Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering: Resource and Infrastructure Geotechnics in Africa: Putting Theory into Practice

PROCEEDINGS OF THE 15TH AFRICAN REGIONAL CONFERENCE ON SOIL MECHANICS AND GEOTECHNICAL ENGINEERING This page intentio...

Author: S.W. Jacobsz | C. Quadros

4 downloads 263 Views 26MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!

Report copyright / DMCA form

DOWNLOAD PDF

PROCEEDINGS OF THE 15TH AFRICAN REGIONAL CONFERENCE ON SOIL MECHANICS AND GEOTECHNICAL ENGINEERING

This page intentionally left blank

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering Resource and Infrastructure Geotechnics in Africa: Putting Theory into Practice

Edited by

Carlos Quadros TÉCNICA-Engenheiros Consultores, Maputo, Mozambique

and

S.W. Jacobsz Department of Civil Engineering, University of Pretoria, Pretoria, South Africa

Amsterdam • Berlin • Tokyo • Washington, DC

© 2011 The authors and IOS Press. All rights reserved. No part of this book may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, without prior written permission from the publisher. ISBN 978-1-60750-777-2 (print) ISBN 978-1-60750-778-9 (online)

Publisher IOS Press BV Nieuwe Hemweg 6B 1013 BG Amsterdam Netherlands fax: +31 20 687 0019 e-mail: [email protected] Distributor in the USA and Canada IOS Press, Inc. 4502 Rachael Manor Drive Fairfax, VA 22032 USA fax: +1 703 323 3668 e-mail: [email protected]

LEGAL NOTICE The publisher is not responsible for the use which might be made of the following information. PRINTED IN THE NETHERLANDS

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved.

v

Preface The Mozambican Geotechnical Society (SMG) organized with great pleasure the 15 th African Regional Conference on Soil Mechanics and Geotechnical Engineering in Maputo. The valuable contribution of the South African Geotechnical Chapter particularly in the review of the abstracts and papers is gratefully acknowledged. The general theme of the conference was Resource and Infrastructure Geotechnics in Africa: Putting Theory into Practice. More than half of the papers submitted by authors are related to the construction of geotechnical works in Africa. Roads, airports, bridges, dams, railways, among other significant works were the subject of these papers. This signals a remarkable growth in the number of infrastructure projects that have been carried out or are under construction in Africa. The increasingly specialized nature of the construction works and some very difficult local conditions demand a deeper knowledge of soil mechanics and geotechnical engineering and the involvement of large numbers of geotechnical engineers, as well as specialists of related areas such as geology, rock mechanics, subsurface investigation and field and laboratory testing. The drastic increase in the number of projects in the mining industry will also create additional opportunities and challenges for geotechnical engineers. The proper training of these individuals must be a priority in Africa. We hope that this conference has made a significant contribution towards this goal. The 94 papers submitted to this Conference are presented in 8 sections namely Roads (17), Foundations (14), Lateral Support and Retaining Walls (11), Materials Testing (16), Site Investigation (20), Environmental Engineering (5), Slopes (3), Dams (2) and General (6). Three Keynote Lectures presented at the Conference on relevant issues for the African continent are included in this volume. The Editors wish to thank the authors for their valuable work in the preparation of the papers and the members of the Organizing Committee and of the Scientific Committee for the assistance and engagement that made this publication possible. The Editors

This page intentionally left blank

vii

Conference Advisory Committee (CAC) Samuel Jean-Louis Pedro Sêco Neil Mounir Carlos Peter Etienne Saturnino

EJEZIE BRIAUD PINTO TAYLOR BOUASSIDA QUADROS DAY KANA CHEMBEZE

Vice-President of ISSMGE for Africa President of ISSMGE Immediate past President of ISSMGE Secretary General of ISSMGE Immediate past Vice President of ISSMGE for Africa President of Mozambican Geotechnical Society (SMG) Past Vice-President of ISSMGE for Africa Co-Chairman of 14th ARC Secretary of Mozambican Geotechnical Society (SMG)

Conference Organizing Committee (COC) Carlos Saturnino Ivan Adozinda Daniel Elis Ernesto Fleyd Sidney Salomão Ilda

QUADROS CHEMBEZE MINDO MANHIQUE TINGA JOSÉ PALAVE CAMBALA DE ABREU JAMBE SANTOS

Conference Scientific Committee (CSC) Alan Ahmed Carlos Deolinda Eduard Esve Etiene Gavin Gerhard Heather John John Kamel Kolawole M-Abdel

PARROCK ELSHARIEF QUADROS NUNES VORSTER JACOBSZ KANA WARDLE HEYMANN DAVIS MUKABI STIFF ZAGHOUANI OSINUBI BENLTAYEF

viii

Michelle Mounir Nico Nicol Peter Phil Protus Richard Samuel Samuel Trevor

THERON BOUASSIDA VERMEULEN CHANG DAY PAIGE-GREEN MURUNGA PUCHNER AMPADU EJEZIE GREEN

List of Exhibitors Organization/Company

Country

ANE APAGEO ARA SUL CETA COBA COLLINS CONTROLLAB DURA SOLETANCHE BACHY FORDIA FRANKI AFRICA GAST INTERNATIONAL GEOCONTROLE GEODRILL GEOMECHANICS GIGSA GUNDLE HUESKER KAYTECH LEM MACCAFERRI MODENA MOTA- ENGIL NAUE SEDIDRILL SOILLAB STEFANUTTI STOCKS TECNICA TEIXEIRA DUARTE ZAGOPE

Mozambique France Mozambique Mozambique Portugal Mozambique France South Africa France South Africa South Africa Portugal Mozambique South Africa South Africa South Africa Germany South Africa Mozambique South Africa Mozambique Portugal Germany France South Africa South Africa Mozambique Portugal Portugal

ix

Main Sponsors ROAD FUND, Mozambique, www.fe.gov.mz TÉCNICA-Engenheiros Consultores Ltd, Mozambique, www.tec.co.mz CETA Construções e Serviços S.A., Mozambique, www.ceta.co.mz ROYAL EMBASSY OF DENMARK, Mozambique TEIXEIRA DUARTE Engenharia e Construções, S.A., Portugal, www.teixeiraduarte.pt FRANKI AFRICA, South Africa, www.esorfranki.co.za Gold Sponsors DURA SOLETANCHE BACHY, South Africa, www.durasb.co.za SOARES DA COSTA, Mozambique, www.soaresdacosta.pt MACCAFERRI Southern Africa, South Africa, www.maccaferri.co.za Sponsors GEOKON, USA, www.geokon.com COBA Consultores de Engenharia e Ambiente, Portugal, www.coba.pt ARQ Consulting Engineers, South Africa, www.arq.co.za COLLINS Sistemas de Águas Ltd, Mozambique, [email protected] MODENA DESIGN Ltd, Mozambique, [email protected] IT.COM Tecnologias de Informação e Comunicação, Mozambique, www.itcom.co.mz HUESKER, Germany, www.huesker.com SINAVIA Sinalização e Pintura, Ltd, Mozambique, [email protected] JONES & WAGENER Consulting Civil Engineers, South Africa, www.jaws.co.za GUNDLE GeoSynthetics (Pty) Ltd, South Africa, [email protected] ARA-SUL, Mozambique, [email protected] UEM Universidade Eduardo Mondlane, Mozambique

This page intentionally left blank

xi

Contents Preface The Editors

v

Committees and Exhibitors

vii

Sponsors

ix

Section 1. Keynote Lectures The Advances in Everyday Geotechnics in Southern Africa over the Past 40 Years Alan Parrock Towards Developing Paving Materials Acceptance Specifications for Lateritic and Saprolitic Soils Mensa David Gidigasu Use of Geosynthetics to Improve Seismic Performance of Earth Structures Junichi Koseki

3

10 40

Section 2. Dams Pathology of Foundation of Ghezala Dam, a Tunisian Case History Mounir Bouassida, H. Karoui and Moncef Belaid

63

Injection of Contraction Joints at Pretarouca Dam António Costa Vilar and Duarte Cruz

71

Section 3. Environmental Engineering The Challenge of Designing & Constructing Steep Landfill Capping Sealing Systems Using Geogrid Veneer Reinforcement Jörg Klompmaker and Burkard Lenze

77

Design of Soil Covers in Tropical Africa: A Perspective Celestina Allotey and Nii Kwashie Allotey

