in ambient air in and around the

© Woodhead Publishing Limited, 2010

Environmental programs and public engagement

703

WIPP site as well as air exiting the WIPP underground. Particulates are collected using fixed air samplers (FAS) placed at strategic locations. All the analyses of the filters are performed according to methods detailed in CEMRC document-controlled, standard operating procedures. Special emphasis is given to the members of the actinide series that are major components of the wastes emplaced at the WIPP. The main objective for the aerosol studies presented, and for the WIPP environmental monitoring program in general, has been to determine whether the nuclear waste handling and storage operations at the WIPP have released radionuclides into the environment around the WIPP. The aerosol program also has included investigations of several non-radioactive, inorganic chemical species because the data for those substances have been found to be useful for interpreting the results of the actinide studies. Summaries of the aerosol monitoring studies have been included in prior Annual Reports from CEMRC starting in 1997 (CEMRC, 1997). Papers specifically based on aerosol research can be found in Arimoto et al. (2002, 2005, 2006). The element of particular interest for the WIPP is Pu, which has been dispersed throughout the global environment mainly by above-ground weapons testing in the 1950s and 1960s. When quantified by alpha spectrometry 239Pu is typically determined together with 240Pu, because isotopes are almost impossible to separate chemically and they have similar alpha particle energies, about 5.25 MeV. Therefore they are represented as 239,240 Pu (239Pu half-life, t1/2 = 24 110 years and 240Pu t1/2 = 6563 years). If necessary, 239Pu can be distinguished from 240Pu by accelerator mass spectrometry. This was done by CEMRC scientists in collaboration with scientists at the Lawrence Livermore National Laboratory to differentiate the general Pu signatures in soils resulting from global fallout by high-yield fusion weapons from signatures in local soils resulting from low-yield fission bombs used in the Ploughshares Program in the 1960s, one of which was carried out several miles from the WIPP at the Gnome Site (CEMRC, 2005/ 2006). Gnome-contaminated soils had a 240/239 ratio of 0.114, soils around the WIPP were 0.175, while worldwide fallout is 0.180 (Mitchell et al., 1997; Warneke et al., 2002). In addition, CEMRC alpha spectroscopy results for soils around the WIPP show ratios of 239,240Pu:241Am of 2.72, while Gnomecontaminated soils have a ratio 3.95 and worldwide fallout is about 3. The isotopic signature of WIPP waste should be closer to Gnome, being produced from defense activities producing low-yield signatures. Therefore, WIPP soils appear to reflect normal worldwide fallout and have not been measurably affected by Gnome or WIPP waste. These types of general environmental studies in the region lead to an ever-growing understanding of contaminated behavior around a repository and help to refine monitoring strategies. Another actinide of interest is 241Am (t1/2 = 432 years), which is not

© Woodhead Publishing Limited, 2010

704

Geological repository systems for safe disposal

directly produced in significant quantities during the detonation of thermonuclear weapons but rather is a daughter of bomb-produced 241Pu (t1/2 = 14.3 years). An important finding of earlier radiological studies in soils was that the activities of Pu and Am were correlated with the concentration of Al in aerosols and that this was driven by the resuspension of dust particles contaminated with radioactive fallout from past nuclear weapons tests. Related studies of soils collected on and near the WIPP site have shown that correlations exist among Al and both naturally occurring and bomb-derived radionuclides including 239,240Pu (Kirchner et al., 2002). The sampling design for the ambient aerosol studies has changed over the course of the project, and detailed information regarding the sampling design has been presented in prior CEMRC reports starting in 1998 (CEMRC, 1998). Samples for the ambient aerosol/radionuclide studies have been collected using high-volume samplers (‘hivols’, flow rate ~1.13 m3/min). Three long-term aerosol sampling stations have been established; these are On Site, Near Field and Cactus Flats (representing the far field), and each supports a hivol sampler for collecting total suspended particulate (TSP) matter (Fig. 22.4(b)). The Near Field and Cactus Flats stations also supported a second hivol sampler for a time, which were used for studies of PM10, particulate matter with an aerodynamic equivalent diameter less than 10 μm. A fourth set of samples was collected at Hobbs over a period of approximately a year and a half, but the sampling there was discontinued in April 2002. Until the end of March 2002, both low-volume samplers (‘lovols’, 10 l/min) and Graseby–Anderson dichotomous samplers (dichots) were used for collection of aerosols for the studies of non-radioactive, inorganic constituents, specifically trace elements and selected water soluble ions. The monitoring program underwent major restructuring in FY 2002, and afterwards sampling for the non-radiological aerosol analytes was done using dichots exclusively. In November 2004, the collection of aerosols by dichots was discontinued as a result of funding constraints. The high-volume samples are analyzed for selected radionuclides, including 238Pu, 239,240Pu and 241Am following 4 h of heating in a muffle furnace at 500 8C, which drives off organics; dissolution of the material on the filters using strong acids (HF, HCl and HClO4); and multiple precipitation, co-precipitation and ion-exchange and/or extraction chromatography steps. The nuclides of interest were precipitated with LaF3, deposited onto filters, mounted on planchettes and counted using an Oxford Oasis alpha spectroscopy system. The sampling strategy for the aerosol/radionuclide studies has been to collect as much particulate material as reasonably practical so as to maximize the chances of detecting the radionuclides of interest. Individual samples typically have been collected over periods of 3 to 5 weeks depending

© Woodhead Publishing Limited, 2010

Environmental programs and public engagement

705

on the rate at which the sample filters become loaded. For these studies, high-volume samples are collected on 20 cm625 cm Gelman A/E™ glass fiber filters. Gravimetric measurements of the glass fiber filters are made to determine the mass of aerosol material that accumulates over the sampling interval. For monitoring of the WIPP underground air, there are three shroudedprobe aerosol FAS samplers at a location designated as Station A. These are located on three separate sampling skids denoted A1, A2 and A3 (Fig. 22.5 (a)). Station A is an above-ground air sampling platform shared with several other groups, including the site contractor (Washington TRU Solutions, WTS) and the New Mexico Environment Department (NMED). Sampling operations at Station A provide a way to monitor for releases of radionuclides and other substances in the exhaust air from the WIPP. In addition, if radioactive materials were to be released from the facility, the Station A data also would be invaluable for reconstructing exposure scenarios. CEMRC commenced sampling of the WIPP exhaust air at Station A on 12 December 1998. In April 2001, primary sampling operations were transferred from Skid A1 to A3 (south skid) to reduce problems associated with water infiltration and precipitation in the exhaust shaft. The FAS sample filters are normally changed daily except on holidays, when a filter will run for multiple days. The aerosol sampling operations at Station A have at times been hampered by filter clogging, and during one interval (24 January 2000 to 28 November 2001) CEMRC and the other organizations changed filters twice daily Monday through Friday. Daily sampling resumed when the mass concentrations decreased and flow rates improved. However, occasionally more than one sample per day is still collected. If the flow rate on any of the sampler legs drops below 1.8 cfm (0.06 cubic meters per minute), a low-flow alarm on the sampler is activated and the filters are changed. After 2003, CEMRC implemented an additional filter, called the Trip Blank, which is a blank filter that accompanies the sample filter through all of the process, including transport to and from the WIPP site, and is placed on the collector for 15 seconds, then removed. Unlike laboratory and reagent blanks, the Trip Blank is meant to reflect sampling errors or field contamination that is independent of laboratory procedures. Samples are collected on 47 mm diameter membrane filters with the use of a shrouded probe, commonly referred to as a fixed air sampler or FAS. The airflow through the FAS is approximately 170 liters per minute. From a practical standpoint, Station A is located where radioactive or hazardous materials would most likely first be detected in the event of a release. After the samples are returned to the laboratory, the individual filters are first weighed to determine mass loadings, and after allowing for the decay of

© Woodhead Publishing Limited, 2010

706

Geological repository systems for safe disposal

© Woodhead Publishing Limited, 2010

22.5 (a) Fixed air samplers (FAS) at Station A continuous sampling of air exiting the WIPP underground after having moved over the nuclear waste containers. BU, backup; SOR, skid of record; XO, extra probe. Time series plots of (b) gross alpha and (c) gross beta activity concentrations of aerosols on Station A FAS filters. (d) Time series plot of aerosol mass loadings on the Station A FAS filters. Closed circles denote predisposal samples (before 26 March 1999) and open circles are for operational samples. Notice the two high points in January 2001 that resulted from an accidental release of a fire extinguisher underground in the WIPP.

Environmental programs and public engagement

© Woodhead Publishing Limited, 2010

707

708

Geological repository systems for safe disposal

short-lived radon daughters, they are counted for gross alpha/beta activities for 1200 minutes using a low-background gas proportional counter (LB4100, Canberra, and more recently starting in April 2006, a Protean MPC9604). During a study to investigate fouling of the sample probes, the count times were reduced to 480 minutes to accommodate additional samples from the experimental unit used in some studies of probe-fouling. In preparation for that study, data from the back-up FAS sampler were collected to determine whether gravimetric and gross alpha/beta data were shown to be comparable to the data obtained with the sampler of record, and showed that they were (CEMRC, 2005/2006). The gross alpha and beta activities are expressed in the following two ways. First, the activity concentration is calculated as the activity per unit volume of air sampled (Bq/m3). Second, activity density is calculated as the activity per unit aerosol mass collected (Bq/g). The minimum detectable activity concentrations and densities for the gross alpha emitters are ≈ 16107 Bq/m3 and ≈ 0.7 Bq/g respectively, while for gross beta emitters the corresponding values are ≈ 26107 Bq/m3 and ≈ 1.7 Bq/g. Elemental and gamma-ray analyses are conducted on weekly composites of the filters. Quarterly composites were initially used for the determination of actinide activities, but monthly compositing was implemented in July 2004 to be able to compare directly with other monitoring programs. Individual FAS filters are digested using a mixture of strong acids in a microwave digestion unit, and weekly composites are prepared from the digestates of the individual filters. Weekly composites are then analyzed for a suite of trace elements with the use of Perkin-Elmer Elan inductively coupled plasma–mass spectrometry (ICP–MS). The ICP–MS methods can provide data for up to ~35 elements, but in practice the concentrations of some elements, including As, Be, Cd, Er, Eu, Sc, Se, Sm, Tl and V, are often below detectable or quantifiable levels, and a second set of elements (notably Ag, Li and Sn) has variable concentrations in blank filters, which makes their quantification difficult. Analyses of gamma emitters are performed on the same weekly composites as used for the elemental studies; the gamma analyses are done using a low-background, high-purity Ge well detector and a count time of 24 hours. Information, photos and procedures on these techniques can be found at www.cemrc.org. Finally, monthly composites are prepared from the weekly composites, and these are used for determination of actinide activities. Only one half of the composite sample is normally used for determination of the actinide activities. The remaining aliquot is archived. The composite sample is evaporated to dryness and the residue is digested in perchloric acid to destroy the black residue, which consists mostly of diesel exhaust particulates from underground operations. This process ensures that fluorine is completely removed and all traces of organic filter residue have

© Woodhead Publishing Limited, 2010

Environmental programs and public engagement

709

been oxidized. The actinides are then separated as a group by coprecipitation on Fe(OH)3. After dissolution, Pu, U and Am are separated by anion exchange and extraction chromatography, and the sample planchettes are finally prepared for alpha spectrometry using rare-earth micro-co-precipitation. Count times for alpha spectrometry are unusually long, five days, in order to reduce the detection limits.

22.6.1 Results from fixed air samplers of WIPP underground air The essence of the strategic design for the WIPP monitoring program, including the studies at Station A, has been to compare pre- versus postdisposal data. The first radioactive waste shipments were received at the WIPP on 26 March 1999, and this is considered the cut-off date separating the pre-disposal phase from the post-disposal or operational phase. The WIPP first received mixed waste on 9 September 2000, and therefore data for samples collected prior to that date compose a pre-mixed waste baseline for the elemental data while those collected afterwards are considered operational. Figures 22.5(b) and (c) show the gross alpha and beta activities in Station A samples from 1998 to 2009. The bulk of the activity in pre-disposal or preoperational samples results from naturally occurring radioactive materials, specifically radon daughters. The pre-operational gross alpha activity densities and concentrations were both high compared with the annual mean values for the next five years. This is in large measure due to the fact that the gross alpha activities exhibit clear seasonal variability with peaks occurring in winter (Fig. 22.5(b)), and the pre-disposal samples were collected at that time of year. An especially pronounced annual cycle in alpha activity concentrations, with high values in December and January and low values mid-year is seen in 2004 to 2005. After 2005, activities appear to have gone back up to pre-operational levels. Generally similar trends occur for gross beta data (Fig. 22.5(c)). One sample that stands out during this time is the maximum beta activity concentration of 0.058 Bq/m3 observed in early 2001. This sample and another collected around the same time (the two highest points in Fig. 22.5(c)) resulted from contamination by material released from an underground fire extinguisher. While the activities of the alpha and beta emitters have not changed greatly since the inception of the studies, the gross alpha activities appeared to decrease slightly after the WIPP became operational and then in 2003 began to increase again to pre-disposal levels. The reported gross alpha and beta activities are normalized by dividing the measured activities by the mass loadings on the sample filters or by the volume of air sampled. Therefore

© Woodhead Publishing Limited, 2010

710

Geological repository systems for safe disposal

trends in the former, that is the activity densities, could either be due to changes in the amount of radioactivity in the sample or the aerosol mass in the samples (the volumes of air sampled, which are not shown, have changed little during the course of the program and so there should be little or no effect on the activity concentrations). Aerosol mass loadings (Fig. 22.5(d)) show a slight trend towards lower sample masses beginning in 2004 and also less scatter in the gravimetric data. The latter trend is also evident in the relative standard error, i.e. the standard error divided by the arithmetic mean and expressed as a percentage, which was ≤ 8.1% in the last five years of the study compared with 10 to 20% in three of the first four years of the program. This decrease in aerosol mass loadings would directly contribute to the high alpha and beta activity densities observed in the most recent years of the monitoring. The sensitivity of the monitoring program at Station A was dramatically demonstrated in January 2001 when the CEMRC found elevated gross beta radioactivity in the FAS sample filters (the two highest points in Fig. 22.5 (c)). Follow-up investigations eventually traced the source of the beta emitters to the discharge of a fire extinguisher underground, but the incident was more notable because it demonstrated for the first time the ability of the monitoring system to detect a non-routine event. A second, more significant incident occurred when scientists from CEMRC reported that they had detected a small quantity of Pu in a composite aerosol sample from the second calendar quarter of 2003. This discovery was later corroborated by contractor compliance monitoring and other environmental monitoring programs, Washington TRU Solutions (WTS) and the Environmental Evaluation Group (EEG), through the analyses of samples that were independently collected and analyzed. The detection of Pu in the exhaust air led to the issuance of a CEMRC report to the US Department of Energy (CEMRC, 2007) and a briefing presented to the New Mexico Environment Department. The activity was extremely low (hundreds of femtograms of Pu) and well within historic background, but indicated the ability of the monitoring program to detect radionuclides of interest at any level above the minimum detectable concentration or activity (MDC or MDA). Recently, CEMRC has again detected a small quantity of Pu in two composite aerosol samples from the first and third calendar quarters of 2007 similar to the 2003 detection, and also corroborated by WTS. Such small occasional detections are to be expected in a Pu repository that is filling and the 2003 and 2007 hits provide a baseline for future events. The concentrations are so low (all values are orders of magnitude below compliance or action levels, tens of counts per 5000 minutes) that it is impossible to determine the origin, whether from dust particulates electrostatically attached to the outside of containers or equipment, external dust from fallout and the nearby Gnome site chromatographically moving through the underground over years, or Pu actually coming from the waste.

© Woodhead Publishing Limited, 2010

Environmental programs and public engagement

711

Like so much involved in nuclear and environmental issues, detection at these levels becomes a philosophical issue – how low is low enough? Society’s obsession with unachievable goals like zero concentrations or zero activities come up against the reality of the physical world. The choice for CEMRC to monitor at levels orders of magnitude below action or compliance levels, even below background, raises the question as to what does this mean? What should be done, if anything, when positive values are observed? At these levels, even laboratory contamination using traditional procedures becomes more important than for normal situations. It is hoped that performing such measurements will spark scientific and societal discussion about the nature of contaminants and environmental monitoring in general.