83

Is There a Future for GCLs in Waste Barrier Systems? Peter Legg and Molly McLennan

89

Design of Hazardous Waste Landfill Liners: Current Practice in South Africa Riva Nortjé, Danie Brink, Jonathan Shamrock, David Johns and Jabulile Msiza

97

Geosynthetic Clay Liners: A Useful New Tool for Environmental Protection in the Engineer’s Toolbox Peter Davies

104

xii

Section 4. Foundations Numerical Modeling of Skirted Foundation Subjected to Earthquake Loading W.R. Azzam

113

Prediction of the Axial Capacity of Bored Piles Using Methods Based on CPT and Static Analysis Approaches Abdul Karim M. Zein and Samah B. Mohammad

119

Case History on the Design of Foundation for Oil Storage Tank on Coastal Plain Sands E.A.J. George, T.J. Atuboyedia and M. Oju

127

Moment-Induced Displacement of Offshore Foundation in the Niger Delta S.U. Ejezie and S.B. Akpila

133

Lateral Response of Suction Caissons in Deep Water Floating Structures off Niger Delta Coast Samuel U. Ejezie and Baribeop Kabari

139

Foundation Design and Construction for an LPG Terminal in a Difficult Geology and Constrained Waterfront in Coastal Lagos Olaposi Fatokun and Gianguido Magnani

145

Variation of Hydrodynamic Forces and Moments on Offshore Piles in the Niger Delta S.B. Akpila and S.U. Ejezie

152

Contribution à l’Analyse du Comportement des Pieux sous Chargement Vertical – Analyse d’Une Base de Données Locale Ali Bouafia and Abderrahmane Henniche

158

The Use of Micropiles as Settlement Reducing Elements H.N. Chang and T.E.B. Vorster

165

Rigid Inclusions in Sand Formation Resting on Compressible Clay Mounir Bouassida

175

Dynamically Loaded Foundations André Archer

183

Construction of a Bridge over the Kwanza River at Cabala in Angola Duarte Nobre, Francisco Caimoto and Baldomiro Xavier

190

Case Studies to Support Recent Advances in Geogrid Technology Clifford D. Hall

196

Shaft Resistance of Model Pile in Wet Soil Mohamed M. Shahin

202

Section 5. Lateral Support and Retaining Structures The Effect of Anchor Post-Tensioning on the Behaviour of a Double Anchored Diaphragm Wall Embedded in Clay Amr Elhakim and Abdelwahab Tahsin

215

xiii

Observed Axial Loads in Soil Nails S.W. Jacobsz and T.S. Phalanndwa

221

Deformations of Soil Reinforced Walls in Relationship with Reinforcement Used Edoardo Zannoni, Marco Vicari and Moreno Scotto

228

Performance Comparison of Vertical-Horizontal with Conventional Reinforced Soil Walls Using Numerical Modelling Binod Shrestha, Hadi Khabbaz and Behzad Fatahi

237

The Behaviour Under Excavation of the Luanda’s Sandy Formation: Case Studies Duarte Nobre, João Pina and Baldomiro Xavier

243

Theoretical Evaluation of the Influence of Cohesion on Lateral Support Design Jacobus Breyl, Gavin Wardle and Peter Day

249

Internally Instrumented Soil Nail Pull Out Tests Jacobus Breyl and Gavin Wardle

255

Reinforced Soil Retaining Wall Systems Reach New Heights in the Middle East Peter G. Wills and Chaido Doulala-Rigby

262

Deep Excavations in Luanda City Centre Alexandre Pinto and Xavier Pita

269

Geotechnical Innovation in Shaft Sinking in the Zambian Copper Belt G.C. Howell

275

The Use of Reinforced Soil to Construct Steep Sided Slopes in Order to Create a Safer Highway – Ruhengeri to Gisenyi Road, Rwanda Peter Assinder, Heribert Schippers and Giuseppe Ballestra

284

Section 6. Materials Testing Characterization of Shear Strength of Abandoned Dumpsite Soils, Orita-Aperin, Nigeria Kolawole Juwonlo Osinubi and Afeez Adefemi Bello The Use of the Crumb Test as a Preliminary Indicator of Dispersive Soils Amrita Maharaj Some Engineering Properties of Fine and Coarse Grained Soil Before and After Dynamic Compaction Brian Harrison and Eben Blom Using Electro-Osmosis Technique in the Improvement of a Ugandan Clay Soil Denis Kalumba, Brenda Umutoni, Robinah Kulabako and Stephanie Glendinning Derived Empirical Relations and Models of Vital Geotechnical Engineering Parameters Based on Geophysical and Mechanical Methods of Testing John Mukabi

293 299

307 313

320

xiv

Quantitative Analysis to Verify the Theory of Soil Particle Agglomeration and Its Influence on Strength and Deformation Resistance of Geomaterials Sirmoi Wekesa, John Mukabi, Vincent Sidai, Sylvester Kotheki, Joram Okado, Julius Ogallo, George Amoyo and Leonard Ngigi

330

Characterizing Bulk Modulus of Fine-Grained Subgrade Soils Under Large Capacity Construction Equipment Joseph Anochie-Boateng

337

Aspects Géologiques et Géotechniques Associés au Projet et à la Construction d’un Tronçon de l’Autoroute de Dakar (Sénégal) Rui Freitas, Virgílio Rebelo, Luís Ferreira and André Cabral

343

Characterization of Granular and Bitumen Stabilised Materials Using Triaxial Testing Kim Jenkins and William Mulusa

349

The Effect of Iron Oxide on the Strength of Soil/Concrete Interface F. Okonta and A. Derrick

355

Moisture Retention Characteristics of Some Mine Tailings S.K.Y. Gawu and J. Yendaw

360

Prediction of Over-Consolidated-Ratio for African Soil Diganta Sarma and Moumy Dsarma

366

The Strength of Compacted Sand in a Modified Shear Box Apparatus F. Okonta and D. Schreiner

376

Experimental Study on Use of Mechanically Stabilized Residual Soils for Pavement Layers in Magoe, Mozambique Raphael Ndimbo

382

Effects of Compaction on Engineering Properties of Residual Soils of Tete – Mozambique Carlos Quadros and Raphael Ndimbo

389

Selection of Pavement Foundation Geomaterials for the Construction of a New Runway Joseph Anochie-Boateng

396

Suggested Improvements in Site Investigation and Numerical Characterization Procedures for House Foundation Design John Terry Pidgeon and Rachael Govender

403

Section 7. Roads Improvement of Unbound Aggregates in Khartoum State O.G. Omer, A.M. Elsharief and A.M. Mohamed Applying the Dynamic Cone Penetrometer (DCP) Design Method to Low Volume Roads Philip Paige-Green

415

422

xv

The Use of a Sedimentological Technique for Assessing the Engineering Performance of Sands in Roads Philip Paige-Green and Michael Pinard Characterization of Pozzolanic Geomaterial for the Construction of Pavement Structures of Songwe Airport in Tanzania Paul Omindo, John Mukabi, Prosper Tesha, Vincent Sidai, Sylvester Kotheki and Leonard Ngigi Correlation Between the Dynamic Cone Penetration Index and the Falling Weight Deflectometer-Determined Subgrade Resilient Modulus Samuel I.K. Ampadu and Emmanuel Klu Okang Fundamental Theory of the ReCap Technique and Its Application in the Construction of Pavement Structures Within Problematic Soils John N. Mukabi, Bernard Njoroge, Tilahun Zelalem, Samuel Kogi, Maurice Ndeda and David Kamau Utilisation des Bétons Compactés au Rouleau (BCR) I.K. Cisse and A. Sall

431

439

446

453

460

Pavement Rehabilitation Options for Developing Countries with Marginal Road-Building Materials Khaimane M.D. de Deus and Wynand Jvd Steyn

468

Applications of Participatory Road Maintenance Using “Do-nou” Technology in Kenya Makoto Kimura and Yoshinori Fukubayashi

476

Modélisation Numérique du Renforcement des Chaussées non Revêtues par Géogrille Mohamed Saddek Remadna, Sadok Benmebarek and Lamine Belounar

482

Reducing the Cost of Road Construction Through Targeted Geotechnical and Geophysical Investigations – A Case Study of Road Section Re-Design in the Hwereso Valley of Ghana C.F.A. Akayuli, S.O. Nyako and J.A. Yendaw

489

Appropriate Engineering Solutions for Rural Roads in Mozambique Luis Fernandes and Irene Simoes

495

Preliminary Studies on the Utilization of Sand Treated with Emulsion Luis Fernandes, Irene Simoes and Hilário Tayob

501

Geosynthetics in Road Pavement Reinforcement Applications Garth James

507

Treatment and Stabilization of the National Road E.N. 379-1 Hillsides, Between Outão and Portinho da Arrábida Jorge Dinis, João Pina and Baldomiro Xavier

518

Contraintes Géotechniques Associées à la Construction de la Deuxième Piste de l’Aérodrome d’Oran en Algérie Vicente Rodrigues, Mário Roldão and António Silva

524

xvi

Effect of Geosynthetic on the Performance of Road Embankment on Algeria Sabkha Soils Sadok Benmebarek, Naima Benmebarek and Lamine Belounar

532

Section 8. Site Characterisation Geotechnical Characteristics of the Portuguese Triassic Mudstones Mário Quinta-Ferreira

541

Hydraulic Conductivity of Compacted Foundry Sand Treated with Bagasse Ash Kolawole Osinubi and George Moses

545

Subsurface Conditions in Central Khartoum Eisa A. Mohamed and Ahmed M. Elsharief

551

An Alternative to the Re-Drive for Determining Rod Friction Exerted in DPSH Testing Charles MacRobert, Denis Kalumba and Patrick Beales

559

Empirical Equivalence Between SPT and DPSH Penetration Resistance Values Charles MacRobert, Denis Kalumba and Patrick Beales

565

The Dynamic Probe Super Heavy Penetrometer and its Correlation with the Standard Penetration Test Brian Harrison and Tony A’Bear

571

The Potential of Using Artificial Neural Networks for Prediction of Blue Nile Soil Profile in Khartoum State H. Elarabi and M. Mohamed

580

Using a Modified Plate Load Test to Eliminate the Effect of Bedding Errors Hennie Barnard and Gerhard Heymann Geotechnical Characterization and Design Considerations in the Moatize Coalfields, Mozambique Gary N. Davis, T.E.B. Vorster and Célia Braga Estimating the Heave of Clays A.D.W. Sparks

587

593 599

Instrumentation and Monitoring During Construction of the Ingula Power Caverns G.J. Keyter, M. Kellaway and D. Taylor

605

Piezocone Investigation of Paleo River Channels at Changane River, Mozambique, for a Railway Embankment H.A.C. Meintjes and G.A. Jones

611

Site Selection of the Mathemele Landfill Carlos Quadros and Ivan Mindo

620

Hazard Assessment on Shallow Dolomite Tony A’Bear and Lindi Richer

626

xvii

Correlations of DCPT and SPT for Analysis and Design of Foundations Dalmas L. Nyaoro and Mwajuma Ibrahim

632

The Effective Porosity Paradigm and the Implications on Empirical Permeability Estimations Matthys A. Dippenaar and J. Louis van Rooy

638

Numerical Modelling of Wave Propagation in Ground Using Non-Reflecting Boundaries S.J. Mbawala, G. Heymann, C.P. Roth and P.S. Heyns

644

Geotechnical Characteristics of the Red Sands of Chibuto, Mozambique H.A.C. Meintjes and G.A. Jones

653

Simple Expansion Model Applied to Soils from Three Sites A. Dereck W. Sparks

663

Correlation Studies Between SPT and Pressuremeter Tests Emmanuel Kenmogne and Jean Remy Martin

669

Section 9. Slopes The Value of Slope Failure Back-Analysis in Open-Pit Slope Design: A Case History from the South African Coalfields Mmathapelo Selomane and Louis van Rooy

679

General Slope Stability Using Interslice Forces and Flow Nets but Avoiding r u Factors A.D.W. Sparks

685

Pit Slope Design Near Tete, Mozambique, Without the Benefit of Previous Slope Performance Experience Phil Clark

691

Section 10. General Équations et Exemple de Calcul Hydrique dans les Sols Non Saturés Abdeldjalil Zadjaoui

701

Processus de la consolidation des sols peu cohérents saturés Mohamed Salou Diane and Salou Diane

709

The African Regional Conferences as an Indicator of Research Trends in South Africa Philip Paige-Green Geotechnical Investigations: Over-Regulated or Under-Investigated? Tony A’Bear and Louis van Rooy Challenges to Geotechnical Engineering Practice in the Urbanization of the City of Accra, Ghana J.K. Oddei

719 726

730

xviii

Soil Improvement Through the Utilization of Agricultural Residues from Nigeria N.L. Obasi and E.B. Ojiogu

736

Subject Index

743

Author Index

747

Section 1 Keynote Lectures

This page intentionally left blank

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-3

3

THE ADVANCES IN EVERYDAY GEOTECHNICS IN SOUTHERN AFRICA OVER THE PAST 40 YEARS Alan PARROCK Managing Director/geotechnical principal of ARQ Consulting Engineers (Pty) Ltd

Abstract. In this paper, the author looks back at the developments in everyday investigations, testing and analysis that have taken place in geotechnical engineering during his 40 year career in the industry to date. Demonstrating how the use of public information freely available on the internet can allow geotechnical practitioners to reduce early project risk, the author goes on to discuss and explore modern equipment and techniques that allow important information to be more-readily and less-intrusively recovered and processed; providing substantially better strength information and predictions of behaviour under load. The use of the computer to reduce human error and involvement in testing is discussed, alongside the obvious benefits now routinely possible through broader and more sophisticated and representative analysis techniques. Looking forward on the basis of past and recent technological progress, the author attempts to explore and predict the developments in geotechnical engineering that we might be likely to see over the coming 4 decades. Keywords. Past, present future, satellite imagery, hyperstectral, fibre optics.

Introduction This paper initially examines the early years in the author’s geotechnical career and how the mode of operation changed from those basic computer starts to what is now the norm. It ends by attempting to make a prediction of what the next 40 years has in store for the geotechnical practitioner.

1. The Early Years The four decades prior to 2011 comprised the 70s, 80s, 90s and the post millennium 2000s. 1.1. The 70s During the 1970s the computer was a monster tucked away in a locked air conditioned room, computer input was via punched cards, programming via FORTRAN and certainly in the early years of that decade, not many people utilised an electronic calculator. The author’s first purchase of a calculator was in 1973 and, as it cost four

4

A. Parrock / The Advances in Everyday Geotechnics in Southern Africa over the Past 40 Years

times his monthly salary, he was forced to share it with his brother. They alternated on six month cycles. The author’s first exposure to some form of desktop computer was when he was employed in the Natal Roads Department and the Materials department owned a Wang. Wikipedia indicates that this was likely to be the LOCI-2 introduced in 1965. It was the first desktop calculator capable of computing logarithms which apparently was quite an achievement as it did not use integrated circuits but was equipped with 1275 discrete transistors. Wang Laboratories (WL) was founded in 1951, peaked in 1981 with annual revenues of $3billion and employed 33 000 people at the time. WL filed for bankruptcy in 1992. The Wang in the Materials department was used to write a program to calculate gradings, Atterberg limits and the A type classifications (A1 to A7). In the absence of what are now readily-available geological maps, not much data could be gleaned in the pre-investigation phase other than that known to locals and available at small-scale in geological literature. The industry thus developed a means to address this and many soil survey firms were active in establishing the geology of routes traversed by roads. Roads were enjoying their heyday at that time [1]. It is of interest to note that the first Bidim geosynthetic was imported from France to RSA in 1971. Local manufacture of the product started in 1978 and during the 70s some 1-2 million m2 were used in civil engineering projects. [2]. The norm for a geotechnical foundation investigation comprised backactorexcavated test pits for shallow deposits while deeper profiles were characterised via core drilling supplemented with Standard Penetration Testing (SPT) and possibly vane shear testing [3]. Undisturbed samples were retrieved from the core via U4 or Shelby tubes. Triaxial testing of undisturbed samples was conducted via hand or machine controlled rates of deformation, which were measured by dial gauges read and recorded manually. Analysis of results and calculations were performed using a slide rule in conjunction with trigonometric tables. The time thus taken, for example, to perform a single circle slope stability evaluation was usually about an hour when the somewhat inaccurate Fellenius solution method of slices was used. This error was reduced when the formulations of Bishop [4] were incorporated, but additional time was required to generate an answer as a process of successive approximation was necessary to obtain a solution to an equation in which the required variable F appeared on both sides of the equation. The first Brink book was published in 1979 and the wealth of information held privately was made available to a much wider audience via reports on case studies. Although the proceedings of the 5th Regional Conference for Africa (ARC) held in Luanda in 1971 and the 6th in Durban in 1975 occupy the author’s bookshelf, he did not attend them as he was no doubt much too young and inexperienced to know about those illustrious authors and occasions. The 7th ARC took place in Ghana in 1979 but South Africans were not permitted to attend. [5] Davis [5] also details that the 1 st ARC was held in Pretoria in 1955, the 2nd in Lourenço Marques, Mozambique (sounds familiar) in 1959, the 3rd in the then-named Salisbury of Southern Rhodesia, and the 4 th in Cape Town in 1967. It certainly is good to have it back here in 2011 after an absence of 52 years.