22.6.2 Ambient aerosol monitoring in the vicinity of the WIPP Data for the high-volume ambient aerosol samples (hi-vols) are presented in Table 22.5 and Fig. 22.6 for 1998 to 2007. 238Pu was infrequently detected, with activity concentrations slightly above minimum detectable levels in only 15 of the 227 samples. 239,240Pu was above detection limits in 223 of the 227 samples. As in prior years, the 239,240Pu activity concentrations showed a strong annual cycle with activities greatest in the spring (Fig. 22.6(a)). During most of the years studied, the peak 239,240Pu activities generally occur in the March to June timeframe, which is when strong and gusty winds in the area frequently give rise to blowing dust. Some samples taken at Cactus Flats in 1999 and 2000 and at On Site in 2004 exhibited slightly higher 239,240Pu activity concentrations (Fig. 22.6(a)) than surrounding data points. The points correspond with higher activity densities as well, i.e. greater activity per gram of dust, indicating changes in dust composition and origin. However, insufficient auxiliary data are available for attributing a cause to this result. The activity concentrations of 241Am in the highvolume samples closely tracked those of 239,240Pu (Fig. 22.6(b)). Most notably, strong springtime peaks in 241Am activity concentrations were evident in the samples from 2001 through 2002 and 2004 through 2005. Data from 2003 do not exhibit these springtime peaks. In general, actinide concentrations were highest for the Cactus Flats station, the farthest from the WIPP site but in an area having higher soils concentrations of actinides than found in the area around the WIPP facility. On the other hand, the aerosol mass loadings at the On Site station were generally the highest of the three stations with comparable data sets (Table 22.5). Aerosol mass loadings at all stations tend to track one another remarkably well, but during several extended periods, most noticeably

© Woodhead Publishing Limited, 2010

712

Geological repository systems for safe disposal

22.6 Ambient aerosol data for collected particulates from high-volume samplers On Site (at the WIPP), Near Field (1 km from the WIPP exhaust shaft) and Off Site (13 km from the WIPP at Cactus Flats): (a) 239,240Pu activity concentration (Bq/m3), (b) 241Am activity concentration (Bq/m3). Note the strong seasonal correlation.

January 1999 to July 2000, July 2001 to January 2002 and January 2004 to January 2006, the mass loadings at On Site were consistently higher than at the other sites (CEMRC, 2007). As a consequence of the similar 239,240Pu activity concentrations at all

© Woodhead Publishing Limited, 2010

Environmental programs and public engagement

713

Table 22.5 Summary statistics for aerosol mass loadings and actinide activities in high-volume aerosol samples Station

Cactus Flats Near Field On Site

Type of sample

TSP

TSP

76 75 28.4 14.2 38 5.06109 3.2 6 109 37 1.86104 5.76105 6 8.26109 1.46109 6 3.56104 6.76104 76 1.66108 1.26108 75 5.36104 2.06104

75 73 25.6 10.3 35 4.06109 2.06109 34 1.56104 6.26105 2 1.26109 4.661010 2 3.86105 5.16106 72 1.26108 7.56109 72 4.56104 1.46104

Number of samples Aerosol mass, micrograms per cubic meter 241

Am activity concentration, Bq/m3 241

Am activity density, Bq/g

238

Pu activity concentration, Bq/m3 238

Pu activity density, Bq/g

239,240 Pu activity concentration, Bq/m3 239,240

*

Pu activity density, Bq/g

N* mean StdDev N mean StdDev N mean StdDev N mean StdDev N mean StdDev N mean StdDev N mean StdDev

TSP 76 73 34.8 14.9 41 4.06109 2.26109 41 1.36104 5.76105 7 2.86109 1.76109 7 7.16105 3.86105 74 1.36108 7.96109 75 3.66104 2.16104

N stands for number of samples with masses or activities above detection limits.

stations and the higher mass loadings at On Site, the activity densities (Bq/g) at On Site tended to be lower than at Cactus Flats or Near Field (Table 22.5). The combination of 239,240Pu and gravimetrics data thus suggest that WIPP operations and mining generate detectable levels of aerosol particles sampled by the On Site station, but those particles contain less 239,240Pu than typical ambient aerosols. These are most probably particles from construction dusts or salt from the underground operations and tend to dilute normal fallout signatures. The results presented here for aerosols strongly indicate that there has been no significant increase in actinide concentrations since WIPP began receiving waste.

22.6.3 Elemental data from WIPP underground air Prior studies at Station A have shown that the concentrations of hazardous metals and various trace elements can be highly variable over time; this was true even in the samples collected prior to receipt of the mixed waste in

© Woodhead Publishing Limited, 2010

714

Geological repository systems for safe disposal

22.7 Concentration of selected elements in WIPP exhaust air from Station A FAS samplers: (a) Al, Mg, (b) U, Th, Pb, Cd.

September 2000. Timeseries plots of selected trace element data are presented in Figs 22.7(a) and (b). No marked differences are evident in the baseline versus operational samples. Aluminum is of interest because of relationships observed between the Al concentrations in ambient aerosols and the activities of 239,240Pu and 241Am (Arimoto et al. 2002, 2005, 2006). Windblown dust is the main source for Al and many other elements (Fe,

© Woodhead Publishing Limited, 2010

Environmental programs and public engagement

715

Mn, Sc and the rare earth elements) and also represents a source for U, some other naturally occurring radionuclides and fallout radionuclides such as Pu and Am. Kirchner et al. (2002) have also shown relationships between Al and various radionuclides, both artificial and naturally occurring, in soils. Magnesium is of interest as it is the primary component in the MgO backfill material that is the only engineered barrier at the WIPP. Several potentially toxic elements, i.e. Pb, Cd, U and Th, which are components of the WIPP mixed waste, were already present in measurable amounts in the WIPP aerosol effluent prior to the receipt of mixed waste. The concentrations of these elements, too, change with season and over the course of the monitoring program. Most important, there is no evidence for a long-term increase in the concentrations of any of these elements that can be linked to the WIPP operations in any way.

22.7

Future trends

The two primary future issues of interest to monitoring are how the repository will be closed and how long does one monitor a closed repository. WIPP is about half-full as of this writing, and can be expected to continue for about 15 to 25 more years. Disposal operations may continue for some time afterwards to accommodate the small amount (2% by volume) of TRU waste generated after the emplacement of the legacy waste. Therefore, final closure is many years away. The closure process proposed by the DOE and accepted by the EPA and the State of New Mexico (DOE, 1996b) called for single-pour concrete structures in the access drifts at the entrance and exit to each panel, each about 30 m610 m64 m. This will be onerous, if not impossible, to achieve in the underground, and should not be necessary considering the creep-closure properties of the salt (DOE, 1997b). Discussions among DOE, the State and the EPA are ongoing and temporary closure includes the construction of a 12-ft thick explosionproof block wall. Whether poured concrete, backfilled salt, or a combination, will fill the remaining access drifts remains to be determined by the relevant parties. Future monitoring of a deep underground geological repository is truly more of a philosophical issue than a technical one. Certainly, continued monitoring for the duration of the operational phase is warranted as monitoring can reveal operational problems, accidental releases or other issues that might be addressed while operations are ongoing. However, once the repository is filled and closed, it is unclear what additional monitoring would accomplish. For a short time, e.g. several years, monitoring might provide confidence that the closure was successful and the repository is performing as desired. In addition, subsequent disposal projects in the region may benefit from continued monitoring. However, probabilistic

© Woodhead Publishing Limited, 2010

716

Geological repository systems for safe disposal

models of performance assessment suggest that there will be no mean releases exceeding compliance levels over the subsequent 10 000 years (DOE, 2009) and that releases are contingent on the penetration of the repository by drilling. Therefore, it is unlikely that monitoring will provide further useful information beyond this confidence period of five or ten years. This decision is best left to the public in concert with the appropriate State and Federal agencies.

22.8

Conclusions

The Salado Formation, a massive bedded salt deposit near Carlsbad, New Mexico, is host to permanently disposed nuclear waste at the WIPP site. The geological, mechanical and chemical properties of the Salado salt are ideal for permanent deep geological disposal. From the standpoint of addressing operational and environmental risk, the WIPP has had extensive monitoring of human health and the environment, beginning from before operations, on the public, waste disposal workers, aerosols, water and soils. Public and regulatory acceptance of the WIPP before and after operations began have been tied strongly to an aggressive monitoring program by several groups, including the site contractor and State and independent groups. Public surveys of the region around Carlsbad showed that the main public concerns were with people and air, with drinking water second and soil and other media a distant third. It is interesting to note that aerosol release to outside air is the only likely vector for contamination from most repositories during the operational phase, suggesting the public was reasonably aware of the main issues on some intuitive level. The Carlsbad Environmental Monitoring and Research Center (CEMRC) at New Mexico State University, located in Carlsbad, New Mexico, is an academic, independent science-based monitoring facility and program developed to monitor the environment around the WIPP to below background levels for radionuclides and other constituents of interest to the regional community. The program was developed in conjunction with Department of Energy, contractor, state and local government, and regional citizens to monitor air, soil, water and people in a 100-mile (160-km) radius around the WIPP. After ten years of continuous operations, there is no evidence of increases in radiological contaminants in the region that could be attributed to releases from the WIPP.

22.9

Acknowledgments

This work is funded by an ongoing grant DEFC29-91AL74167 from the US Department of Energy through their Carlsbad Field Office. Additional support has been provided by the City of Carlsbad. CEMRC would like to thank the people of Carlsbad, Hobbs and the region, the State of New

© Woodhead Publishing Limited, 2010

Environmental programs and public engagement

717

Mexico, the US Department of Energy and the US Environmental Protection Agency for strong, active and critical support in a multigenerational and multi-disciplinary effort.

22.10 References Arimoto R, Kirchner T, Webb J, Conley M, Stewart B, Schoep D and Walthall M (2002), ‘239,240Pu and inorganic substances in aerosols from the vicinity of the Waste Isolation Pilot Plant: the importance of resuspension’, Health Physics, 83, 456. Arimoto R, Webb J and Conley M (2005), ‘Radioactive contamination of atmospheric dust over southeastern New Mexico’, Atmospheric Environment, 39, 4745–4754. Arimoto R, Stewart B, Khaing H and Tatro D P (2006), ‘Biogeochemical recycling on aerosol particles’, Eos Trans. AGU, 87(52), Fall Meeting Supplement, Abstract A53A-0175. Beauheim R L and Roberts R N (2002), ‘Hydrology and hydraulic properties of a bedded evaporite formation’, Journal of Hydrology, 259, 66–88. CEMRC (1996), Carlsbad Environmental Monitoring and Research Center Annual Report, New Mexico State University, Carlsbad, New Mexico, http://cemrc. org/reports/96rept/1996Report.pdf. CEMRC (1997), Carlsbad Environmental Monitoring and Research Center Annual Report, New Mexico State University, Carlsbad, New Mexico, http://cemrc. org/reports/97rept/1997Report.pdf. CEMRC (1998), Carlsbad Environmental Monitoring and Research Center Annual Report, New Mexico State University, Carlsbad, New Mexico, http://cemrc. org/reports/98rept/1998Report.pdf. CEMRC (2005/2006), Carlsbad Environmental Monitoring and Research Center Annual Report, New Mexico State University, Carlsbad, New Mexico, http:// cemrc.org/reports/0506rept/index.html. CEMRC (2007), Carlsbad Environmental Monitoring and Research Center Annual Report, New Mexico State University, Carlsbad, New Mexico, http://cemrc. org/reports/07rept/index.html. Conca J L, Apted M J and Arthur R C (1993), ‘Aqueous diffusion in repository and backfill environments’, in Scientific Basis for Nuclear Waste Management XVI, Materials Research Society Symposium Proceedings, La Grange, Illinois, 294, 395–402. DOE (1996a), Title 40 CFR Part 191 Compliance Certification Application for the Waste Isolation Pilot Plant, US Department of Energy, Washington, DC. DOE (1996b), Detailed Design Report for an Operational Phase Panel-Closure System, US Department of Energy, DOE/WIPP 96-2150, Carlsbad, New Mexico. DOE (1997a), WIPP Disposal Phase Supplemental Environmental Impact Statement, US Department of Energy, DOE/EIS-0026-S-2, Washington, DC. DOE (1997b), Waste Isolation Pilot Plant Safety Analysis Report, US Department of Energy, DOE/WIPP-95-2065, Revision 1, March, Carlsbad, New Mexico. DOE (2009). Supplemental Analysis for the Waste Isolation Pilot Plant Site Wide

© Woodhead Publishing Limited, 2010

718

Geological repository systems for safe disposal

Operations, US Department of Energy, DOE EIS-0026-SA-07, Carlsbad, New Mexico. Federal Register (1985), 50(No. 182; September 19), 38066. Federal Register (1991), 56(No. 196; October 9), 50923. Federal Register (1993a), 58(No. 26; February 10), 7924. Federal Register (1993b), 58(No. 242; December 20), 66398. Federal Register (1996), 61(No. 28; February 9), 5224. Federal Register (1998), 63(No. 95; May 18), 27354. Griffith J D, Willcox S, Powers D W, Nelson R and Baxter B K (2008), ‘Discovery of abundant cellulose microfibers encased in 250 Ma Permian halite: a macromolecular target in the search for life on other planets’, Astrobiology, 8, 1–14. Holdoway K A (1972), ‘Petrofabric changes in heated and irradiated salt from Project Salt Vault, Lyons, Kansas’, Union Carbide Corp., Oak Ridge, Tennessee. Kirchner T, Webb J, Webb S, Arimoto R, Schoep D and Stewart B (2002), ‘Variability in background levels of surface soil radionuclides in the vicinity of the Waste Isolation Pilot Plant’, Journal of Environmental Radioactivity, 60, 275–291. Mitchell P I, Vintro L L, Dahlgaard H, Gasco C and Sanchez-Cabeza J A (1997), ‘Perturbation in the 240Pu/239Pu global fallout ratio in local sediments following the nuclear accidents at Thule (Greenland) and Palomares (Spain)’, Science of the Total Environment, 202, 147–153. NAS–NRC (1957), ‘Disposal of radioactive waste on land’, Report by the Committee on Waste Disposal, Division of Earth Sciences, The National Academies Press, Washington, DC. ORNL (1973), ‘Site selection factors for the bedded salt pilot plant’, ORNL-TM4219, Oak Ridge, Tennessee. Public Law 102-579 (1992), Waste Isolation Pilot Plant Land Withdrawal Act, Washington, DC. Public Law 104-201 (1996), National Defense Authorization Act for Fiscal Year 1997, Washington, DC. Sandia Laboratories (1978), ‘Geological characterization report: Waste Isolation Pilot Plant (WIPP) Site, southeastern New Mexico’, SAND78-1596, Albuquerque, New Mexico. Sandia National Laboratories (1991), ‘The advantages of a salt/bentoninte backfill for Waste Isolation Pilot Plant disposal rooms’, SAND90-3074/UC-721, Albuquerque, New Mexico. State of New Mexico–US Department of Energy (1987), Second Modification to the 1981 Agreement for Consultation and Cooperation with DOE and the State of New Mexico on WIPP, Santa Fe, New Mexico. Warneke T, Croudace I W, Warwick P E and Taylor R N (2002), ‘A new groundlevel fallout record of uranium and plutonium isotopes for northern temperate latitudes’, Earth and Planetary Science Letters, 203, 1047–1057. Webb J L and Kirchner T B (2000), ‘An evaluation of in vivo sensitivity via public monitoring’, Radiation Protection Dosimetry, 89(3–4), 183–191. Wentz C J (1999), Chronology of WIPP, Radioactive Waste Consultation Task Force, State of New Mexico, Available from http://www.emnrd.state.nm.us/ wipp/chronolo.htm (Accessed 11 November 2009).

© Woodhead Publishing Limited, 2010

23 Methods for social dialogue in the establishment of radioactive waste management programmes K . A N D E R S S O N , Karita Research AB, Sweden

Abstract: This chapter shows how, and why, efforts have been made worldwide to promote a social dialogue about radioactive waste management that can lead to high-quality and broad-based decision-making. Different processes of participation and initiatives to enhance transparency are described within a context of social dialogue that includes different approaches to risk governance. It is concluded that arenas of participation and transparency need to be connected to national and local forms of democratic systems within which decisions on the final disposal of nuclear waste are ultimately taken. It is recognized that social dialogue must cover both technical as well as non-technical issues. Key words: transparency, radioactive waste management, risk governance, public participation, environmental impact assessment.

23.1

Introduction

Radioactive waste management is an icon for scientifically complex problems that need both technical and political solutions. It covers a wide range of disciplines in natural sciences and technology, such as radiological health physics, geology, hydrology, mechanics, chemistry and metal corrosion. It must also embody social sciences and a high level of ethical and value-laden considerations. There have been many failures but also success stories over a thirty-year period, during which numerous radioactive waste management programmes have commenced, made some progress, been halted and failed (mostly on the basis of lack of social acceptance), and then been re-started with new approaches sometimes only to encounter new sets of problems. The issue of nuclear waste, being part of the overall debate about nuclear energy, has at 719 © Woodhead Publishing Limited, 2010

720

Geological repository systems for safe disposal

times been controversial – it has even contributed to the fall of governments, as was the case in Sweden during the late 1970s. Complex societal issues often have a tendency to divide rather than unite the various different stakeholders concerned. The issues cover so many aspects that it is impossible for any one person – experts are no exceptions – to really grasp all implications. Therefore, complex issues can be debated for a long time without really developing or progressing in understanding and resolution. Debates on nuclear waste management, as is the case also in many other modern scientific-age areas, such as genetic modification and radiation from telecom antennas, tend to see participants fixed in their positions already at an early stage. Some of the problems in society’s dealings with these issues are: .

.

.

The narrow perspective of some stakeholders, including the expert community, who try to make the issues framed as only technical/ scientific, while the issues also contain value-laden and ethical aspects, which are equally relevant and valid in the social perspective and the political context. Stakeholders from both advocate and opposition positions are well aware that it is indeed difficult to form a comprehensive picture of a complex problem – therefore, they can present a skewed or partial picture which may favour their own position in the societal debate. Then the public’s possibility to form its own opinions can be restricted. Media saturation, in which advertising and entertainment dominate everyday life in modern society, where thousands of messages via numerous channels flood citizens. As people’s attention span is limited, complex messages become more or less ignored.

In this chapter we show how efforts have been made and how methods have been developed to overcome these problems to promote a social dialogue about radioactive waste management that can lead to high-quality and broad-based decision-making. It is important to recognize that this social dialogue will cover both technical as well as non-technical issues. There are socioeconomical factors of importance for stakeholders but safety issues are a basic concern for all as well as issues such as time scales of the radioactive waste management programmes and what is economically reasonable. The core message is that different concerns are listened to and taken into account, and that a fruitful dialogue with insight and transparency is established between the technical and scientific expertise, on the one hand, and ‘lay people’, on the other hand. The overall framing should not be narrowly done by expertise only, and not only by other stakeholders either, so that it becomes socially narrowly framed. Also, this chapter is not restricted to issues dealing with long-term safety having to do with final disposal but the social dialogue also needs to cover, for example, more

© Woodhead Publishing Limited, 2010

Establishment of radioactive waste management programmes

721

short-term issues, both technical ones, such as operational safety, and nontechnical ones, such as the social impact of a repository siting.

23.2

The emergence of participation in nuclear waste management

During the 1960s and 1970s, decision-makers in industry and government met increasing concerns among the public about environmental impact and sustainability (see, for example, Dunlap, 1997). Large industrial facilities and infrastructure programmes met strong opposition. There was a lack of understanding of the new social environment that resulted in a distrust of industry and government. Failures of scientific assessment and arguments to persuade stakeholders resulted in a range of outcomes from public protest and the expensive deferment, reversal and review of projects, to the longerterm erosion of trust and confidence in both industry and regulators. The traditional ‘DAD’ (decide, announce and defend) approach to decisionmaking was found inadequate and even unacceptable in many circumstances. This general trend in moving away from DAD has been certainly true for nuclear waste management as well. Social science research described this change early but it took some time before legislators and industry took it seriously. Initially, more and better technical information was seen as the solution. This strategy also failed to overcome concerns because the approach was often reduced to a matter of ‘us and them’ and involved no sharing of values or participation and ownership by concerned people in the decision-making process. The need to get away from the DAD approach triggered a move towards participation and inclusiveness. Now, waste management organizations have progressively changed their attitudes and interaction with concerned publics. New approaches to participation and transparency have emerged in this sector, which is now being disseminated to other fields. At the same time, however, several countries are still looking for a way forward and for processes to reach solutions with public support. During recent years, progress towards solutions on waste management has been made in several national programmes, most notably the repository siting programmes in Finland and Sweden, but also the deliberations of the UK Committee on Radioactive Waste Management (CoRWM), followed by consultations on the UK ‘Managing Radioactive Waste Safely’ report (Department for Environment, Food and Rural Affairs, 2006, 2008), as well as new French laws in 2006 on nuclear transparency and repository development. When considering recent progress in Europe it should be noted that in the United States, the Waste Isolation Pilot Plant (WIPP) has been in operation

© Woodhead Publishing Limited, 2010

722

Geological repository systems for safe disposal

since 1999 with support from local authorities and the majority of the population. The WIPP is the first facility in the USA for deep geological disposal of transuranic radioactive waste (TRU) coming from the US defense nuclear programme. It has been developed by the US Department of Energy in southeastern New Mexico, east of Carlsbad. The operation was able to start as the US Environmental Protection Agency (EPA) had certified in May of 1998 that the WIPP should comply with EPA standards for long-term isolation of transuranic waste. Now, both the Federal Government and the State Government (New Mexico Environment Department, on NMED) regulate the WIPP as the EPA has shifted authority for certain aspects of the WIPP to the State. Both the EPA and NMED hold public hearings before any major decisions on modifications and occasional oversight are made. Also, every fifth year the EPA verifies that changes in the facility are still in compliance with EPA standards. Also the WIPP itself maintains a public website with access to current issues as well as historical documents. In Canada a report, the ‘Seaborn Report’ (Environmental Assessment Panel, 1998), concluded in 1998 that even if the disposal concept was technically sound, social concerns had to stop the siting programme. This made the Canadian Government reorganize the radioactive waste management programme and a new organization, the Nuclear Waste Management Organization (NWMO), was established in 2002 under the Nuclear Fuel Waste Act (NFWA) to investigate approaches for managing Canada’s used nuclear fuel. The purpose of the NWMO is to ‘develop and implement collaboratively with Canadians a management approach for the safe longterm care of Canada’s used nuclear fuel that is socially acceptable, technically sound, environmentally responsible and economically feasible’ (NWMO website, www.nwmo.ca). Transparency is a key component of the NWMO approach and the NWMO engages Canadians in formal public engagement activities, round tables, public meetings and other forms of dialogue. On the societal research side, a number of international projects have been completed, such as RISCOM II (Andersson et al., 2004), COWAM II (He´riard Dubreuil et al.) and CARL (Bergmans et al, 2008), and new ones have been formed under the Sixth Framework Programme of the European Commission, such as ARGONA, dealing with arenas for risk governance, COWAM in Practice (a continuation or COWAM II) and OBRA, which was a feasibility study for a ‘European observatory for long-term governance of radioactive waste management’. Furthermore, numerous documents have been released from the Forum for Stakeholder Confidence (FSC) of the OECD/Nuclear Energy Agency (2001), including principles for stepwise decision-making and approaches to fostering a durable relationship between a waste management facility and its host community.

© Woodhead Publishing Limited, 2010

Establishment of radioactive waste management programmes

23.3

723

Rationales for participation in nuclear waste management programmes

From the previous section you can draw the conclusion that public participation is something that has naturally emerged as a possible solution to social acceptance issues that have been met by many radioactive waste management programmes. Experience tells us that progress has been made but also that the results of increased and broader societal participation have not always met expectations. Also, it cannot be taken for granted that the practices of participation are sustainable and that the DAD approach has been superseded by a more inclusive approach. Furthermore, there is a need for better understanding of what different modes of participation really mean, and how such modes should be practised. As a first step towards such an understanding, we take a look at the rationales for broader and more inclusive participation, first on a level of principle and then in the real world. In general, three rationales for the desirability of public involvement are often given in social science literature. The first rationale is ethical, which means that the public should be involved because they are the ultimate source of values in society, and these values should be expressed in decisionmaking. In other words, we should respect our citizens’ right to selfdetermination and autonomy. The second rationale is political. This means that public involvement strengthens the legitimacy of decisions and provides a broader responsibility for them, which also increases the stability of decisions over time. The third rationale concerns knowledge. The public should be involved because citizens have knowledge, which is different from the knowledge of experts and politicians. This lay-knowledge is often of essential importance for the issue at hand, at the very least it means that the knowledge base becomes broader. When considering local projects, or local effects of regional, national or global issues, it is usually the case that local residents have important knowledge that might otherwise be neglected by decision-makers (see, for example, Irvin, 1995, or Irwin and Wynne, 1996). It is interesting to see these rationales in the light of how actors in the nuclear waste management area actually motivate their efforts to involve stakeholders including the non-technical general public. A study was made within the European Union research project ARGONA where a questionnaire on policy-making structures was issued and sent to key nuclear waste management organizations and stakeholders at national and local levels in several EU member countries. The questionnaire included questions about legislative frameworks, reasons, current practices and future needs of participation and transparency. Responses from the nuclear waste management organizations showed mainly three types of reasons (Andersson et al., 2008b, p. 23). One major reason mentioned was to build confidence. Another is the strategic aim of making the process acceptable to

© Woodhead Publishing Limited, 2010

724

Geological repository systems for safe disposal

key stakeholders. Participation and transparency is also associated with different improvements, in direct relation to the project but also more in general, such as increased safety or making people more well-informed. Obviously, there is also a need to meet the requirements in the legislation, and this was, in certain cases, mentioned as the only reason for participatory activities. Most countries consider several of these guiding principles, but the mix and emphasis differs rather much from country to country. In relation to the more philosophical rationales, clearly building confidence and acceptance is mostly related to the political (sometimes called instrumental) rationale, as the aim is to increase the legitimacy and stability of decisions that have been made or are foreseen to come. The aim to improve the programmes is consistent with the knowledge rationale and is certainly also the key responsibility of a radioactive waste management organization. The reason for simply meeting legal and regulatory requirements can seem simply pragmatic, but then it is beneficial to examine the basis behind the requirements. The ARGONA project did not include this aspect, but it could be assumed that the legal requirements, such as the Environmental Impact Assessment (EIA) Directives, are based on a combination of ethical, political and knowledge-based rationales. The ethical rationale is not mentioned explicitly by acting radioactive waste management organizations, but this may be because of a presumption that the legal and regulatory requirements implicitly subsume ethical concerns as well. The ARGONA project also provides examples of what has triggered public participation activities to take place. Obviously, laws and regulations are often initial triggering factors, as well as practical reasons for participation. The consultations required in the EIA Directive is an important example. National legislation, such as the decision to allow the Swedish environmental movement to apply for funding from the nuclear waste fund, is another example of legislation of importance. In a similar way, the US Department of Energy gave no-strings-attached grant money to local community stakeholders in Carlsbad and New Mexico to hire their own experts, develop their own analyses, etc., in relation to the WIPP repository for US defense waste. Other government initiatives have taken place without any changes in the law, such as the CoRWM consultations in the UK (COWAM 2, Work Package 5, Final Report , pp. 108–122) and the partnership initiatives in Belgium (pp. 5–19). Other political events and decisions have also been of great importance, such as referendums or elections with nuclear power as a central issue; statements, pointing out the importance of participation and local acceptance; the initialization of programmes for local participation in the final disposal process; and agreements, as the UK Memorandum of Agreement for stakeholder cooperation. Demonstrations and protests

© Woodhead Publishing Limited, 2010

Establishment of radioactive waste management programmes

725

against nuclear power and final disposal have also contributed to the process. It should also be noted that important activities have been initiated by EU research projects such as RISCOM, ARGONA and the COWAM series of projects. Within these projects processes of inclusive governance and transparency have been implemented in certain participating countries, either as the core idea with the project itself (COWAM projects) or as an implementing part of a larger integrated project (ARGONA). In the end, the increased focus on participation and transparency can also be viewed as a social trend, something that is demanded by society to be able to proceed with nuclear waste management projects, as this quote from the ARGONA survey response shows: Since the end of the 1980s it became very clear to SKB that an open process with open information offering various and great possibilities for participation and consultation for all groups is the only way that acceptance and success could be reached for nuclear waste management projects. It is of special importance to have frequent contacts with nearby residents and land owners.

23.4

The Swedish dialogue and transparency process

In Sweden, a series of initiatives have been taken both at the national and local levels over a period of almost two decades with the ‘Dialogue Project’, RISCOM projects, regulator hearings, the Oskarshamn model and most recently the ‘Transparency Program’ (Andersson, 2007). These activities have taken place in parallel with the SKB formal EIA consultations and have been initiated and hosted by other stakeholders than SKB (the regulatory body SKI, Oskarshamn municipality and the Swedish National Council for Nuclear Waste). These dialogue-related activities have not been triggered by specific events, legal requirements or government initiatives, but can rather be seen as proactive spontaneous initiatives taken by organizations having their own roles in the radioactive waste management programme. These initiatives, which have been described in one of the ARGONA reports (Elam et al., 2008), started with the Dialogue Project, which took place from 1990 until 1993. At that time the SKI proactively realized that new activities were needed to communicate the safety assessments to lay persons and local stakeholders as the SKB radioactive waste management programme was soon to leave the more pure research stage to enter a phase of site selection. The Dialogue Project was organized as a simulated review process (Andersson et al., 1993) of an application for the final disposal of spent nuclear fuel (SKI, 1993a). SKI initiated and funded the project.

© Woodhead Publishing Limited, 2010

726

Geological repository systems for safe disposal

Perhaps the most important result was that the participants, who included licensing authorities, municipalities and environmental groups, received a ‘preunderstanding’ of issues and arguments in the coming decision-making process. The participants also wrote a joint letter to the Swedish government about issues in the decision-making process (SKI, 1993b). In the mid-1990s, SKI and SSI saw the need for a broader participatory consultation, which led to the RISCOM projects. The RISCOM model was originally developed during 1996–1998 in the RISCOM Pilot Study (Andersson et al., 1998), which can be seen as a continuation of the work initiated in the Dialogue Project. SKI hosted and funded the project together with SSI as the authorities realized that they needed a new, more open, public profile. The aim was to develop procedures to increase transparency in decision-making and the model was developed by a group of three researchers on contract from SKI and SSI. A central concept of the RISCOM model, which is briefly described in Section 23.6.3, is stretching, which refers to a practice where central actors in a decision-making process are gathered in front of a wide audience that challenges their claims to truth, validity and authenticity by posing questions from different perspectives. RISCOM II (Andersson et al., 2004) was a European Union project which included the evaluation and testing of various forms of dialogue processes with the involvement of a wide range of stakeholders. The hearings held in Sweden in 2001 in a critical stage of site selection organized by SKI were directly linked to the RISCOM II project. This was the first time the RISCOM model was implemented in a real decision-making situation (see Andersson et al., 2003, and Drottz Sjo¨berg, 2001). The Oskarshamn model developed soon after the Dialogue Project had ended (A˚hagen et al., 2003). The activities in the Oskarshamn municipality started in 1992 when SKB presented its plan to expand the interim storage of spent fuel in Oskarshamn to also include a facility for encapsulation of spent fuel. Soon the municipality leadership could foresee that Oskarshamn was also to become a candidate for hosting the spent nuclear fuel repository. There are clear links between the Oskarshamn work and the Dialogue– RISCOM projects, as the great involvement of the municipality as a key actor in the radioactive waste management programme was inspired by the majors’ participation in the Dialogue Project and later the municipality work was organized and inspired by the ideas in RISCOM. Later Oskarshamn and O¨sthammar, the other municipality being subject for detailed site investigation, cooperated in setting up meetings and hearings. The Swedish National Council for Nuclear Waste initiated its transparency programme in 2006 (Andersson, 2007) in response to the general agreement among central stakeholders in the Swedish nuclear waste management programme that there was a need for activities by the Council leading to more transparency. The objectives were to increase the

© Woodhead Publishing Limited, 2010

Establishment of radioactive waste management programmes

727

quality of the decision-making process by contributing with more clarity and awareness, to prepare the Council for its advisory role to the Swedish government and also be a resource for all stakeholders, the political decision-makers and concerned citizens who wished to deepen their insight into the issues addressed. The programme was largely built on RISCOM ideas and regular meetings have been held with central stakeholders in the Swedish programme and hearings with ‘stretching’ on selected topics have been held. This Council initiative was taken at a time when the profile of the Oskarshamn municipality as hosting ‘transparency arenas’ decreased as the municipality itself became an obvious stakeholder, having an outspoken will to become the selected host municipality, as did the other candidate, the O¨sthammar municipality. In 2010 SKB will, according to plans, formally apply for a final repository in O¨sthammar, and then a review period, estimated to about three years, starts. The newly formed Swedish Radiation Safety Authority (SSM), which is the merger of SKI and SSI, will then be in focus. It can be foreseen that the results from earlier phases for research and development of transparency and public involvement will then be implemented and used as an integrated part of the licensing process.

23.5

Public participation processes in nuclear waste management programmes

There are processes for participation, which have been formalized in legal systems, that can be implemented from the very beginning of a project for nuclear waste disposal and maintained over a long period of time. They can be seen as overarching processes that could be umbrellas to host a number of related activities. One such process is consultation as part of Environmental Impact Assessment (EIA). The development of EIA started in the United States with the National Environmental Policy Act of 1969. Now there is an EU Directive on EIA (1985 and 1997) – European Union, Directive 85/337/EEC, 1985, as amended by Directive 97/11/E, 1997 – as well as national legislation on EIAs in EU countries. The Directive requires public participation to occur as part of the EIA process for certain projects, including disposal facilities and facilities for long-term storage of radioactive waste. Even if there are many versions of EIA they all include two core principles, first that the EIA process should start early (participation must take place before real decisions are made) and second the requirement of assessing alternatives to the suggested action (including the ‘no action’ option), which can be seen as a means for opening up the assessment. Another potentially important development is the Strategic Environmental Assessment (SEA) (see, for example, Directive 2001/42/EC

© Woodhead Publishing Limited, 2010

728

Geological repository systems for safe disposal

of The European Parliament and of The Council, 2001), which basically means using the EIA principles for projects on the higher levels of policy and planning. In Europe, EIA and SEA Directives are incorporated in the legislation in member states, with the strategic type of directive being made more recently. SEA may be important from a state/region-wide perspective, because it can define the framework of a field in longer-term perspectives, whereas a project EIA is more locally focused. Participatory Technology Assessment (PTA) is a broader concept than EIA and SEA, having the aim of ‘finding solutions together’ or ‘generating dialogue’ (see Klu¨ver et al., 2000). There are also international conventions where countries have committed themselves to certain levels of citizen involvement in technical and environmental assessment. Perhaps the most important one in our context is the Aarhus Convention on Access to Information, Public Participation in Decision-Making and Access to Justice in Environmental Matters (United Nations Economic Commission For Europe, 1998), which came into force on 30 October 2001. The Convention consists of three pillars: 1. 2. 3.