A. Parrock / The Advances in Everyday Geotechnics in Southern Africa over the Past 40 Years

5

1.2. The 80s Much progress had taken place on the computer front especially during the last few years of the 1980s. Word processing, which had started out with WordStar, had progressed to WordPerfect. The spreadsheet of choice was Lotus 123. Certainly, the technical computer programs were still dominated by those based on the programming language FORTRAN and it was only in the later stages of the 80s that the platforms that these ran on became PCs as opposed to the mainframe VAXs and PRIMES which appeared to rule the roost technically. Local geological maps were more readily available with the Geological Map of Johannesburg at a scale of 1:5 000 being prepared by JH de Beer in 1985. In addition, detailed data for the area was available from records held by the Johannesburg Data Bank. Volume 2 of Brink was published in 1981, Volume 3 in 1983 and the final Volume 4 in 1985. The use of the pressuremeter as an investigation tool was introduced to South Africans in 1980 by Professor CP Wroth of Cambridge University [5] and locally Michael Pavlakis was a proponent of its use. It was used by the author during 1982 as part of the investigation for a 26m deep basement for the planned SA Transport Services Computer Centre located in the Ventersdorp lava of the Johannesburg graben. Probabilistic analysis methods were first mooted in RSA by Milton Harr in 1980 and this was followed in 1982 by Dimitri Grivas who expanded on Harr’s initial approaches. The attributes of the beta distribution and the point-estimate method were certainly employed by the author in many applications, especially as the computer was becoming more useable for everyday analyses. The development in the 80s was frenetic and this was reflected in the number of courses, symposia and conferences which were organized: grouting; ground anchors, slope stability and piling to name a few. The problem materials, collapsible and dispersive soils, soft and heaving clays and dolomites and their residuum were also very well covered. On the investigation front, other than the Pressuremeter, the Dutch probe and the later derivative, the Piezocone, were enjoying much success especially when used as an investigation tool for the soft alluvial deposits of the Kwa-Zulu Natal coastline. Essentially the same techniques employed in the 70s were used in the 80s for shallow and deeper drilling projects. The now ubiquitous 1:250 000 geological maps issued by the Council for GeoScience were also making their appearance. Initially confined to the more populated areas, the series was later expanded to include all of RSA. It was supplemented on a regional basis by 1:50 000 scale versions for the Pretoria region. The 8th ARC took place in Salisbury in 1983 and the 9th in Lagos Nigeria in 1987 (again South Africans were not permitted to attend). 1.3. The 90s The Lateral Support Code, although dated 1989, was released in 1990 and offered many opportunities to those involved in this exciting field. The XT computers of the late 80s were replaced by the 286s and 386s and most engineers had one on their desks. The DOS operating system gave way to Windows and Quattro Pro was the spreadsheet of choice at the beginning of the decade later to be replaced by Excel. On the word processing front, WordPerfect was superseded by

6

A. Parrock / The Advances in Everyday Geotechnics in Southern Africa over the Past 40 Years

Word, AutoCad was the draughting package used by most and programming was a mixture of C, Pascal and other languages. Finite element analyses, almost exclusively the domain of larger organisations, universities and research establishments, were now being used more and more as the PC become more powerful. Certainly the author’s first stab at a geotechnical FE package came about in 1994 running SOILSTRUCT on a 286. This used the non-linear hyperbolic Duncan Chang [6] and Duncan et al [7] formulations as the basis for the code which was written in FORTRAN. The Prokon geotechnical computer programs were released in 1994 running under the DOS operating system. These packages were a joint venture between ARQ and Prokon and all incorporated probabilistic modules which were initially written in the late 80s and early 90s in FORTRAN. The programs operated under DOS and it was not unusual during the simulation routines which often comprised 10 000 iterations, that the computer would be busy for 5-10 minutes. The electronic aspects of geotechnical engineering certainly came of age in the 90s. The first e-mail was installed at ARQ in 1996 and in 1998 the Geotechnical Division of SAICE established their web site. Reports with many pages in colour became the norm, although drawings were almost exclusively issued in black and white. The issue of reports to Clients was however usually only done in hard copy paper format. The use of Bidim geosynthetic had increased to 5million m 2 per year. Recycled two litre cool drink bottles were initially used in the manufacture of this product starting at a rate of 10% and reaching 100% in 1995. The first high strength geosynthetics were imported from overseas sources circa 1995 which was supplemented later by local manufacture The 10th ARC took place in Lesotho in 1991 and as political change was about to happen, South Africans were permitted to attend. The 11 th ARC was held in Egypt in 1995. The highlight of the 90s, certainly from a personal knowledge point of view, was attending the International Conference on Soil Mechanics and Foundation Engineering hosted in Hamburg Germany in 1997. At that conference it was decided that the “Foundation” part of the title would be replaced and that the organisation would in future be known as the International Society for Soil Mechanics and Geotechnical Engineering or ISSMGE. Here the most significant part of the proceedings which impacted the author was the work which had been conducted by Oshima and Tokada [8] on dynamic/ram compaction and that by Mark Randolph on the beauty of using piled rafts to equalise settlements under large structures. This occasion was complemented two years later when attending the 12th African Regional Conference held in 1999 in Durban. The information provided at the mini symposium on the Sunday preceding the conference by Chris Clayton on the SPT has been used on numerous occasions in the intervening 12 years.

2. The New Millenium The start of 2000 was meant to be the time when the Y2K pandemonium reigned. Of course, it was only a perceived threat dreamed up by the computer guys to increase revenue. The effect on the geotechnical fraternity was minimal. Perhaps the defining moment in 2003 for the author was the 2nd Jennings lecture delivered by Harry Poulos entitled ‘Foundation design: the research practice gap’ in

A. Parrock / The Advances in Everyday Geotechnics in Southern Africa over the Past 40 Years

7

which this eminent pragmatist demonstrated that much common sense is necessary in interpreting high level (theoretical) analysis. This was aptly illustrated when in 2003 the ARQ foundation design for a 20 storey building in Luanda, comprising a piled raft solution, was challenged by an international expert (IE) as to the settlement predictions made. ARQ predicted the central raft would settle between 11 and 20mm while the IE was of the opinion that the value would be some 175mm. Serious political fallout followed this assertion and thousands of additional hours and R1m extra was spent in ensuring a deflection of this magnitude could be accommodated by the building. The deflection of the building was monitored and needless to say when, at the end of construction, deflection was only 12mm, the IE was nowhere to be seen. The 14th ARC held in December 2003 was attended in beautiful Marrakech, Morocco. Many delegates had a nightmare trip to get there [9] as most either had first to fly to Paris or jet in from Dubai. The author had a most memorable return trip 1 st class on Air France due to a mix up in booking. The six course meal (with a different wine for each course) was something to behold and when he awoke (somewhat groggily) the next morning, the plane was directly over an airstrip which he had built in 1978/79 in the central Caprivi of Namibia. The memories flooded back and who says it is not fun being a geotechnical engineer? In 2005 a personal highlight was being asked to be the Godfather to the Young Geotechnical Engineers Conference held at the Swadini Spa. Much useful information was gained from the many and divergent papers presented and it was a delight when a Black man and a young lady were adjudged to have the best technical paper and the best presentation respectively. The prize for this was a trip to attend an international conference in Tokyo and for the recipients this was one of the highlights of their lives. The Commemorative Journal of the SAICE Geotechnical Division was published and much of the data contained in this presentation comes from that publication. the author gives eternal thanks, to his friend and colleague, Heather Davis. Computers became faster (they never get cheaper), colour reports and drawings were the order of the day, although most reports and plans are exchanged electronically in .pdf format. The cost of finite element (FE) software for the modeling of complex geotechnical solutions enabled most geo-practitioners to at least own a 2-d version. Google became part of our lives. It was established in 1998, and its initial public offering followed in 2004. The company’s stated mission from the outset was “to organize the world’s information and make it universally accessible and useful.” Google Earth is used on every ARQ geotechnical report for the location of the project. The 3d viewing facility enables geological formations to be spotted with ease by the trained eye and the Street View facility provides the ultimate in gaining information at the desk top study stage. This latter facility has been used extensively to provide input to the designers of fibre-optic cable routes in establishing quantities of hard and soft material. On the investigation front, continuous surface wave (CSW) testing is now the norm for most projects where knowledge of the stiffness of material at depth is required. Recent advances in interpreting the data have eliminated the hard layer overlying a soft one conundrum.