Access to information Public participation in certain decisions relating to the environment Access to justice in matters pertaining to the environment

The third pillar stipulates that a person has the ability to go to court or another independent and impartial review body to ask for a review of potential violations of the Convention, e.g. if his or her request for information has been ignored, wrongfully refused or inadequately answered.

23.5.1 Methods of participation Within (or outside) the scope of EIA and SEA regulations, as well as international conventions, there is a large field of methods and instruments that can be used for social dialogue. They can be implemented over a short or longer periods of time and events can be scheduled at optimal points to give the best possible input to the overall decision-making process. A number of them are briefly described and analysed in Andersson (2008a). One method is consensus conferences (Klu¨ver, 1995). Here ordinary people are given the opportunity to assess a given technological development and make up their minds about its possibilities and consequences. Participants are found by sending out invitations to randomly selected citizens. The idea of science shops started in Netherlands in the 1970s as part of the Dutch radical science movement (Dickson, 1984, p. 329). The aim was to increase the influence of civil society on the universities, to make contact between citizen groups and scientists and to make use of the knowledge available at the universities. In a study financed by the European

© Woodhead Publishing Limited, 2010

Establishment of radioactive waste management programmes

729

Commission, Gnaiger and Martin (2001) concluded that science shops are a cost-effective way to provide research and information to a wide range of civil society organizations. A focus group is an informal technique that can help assess people’s opinions and feelings in a certain area (see, for example, Krueger and Casey, 2000). In a focus group, you bring together a limited number of people, usually between six and ten, to discuss a selected issue. Compared to surveys, focus groups provide more in-depth qualitative insights coming from a relatively small number of people. Focus groups can be maintained over significant time intervals, so that evolution in attitudes and knowledge arising can be evaluated. One method of participation for decision-making, which allows experts and lay people to meet in a joint effort, is multi-attribute utility analysis. This is a quantitative decision analysis method that arrives at a preferred decision among a number of alternatives based on the importance and values of different factors. As the weighting of the importance between different factors is often more a matter of judgement than scientific analysis, this is where experts and lay people can interact. The method is often used successfully in the United States in various nuclear and hazardous social dialogue programmes. The critical point for success is actually to make lay people actively involved. If they are not, meaning that the experts in fact also do the weighting more or less by themselves, there is a risk of failure as other stakeholders may later show that another, equally justifiable, weighting would have given another result. Each one of these and other methods and instruments has its own characteristics and its own niche of application. They also have their own limitations in terms of involvement and transparency, and there can be concerns such as the possible remaining expert dominance over the process, limited impact on public awareness, accountability for decision advice, etc. Therefore, before entering into any application of these processes, specification of objectives for everyone involved is crucial. However, to select the most appropriate method for a particular set of circumstances is not easy as ideally this should be done using a systematic knowledge base of methods that does not seem to exist. Despite the large body of practical insights from case studies of public participation (see, for example, Rowe and Frewer, 2000; Beierle and Cayford, 2002; Bergmans et al., 2008; Elam et al., 2008), there is still a lack of conceptual and theoretical understanding in order to help match social dialogue methods with context and purpose. There have been a number of approaches to evaluation and mapping of participatory methods. For example, in a study for Resources for the Future, Beierle and Cayford (2002) made an extensive review of public participation processes in the Unites States. They used five ‘social goals’ in their review. Among the five social goals ‘educating and informing the

© Woodhead Publishing Limited, 2010

730

Geological repository systems for safe disposal

public’ was the easiest one to comply with whereas, not surprisingly, ‘building trust’ scored lowest. Rowe and Frewer (2000) used nine criteria for the evaluation of processes: representativeness (participants should comprise a representative sample of the population), independence, early involvement, influence, transparency, resource accessibility (participants should have adequate resources for their participation), task definition (a clearly defined process), structured decision-making and cost effectiveness. Rowe and Frewer have also reviewed the entire field of public participation exercises (Rowe and Frewer, 2004) and came to the conclusion that research in this area has been disorganized and sporadic; they suggest a more systematic research agenda. Andersson (2008a, pp. 150–155) added elements of transparency to existing criteria sets but concluded that there is not yet enough understanding of how a systematic framework for public participation processes should be developed.

23.6

The context of social dialogue in nuclear waste management programmes

In this section we put the processes of participation and dialogue into a wider framework. First, we introduce the concept of risk governance and relate processes of participation to risk governance approaches. Then we examine the strengths and limitations of participation as part of a democratic society that has representation as a core element.

23.6.1 Risk governance approaches Radioactive waste management is part of societies’ overall risk governance. In the EU research project CARGO (Andersson et al., 2008a), three approaches to risk governance (risk informed decision making, precaution and risk deliberation) were compared by using a number of example areas, such as genetically modified organisms (GMOs), mobile telephone risk assessment and remediation of chemically contaminated sites. Risk-informed decision-making relies on quantitative risk assessment (QRA), which is a systematic methodology for the application of a mathematical construct of risk. It tries to identify all possible events that can lead to an undesired end state, thus evaluating an overall estimate of risk for the system being analysed (e.g. a nuclear reactor). The Precautionary Principle comes into force when there are possible serious risks with large scientific uncertainty. The Precautionary Principle is listed as Principle 15 of the Rio Declaration of 1992 among the principles of general rights and obligations of national authorities (Robinson, 1992): In order to protect the environment, the precautionary approach should

© Woodhead Publishing Limited, 2010

Establishment of radioactive waste management programmes

731

be widely applied by States according to their capabilities. Where there are threats of serious or irreversible damage, lack of full scientific certainty shall not be used as a reason for postponing cost-effective measures to prevent environmental degradation. Since 1992, the principle has been implemented in various environmental instruments for areas such as global climate change, ozone-depleting substances and biodiversity conservation. It is action-oriented, meaning that persistent dissent among scientists cannot be taken as an excuse not to take action. The principle as cited above deals with protection of the environment; however, in different forms it is used in many other areas, especially in human health protection, and it has been adopted as a legal requirement in many countries. However, for certain situations is has been argued that the Precautionary Principle is in conflict with pre-existing legal, regulatory and logical precedents (Okrent, 1999; Belzer, 2000). Risk-informed decision-making can be seen as an expert domain whereas the Precautionary Principle is often a legal requirement handled by state authorities which stakeholders sometimes argue should be applied more stringently (for example by introducing moratoria for certain new technologies). The risk deliberation approach (Reynolds et al., 2008) is close to what is meant by social dialogue in this chapter, as it involves participation by stakeholders and citizens. What is perceived as ‘good risk governance’ in the EU is explicitly announced to involve deliberative participatory elements. The CARGO project concluded that there are no precise borders between the areas of application for risk-informed decision-making, precaution and deliberation – instead they are complementary. There is a place for elements of deliberation in risk-informed decision-making, in order to generate scenarios to be included in the risk assessment and to clarify the limits in using the results. Exposing technological elites to discussions with lay people and other stakeholders can broaden their thinking and make them reflect on the limitations of their methods, especially in a policy-making context, what aspects they cover and what aspects they leave out. The role of deliberation for the application of the Precautionary Principle is even more obvious as value-laden issues play a key role (von Schomberg, 2004; Andersson, 2008b). Precautionary considerations may help in risk-informed decision-making so that great and conceptual uncertainties are not downplayed, and the Precautionary Principle is often a central theme in deliberation. When applying the Precautionary Principle the question must be posed as to whether there is enough knowledge for risk-informed decision-making, in which case the Precautionary Principle might not be needed. Also, expert knowledge and therefore elements of risk-informed decision-making should

© Woodhead Publishing Limited, 2010

732

Geological repository systems for safe disposal

be part of the deliberation – otherwise it could be narrowly framed as a purely value-laden and social process and science would be down played. Instead of relying on the dichotomy between analytical expert-based approaches and more open and flexible participatory approaches, Stirling (2005) suggests that we look at distinctions that are cross-cutting and that may be equally relevant to both. One such cross-cutting distinction is that between opening up or closing down the process of technological choice. When the social appraisal process is about closing down, the aim is instrumentally to assist policy-making: ‘Whether analytical or participatory, the role of social appraisal process lies in cutting through the messy, intractable and conflict-prone diversity of interests and perspectives to develop a clear authoritative, prescriptive recommendation to inform decisions’ (Stirling, 2005, p. 228). The outcome of a process aiming to close down is a unitary and prescriptive policy advice, presenting only a small number of choices or courses of action to be explored, which appear to be favourable in the light of how the process has been framed. To return to social dialogue and public participation in radioactive waste management, it seems evident that the methods to be used must be selected taking the particular risk governance situation into account, e.g. if it is a situation where risk-informed decision making is the dominating mode (with dialogue essentially as a supporting element) or if the nature of risk and uncertainties make the Precautionary Principle come into action. Obviously, a basic factor is if the nature of a nation’s legal and regulatory framework regarding risk assessment and protection of public safety encourage or require that the Precautionary Principle should be used. Also of importance is the stage of the risk management process, whether it is an initial stage of opening up, expanding and enriching the scope of the debate or if it is in a later, licensing stage, for example, in which extensive debate and dialogue have been conducted leading to a resolution of core issues prior to closing down and finalizing important decisions. On the other hand, specific methods of participation can be used in different situations. For instance, consensus conferences can open up discussions over technology and enrich the political debate, but they can also be used to provide input and support to aid in resolving the issue before formal decisions are made. In the same way, transparency arenas (see Section 23.6.3), although as yet presented mainly as a support to political decisionmaking, can be an essential element in identifying issues for debate. Focus groups can also be used in either opening up or closing down phases, depending on how the results are presented and fed into a debate. It is always important, of course, to define the purpose of the process clearly in the initial phase.

© Woodhead Publishing Limited, 2010

Establishment of radioactive waste management programmes

733

23.6.2 Limits of participation Ultimately, nuclear waste management involves governmental decisions. For approving the siting and construction of a repository, for example, decisions must be taken, sometimes at different levels – national, regional or local – depending on the legal and administrative system. Even if these decisions ordinarily are preceded by reviews by licensing and regulatory bodies and court procedures, in the end there is likely to be an explicit or implicit political dimension. The legitimacy of the processes leading up to these decisions, therefore, are crucial and the levels of participation, transparency and accountability become core issues. Even if participation is a requirement, such participation can be implemented in different ways and with different levels of ambition. The value of the social dialogue that has previously taken place will be examined; therefore such dialogue is of importance to provide a broad stakeholder context in relation to the pending governmental decision. It is necessary to set public participation into the context of the overall political decision-making system in society. It is also necessary to evaluate critically whether participation really helps to build the awareness and transparency needed for dealing with complex technological and environmental issues under democratic, representational governments. As we have observed in Section 23.3, different organizations often indicate reasons such as confidence building and acceptance for their social dialogue activities. There are potential problems with such restricted attitudes (Andersson, 2008a, pp. 170–174). If participating citizens believe that their concerns are important for the project and that there is a real chance for those concerns to be addressed, but instead the outcome is decided in advance by the controlling body, then trust will be lost and citizens will withdraw their engagement. Christian Vergez, Principal Administrator of the OECD Directorate for Public Governance and Territorial Development, has expressed such concerns (Vergez, 2003): While the benefits of engaging citizens in policy-making may be considerable, governments should not underestimate the risks associated with poorly designed and inadequate measures for information, consultation and active participation. They may seek to inform, consult and encourage active participation by citizens in order to enhance the quality, credibility and legitimacy of their policy decisions. However the opposite effect may be achieved if citizens discover that their efforts to be informed, provide feedback and actively participate are ignored or have no impact at all on the decisions reached. The more philosophical arguments for participation (the ethical, political and knowledge rationales) can be summarized in the sense that participation

© Woodhead Publishing Limited, 2010

734

Geological repository systems for safe disposal

has a value in itself – that it creates a ‘common good’. There is an ideal of ‘deliberative democracy’, which means that participation increases the legitimacy of societal decisions (Held, 2002). Participation is good not only for the society as a whole but also for individual citizens who, according to the proponents of deliberative democracy, are rewarded by self-fulfilment and, according to the Harvard political philosopher John Rawls (1971), deliberation can also bring justice and more equality between groups in society. The main characteristic of deliberative democracy is a focus on process as much as results. However, considering the essentially unlimited amount of information a citizen always has to manage in everyday life, there may not be enough personal time and attention available for participation in the full range of controversial projects. As the political scientist David Held puts it in a critical comment, ‘What if they do not wish to participate in the management of social and economic affairs? What if they do not wish to become creatures of democratic reason?’ (Held, 2002, p. 272). There are also questions about democratic accountability. Participative or deliberative democracy means that the elected assemblies and governments will lose some of their power. Then it will be more difficult to understand who should be held accountable for decisions made (Andersson, 2008a, p. 153). Real political accountability means that there is clarity about the issues, which social dialogue in a general sense may not necessarily provide. Although the result of social dialogue could be a richer flow of information, without a structured way to systematize and deal with the information, it might not enhance transparency and awareness.

23.6.3 The importance of transparency To overcome some of the concerns with deliberative democracy, such as accountability, and to increase clarity before decisions are made, the ‘transparency arena’ approach for political insight and accountability has been suggested (Andersson and Wene, 2008). This is an arena where all arguments are put forward and challenged in a structured way. Here the RISCOM model (Andersson et al., 1998) is a cornerstone. It is aimed to ensure that decision-makers and the public can validate claims of truth, legitimacy and authenticity. It has emerged as an outcome of Habermas’ theory of communicative action (Habermas, 1987) and Stafford Beer’s organizational theory (Beer, 1994). The RISCOM model also gives clarity and structure to the debate in another sense. Decision processes must often deal with different levels of discussion. For example, in selecting a site for high-level nuclear waste, the expert work at the ground level (geological investigation, risk assessment, etc.) takes place within a broader framework for managing the programme at the national level. However, the site-selection programme itself depends

© Woodhead Publishing Limited, 2010

Establishment of radioactive waste management programmes

735

on a waste management method decided at a higher societal level. The RISCOM model can help bring organization and order into the debate, since claims of truth, legitimacy and authenticity are made at each level of debate. The transparency arena approach does not aim for consensus per se, although the approach can lead to increasing consensus. Equally, the approach is not looking for ‘acceptance’ since that would mean pre-judging from the beginning what should be the result. On the other hand, challenging arguments is not aimed at creating a more polarized debate, but to promote increased clarity and understanding. The key aim is to create an arena where all stakeholders share and increase their mutual awareness. The main events in the transparency arena are hearings with stretching, where stakeholder arguments are challenged from different angles. This is a tailored hearing format to gain transparency. It should be noted that the stretching is for all stakeholders, not just the implementer or a regulator. Everyone who is advocating an argument, also against the suggested way forward or solution, should be prepared to have his or her arguments stretched for the sake of clarity. The actual hearings may be combined with other kinds of dialogue formats including group-work activities.