8

A. Parrock / The Advances in Everyday Geotechnics in Southern Africa over the Past 40 Years

3. Summary Thus whereas the practice of test-pitting, core-drilling and seismic testing have essentially remained constant over the past 40 years, other field investigation techniques including piezocone, pressuremeter, CSW, resistivity, gravity and the like have seen the light of day and are now considered the norm. However, with the advent of fast computers, analysis techniques have progressed in leaps and bounds.

4. The Next 40 Years The past has been easy to document, but what about the future? No doubt, computing power will increase exponentially as it always has. This will enable finite element and/or finite difference models to be constructed, probably in three dimensions, and analyses to be performed in static and dynamic modes and the outputs represented either deterministically or as single-valued solutions. Alternatively, it may well be more common to have the answers registered in a probabilistic sense where the solution will be depicted in a band of values with variable probabilities assigned. Remote sensing will in all likelihood become the order of the day. It is not difficult to imagine “electronically flying” to your site of choice, requesting information such as elevation, slope-angles, rainfall, geology at surface and depth, geothermal attributes (conductivity) and any other available attributes which have been put together in a public domain data base populated from information gathered during numerous satellite passes over the site. Already change in groundwater depth is determined by mapping, on successive satellite passes, the change in surface elevation [10]. It is not difficult to comprehend why. A change in, say, 10m depth of water table induces an effective stress change of some 100kPa. 100kPa acting over a soil profile with an E-value of, say, 50MPa would induce a surface deflection change of some 20mm. This is well within the accuracy of satellite predictions at present. Permeability of the world’s surface to depths of 100m has also recently become the norm [11]. Imagine the benefit to groundwater studies. Hyperspectral imagery [12] obtained from an airborne platform enables spectral signatures of various minerals e.g. quartz and kaolinite, plant types and salts, to be mapped over vast areas. These, in turn, can be interpreted to yield probable performance in terms of suitablity for road aggregates, expansivity and salt damage potential, to mention but a few. The performance of structures will be monitored, especially during extreme events, via fibre-optic cables installed within structural elements embedded in the earth. Compressive and tensile forces in foundation elements would be able to be monitored under, say, earthquakes or tsunamis. The propensity for movement of high rock slopes in open pit mines or railway/road cuttings would be monitored remotely and if a danger to personnel or the public was imminent, this could be communicated to them via variable message signs or SMSs on cell phones. Top-of the-range construction machinery will become larger and more powerful although, as has been demonstrated in the airline industry, the majority of the work will in all likelihood be done by a much more modest machine. It is, however, not difficult

A. Parrock / The Advances in Everyday Geotechnics in Southern Africa over the Past 40 Years

9

to imagine that auger machinery of the future may well be able to install piles in excess of 3m diameter to depths which may be as deep as 100m. Geothermal drill holes to some 100m depth will be done with specialised multi-casing percussion rigs such that pollution of the substrata does not result. Intelligent geosynthetics will be the order of the day. They will be able to sense pollutants and, via either chemical injection or electrical change, alter the pollutants to render non-toxic end products. The day-to-day investigations may well be accomplished in a non- destructive manner. Prior planning based on Google Street View images will enable estimates to be made of depth to hard material by examining the plant types present and knowing root penetration potential. Waves will be injected into the ground and response measured. Here one or more of the following: CSW, infra-red, ground penetrating radar, resistivity, magnetics and the like, will probably form the core of what will be done. However one asks, “will the backactor be made superfluous ?” Probably not.

5. Conclusion This note has attempted to span some 80 years, the past 40 and those that lie ahead. Future predictions are notoriously arbitrary and it may well be that the predictions made by the author could be way off and a technology that does not even exist at present, could become the norm. Watch this space.

References [1] [2] [3] [4]

[5]

[6] [7]

[8]

[9] [10] [11] [12]

Taute A 2011. Personal communication. Speech delivered at the offices of Vela VKE during the goingaway ceremony for a retiring staff member. James G. 2011. Personal communication. Telephone conversation with the marketing director of a large geosynthetics company . Blight, GE. 1970. In situ strength of rolled and hydraulic fill. ASCE Journal of Soil Mechanics and Foundations Division. May. pp. 881-899. Bishop AW. 1955. The use of the slip circle in the stability analysis of earth slopes. Geotechnique No 4 pp 128-152. Bishop AW. 1955. The use of the slip circle in the stability analysis of earth slopes. Geotechnique No 5 pp 7-17. Bjerrum L. 1963. Discussion, Proceedings of the European Conference on Soil Mechanics and Foundation Engineering, Wiesbaden. Volume 3. Davis, H. 2006. Concise history of the geotechnical division of the South African Institution of Civil Engineering. pp xi-xxix. Extract from the Commemorative Journal of the Geotechnical Division of the South African Institution of Civil Engineering. Duncan, JM and Chang, C-Y. 1970. Nonlinear analysis of stress and strain in soils. Journal of the Soil Mechanics and Foundation Division of the ASCE. Volume 96 Number SM5 September pp 1629-1653. Duncan, JM, Byrne, P, Wong, KS and Mabry, P. 1980. Strength, stress-strain, and bulk modulus parameters for finite element analyses of stresses and movements in soil masses. Report No UCB/GT/80-01 of the Charles E. Via, Jr. Department of Civil Engineering, Virginia Polythechnic Institute and State University. 70 pp plus Appendix detailing FORTRAN computer printout listing. Oshima, A and Takada, N. 1997. Relation between compacted area and ram momentum by heavy tamping. Proceedings of the 14th International Conference on Soil Mechanics and Foundation Engineering, Hamburg 6-12 September Volume 3. pp. 1641 - 1644. Vermeulen N 2003. Personal communication during an airport meeting to the 14th ARC in Morocco. Young, Susan. 2011. Monitoring groundwater aquifers in agricultural regions. www.stanford.eu Balma, Chris. 2011. Global map of surface permeability. [email protected] Fortescue, Alex. 2011. Hyperspectral imagery solutions. Position IT March 2011 pp. 54-58.

10

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-10

Towards Developing Paving Materials Acceptance Specifications for Lateritic and Saprolitic Soils Mensa David Gidigasu1 Comptran Engineering and Planning Associates, Accra, Ghana, Formerly: Director, Building and Road Research Institute (BRRI/CSIR) Kumasi-Ghana

Abstract. The principle of ideal grading, low plasticity and higher compactive effort producing higher density and higher bearing strength of the compacted material for satisfactory pavement performance has characterized pavement materials acceptance specification requirements and practices related to the temperate zone countries. Investigations of cases of premature distress and deteriorations of pavements in some tropical environments have revealed that in addition to selecting well-graded gravels and aggregates to produce high compaction densities and bearing strengths for design, serious attention should also be given to the influence of the nature, geo-chemical, chemical and mineralogical compositions of the materials, testing and geomechanical rating procedures, construction techniques, as well as pavement maintenance history and environmental conditions. For tropically weathered materials formed in diverse climatic and drainage conditions, there is the need for materials oriented approach that integrates relevant aspects of such fields as engineering geology, geomorphology, geochemistry, petrography, pedology, climatology, rock and soil mechanics, innovative roadway design and construction methods as well as costeffective roadway management and maintenance strategies, etc. A key component of this approach would be the construction and instrumentation of road test sections in relevant climatic, geologic, soils and drainage conditions for long-term serviceability and structural integrity assessment and evaluation. The objective of this lecture is to highlight the key factors, characteristics and parameters useful for developing materials oriented paving materials acceptance specifications for lateritic and saprolitic soils. Keywords. Geomechanical rating, pedology, acceptance specification, lateritic soils, saprolitic soils.

1. Introduction Highway geomechanical engineering and roadway construction in the tropics are very important as many countries are expanding their road networks to improve communication and infrastructural developments. As part of this development, roads are built to a wide range of standards from simple earth roads to provide rural access to all-weather gravel roads and to paved roads usually with bituminous surfacings which are designed to carry heavier traffic. 1

Corresponding Author.

M.D. Gidigasu / Towards Developing Paving Materials Acceptance Speciﬁcations

11

The design standards of the roadways need to be appropriate to the type of road that is being built so that total transportation costs can be minimized. One way of helping to achieve this is to ensure that best use is made of the locally occurring materials and aggregates that are available. The development of specifications for temperate zone paving materials has been the results of tedious and long-term laboratory and field studies. The process has been a combination of theoretical and practical definition of optimum grading characteristics of materials that would yield the highest compaction density (e.g. Fuller and Thompson, 1907; Zemour and Durrier, 1966) and the relation between the fines, gravel contents and plasticity and the desired grading curves (e.g. Dunn, 1966). The strength and breakage behaviour of aggregates during construction and under traffic loads have also been extensively studied both in the laboratory and during pavement construction and in-service (e.g. Shelburne, 1939, 1941; Shergold, 1948; Shergold and Hosking, 1963; Melville, 1948; Dunn 1966; Day, 1962; Farrah and Thenoz, 1960). The effects of geological, petrographical, physical, chemical and mineralogical factors on the laboratory and field test data for paving gravels and aggregates have also received serious studies (e.g. Hartley, 1974; Lee and Kennedy, 1975; Reed, 1967; Scott, 1955; Wylde, 1975, 1976). The results of these and other investigations have resulted in the formulation of useful specifications for paving gravels and aggregate materials in different temperate zone countries (e.g. Zemour and Durrier, 1966). The development of paving materials specifications for tropical materials has not resulted from systematic European and North American methodologies of long-term laboratory and field construction and in-service performance studies. In fact, the temperate zone paving materials specifications have in some cases been transferred to tropical environments without local assessment for application in varied tropical and sub-tropical climate conditions. The use of non-traditional tropical lateritic and saprolitic materials in pavement construction has posed many problems. Some light has been thrown on the difficulties involved in utilizing other equally abundant and unpredictable tropical and residual materials. For example, collapsing residual and transported materials constitute problem paving materials in different parts of the tropics (e.g. Knight and Delhen, 1963). Similarly, the salt bearing soils are problem road materials in the Mediterranean areas and extensive studies on these materials have resulted in developing some useful guidelines relating to their utilization (e.g. Fookes and French, 1977). Failure resulting from the use of natural aggregates containing soluble salts has also been reported by Blight (1976). Pavement performance on expansive soils has also been a source of concern in many parts of the tropics. For example, cases of heave of pavements on these soils have been extensively reported (e.g. Williams, 1965). As regards lateritic and saprolitic paving materials a lot of published information is available scattered in various sources. Attempts have been made to summarize relevant information relating to developments in road way construction practices using some of these problem materials in the tropics (e.g. ISSMFE, 1982-1985). It has been shown (e.g. Little, 1969; Lohnes et al., 1971, 1976; Gidigasu, 1974, 1976; Brand and Philipson, 1985; ISSMFE 1982, 1985, 1988) that lateritic and saprolitic materials constitute a chain of materials ranging from decomposing rocks to lateritic (pedogenic) rocks. These materials differ from one another in many respects; compositional (physically, chemically, structurally and mineralogically) and useful methods of testing and evaluating each group or grade of these materials for construction have been shown to be different in many respects (e.g. Gidigasu, 1976).

12

M.D. Gidigasu / Towards Developing Paving Materials Acceptance Speciﬁcations

The purpose of this lecture generally stems from the recognition that a need exists to build knowledge of problem and unstable tropically weathered and residual soils relative to highway geomechanical practice in the tropics. The lecture attempts to highlight elements of good (acceptable) and poor (unacceptable) aggregates, gravels and soil specification practices It is hoped that this and other contributions will engender a renewed appreciation of the importance of soil science (pedology), geology and mineralogy to understanding the engineering behaviour of major soils, tropical and non-tropical” (Clemente, 1981).

2. PROBLEMS OF DEVELOPING SPECIFICATIONS FOR TROPICAL PAVING GRAVELS AND AGGREGATES Generally, many local gravels abound in the tropics. The characteristics of some lateritic gravels and stones are discussed elsewhere (e.g. Hammond, 1970).. The degree of desiccation, clay mineralogy and the cementing effects of salts (Al2O3 or Fe2O3) have significant influence on the grading, Atterberg limits and strength of some lateritised soils (Lohnes et al., 1971, 1976). For detailed discussion on this subject one could refer to other sources (e.g. De Graft-Johnson, Bhatia and Hammond, 1972; De Graft-Johnson, Bhatia and Gidigasu, 1969; Gidigasu, 1976; Millard, 1962; Nanda and Krishnamachari, 1952; Philip, 1952). Careful choice of pretesting preparation of samples and testing procedures are required to obtain reproducible results.. Most of the standard aggregate tests are applicable to most tropical decomposing rocks, soft aggregates, lateritic gravels, and crushed lateritic stones (e.g. De Graft-Johnson et al., 1972; Gidigasu, 1976). There are, however, cases where these tests are unable to provide good prediction of their behaviour in pavements. Sometimes, climatic conditions and rapid rate of chemical weathering of pavements negate the usefulness of these tests. Consequently, attempts have been made to evolve new and non-traditional test procedures which are more predictive of their in-service behaviour. For example, the so-called modified aggregate impact test, ten percent fines test, drying and wetting test, acidity soundness tests have been found very useful (e.g. Tubey and Beaven, 1966; De Graft-Johnson et al., 1972; Hosking and Tubey, 1969; Netterberg, 1971). The most significant contribution to the study of doubtful tropical and sub-tropical aggregates have been made in Australia by Wylde (1975, 1976), and in South Africa by Weinert (1961, 1964, 1965, 1968, 1980; Weinert and Clauss, 1962, 1967). Other areas of significant contributions have been shrinkage, specific surface tests (e.g. Roper, 1950), Methylene absorption test (e.g. Davidson, 1972). Washington degradation test (e.g. Davidson, 1972) and secondary mineralogical studies (e.g. Weinert, 1964, 1980; Scott, 1955) as well as petrographical and mineralogical tests (e.g. Wylde, 1976). Typical results of factors affecting the compaction results are also reported elsewhere (e.g. Gidigasu, 1976; De Graft-Johnson et al., 1972). The genetic variability and influence of compositional factors have also been shown to influence correlations between properties for some soil deposits and no correlations for similar deposits (e.g. Gidigasu and Bhatia, 1971). Climatic conditions of the formation of the soils have also been found to influence correlations between index and significant highway geotechnical properties (e.g. Gidigasu and Mate-Korley, 1984). Consequently, it is appropriate to emphasize the need to introduce climatic indicators in evaluating paving materials in the tropics.