23.7

Conclusions

During recent decades much progress has been made with regard to methods of social dialogue in the area of nuclear waste management. Processes of participation and transparency have been developed. For example, the European Union has launched research projects (COWAM I and II, RISCOM II, CIP, OBRA and ARGONA). There have also been other cooperative projects such as CARL and the OECD/Nuclear Energy Agency launched the Forum for Stakeholder Confidence. Some of these new processes have started to be used in practice, again notably within EU projects. Platforms have been established for stakeholders, especially local stakeholders, to meet, share experiences and reach common ideas and proposals. In the USA, the WIPP facility is subject to outreach activities by a number of involved bodies. The transparency approach with the RISCOM model provides a framework for dialogue which prioritizes clarification of complex issues so that decisions can be made on the ‘best possible basis’. Experience has shown that existing governmental and regulatory authority institutional settings can be used for initiating and guiding social dialogue. There is a high degree of freedom inside the current national legislation regarding nuclear waste management for participation and transparency initiatives and improvements. In other words, there are no

© Woodhead Publishing Limited, 2010

736

Geological repository systems for safe disposal

inhibitions or restrictions that hinder increased participation and transparency if properly pursued. Furthermore, there are probably opportunities to make improvements in the social dialogue process inside and beyond the requirements of existing legislative frameworks. One lesson learned from the Belgian and Swedish examples (see, for example, Bergmans et al., 2008, and Elam et al., 2008) is that in reality new processes for achieving social dialogue can be implemented within the existing political and legal systems. For example, the achievements made in Oskarshamn using ‘EIA best practices’ showed the potential way forward for enhanced social dialogue long before binding legislation came into being (Andersson, 2008a, pp. 159–161). Many of the good examples of public participation, such as the Belgian partnerships, have been developed and used without new laws or conventions. However, regulations for access to resources can have impacts on the results of the processes of participation and transparency. ‘Access to participation’ must mean financial resources for impacted stakeholders in order to assure their empowerment to be able to participate in social dialogue in an active and fair way. We have seen that in fact there is a mixture of driving forces – events, laws and regulations, as well as spontaneous, governmental and research initiatives – that has triggered and developed processes of participation and transparency in social dialogues related to nuclear waste management (Elam et al., 2008). The initiatives for such processes have come from a mixture of factors and stakeholders, some initiatives arising from legislation to do it, but many other initiatives have been voluntarily taken beyond the reason of fulfilling formal legal or regulatory requirements. The development of processes for participation and transparency has, thus, been based on a diverse mix of factors, forces and reasons, and with the expressed future needs for safely managing nuclear waste in mind. It could be suggested that, for improvements of processes of participation and transparency in the future, opportunities for a diversified and balanced mix of stakeholder involvement need to be maintained and enhanced in several ways. Many such improvements can take place inside the current legal and regulatory framework for most countries, rather than necessitating any formal legal or organizational changes. On the other hand, in countries without formal requirements, no level of participation is guaranteed; the basic existence of regulations and international conventions makes an argument for participation and dialogue for groups who feel they want to participate in the review and oversight of nuclear waste management plans. Even if international conventions have an important role, their importance should not be overestimated. Because of the consensus-building process behind most international conventions, the resulting document is often general and broadly permissive of a wide range of approaches, giving little practical advice about what kinds of participation processes could or

© Woodhead Publishing Limited, 2010

Establishment of radioactive waste management programmes

737

should be used, how such social dialogue programmes should be set up and how the results of social dialogue programmes should be integrated in national nuclear waste management programmes. In practice, as is the nature of international agreements, much of this will be up to the definition of the specific national programmes within their own countries. How to implement new processes is not self-evident. There are legal, organizational, historical and cultural factors and contexts that set conditions and may limit or inhibit innovation. This is true already within the European Union where projects such as ARGONA and COWAM have been accomplished, but certainly such factors differ also between the EU countries, North America and Asian countries. Certainly there are findings from social sciences research that have a general character and there are principles that should be relevant over a wide context, but when it comes to implementation these factors may have significant impact. We have to understand this better for effective implementation. Even if this issue is addressed, for example, in the ARGONA project, it seems to be an area where more research is needed at a broader international level. Furthermore, the arenas of participation and transparency for social dialogue need to be connected to national and local forms of representative democracy, within which decisions on the final disposal of nuclear waste are ultimately taken. We must also understand how deliberative methods and the transparency approach relate to each other, and to formal decisionmaking in representative democracy. There is still a large field of remaining research. As Rowe and Frewer (2000) concluded, research in this area has been disorganized and sporadic. They suggested a more systematic research agenda, a demand that still seems to be relevant. On the other hand, there is a large knowledge base about governance and dialogue that should be further explored and implemented as appropriate.

23.8

References

A˚hagen H, Andersson K, Carlsson T, Eklind R, Nilsson K and Wretlund P (2003), ‘Linking the Oskarshamn model for public participation with the RISCOM model of transparency’, in VALDOR Symposium, Proceedings, Stockholm, June 2003, pp. 405–410. Andersson K (2007), ‘Making the decision-making basis for nuclear waste management transparent, English summary’, KASAM Contract 38/06, Karita R: 07:01, ISBN 978-91-976858-0-1. Andersson K (2008a), Transparency and Accountability in Science and Politics – The Awareness Principle, Houndmills, Basingstoke and Palgrave Macmillan, New York, ISBN-13: 978-0-230-54217-4. Andersson K (2008b), ‘Precaution and risk reduction – politics and expertise. Some reflections on the precautionary principle, CARGO Deliverable 5’. Andersson K and Wene C-O (2008), ‘The ITA process. The institutionalised

© Woodhead Publishing Limited, 2010

738

Geological repository systems for safe disposal

transparency and accountability process for clarity in policy making’, Karita R: 08:01, ISBN 978-91-976858-1-8. Andersson J, Andersson K and Wene C-O (1993). ‘The Swedish Dialogue Project. An attempt to explore how different actors may take part in the decision process related to disposal of radioactive waste’, in High Level Radioactive Waste Management, Fourth Annual International Conference, Las Vegas, Nevada. Andersson, K, Espejo R and Wene C-O (1998), ‘Building channels for transparent risk assessment’, SKI Report 98:5, RISCOM Pilot Study, Stockholm, Sweden. Andersson K, Wene C-O, Drottz Sjo¨berg B-M and Westerlind M (2003), ‘Design and evaluation of public hearings for Swedish site selection’, SKI Report 2003:32 (RISCOM II Deliverable 5.3), Stockholm, Sweden. Andersson K, Westerlind W, Atherton E, Besnus F, Chataıˆ gnier S, Engstro¨m S, Espejo E, Hicks T, Hedberg B, Hunt J, Laciok A, Leskinen A, Lilja Ch, O’Donoghue M, Pierlot S, Wene C-O, Vira J and Yearsley R (2004), ‘Transparency and public participation in radioactive waste management’, RISCOM II Final Report, SKI Report 2004:08, Stockholm, Sweden. Andersson K, Drottz Sjo¨berg B-M, Reynolds L, Serbanescu D, Soneryd L, Szerszynski B and Vetere A (2008a), ‘Comparison of approaches to risk governance’, CARGO Final Report, EU Contract FP6-036720. Andersson K, Falck E and Lidberg M (2008b), ‘Policy making structures in the EU and participating countries’, ARGONA Deliverable 2, Contract FP6-036413. Beer S (1994), Beyond Dispute – The Invention of Team Syntegrity, Wiley and Sons Ltd, Chichester. Beierle T and Cayford J (2002), ‘Democracy in practice. Public participation in environmental decisions’, in Resources for the Future, RFF Press Book, Washington DC, ISBN-1-891853-53-8. Belzer R B (2000), ‘Discounting across generations: necessary, not suspect’, Risk Analysis, 20(6), 779–792. Bergmans A, Elam M, Kos D, Policˇ M, Simmons P, Sundqvist G and Walls J (2008), ‘Wanting the unwanted: effects of public and stakeholder involvement in the long-term management of radioactive waste and the siting of repository facilities’, Final Report CARL Project. COWAM 2, Work Package 5, Final Report, National Insights, http://www.cowam. com/spip.php?article71. Department for Environment, Food and Rural Affairs (2006), ‘Response to the Report and Recommendations from the Committee on Radioactive Waste Management (CoRWM)’, by the UK Government and the devolved administrations. Department for Environment, Food and Rural Affairs (2008), ‘Managing radioactive waste safely – a framework for implementing geological disposal’, A White Paper by Defra, BERR and the devolved administrations for Wales and Northern Ireland, Presented to Parliament by the Secretary of State for Environment, Food and Rural Affairs by Command of Her Majesty, June 2008. Dickson D (1984), The New Politics of Science, Pantheon Books, New York. Drottz Sjo¨berg, B-M (2001), ‘Evaluation of hearings with questionnaires and

© Woodhead Publishing Limited, 2010

Establishment of radioactive waste management programmes

739

interviews’, SKI Rapport 01:39, in Swedish with a two page English summary (RISCOM II Deliverable 5.4), Stockholm, Sweden. Dunlap R E (1997), Trends in public opinion toward environmental issues 1965– 1990’, in Green Management, edited by Pierre McDonagh and Andrea Prothero, The Dryden Press, London, pp. 55–85. Elam M, Lidberg M Soneryd L and Sundqvist G (2008), ‘Demonstration and dialogue: mediation in Swedish nuclear waste management’, ARGONA Deliverable 10, EU Contract FP6-036413. Environmental Assessment Panel (1998), Report of the nuclear fuel waste management and disposal concept assessment panel’, Canadian Environmental Assessment Agency, Minister of Public Works and Government Services, Canada. Gnaiger A and Martin E (2001), ‘Science shops: operational options SCIPAS Report 1’, Study financed by the EC-DG XII programme ‘Improving the human research potential and the socio-economic knowledge base (IHP)’ and ‘Strategic analysis of specific political issues (STRATA)’, HPV1-CT-199900001. Habermas J (1987), Theorie des kommunikativen Handelns, 2 vols, Suhrkamp, Frankfurt (1981); Translated by Th McCarthy as The Theory of Communicative Action, 2 vols, Polity Press, Cambridge (1987). Held D (2002), Models of Democracy, 2nd edition, Polity Press, Oxford. He´riard Dubreuil G (Project Coordinator), Espejo R, Flueler T, Gadbois S, Mays C, Paixa A and Schneider T, ‘Cooperative research on the governance of nuclear waste management’, COWAM II Final Synthesis Report. Irwin A (1995), Citizen Science: A Study of People, Expertise and Sustainable Development, Routledge, London and New York. Irwin A and Wynne B (eds) (1996), Misunderstanding Science? The Public Reconstruction of Science and Technology, Cambridge University Press, Cambridge. Klu¨ver L (1995), ‘Consensus Conferences at the Danish Board of Technology’, in Public Participation in Science: The Role of Consensus Conferences in Europe, edited by S Joss and J Durant, Science Museum with the support of the European Commission Directorate General XII (c), pp. 41–49. Klu¨ver L, Bellucci S, Bu¨tschi D, van Eijndhoven J, van Est, Gloede F, Hennen L R, Joss S, Nentwich M, Peissl W and Togersen H (2000), ‘EUROPTA. European Participatory Technology Assessment – participatory methods in technology assessment and technology decision-making’, The Danish Board of Technology, Copenhagen. Krueger R A and Casey M A (2000), Focus Groups: A Practical Guide for Applied Research, 3rd edition, SAGE Publications. OECD/Nuclear Energy Agency (2001), Forum on Stakeholder Confidence (FSC), Strategic Directions of the RWMC Forum on Stakeholder Confidence’, EA/ RWM/FSC(2001)2/REV2, Paris. Okrent, D. (1999), ‘On intergenerational equity and its clash with intragenerational equity and on the need for policies to guide the regulation of disposal of wastes and other activities posing very long-term risks’, Risk Analysis, 19(5) 877–901. Rawls J (1971), A Theory of Justice, Harvard University Press, Cambridge, Massachusetts.

© Woodhead Publishing Limited, 2010

740

Geological repository systems for safe disposal

Reynolds L, Soneryd L and Szerszynski B (2008), ‘Risk deliberation’, CARGO Deliverable 6, January 2008. Robinson, N. (ed.) (1992), Agenda 21 and the UNCED Proceedings, vol. 1, Oceana Publications, New York. Rowe G and Frewer L J (2000), ‘Public participation methods: a framework for evaluation’, Science, Technology and Human Values, 25, 3–29. Rowe G and Frewer L J (2004), ‘Evaluating public participation exercises: a research agenda’, Science, Technology, and Human Values, 29, 512–556. SKI (1993a), ‘DIALOG-projektet – spelgruppens rapport om projektets uppla¨ggning och inneha˚ll’, SKI Report 1993 No. 35 SKI (1993b), ‘DIALOG-projektet – akto¨rsgruppens slutrapport’, SKI Report 1993 No. 34. Stirling A (2005), ‘Opening up or closing down? Analysis, participation and power in the social appraisal of technology’, in Science and Citizens. Globalization and the Challenge of Engagement, edited by Melissa Leach, Ian Scoones and Brian Wynne, Zed Books, London, pp. 218–231. United Nations Economic Commission For Europe (1998), Convention on Access to Information, Public Participation in Decision-Making and Access to Justice in Environmental Matters, Aarhus, Denmark, 25 June 1998. Vergez C (2003), ‘Evaluating public participation exercises – PUMA findings’, NEA/ RWM/FSC(2003)10. von Schomberg R (2004), ‘The normative dimensions of the precautionary principle and its relation to science and risk management decisions’, in Microscopic Modification and Big Politics, edited by Th Achen, Linkoeping Studies in Arts and Science, Vadstena, Sweden.

© Woodhead Publishing Limited, 2010

Index

abnormal pressures, 166 acid mine drainage, 60, 142 active fault mapping, 199–202 active faults, 199–202 activity concentration, 708 activity density, 708 admixtures, 308 Advanced Fuel Cycle Initiative, 31, 33 advection, 370, 472 advective transport, 10 aeromagnetics, 206–7 AFRA project, 65 air monitoring, 702–15 fixed air samplers, 705–6 AISI Type 316L stainless steel, 398 Aladdin, 630 aleatory uncertainty, 547–8 representation in 2008 Yucca Mountain PA, 552 alkaline plume, 294 alkaline reserve, 288 Alloy 22, 408 Amargosa Valley, 670–1 ambient aerosol monitoring, 711–13 aerosol mass loadings and actinide activities, 713 high-volume aerosol sample data, 712 americium, 703–4 anchor, 588 anchoring, 588 ANDRA, 89 anionic exclusion, 475 aqueous chemical processes deep porewater compositions, 460 defining the relevant aqueous chemical species, 459–66 aqueous complexation, 462–4 hydrolysis reactions, 460–1

precipitation and coprecipitation, 464–6 redox reactions, 462 Ep–pH diagram for Pu in groundwater, 464 evolution of neptunium solubility as function of redox potential, 465 Pourbaix diagram of neptunium, 463 radionuclides aqueous form, 461 aqueous complexation, 462–4 aqueous technologies, 34 UREX + process and waste/storage products, 35 argillaceous rocks, 251–2 see also clay geological repository systems ARGONA, 723–5 Argonne National Laboratory, 35 argumentation network approach, 619 artificial neural networks, 629 A¨spo¨ Hard Rock Laboratory, 90, 91, 98, 147, 148 ASTM D6527, 690 Atomic Energy Commission, 683 austenitic stainless steels, 395 atmospheric corrosion rates, 397 CCT and CPT in ferric chloride solution, 397 composition, 396 general corrosion rates, 396 backfill materials development and application, 323–51 long-term performance, 344–51 types, properties and fabrication, 341–4 clay/ballast microstructure schematic, 341

741 © Woodhead Publishing Limited, 2010

742

Index

clay block masonry with smectite pellets or clay mud, 343 compaction curves, 342 grain size distribution, 342 placement and compaction using inclined layer principle, 343 backfilling, 70 backward chaining, 629 Barcelona basic model, 338 barium disilicate glass, 428–9 Basic Safety Standards, 48, 68 Bayesian approach, 598 bedrock, 196–7 behavioural aggregation, 598 Belgian supercontainer concept, 388 bentonite, 104, 193, 294, 356, 385 best available technology, 658 vs optimisation, 658–9 biologic markers, 204 bituminisation, 455–6 blackboard architecture, 630 blister formation, 394 Boom Clay, 154, 163, 403, 407, 460 borehole disposal borehole facilities for large- and small-volume waste packages, 62–4 current role in radioactive waste management, 44–5 future trends, 73–4 borehole disposal of sealed sources, 46, 65–6 borehole drilling, 124 borehole facilities, 62–4 borehole flow logging tool, 132 borehole pulse tests, 175 boreholes long-term monitoring equipments, 177 systematic analysis of breakouts, 176–7 boreholes slug tests, 174 boron mass balance equation, 435 borosilicate glass, 427 borosilicate glasses, 265, 271–3 compositions, 272 BOSS see borehole disposal of sealed sources bottom-up approach, 95 BRAGFLO code, 506 BRAGFLO_DBR, 508 Bure URL, 89 C-steels see carbon steel

CACs see calcium aluminate cements caesium-137, 68 calcium aluminate cements, 301–3 calcium silica hydrate, 454 calcium sulphoaluminate cements, 303–4 calibration, 599 Callovo-Oxfordian argillite, 460, 472, 474 Callovo-Oxfordian clay, 154, 156 Canada Nuclear Safety Commission, 209 Canadian repository program, 19 canister containment, 7–8 impact of a 1000-year canister, 8 canisters, 356 see also nuclear waste canister materials capillary barrier, 358 carbon-14, 58 carbon steel corrosion behaviour, 391–4 anaerobic corrosion rate time dependence in simulated ground water solution, 392 differences between anaerobic corrosion behaviour in bulk solution and in compacted bentonite, 393 CARGO, 731 Carlsbad Environmental Monitoring and Research Centre, 679 history and current status, 689–93 whole body counter, 691 Cartesian co-ordinate system, 367 cast iron corrosion behaviour, 391–4 anaerobic corrosion rate time dependence in simulated ground water solution, 392 Castile Formation, 685 caverns, 57–60 CCDF_GF, 508 CCT see critical crevice temperature cements, 279–80 CEMRC see Carlsbad Environmental Monitoring and Research Centre Centre de l’Aube NSD facility, 68 CeO2, 428 ceramic waste forms, 266–7 ceramics, 410 40 CFR 191, 500 Chalk River Laboratories, 60

© Woodhead Publishing Limited, 2010

Index chemical disturbance, 143–4 chloride, 395 chromium, 395 clairvoyance test, 590, 651–2 Classical Approach, 593, 596, 598–9 clay geological repository systems characterisation and site surveying technologies and techniques, 153–80 core lab analysis, 178–9 future trends, 180 geological mapping, 166–7 geophysical seismic surveys, 167–8 integration of results, 179 technologies, 166 underground structures surveys, 178 clay site survey specific features, 155–9 characteristics required from geological environment, 157–9 chloride concentration profile in Opalinus Clay layer, 158 lithology and consequences for confining properties, 155–6 mechanical and thermomechanical properties of clay rocks, 156–7 potential implantation of disposal site in Benken region, 160 Zu¨rcher Weinland burial history of sedimentary formations, 158 drilling, 168–78 borehole long-term monitoring equipments, 177 geomechanical tests, 175–7 geophysical wireline logging, 170–1 hydrogeological logging tools, 172–4 hydrogeological logs interpretation, 172 natural gamma-ray logs interpretation, 171 packer tests, 174–5 programme, 168–9 reverse-air percussion drilling technique, 170 in situ stress profile at Bure, 176 workover operations, 178 survey strategy, 162–6 data analysis, 162 extent of Boom Clay and locations of investigation boreholes, 163