M.D. Gidigasu / Towards Developing Paving Materials Acceptance Speciﬁcations

13

2.1. Importance and Limitations of Ideal Grading Specifications for Paving Aggregates and Gravels Grain size distribution is a key property of aggregates. It affects the stability and durability of bituminous concrete, as well as the stability and drainage of pavement layers. Aggregates may be dense, well-graded, uniform, open, gap or skip graded. The densest aggregate gradation provides the greatest durability by minimizing air voids, but sufficient room will not be available for traffic compaction, and the asphalt cement may flow to accumulate at the surface of the mix, a phenomenon known as ‘bleeding’. Of the many methods of expressing size distribution, the most important one relates to Equation 1 where d represents the sieve size in question, P is the percent finer than the sieve, D is the maximum size of the aggregate, and n is a coefficient which adjust the curve in a finer or coarser position: n ⎛d⎞ Equation 1 P = 100 ⎜ ⎟ ⎝D⎠

Studies by Fuller and Thompson (1907) have indicated that a maximum density may be achieved for an aggregate when n = 0.5. The “Fuller’s Curve” is only an approximation of maximum density, since actual gradation required for maximum density depends partly on the nature of the materials. However, it is a remarkably useful point of reference for designing aggregate blends for maximum density. Control of gradation to yield the type of base sought, whether it be densely graded for maximum stability or open graded for maximum drainage is of particular importance. Relating to this control is the hardness of the aggregate, since soft or weak aggregates may undergo degradation, a process whereby fines are generated by aggregate breakdown during placement and use. The aggregate property most important to base is gradation, including per cent fines or ‘binder’. Theoretically, for a maximum stability, a base course aggregate should have sufficient fines to just fill the voids among aggregate particles, with the entire gradation representing a very dense mixture resembling that of Fuller’s maximum density curve. The extent to which fines may increase or reduce stability are discussed elsewhere (Yoder and Woods, 1946; Dunn, 1966). The fines content of base-course aggregate may be considerably influenced by changes in aggregate gradation caused by physical and chemical action during storage, transportation, construction, and in service (Wylde, 1976). Most paving material specifications are based on the Fuller gradation curve. A critical evaluation of the formula in a more generalized form in relation to the performance of gravel in roads in some tropical environments (Fossberg, 1963) revealed that usually where “n” is less than 0.25 the fines content is excessive and the gravel often lacks stability, particularly in the wet weather conditions. Where “n” is greater than 0.5, the gravel tends to be stony and porous and usually requires additional soil binder for satisfactory behaviour, particularly in dry weather conditions. Apparently, the desired grading envelope for a particular climatic condition has to be determined in the light of local experience and local pavement performance records. Adequate cohesion of pavement materials is achieved by also specifying the plasticity index, a parameter which is roughly proportional to the amount of fines in the material.

14

M.D. Gidigasu / Towards Developing Paving Materials Acceptance Speciﬁcations

3. ELEMENTS OF STANDARD SPECIFICATION REQUIREMENT FOR SUBBASE AND BASE MATERIALS 3.1. The Implications of Ideal Grading Requirements The most stable soils and aggregates in the pavement structure are those possessing high degree of mechanical interlock together with good cohesion. Good interlocking is obtained when the larger particles are angular with rough surfaces and cohesion is dependent on the fines and clay size content. To achieve maximum stability of road pavements attempts have been made to select materials that satisfy these requirements. The grading limits adopted by some Highway Authorities in Standard specifications in Europe and North America for paving aggregates and gravels approximate the Fuller and Thompson (1907) formula (i.e. ASTM, 1964; AASHO, 1966). Similar grading specifications have been proposed by the British Road Research Laboratory (1952) on the basis of theoretical considerations and the Fuller-Thompson curves (Fig. 1).

Fig. 1

Ideal grading envelope for selection of paving material

The combinations of the ASTM, AASHO and British Standard specifications are used in many temperate as well as tropical countries for selecting pavement construction materials. It is usual to limit the maximum size in order that the material can be laid by machine and, for the top layers, to give a smooth finish suitable for traffic or for sealing. It is also usual to require that the particles be approximately cubical for good packing. Elongated or round particles are not easy to compact into a dense mass and long, thin particles may fracture during placing and compaction altering the grading, usually detrimentally. Fines content is rather easier to control; with the much fine material, interlock between the larger particles is prevented and shear strength much reduced; with too little fine material, the material will be harsh to work, difficult to compact (and the resulting loss of density will reduce strength) permeable to moisture and likely to have a coarse open surface. The risk of segregation during construction is also much increased. The most relevant property of the fines is essentially that they should not be susceptible to the action of water, that is, they should not swell or shrink to excess with change in water content. The limitation of this susceptibility are usually by means of a restriction on the nature of the clay content of the fines, the presence of highly plastic fines being undesirable. It is common, therefore, to place restrictions on the Atterberg limits of the fines (the material smaller than 0.425mm).

M.D. Gidigasu / Towards Developing Paving Materials Acceptance Speciﬁcations

Fig. 2

15

General properties of mechanically stable gradings (from Ingles and Metcalf, 1972)

These limits must be treated with caution because as Morgan, (1972) noted, the test is carried out on only the fine portion of the material which is often less than 20 per cent of the mass and of which the plastic fines (clay) content might be one-quarter, or 5 to 10 per cent of the total. Morgan showed that the compressive strength of a crushed rock was insensitive to plasticity, however, the CBR tended to decrease as plasticity index increased for samples at optimum moisture content and laboratory maximum density. But if a material with a high plasticity index is kept dry it has a high crushing strength and it is possible to use such materials in well-drained and dry environments.

4. REVIEW OF STANDARD PAVING MATERIALS ACCEPTANCE SPECIFICATION REQUIREMENTS 4.1. General Wooltorton (1954, 1968) who has been associated with paving materials specification development, pavement design, and construction quality control in many climatic areas of the world including the United States, United Kingdom, Africa, Asia and Australia has emphasized that the definition of plasticity index (or potential swell) should in theory be modified to suit specific climatic and drainage conditions. Wooltorton explained that the upper and lower moisture content limits within which potential swell would take place should be the maximum and minimum moisture contents likely to be found under a given climatic and drainage condition. 4.2. Plasticity and Shrinkage Properties Requirements In a theoretical explanation of existing specifications, Wooltorton (1954) suggested that for no overall swelling of a coarse granular system, the product of the plasticity index and the fines content should not be greater than the volume of voids between granular aggregates to accommodate swelling. On this basis, he established the following relationship: X.Ip Va Equation 2 p 100

γd

16

M.D. Gidigasu / Towards Developing Paving Materials Acceptance Speciﬁcations

Where: X=percentage of fines in 100gm total mix;, Ip=plasticity index %; Va=% of entrapped air between fines and coarse material (air voids); γd=apparent dry density of compacted mixture (gm/cc); (X).Ip=“binder plasticity index product”. For example, one state in Australia specifies the maximum value of the binder plasticity product of 200 for base course materials and 360 for sub-base course materials (Frost, 1967). A similar approach was reported in the determination of maximum permissible value of liquid limits. The most important assumption above is that of the definition of plasticity index (or potential swell) and Wooltorton suggested that this should, in theory, be modified to suit local conditions. For example, the upper and lower moisture content limits within which the potential swell may take place would be the maximum and minimum moisture contents characteristic of given materials, as well as climatic, physical and drainage conditions. This means that considering the four significant moisture content phases in soil-air-water system of the liquid limit (W L, plastic limit (Wp), field moisture equivalent (FME), and shrinkage limit (Ws), the plasticity index Ip (or potential swell) may be defined as (WL – Wp) or (FME – Ws) depending upon the site conditions. For example, in a temperate zone condition with no appreciable cementation and with possibility of frost action, the maximum moisture content would be the liquid limit and the minimum moisture content, the plastic limit which gives the well-known definition for plasticity index as liquid limit minus plastic limit. Frost (1967) emphasized that there are many natural soils which would appear to be troublesome on the normal basis for determining the Ip , (WL – Wp) but which in fact make excellent road sub-bases. For example, he noted that the desiccated soils of Burma have a Ip of 48 on the basis of (WL – Wp) but only 10 on the basis of (FME Ws) in which case the latter value of Ip more closely represented the true plasticity index. The importance of soil fines in evaluating the strength and durability properties of pavement construction materials are also illustrated by the inter-relationships between the maximum dry density and optimum moisture content, triaxial shear strength and the California Bearing Ratio on the one hand, and the fines content and plasticity index on the other (Figs. 3/4).

Fig. 3a Effect of fines on Compaction (from Yoder and Witczak, 1975)

Fig. 3b Effect of fine content on triaxial strength of a gravel (maximum aggregate size is 25mm) (from Yoder and Witczak, 1975)

M.D. Gidigasu / Towards Developing Paving Materials Acceptance Speciﬁcations

17

Fig. 4 Effect on soaked CBR of the plasticity of the 20% passing No. 36 sieve contained in crushed basalt aggregate compacted at modified AASHO, OMC (from Dunn, 1966)

It is noted that the higher the fines content the lower the strength and bearing properties. Similarly, the plasticity index significantly influences the bearing strength of compacted soil mass, apparently, here lies the need to control the fines content and their plasticity index in paving materials acceptance specifications. The product of the fines content and the plasticity index has also been known to affect the compaction density, strength and the compressibility ratio (Fig. 5). Field experimental evidence of the influence of fines on the suitability of aggregate bases has also been investigated and is illustrated in Fig. 6.