743

map covering the sector studied in France, 164 pressure and permeability profiles obtained within Paleozoic sedimentary sequence, 167 specific features, 165–6 survey grid and adapted key parameters of tools, 164–5 survey zone boundaries definition, 162–4 survey tools, 159–62 drillings and borehole measurements, 161–2 surface methods, 160–1 underground structures, 162 clay minerals, 466–7 clay site see clay geological repository systems CLAYTRAC project, 157 coffinite, 451 cold-crucible melting, 272 collaborative software, 632 colloid and radionuclide retardation, 102–4 commercially pure titanium, 403–5 Committee on Radioactive Waste Management, 721, 724 common KADS, 630 complementary cumulative distribution function, 549, 556 complete variance decomposition, 573 complexation, 462 conceptual model, 534, 543 conceptual model uncertainty, 582–3 concrete, 286–9 see also low-pH concrete concrete plugs, 291–2 consensus conferences, 728 conservative bias, 535 CoolRep, 622–3, 628, 633, 634 copper and copper alloys corrosion behaviour, 399–403 corrosion potential and pitting potential, 403 general corrosion reaction mechanism, 401 potential-pH diagram for Cu–H2O–Cl– system, 400 roughened surface after exposure to simulated repository environment, 402 core drilling, 124 core lab analysis, 178–9

© Woodhead Publishing Limited, 2010

744

Index

CPT see critical pitting temperature creep, 345–6 crevice corrosion, 398 critical crevice temperature, 395 critical pitting temperature, 395 CRR see colloid and radionuclide retardation CrunchFlow, 371 crustal deformation, 198–9 crystalline geological repository systems characterisation, site surveying and construction technologies, 121–50 future trends, 148–50 disturbance by excavation or waste emplacement, 140–4 chemical disturbance, 143–4 hydrogeochemical disturbance, 142 mechanical disturbance, 141–2 thermal disturbance, 143 feasibility of construction, 146–8 construction of methods, 147 design considerations, 148 stabilising the excavations, 147–8 geochemistry, 133–8 geochemical information application, 135–8 multidisciplinary approach, 134–5 geological structure, 125–7 deformation zones characterisation, 126 discrete fracture network models, 127 fractures and fracture networks characterisation, 126–7 length scales, 125 hydrogeology, 128–33 Borrowdale Volcanic Group rocks, 132 deterministic and stochastic treatment of hydrogeological features, 130 groundwater flow system, 129 hydraulic parameters, 129 hydrogeological conceptual model, 130–1 hydrogeological measurements, 131–3 model domain for groundwater flow calculations, 133 testing the conceptual model, 133 lithologies, 123–5 borehole drilling, 124 geophysical surveys, 123

role of rock units, 124–5 surface mapping, 123–4 radionuclide transport, 138–40 description of transport characteristics, 140 flow data, 139–40 transport data, 138–9 rock mechanics and geotechnical properties, 127–8 mechanical properties, 127–8 in situ stress, 128 stability, 144–6 geoscientific understanding, 144–5 geosphere stability in safety cases, 145–6 implications for repository design, 146 crystalline rocks, 121–2 crystallisation, 428–9 Culebra dolomite aquifer, 508 cumulative distribution function, 549, 556, 592 CUTTINGS_S, 508 DAD approach, 721 Darcy equations, 367–8 Darcy’s law, 472 decide–announce–defend approach, 613 decision analysis, 583 probability encoding, 584–6 quantifying uncertainties, 584–6 deep geological disposal, 59–60 IAEA requirements, 75–9 safety assessment of HLW in geological repository systems, 497–518 deformation zones, 125 characterisation, 126 deliquesce, 385 Delphi Method, 598 Department of Energy’s In Vivo Laboratory Accreditation Program, 690–2 Dewey Lake Formation, 685 3DHYDROGEOCHEM, 371 Dialogue Project, 725–6 DIFFU-CA model, 454 diffusion, 370, 472–3 diffusive transport, 9 direct assessment, 585 Directive 97/11/E, 727 Directive 2001/42/EC, 727–8 Directive 85/337/EEC, 727

© Woodhead Publishing Limited, 2010

Index discharge point, 71 discrete fracture network, 127, 233 dissolution rate, 13 disturbance, 140–4 Dounreay nuclear power station, 195 Drigg facility, 43, 44, 57, 68, 72 drilling, 168–78 programme, 168–9 cored boreholes, 168 geological programme, 169 percussion-air drilling, 168–9 drip shield, 358 duplex stainless steels, 395 CCT and CPT in ferric chloride solution, 397 composition, 396 earthquakes, 202–4 Electric Power Research Institute, 512, 585 electrical conductivity logging, 235–6 electromagnetic surveys, 123 encapsulation, 70 energy conservation equation, 368 Energy Policy Act (1992), 643, 645 engineered barrier system, 69–70, 106–10, 288, 355–9 engineered vaults, 54–6 engineering studies and demonstration of repository design, 110–11 environmental coupling, 449–52 Environmental Evaluation Group, 710 Environmental Impact Assessment, 724, 727–8 environmental monitoring Waste Isolation Pilot Plant, 678–716 Epidemiology and Cancer Control Program, 693 epistemic uncertainties, 536–7 epistemic uncertainty, 548 equilibrium constant, 13 equilibrium equation, 368 equilibrium thermodynamics, 470 ESDRED see Engineering studies and demonstration of repository design excavation, 140–4 excavation damage, 141 excavation damaged zone, 111–12, 364 excavation disturbance, 141 Experimental Breeder Reactor II, 35 expert judgment aggregating assessments from multiple experts, 598–600

745

comparing aggregation methods, 600 scoring rules and differential weighting of expert assessments, 598–600 assessment with multiple experts, 593–7 common elements of assessment with multiple experts, 594 when experts’ assessment differ, 593–4 biases in judgments and formal probability elicitation protocols development, 587–92 degrees of formality in assessments, 603 degrees of rigor and formality in assessing expert judgments, 600–602 formal elicitation protocols, 589–92 future trends, 602, 604–5 judgmental biases, 587–9 anchoring and adjustment, 588 availability, 587–8 overconfidence, 588–9 representation, 589 key differences in approaches to assessments with multiple experts, 595–7 feedback to experts, 597 process and expert interactions, 596–7 process of assessing judgments from multiple experts, 595 safety analyses and performance assessment of geological repository systems, 580–605 structure and assessment, 583–4 expert system approach, 622 Exploratory Studies Facility, 89 extended x-ray absorption fine structure spectroscopy, 468 far field, 222 far-field process analysis and radionuclide transport modelling in geological repository systems, 222–52 argillaceous sedimentary formations transport and retardation, 227–32 crystalline-basement environments transport and retardation, 232–8

© Woodhead Publishing Limited, 2010

746

Index

emerging trends, 249–52 framework, 222–7 quantifying radionuclide transport, 238–49 Faraday constant, 470 FAS see fixed air samplers fault rupture, 194 fault trenching, 201 features, events & processes, 212, 239–41, 503–5, 514, 531, 582, 672–5 categories, 532–3 evaluation results for repositories in Opalinus Clay, 242–3 process, 531 screening techniques, 532 FEBEX see full-scale engineered barriers experiment FEP see features, events & processes Fick’s law, 9, 454, 456, 472 finite element approach, 212 fitted distribution, 592 fixed air samplers, 702 flow focusing, 585 flow-wetted surface, 232 flow-diversion barrier see Richards barrier fluid conductivity logging, 173 fluvial terraces, 205 FMT, 506, 508 focus group, 729 Fo¨rsmark facility, 44, 57 FORTRAN, 543 forward chaining, 629 French vitrified glass, 423 French Waste Management Act (June 2006), 427 Freundlich isotherm, 469 fuel matrix release, 448 full-scale engineered barriers experiment, 98, 106–9 current layout, 107 large-scale process testing, 106, 108–9 fuzzy logic theory, 212 fuzzy set theory, 537 gas, 349–50 gas migration, 156 gas migration test, 98 large-scale process testing, 109–10 gehlenite, 302 gel, 432 generic URLs, 85–6, 87–8 geochemistry, 133–8

geochemical information application, 135–8 concentration and fluxes of analogue solutes and mobile species, 137 controls on groundwater composition, 136 establishing baseline hydrochemical conditions, 137 groundwater ages and solute travel times evaluation, 136–7 hazardous or anomalous conditions or scenarios identification, 138 identification of objectives, 135–6 palaeohydrogeological stability, 137 redox model development, 136 surface compartments of hydrogeological model, 137 multidisciplinary approach, 134–5 Geochemist’s Workbench, 371 geodetic surveying, 204 geological disposal, 4 safety, 642–6 NAS report impact in Sweden, 644–5 non-human species protection, 646 Swedish radiation protection standard, 645–6 Swedish standard development, 643 US National Academy of Sciences, 643–4 spent nuclear fuel recycling, 36–41 geological environments, 188–213 geological mapping, 166–7, 196–8 geological repository, 642 geological repository systems aqueous chemical processes in defining the relevant aqueous chemical species, 459–66 aqueous complexation, 462–4 deep porewater compositions, 460 hydrolysis reactions, 460–1 precipitation and coprecipitation, 464–6 redox reactions, 462 argillaceous sedimentary formations transport and retardation, 227–32 Cl– distribution in pore water of Opalinus Clay, 231 geometric properties and transport parameters, 228

© Woodhead Publishing Limited, 2010

Index assessment of probability for scenarios, 653–4 completeness, 653–4 omega of risk, 654 candidate waste forms and disposition schemes, 271–8 borosilicate glass compositions, 272 borosilicate glasses, 271–3 glass–ceramics, 276–7 phosphate glasses, 273 silicate, aluminate and phosphate ceramics, 273–4 spent fuel, 277–8 synroc-C titanate ceramics composition and phase abundances, 275 titanate ceramics, 274–6 coupling with transport processes, 471–7 anionic exclusion origin within clay-rich rocks, 475 Cl and I experimental diffusion coefficient, 476 classical retardation factor values for bentonite, 474 diffusion and advection processes, general transport equation, 471–3 ionic retardation and anionic exclusion, 473–5 migration simulations and confrontation with experimental results, 476–7 crystalline-basement environments transport and retardation, 232–8 connected porosity experiment at Grimsel Test Site, 233–4 electrical conductivity logging in boreholes, 235–6 Grimsel granodiorite connected porosity, 234 matrix fluid studies, 234–5 in situ diffusion experiments, 236–7 treatment of matrix diffusion in geosphere-transport calculations, 237–8 dose and risk, 646–9 nominal risk coefficients, 647 risk definition, 646 Swedish risk standard, 647–8 Swedish standard’s dose or risk limit, 648–9 emerging trends, 249–52

747

argillaceous rocks microstructure, 251–2 matrix pore fluids chemical characterisation, 250–1 in situ studies of porosity and transport parameters, 249–50 environmental monitoring and public engagement for siting and operation of WIPP, 678–716 air monitoring, 702–15 CEMRC history and current status, 689–93 contaminant exposure and environmental risks perceptions survey, 693–5 future trends, 715–16 history of salt and site selection, 682–9 internal dosimetry and whole body monitoring, 696–702 expert judgment assessment for safety analyses and performance assessment, 580–605 far-field process analysis and radionuclide transport modelling, 222–52 framework, 222–7 radionuclide inventories of Finnish and Swiss waste types, 226 rock types as potential host formations, 224–5 role of geosphere in a repository system, 223–4 time scales of concern, 225–7 future trends, 659–60 geological disposal, 280–4 deep geological repository schematic diagram, 281 waste form research impact on future fuels and future trends in HLW management, 283 historical waste-form development for processing, 265–71 design drivers for high-level nuclear waste forms, 267–8 ionic concentrations in water in Canadian Shield, 270 radiation damage, 270–1 waste form durability, 268–70 HLW deep geological disposal safety assessment, 497–518 HLW generation from nuclear fuel, 261–5

© Woodhead Publishing Limited, 2010

748

Index

fission products abundances relative distribution, 262 fission yield vs mass number for uranium-235, 264 half-lives of key fission product radionuclides, 263 types of radioactive waste, 263–5 IAEA proposed waste disposal depth identification, 527 IAEA waste classification system, 526 inert matrix fuels, 278–80 cements and geopolymers, 279–80 long-lived FPs transmutation, 280 KMS application for safe geological disposal of radioactive waste, 610–34 long-term nuclear glass performance, 427–37 glass behaviour in closed system, 428–30 glass behaviour in water-saturated open system, 430–7 low- and intermediate-level waste performance overview, 452–8 bitumen, 455–6 cement and concrete, 452–4 compacted hulls and end-pieces, 456–8 low pH concrete development and application for structural purposes, 286–316 functional cementitious material requirements, 290–4 long-term durability, 311–16 low-pH cements design and properties, 294–304 low-pH concrete development and production, 304–11 multiple-barrier design and safe disposal operation strategies for radioactive materials, 3–25 basic disposal strategies, 7 concentration constraints, 12–24 concepts, 5–6 containment, 7–12 near-field processes evolution and performance assessment, 353–76 nuclear waste canister materials corrosion behaviour and long-term performance, 379–413 optimisation and best available technology, 657–9 BAT, 658

justification, 657 optimisation, 657–8 optimisation vs BAT for waste disposal, 658–9 use of optimisation and BAT in regulation, 659 post-containment performance, 421–79 different waste forms relative performance, 458 main radionuclides behaviour, 477–8 radionuclide fate after release, 458–9 waste form degradation, 423–7 probability and risk, 649–53 alpha and omega, 650 clairvoyant test, 651–2 dilution and intergenerational distribution, 652–3 objections to usage, 651 regulator and operator, 650 quantifying radionuclide transport, 238–49 features, events and processes, 239–41 FEP evaluation results for repositories in Opalinus Clay, 242–3 geosphere treatment in safety assessment calculations, 241, 244–9 radionuclides release rates from bentonite buffer and host rock, 244 spent fuel disposal concepts, 239 radiation protection principles and standards development, 641–60 retention processes as net retardation effect, 466–71 amplitude of retention processes, 466–7 attempts to derive more mechanistic and predictive models, 470–1 empirical modelling of retention, 469–70 origin of retention processes, 467–9 risk-informed performance-based regulations development, 663–75 future trends, 675 methodologies development and application, 668–75

© Woodhead Publishing Limited, 2010

Index safe disposal regulatory principles and methodologies, 664–8 smectitic buffer and backfill materials development and application, 323–51 backfill types, properties and fabrication, 341–4 buffer design and performance, 331–9 buffer types, properties and fabrication, 326–30 clays for isolation of high-level waste, 323–4 long-term performance, 344–51 smectite clays microstructural constitution, 324 smectite minerals, 324 spent nuclear fuel and high-level radioactive waste immobilisation for safe disposal, 261–84 spent nuclear fuel long-term behaviour, 437–52 evolution of spent nuclear fuel in closed system, 440–4 evolution of spent nuclear fuel in repository, 444–52 fuel pellet condition after irradiation, 437–40 spent nuclear fuel recycling practices, technologies and impact, 29–42 advanced technologies, 33–6 current technologies, 32–3 future trends, 41–2 recycling impact on geological disposal, 36–41 time scales, 654–6 cut-off time justification, 656 human vs geological time, 654–6 treatment of uncertainty in performance assessment, 547–74 underground research facilities and rock laboratories for geological concepts disposal development, 82–116 case studies, 102–15 future trends, 115–16 planning and designing, 98–101 public outreach and role of URLs as training forms, 101–2 URLs definition and roles, 82–5 URLs types and role, 85–98 geomechanical tests, 175–7 sleeve fracturing test, 176

749

systematic analysis of breakouts in boreholes, 176–7 traditional hydraulic fracturing test and HTPF method, 175–6 geomorphic markers, 204 geomorphological evolution, 241 geophysical seismic surveys, 167–8 geophysical surveys, 123 geopolymers, 279–80, 452 geosphere, 222 role in a repository system, 223–4 geosynthesis, 179 glaciation, 195–6 glacio-isostatic adjustment, 196 glass-ceramic waste forms, 267 glass waste forms, 266 glass–ceramics, 276–7 Global Nuclear Energy Partnership, 31 global positioning system, 198–9, 204 global warming, 194–5 Gordian Knot, 611 GRAAL mechanistic model, 434–6 Graseby–Anderson dichotomous samplers, 704 GRASP-INV, 508 gravity surveys, 207 Grimsel Test Site, 90, 96, 97, 98 connected porosity experiment, 233–4 groupware, 632 grouts, 293–4 Habermas´ theory of communicative action, 734 HADES URL, 93 half-life, 7 halite, 685 Hastelloy C-4, 407 Hastelloy C-22, 407 Hastelloy C-276, 407 heat flow, 208 heat-pipe effect, 373 helium, 441–3 heterogeneous waste, 452 HG-A experiment, 112 high-level waste, 44, 353, 360, 380, 614 deep geological disposal safety assessment, 497–518 acknowledging uncertainty, 509–12 applications, 513–16 future trends, 516–18 goals, 499–501 steps in safety assessment, 501–9

© Woodhead Publishing Limited, 2010

750

Index

immobilisation for safe disposal, 261–84 tunnel backfill, 356 high-solubility radioelements, 18 highly compacted bentonite, 385, 387, 389 HLW see high-level waste homogeneity, 91 homogeneous waste, 452 host rock, 222, 223–4, 294, 359–65 hot isostatic pressing, 274 HTPF see hydraulic test on pre-existing fracture human intrusion, 50 hydraulic fracturing, 175 hydraulic test on pre-existing fracture, 175–6 hydrogen-related cracking, 394 hydrogeochemical disturbance, 142 hydrogeological logging, 172–4 electric conductivity fluid logging, 173 flow logging, 172–3 heat-pulse flowmeters, 173 samplers, 173–4 hydrogeology, 128–33 conceptual model, 130–1 hydraulic parameters, 129 measurements, 131–3 characterising fractures and fracture networks, 131–3 hydrolysis reactions, 460–1 IAEA see International Atomic Energy Agency ICRP see International Commission on Radiological Protection ICRP Publication 81, 643 Idaho National Laboratory, 36 ILW see intermediate-level wastes immobilisation, 70 importance analysis, 536 in vivo bioassay, 690–2, 696–7 inclined layer principle, 343 Inconel 625, 407 indirect assessment, 585–6 inert matrix fuels, 278–80 information management, 539–40 Information Synthesis and Interpretation System project, 630, 631 developed and applied expert systems, 630 information technology, 611

initial dissolution rate, 432 InSAR, 204 instant release fraction, 17, 355, 365, 445–8 estimated IRF for various radionuclides in PWR UO2 fuel, 447 evolution of published data for Cs, 446 institutional control, 52–3 interdiffusion, 431–2 intergenerational risk distribution, 652–3 intermediate depth disposal, 57–9, 72–3 intermediate depths, 44, 73 intermediate-level wastes, 44, 67 intermetallics, 406 internal dosimetry, 696 International Atomic Energy Agency, 644 BOSS concept, 65–6, 74 fundamental safety principles, 46–8 requirements for deep geological disposal, 75–9 Safety Requirements, 664–5 International Commission on Radiological Protection, 643 International Tectonics Meeting methodology, 150 iodine-129, 51 ionic sorption, 473–5 IRF see instant release fraction irradiation, 429–30 ISIS project see Information Synthesis and Interpretation System project ISO 9001, 621 JAEA knowledge management system, 615–23 application to safety case development, 618–23 basic concepts, 615–18 Japanese HLW repository, 613 justification, 657 KADS see knowledge acquisition system Kaplan/Garrick ordered triple representation, 558–9 KBS-3 type repository, 357, 372, 374 KBS-3H, 339 KBS-3V, 246, 331, 339 KNetwork 2, 627

© Woodhead Publishing Limited, 2010

Index knowledge, 612 knowledge acquisition system, 630 knowledge asymmetry, 613, 617 knowledge engineering, 624 knowledge management systems application for safe geological disposal of radioactive waste, 610–34 communication, multidisciplinary collaboration and efficient use of resources, 632–3 constructing and visualising safety case arguments, 626–8 critical problems identification and solutions development, 614–15 disposal programme structures and knowledge flows, 612–14 future trends, 633–4 knowledge engineering and advanced information technology, 624 compiling, synthesising and organising knowledge, 628–32 expert system development, 629 integration of multiple approaches, 631 intelligent systems, 630–31 concept of Pairs, 632 definitions and nomenclature, 611–12 expert systems applied in ISIS project, 629–30 hierarchy of knowledge creation, synthesis and application, 624 JAEA knowledge management system, 615–23 application to safety case development, 618–23 basic concept, 615–18 knowledge base compilation and search engine functions, 629 safety case support, 625 SEA top levels, 620 structure and elements, 616 Korea Atomic Research Institute, 36 Kristallin-I safety assessment, 21 KURT-KAERI, 93 La Hague, 32 Lame’s constants, 369 Land Withdrawal Act (1992), 688–9 landslides, 197 Langmuir isotherm, 469 Latin hypercube sample, 559, 567

751

Lawrence Livermore National Laboratory, 276 leaching, 422–3 mechanisms, 430–1 Level IV approach, 601 License Application, 38 Lie Down and Be Counted Program, 697–702 demographics, 698 minimum detectable activities 20072008 calibration, 699–700 results, 701 LILW see low- and intermediate-level waste linear opinion pool, 597 LLW see low-level waste London Dumping Convention, 43 long-term behaviour science, 427 long-term diffusion, 104–6 low- and intermediate-level waste, 59–60, 68, 70, 353, 380 disposal silo in Project SAFE, 357 near-surface disposal safety assessment, 522–43 performance overview, 452–8 B-type waste, 453 bitumen, 455–6 cement and concrete, 452–4 compacted hulls and end-pieces, 456–8 compacted hulls and end-pieces for welded drums, 457 safety assessment application for near-surface disposal, 531–9 low-level waste, 44, 45, 67, 353 low-pH concrete development and application for structural purposes, 286–316 cementitious materials location in HLW repository, 289 pore structure and alkaline liquid phase, 287 development and production, 304–11 aggregate grading and bandwidth limits, 307–8 compressive strength and w/c ratio relationship, 309 concrete design, 308–10 concrete proportions and properties in fresh and hardened state, 310 flowchart for paste constituents selection, 306

© Woodhead Publishing Limited, 2010

752

Index

functional requirements for shotcreted concrete, 305 mix design procedure, 305–8 relative strength evolution vs log time for different types, 309 shotcrete properties and functional requirements, 311 shotcrete trials and sampling, 310–11 trial mix and properties in fresh and hardened state, 310 functional cementitious material requirements for geological disposal, 290–4 host rock, 294 low-pH cement-based grouts required properties, 293 low-pH concrete plugs, 291–2 low-pH injection grouts, 293–4 low-pH rock support, 292–3 rock support functional requirements, 293 specified values for low-pH concrete plugs functional requirements, 292 long term durability, 311–16 EDX microanalyses profiles in low pH samples, 314 groundwater interaction with lowpH concrete, 312–16 rebar corrosion rate evolution, 315 low pH cements design and properties, 294–304 based on other cement types, 303–4 CAC-based cement pastes pore fluid pH, 303 calcium aluminate cement-based, 301–3 decrease of pH in pores of OPC paste, 295 evolution of Na, K and Ca contents, 299 influence of Na2-Oeq on pH, 300 low-pH pastes pore fluid composition, 299 ordinary Portland cement-based, 295–301 pore fluid pH evolution, 298 portlandite content of cement pastes, 300 silicate content of cement formulation influence on pore fluid, 298

use of low-pH cements in reinforced concretes, 316 low-solubility waste form, 17 magnesia phosphate cements, 304 magnetic surveys, 123 magnetollurics, 208 Manta, 631 marine planation surfaces, 205 marine terraces, 205 mass conservation equations, 366–7 mathematical aggregation, 598 Matrix Alteration Model, 449 matrix diffusion, 12, 21, 104 MCC-1 test, 269 mechanical disturbance, 141–2 metastability, 16 Meuse/Haute-Marne URL, 94 microbiologically influenced corrosion, 394, 404 Milankovich cycles, 194 milling, 51–2 milling wastes, 60–2 Mindmap software, 628 mining, 51–2 mining wastes, 60–2 Mizunami URL, 92 model parameter uncertainty, 583 Mont Terri Rock Laboratory, 91, 112 Monte Carlo simulation, 508, 512, 538–9 montmorillonite, 356 Moore’s law, 615 Mount Waldon Intractable Waste facility, 64 MOX fuel, 424, 438–40 multi-attribute utility analysis, 729 multibarrier strategy, 281 multiple-barrier system, 6 multiscale T-H modelling methodology, 373 MX-80, 336, 337, 348 National Research Council, 31 natural barriers, 71 naturally occurring radioactive material, 51–2 near-field processes components, 355–65 containment and isolation description, 365–6 engineered barrier system, 355–9

© Woodhead Publishing Limited, 2010

Index drip shield used in US Yucca Mountain repository, 358 HLW tunnel backfill, 356 LILW disposal silo in Project SAFE, 357 Richards barrier, 359 evolution and performance assessment in geological repository systems, 353–76 future trends, 374–6 host rock, 359–65 dry-out zone, 360 excavation damaged zone, 364 extended dry-out zone, 361 near field stress distribution after waste package emplacement, 362 water saturation contours, 363 modelling overview, 366–74 calculated buffer porosity changes, 372 near field of deep geological repository for HLW, 354 near-surface disposal facility for LLW, 354 near-surface disposal application of safety assessment, 531–9 elements of consequence analysis, 534–5 elements of scenario analysis, 531–4 uncertainties and decisions, 535–9 current role in radioactive waste management, 44–5 definition, 45–6 definition and performance measures, 523–5 future trends, 539–43 elements supporting safety case, 541 information management, 539–40 safety case, 540–2 software advances, 542–3 historical background, 43–4 key issues and development of safety assessment, 525–8 safety assessment methodology, 528–31 safety assessment of LILW, 522–43 safety requirements, 46–8 IAEA safety principles and requirements, 46–8 silos, caverns and tunnels, 57–60

753

deep disposal of low- and intermediate-level waste, 59–60 intermediate depth disposal, 57–9 styles, 53–66 mining and milling wastes, 60–2 trenches and engineered vaults, 54–6 neptunium Pourbaix diagram, 463 solubility evolution as function of redox potential, 465 net retardation effect, 466–71 neutron source tools, 171 Nevada test site, 64 nickel alloy, 457 nickel alloys composition, 407 corrosion behaviour, 407–9 crevice repassivation potential chloride concentration dependence, 409 temperature dependence, 408 Ni–Cr–Mo alloy, 407 Ni–Fe–Cr–Mo alloy, 407 Nominal Group Technique, 598 nominal risk coefficients, 647 non-parametric regression, 574 NORM see naturally occurring radioactive material normal residential intrusion zone, 526 Np-237, 22–4 Nuclear Energy Agency, 650 Nuclear Fuel Waste Act, 722 nuclear glass behaviour in closed system, 428–30 self-irradiation behaviour, 429–30 thermal stability, 428–9 behaviour in water-saturated open system, 430–7 glass alteration regimes and predominant mechanisms, 430–1 GRAAL mechanistic model, 434–6 initial dissolution rate, 432 initial interdiffusion, 431–2 modelling, 434 rate drop, 432–3 residual alteration rate in closed system, 433 resumption of alteration, 433–4 V0→ Vγ operational model, 436–7 calculated altered glass mass, 436 kinetic regimes, 431 long-term performance, 427–37

© Woodhead Publishing Limited, 2010

754

Index

R7T7-type glass density vs alpha decay dose, 430 nuclear power, 261–2 Nuclear Regulatory Commission, 30 nuclear waste canister materials advantages and disadvantages of various candidate materials, 411 Belgian supercontainer concept illustration, 388 corrosion behaviour, 391–410 carbon steel and cast iron, 391–4 ceramics, 410 copper and copper alloys, 399–403 nickel-based alloys, 407–9 stainless steels, 395–9 titanium alloys, 403–7 environmental aspects, 380–9 importance of near-field environment, 387–8 maximum temperature and ground water chloride concentration comparison, 382 near-field environment evolution, 381 near-field environment managing through engineering design, 388–9 repository environment evolution, 386–7 time dependence of canister temperature and relative humidity in drift, 384 environmental factors, 380–6 ground water and pore water composition, 381–3 mass transport, 385 microbial activity, 383–4 redox conditions, 383 residual and applied stress, 385–6 saturation, 384–5 temperature, 381 future trends, 412–13 in geological repository systems, 379–413 long-term performance, 410–12 comparison of candidate materials, 410 lifetime predictions, 410–12 selection, 389–91 canister failure definition, 390–1 compatibility with other engineered and natural barriers, 390

corrosion-allowance vs corrosionresistant materials, 390 target lifetimes, 389–90 Nuclear Waste Management Organisation, 722 Nuclear Waste Policy Act (1982), 30, 41, 665–6 nucleation, 428 NUMO structured approach, 619 NUREG/CR-6372, 593 NUTS, 506, 508 Oak Ridge National Laboratory, 683 Ochoan, 685 OECD-NEA International Thermochemical DataBase project, 466 off-diagonal processes, 459 Oklo natural reactor, 281 ONKALO, 89, 99 ontology cleaning, 628 Opalinus Clay, 154–5, 156, 158, 224, 230, 385 OPC see Ordinary Portland Cement OPC-based concrete buffer, 388 operational models, 426 operational safety, 48–9 optimisation, 657–8 vs BAT, 658–9 Ordinary Portland Cement, 287, 294, 295–301 OS3D/GIMRT, 371 Oskarshamn model, 726 Ostwald step-rule, 16 overcoring, 128 overpacks, 356 packer tests, 174–5 Pairs, 630 concept, 631 paleoseismology, 199–202 PANEL, 506, 508 paradigm shifts, 653 Part 60 subsystem criteria, 665–7 partial rank correlation coefficients, 569 partial rank regression, 573 Participatory Technology Assessment, 728 passivating reaction interphase, 431 Peclet number, 229 peneplains, 205 percussion-air drilling, 168–9 percussion drilling, 124

© Woodhead Publishing Limited, 2010

Index performance assessment see also safety assessment computational design, 565–9 expected dose decomposition into expected incremental dose, 566 expected dose to RMEI, 568 integration procedures to obtain expected incremental dose, 567 conceptual structure, 549–57 aleatory uncertainty representation, 552 component models and associated connections, 553 epistemically uncertain variables, 555 near-field processes in geological repository systems, 353–76 sensitivity analysis of geological repository systems, 569–72 sensitivity analysis results for expectations over aleatory uncertainty, 571 uncertainty and sensitivity analysis results for NCSFL, 570 treatment of uncertainty in geological repository systems, 547–74 uncertainty analysis results DN(τ |aN, eM), 561 DSG(t|a, eM), 562 uncertainty propagation, 557–65 mean and quantile curves, 565 Perkin-Elmer Elan, 708 Permian sediments, 685 phosphate glasses, 273 phosphocalcic cements, 304 pitting, 398 pitting resistance equivalent number, 395 plutonium, 33, 703, 711 Poisson process, 554, 571 Poisson’s ratio, 369 pore waters, 250–1 Portland cement, 452 portlandite, 288 post-closure safety, 49–51 powellite, 428 pozzolan, 302 precipitation and coprecipitation, 464–6 probabilistic seismic hazard analysis, 202, 600 probabilistic volcanic hazard analyses, 596, 601 probability, 584, 651

755

encoding, 584–6 probability density functions, 209 probability measure, 549 probability space, 549, 550, 554 product consistency test, 269 Project Salt Vault, 683 Prototype Repository Project, 336 PUREX process, 32–3 pyroprocessing technologies, 35–6 flow chart, 37 Q-brines, 407, 409 quality management system, 620, 627 quartz, 348 radiation, 350 radiation damage, 270–1 radiation protection principles standards development for geological repository systems, 641–60 assessment of probability, 653–4 dose and risk, 646–9 future trends, 659–60 geological disposal safety, 642–6 optimisation and best available technology, 657–9 probability and risk, 649–53 SSI and SKI major regulatory activities, 642 time scales, 654–6 radioactive decay, 7 radioactive materials basic disposal strategies, 7 concentration constraints, 12–24 additional waste-form considerations, 16–18 cumulative effect of concentration constraints, 21–4 far-field transport, 20–1 sequential and cumulative effects of multiple concentration constraints, 22 spatially distributed containment failure, 19–20 temporally distributed containment failure, 18–19 time/space function in calculated distribution of Np-237, 23 waste-form dissolution and radioelement solubility, 13–16 concepts, 5–6 representative illustration, 6 material containment, 7–12

© Woodhead Publishing Limited, 2010

756

Index

additional issues, 11–12 canister containment, 7–8 transport time, 8–11 multiple-barrier geological repository design and safe disposal operation strategies, 3–25 radioactive waste assessing long-term stability of geological environments for safe disposal, 188–213 future trends, 210–12 geochemical stability issues, 193–4 long-term stability evaluation process steps, 210 modelling long-term stability, 209–10 potential climate change issues, 194–6 designing for safety, 66–72 disposal environment, 69 engineered barriers, 69–70 natural barriers, 71 safety functions, 71–2 stakeholder views, 66–7 waste acceptance criteria, 67–8 future trends, 72–4 borehole disposal, 73–4 Drigg LLW repository showing past, present and possible future disposals, 73 intermediate depth disposal, 72–3 remediation of historical nearsurface disposal facilities, 72 geological, geophysical and geochemical techniques for quantifying stability for safe disposal, 196–23 active fault mapping and paleoseismology, 199–202 Alpine Fault, 200 current crustal deformation measurement, 198–9 detecting crustal structure and volcanic intrusions, 206–8 geological mapping, 196–8 historical seismological record, 202–4 indicators or tectonic uplift or subsidence, 204–6 size and depth of earthquakes in New Zealand, 203 IAEA safety principles and requirements, 46–8

long-term volcano-tectonic stability issues for safe disposal, 189–92 Indian continental plate northward migration, 192 tectonic plate boundaries of the earth, 191 near-surface, intermediate depth and borehole disposal, 43–79 current role of near-surface and borehole disposal, 44–5 defining near surface, 45–6 historical background to nearsurface disposal, 43–4 outline, 46 near-surface disposal safety requirements, 46–8 safety of disposal facilities, 48–53 operational safety, 48–9 post-closure safety, 49–51 safety of mining and milling wastes, 51–2 significance of institutional control period, 52–3 styles of near-surface disposal, 53–66 borehole at MosRadon site near Moscow, 63 borehole facilities for large- and small-volume waste packages, 62–4 BOSS concept schematic, 65 closed near-surface repository at Centre de la Manche, France, 56 Fo¨rsmark LILW repository, 58 Greater Confinement Test facility schematic, 64 IAEA BOSS concept for disused sealed sources, 65–6 investigatory tunnel at Rokkasho, Japan, 59 mining and milling wastes, 60–2 operations at Drigg, UK, 57 sectioned waste package, 55 silos, caverns and tunnels, 57–60 types, 263–5 radioactive waste management social dialogue methods, 719–37 context, 730–5 emergence of participation, 721–2 public participation, 727–30 rationale for participation, 723–5 Swedish dialogue and transparency process, 725–7 radioactivity, 57

© Woodhead Publishing Limited, 2010

Index radioelement solubility, 16–17 radiolytic dissolution, 448–9 radiometric surveys, 123 radionuclide migration, 423, 477, 479 radionuclide transport modelling and far-field process analysis in geological repository systems, 222–52 argillaceous sedimentary formations transport and retardation, 227–32 crystalline-basement environments transport and retardation, 232–8 emerging trends, 249–52 framework, 222–7 quantifying radionuclide transport, 238–49 radionuclides aqueous form as function of charge, radius and electronegativity, 461 behaviour, 477–8 colloid and radionuclide retardation experiment, 102–3 estimated IRF in PWR UO2 fuel, 447 fate after release, 458–9 half-life period, porosity accessible to diffusion, effective diffusion coefficient and retardation factor, 474 key fission product half-lives, 263 location within the fuel rod, 439–40 long-term diffusion tests, 104–6 mobility and migration in geosphere, 459 nature and location in UO2 fuel, 440 release mode from spent nuclear fuel, 445 release rates from bentonite buffer and host rock, 244 transport, 138–40 description of transport characteristics, 140 flow data, 139–40 quantification, 238–49 transport data, 138–9 radiotoxicity, 40 radium-226, 58, 68 RADON, 62, 72 radwaste see radioactive waste rate drop, 432–3 reasonably maximally exposed individual, 649, 672

757

reasonably minimally exposed individual, 561 recoil nuclei, 429–30 red queen’s race, 614 redox reactions, 462 reference lotteries, 584–5 repository cap, 50, 56, 68 repository concept, 619 representational model uncertainty, 583 representativeness, 91 residual rate, 433 Resource Conservation and Recovery Act, 690 retardation factor, 474 retardation process, 232 retention, 466 retention processes Am sorption of bentonite, 471 amplitude, 466–7 attempts to derive more mechanistic and predictive models, 470–1 classical distribution coefficient on smectite, 468 Cs, Np, Am and Se sorption evolution on montmorillonite clay mineral, 467 empirical modelling, 469–70 as net retardation effect, 466–71 origin, 467–9 Richards barrier, 358–9 Rio Declaration (1992), 730 RISCOM II, 726 RISCOM model, 726, 734–5 risk, 646 risk dilution see intergenerational risk distribution risk governance, 730–2 deliberation approach, 731 precautionary principle, 730–1 risk-informed decision-making, 730 risk-informed performance-based regulations geological repository systems, 663–75 methodologies development and application, 668–75 average deep percolation rates, 674 average dose based on deep percolation rates, 674 biosphere characteristics, 669–72 development costs, 671 features, events & processes, 672–5 river terraces, 205

© Woodhead Publishing Limited, 2010

758

Index

RMEI see reasonably maximally exposed individual rock support, 293 RSC ceramic, 274 R7T7, 423 Rustler Formation, 685 safety, 48–53, 642–6 safety assessment acknowledging uncertainty, 509–12 design of probabilistic safety assessments, 511–12 probability and deterministic approaches to account for uncertainty, 510–11 applications, 513–16 direct comparison to regulatory standards, 513 enhancing scientific understanding of system behaviour, 516 evaluating design alternatives, 514–16 providing guidance to research and model development, 513–14 definition, 498–9 feature, event and process list for scenarios development, 504 four steps in typical safety assessment, 502 future trends, 516–18 goals, 499–501 HLW deep geological disposal, 497–518 LILW near-surface disposal, 522–43 definition and performance measures, 523–5 future trends, 539–43 key issues and development, 525–8 methodology, 528–31 methodology features used extensively internationally, 528 uncertainty analysis, 537 near-surface disposal application, 531–9 consequence analysis, 534–5 scenario analysis, 531–4 uncertainties and decisions, 535–9 scenarios and Monte Carlo simulation, 513 steps, 501–9 developing computational models for relevant processes, 505–6

disposal system characterisation, 502 iterative nature of safety assessments, 509 selection of scenarios for analysis, 502–5 system-level analysis, 506–9 safety case, 540–2, 618 formal development of arguments and structured utilisation of knowledge base, 626 supporting elements, 541 Safety Requirements, 664–5 Salado Formation, 683, 685–7 San Andreas Fault, 208 SANTOS, 508 saturation, 13 saturation index, 370 Scarab, 627–8 scenario, 549, 645 scenario-generating -FEPs, 532 scoring rules, 599 SEA see Strategic Environmental Assessment Seaborn Report, 722 SECO-FL2D, 508 SECOTP2D, 508 seed variables, 597, 599 seismic tomography, 207–8 shadow zone, 362 shallow knowledge, 629 shear strain rate, 346 shear strength, 345 shotcrete properties and functional requirements, 311 trials and sampling, 310 shotcrete plug, 304–11 silicon mass balance equation, 435 silos, 57–60 simple random sampling, 557–9 site descriptive models, 624 site-specific URLs, 86–7, 89 SKI, 642 sleeve fracturing test, 176 slow slip events, 199 smectite clays hydraulic conductivity, 326 measured ion effective diffusivities, 329 microstructural constitution, 324 microstructural units, 325

© Woodhead Publishing Limited, 2010

Index recorded and predicted wetting rate, 339 swelling pressure and density relationship, 328 smectite minerals, 324 smectite powder, 334 smectitic buffer design and performance, 331–9 24-hour old clay plug, 340 bentonite powder in deposition hole, 332 design principle, 331–2 engineered barrier components, 333 experimental, 332–7 KBS-3H type supercontainer components, 340 microstructural changes in porewater redistribution, 335 modelling, 337–9 MX-80 powder maturation, 337 other design principles, 339 precipitated salt in hydrothermal experiment, 334 SKB’s concept KBS-3V, 331 test arrangement, 336 total pressure evolution at midheight of buffer, 338 development and application, 323–51 clays for isolation of high-level waste, 323–4 microstructural constitution of smectite clays, 324 smectite clay microstructural units, 325 smectite minerals, 324 stack of smectite lamellae, 325 long-term performance, 344–51 chemical stability, 347–9 impact of gas under pressure, 349–50 influence of radiation, 350 issues, 344 microbial effects, 350–1 mineralogical changes in hydrothermal tests, 348 model test with loaded canister, 347 nitrogen gas movement in bentonite clay, 350 predicted conversion of smectite to illite, 349 time-dependent mechanical strain, 344–7

759

types, properties and fabrication, 326–30 chemical, biological, and radiological impact, 330 hydraulic processes, 330 mechanical properties, 330 oedometer for hydraulic conductivity and swelling pressure determination, 327 thermal impact, 330 SNF see spent nuclear fuel sodium zirconium phosphate, 273 SON68 nuclear glass dissolution kinetics, 434 spent fuel, 277–8, 380 see also spent nuclear fuel Spent Fuel Stability under Repository Conditions project, 446 spent nuclear fuel, 353, 360, 515 alteration rates vs specific alpha activity, 450 evolution in closed system, 440–4 30-year-old PuO2 pellet stored in inert atmosphere, 444 chemical evolution, 441 helium behaviour, 441–3 helium produced over time by alpha decay, 442 physical evolution, 443–4 residual activities in UO2 fuel, 441 evolution in repository, 444–52 fuel matrix release, 448 instant release fraction, 445–8 processes governing matrix radiolytic dissolution, 449 published IRF data for Cs, 446 radiolytic dissolution, 448–9 radionuclides release mode, 445 transition to corrosion and chemical dissolution, environmental coupling, 449–52 fuel pellet condition after irradiation, 437–40 chemical composition, 439 pellet and microstructure state after irradiation in the reactor, 439 physical condition, 438 radionuclide location within the fuel rod, 439–40 immobilisation for safe disposal, 261–84 long-term behaviour, 437–52 spent nuclear fuel recycling

© Woodhead Publishing Limited, 2010

760

Index

advanced technologies, 33–6 aqueous technologies, 34 pyroprocessing technologies, 35–6 UREX + process and waste/ storage products, 35 current technologies, 32–3 impact on geological disposal, 36–41 practices, technologies and impact on geological repository systems, 29–42 sprayed concrete see shotcrete SRI protocols, 589 SSHAC Approach, 593, 595, 596 SSI, 642 stability, 144–6 stainless steel, 457 stainless steels atmospheric corrosion rates, 397 CCT and CPT in ferric chloride solution, 397 corrosion behaviour, 395–9 general corrosion rates, 396 pitting and pit repassivation potentials for Type 316L stainless steel, 398 standard, 642 standardised rank regression coefficients, 569 stepwise rank regression, 573 Stipulation Act, 643 Stokes–Einstein relation, 473 STOLA see STudie- en Overleggroep Laagactief Afval–Dessel Strategic Environmental Assessment, 618–19, 727–8 top levels expressed as argumentation network, 620 stress corrosion cracking, 394, 401–2 stretching, 726 STudie- en Overleggroep Laagactief Afval–Dessel, 67 subjective probability, 593 subsidence, 204–6 subsystem requirements, 666 sulphoaluminate cements, 452 surface mapping, 123–4 Swedish National Council for Nuclear Waste, 726–7 Swedish Nuclear Fuel and Waste Management Company, 124 Swedish Project SAFE, 357 Swedish Radiation Safety Authority, 727

Swedish standard radiation protection principles and geological repository systems, 641–60 assessment of probability, 653–4 dose and risk, 646–9 future trends, 659–60 geological disposal safety, 642–6 optimisation and best available technology, 657–9 probability and risk, 649–53 time scales, 654–6 Swiss PEGASOS expert elicitation, 605 synroc, 274 synroc-C ceramic, 275 T-H coupling model, 373 tailings, 61–2 technically enhanced NORM, 52 tectonic uplift, 204–6 Terzaghi concept, 344 thermal disturbance, 143 thermal models, 39 Thermal Oxide Reprocessing Plant, 32 thermal pulse, 381 thermal rock spalling, 143 Think Tank approach, 619, 634 tide gauges, 204 time-resolved laser-induced fluorescence, 468 titanate ceramics, 274–6 titanium alloys corrosion behaviour, 403–7 passive film properties on titanium, 406 potential-pH diagram for Ti–H2O system, 405 relationship between various grades of titanium alloys, 404 Tongariro Volcanic Centre, 207 top-down approach, 95 Total Systems Performance Assessment, 37–8, 548, 646 TOUGHREACT, 371, 373 trace elements, 18 transition-state theory, 370 transmutation damage, 271 transparency, 734–5 transport equation, 473 transport process, 232 transport time, 8–11 impact on reduction/elimination of

© Woodhead Publishing Limited, 2010

Index the initial inventory as radionuclide half-life function, 10 transuranic radioactive waste, 722 trenches, 54–6 Trip Blank, 707 TRIZ, 622 TRLIF see time-resolved laser-induced fluorescence tunnels, 57–60 uncertainty analysis, 535–9 results DN(τ |aN, eM), 561 DSG(t|a, eM), 562 and sensitivity analysis results for NCSFL, 570 treatment in geological repository systems performance assessment, 547–74 uncertainty propagation, 557–65 underground research laboratories case studies, 102–15 CRR test side three-dimensional view, 103 demonstration experiments, 110–11 engineered barriers, 106–10 ESDRED shotcrete plug layout, 110 FEBEX current layout, 107 gas migration test set-up, 109 gas path through the host rock and along seal sections, 113 hydromechanical evolution of backfill and host rock, 111–15 microtunnel and site instrumentation, 113 post excavation experimental sequence, 114–15 in situ characterisation and testing, 102–6 in situ monopole diffusion experiment, 105 future trends, 115–16 future needs, 116 what has been obtained, 115–16 geological disposal concepts and repository systems development, 82–116 definition and roles, 82–4 laboratory studies, in situ experiments and natural analogues, 83

761

part of overall waste management strategy, 84 planning and designing, 98–101 participation and financing schemes, 101 requirements for implementation of site-specific URL, 99–100 timing of URL development and resources required, 100–1 public outreach, 101–2 role as training platforms, 102 types and role in geologic repositories staged development, 85–98 A¨spo¨ Hard Rock Laboratory, 90 bounding approaches for developing URL programme, 95–6 different types, 85–8 generic URLs overview, 87–8 generic URLs vs site-specific URLs, 86 Grimsel Test Site, 90 Grimsel Test Site example, 97 HADES URL, 93 KURT-KAERI underground research tunnel, 93 Meuse/Haute-Marne URL, 94 Mizunami URL, 92 Mont Terri rock laboratory, 91 ONKALO underground rock characterisation facility, 95 past and present URLs, 88–95 programmes evolution, 96–8 site-specific URLs overview, 89 Yucca Mountain Exploratory Studies Facility, 94 UO2 fuel, 424 uranium, 52 uranium-235, 264 uranium-238, 68 uranium oxide, 278 UREX + process, 34 URLs see underground research laboratories US Energy Policy Act, 643 US National Academy of Sciences, 643–4 US Nuclear Regulatory Commission, 663–4 V0→ Vγ operational model, 436–7 vertical seismic profiling tool, 126 very low-level waste, 44

© Woodhead Publishing Limited, 2010

762

Index

vitreous waste forms, 266 vitrification, 427 vitrified waste, 515 see also nuclear glass VLLW see very low-level waste volcanic intrusions, 206–8 volcano-tectonic stability, 189–92 Washington TRU Solutions, 710 waste emplacement, 140–4 waste-form dissolution, 13–16 schematic diagram, 14 waste forms, 355 degradation, 423–7 integrated approach of science of long-term behaviour, 425 specificity of predicting long-term evolution in repository, 424 difficulties in waste confinement studies, 425–6 disposition schemes, 271–8 borosilicate glasses, 271–3 glass–ceramics, 276–7 phosphate glasses, 273 silicate, aluminate and phosphate ceramics, 273–4 spent fuel, 277–8 titanate ceramics, 274–6 generic boundary conditions, 427 historical development for processing, 265–71 design drivers for high-level nuclear waste forms, 267–8 durability, 268–70 radiation damage, 270–1 impact of research on future fuels, 283 relative performance, 458 Waste Isolation Pilot Plant, 31–2, 500, 680–1, 721–2 1996 WIPP performance assessment linkage of major codes, 507 system model, 507 air monitoring, 702–15 ambient aerosol monitoring, 711–13 elements in exhaust air, 714 fixed air samplers results, 709–11 underground air elemental data, 713–15 contaminant exposure and

environmental risks perceptions survey, 693–5 environmental monitoring and siting for geological repository systems, 678–716 history of salt and site selection, 682–9 between construction and operation, 688–9 geologic column, 686 salt deposits locations, 684 salt repository construction, 687–8 selection criteria for salt repositories, 682 site characterisation and geology, 684–7 site selection, 682–4 internal dosimetry and whole body monitoring, 696–702 Lie Down and Be Counted Program, 697–702 waste packages, 356 behaviour in environmental conditions, 426–7 closed system, 426 open unsaturated medium, 427 water-saturated open system, 426 water management, 56 WIPP see Waste Isolation Pilot Plant WIPP Compliance Certification Application (1996), 506 Young’s modulus, 369 Yucca Mountain, 30–1, 38–41, 392 biosphere, 669–72 development costs, 671 Review Plan, 668 site groundwater, 460 Yucca Mountain Exploratory Studies Facility, 94 Yucca Mountain Total System Performance Assessment (2008), 509, 548 Z2/R ratio, 460–1 ZEDEX, 141 zinc chromite spinel, 428 Zircaloy, 456, 458 zirconium alloy, 456, 457

© Woodhead Publishing Limited, 2010