Experimental evidence of the Fig. 5 (a) Effect of fines on density achieved during compaction Fig. 6 on test track, relative to standard MDD obtained by vibration. influence of fines on the suitability of (b) Relationship between compressibility ratio and product of % aggregates for base (from Dunn, 1966) minus No. 40 U.S.sieve and PI illustrating that fines tend to reduce air voids (from Dunn, 1966)

18

M.D. Gidigasu / Towards Developing Paving Materials Acceptance Speciﬁcations

Consequently, most current paving materials specifications in different countries have stated maximum limits for the fines content (Passing No. 200 sieve size), the liquid limit and the plasticity index (Table 1). Table 1. Typical Temperate zone acceptance specifications for surfacing, base and sub-base course gravels (from Zeymour and Durrier, 1966) U.S.A.

Country United Kingdom

Germany

4
4
4
WL<35 <50 25

WL<35 <65 25

AASHO specifications 5-25

WL<35 <40 0.5-0.33 of layer thickness Empirical grading envelopes 10-15

Ip<6

Ip<6

Ip<6

WL<25 <50 <76.8 AASHO specifications 2-25

WL<25 <40 <76.8 Empirical grading envelopes 0-10

WL<25 <67 <76.8 Mathematical grading envelopes 0-8

Ip<6 WL25 <50 <76.8 AASHO specifications % passing sieve No.200 5-25 **L.A.A.V. = Los Angeles Abrasion Value (%)

Ip<6 WL25 <40 <76.8 Empirical grading envelopes 0-10

Ip<6 WL25 <67 <50.8 Mathematical grading envelopes 0-8

Properties of materials for different layers Sufacing: Plasticity of fraction passing BS sieve No. 36 (US Sieve No.40) Aggregate Strength (L.A.A.V.**) Max. size of course fraction (%) Particle size distribution % passing sieve No.200 Base Course: Plasticity of fraction passing BS sieve NO. 36 (US Sieve No.40) Aggregate Strength (L.A.A.V.) Max. size of course fraction(%) Particle size distribution % passing sieve No.200 Sub-base Course: Plasticity of fraction passing BS No. 36 (US Sieve No.40) Aggregate Strength (L.A.A.V.) Max. size of course fraction(%) Particle size distribution

Mathematical grading envelopes 10-21

4.3. Compaction Requirements of the Soil Mass The various road departments have specified levels of laboratory and field compaction for materials to be used in various pavement layers to ensure long-term stability of the pavement. The compacted materials when tested at stipulated moisture content and densities are supposed to attain certain durability and strength values. The level of laboratory compaction requirements stipulated range from Standard Proctor level to Modified AASHO level with other intermediate levels. Many countries have stipulated procedures for sample preparation for laboratory strength test at the optimum moisture content of a given compactive effort while some countries define the equilibrium moisture content at which the compaction is carried out in the laboratory (e.g. Tanner, 1963). Some West African countries including Ghana have adopted Standard CBR method of soil strength evaluation using 24 to 96 hours of soaking before carrying out the test. Literature survey on the subject (Gidigasu, 1988; 1991) revealed that the periods of pretest soaking normally vary between 24 hours and 96 hours; however, the soaking periods have not been based upon any laboratory and field experimental studies. Insitu CBR testing of base, sub-base and sub-

M.D. Gidigasu / Towards Developing Paving Materials Acceptance Speciﬁcations

19

grade courses at optimum moisture content or local insitu equilibrium moisture content of the project sites has been suggested as alternative solution (e.g. Gidigasu, 1980). Indeed, most of the temperate zone material specifications have not been modified in relation to specific local pavement performance data and experiences. Because there is limited knowledge of geotechnical characteristics of many unusual tropical materials and their performance in pavements for most tropical environments, some tropical paving materials specifications tend to reflect those of temperate zone countries with which local engineers are familiar. However, because differences in soil genesis, nature of the materials, climate and drainage conditions produce different lateritic materials and pavement construction and quality control difficulties, specifications have to be tailored for specific materials occurring in specific environments to meet local road construction challenges. There is a real need to develop materials oriented paving materials acceptance specifications requirements, and roadway construction methods which take cognizance of the unique genetic and geotechnical characteristics of the material (i.e. gravels and aggregates) and the construction equipment available for specific climatic and drainage conditions

5. ELEMENTS OF SPECIFICATION REQUIREMENTS FOR NONSTANDARD AGGREGATES Materials that do not accord with one or more of the temperate zone requirements for a first-class base material are non-standard (Wylde, 1979). Temperate zone current standard requirements were developed by an ad-hoc process of excluding materials to which have been attributed some inadequacy in performance in the pavement or some difficulty during construction. Thus, a ‘standard’ material is one which has conservative properties of the major performance (or, rather classification) parameters. It will be tolerant of construction mishandling and environmental conditions, and probably, will perform well in most instances. It is also contended that almost any earthen material can be used for pavement construction, provided the appropriate design, construction and maintenance procedures are applied and the resulting performance assessed with proper regard for overall economy. 5.1. Durability and Strength Specifications for Concretionary Lateritic Aggregates and Gravels A critical parameter for evaluating laterite gravels for road construction is the durability of the coarse particles (Bhatia and Hammond, 1970). Other significant properties of the coarse particles are the chemical composition, specific gravity, and water absorption (Ackroyd 1967; USAID/BRRI, 1971; De Graft-Johnson et al., 1972). Concretionary lateritic boulders may be used in pavement construction as long as they are sufficiently durable. For example, the use of lateritic rock pieces as road base is shown in Fig. 7. Studies have also shown (Bhatia and Hammond, 1970) that in addition to the aggregate tests, the pH and heat treatment tests could be used for assessing probable performance of lateritic rock aggregates and pisoliths in road pavements. Experience in the use of concretionary lateritic gravels for pavement construction has also shown (Ackroyd, 1985, 1967) however, that the durability is very variable.

20

Fig. 7

M.D. Gidigasu / Towards Developing Paving Materials Acceptance Speciﬁcations

The use of lateritic crushed stones for road base construction (from Persons, 1970)

For example, some of the materials do breakdown during field compaction (Arulanandan, 1969) rapidly losing strength on wetting and most of such aggregates are unacceptable as base material. Attempts have been made to define durability criteria for selecting lateritic gravel sizes for pavement construction. Criteria based upon modified Aggregate Impact value was also suggested for selecting lateritic gravels for pavement construction (De Graft-Johnson et al., 1972) (Table 2). Millard (1962) and Ackroyd (1967) have reported that the durability of lateritic concretions and their probable performances in pavements depends on the content of sesquioxides, especially, on the iron oxide content (Table 3). Clearly, the rating system that is based upon the durability and weathering characteristics are useful for distinguishing good, critical and poor concretionary aggregates for pavement construction Table 2. Recommended criteria for rating West African lateritic rock aggregates and pisoliths for pavement construction (from De-Graft Johnson et al., 1972) Specific gravity

Water absorption after 24 hours soaking (%)

Aggregate Impact Value (%)

Los Angeles abrasion value (%)

>2.85 2.85-2.75 2.75-2.58 <2.58

<4 4-6 6-8 >8

<30 30-40 40-50 >50

<40 40-50 50-60 >60

Rating based on probable in-service performance excellent good fair poor

Table 3. Relation between the chemical composition of Nigerian pisoliths and probable performance in road pavements (from Ackroyd, 1967) Chemical Composition SiO2 Fe2O3 8.4 65.1 15.5 53.9 19.1 50 39.3 30.4 48.1 25.4 38.1 28.5 41.2 28.6 40.8 5.8

Al2O3 17.4 19.7 27.3 20.2 16 20 16.8 41.6

Nature of the Material

Probable Performance Rating

Hard concretionary gravels

Good base and surfacing materials

Weak concretionary gravels Nodules in clay matrix

Sub-base materials and fills Groundwater laterites: sub-base and sub-grade materials

5.2. Linear Shrinkage Specification Requirements Ackroyd and Rhodes (1960) and Easterbrook (1961) have suggested the use of linear shrinkage test for the selection of lateritic soils for unstabilized road base and sub-base

M.D. Gidigasu / Towards Developing Paving Materials Acceptance Speciﬁcations

21

construction. On the basis of studies on over 800 lateritic soils in Nigeria, upper linear shrinkage limits of 6% and 7% were suggested respectively for accepting sand-clays and gravel-sand-clays for base courses. For the sub-base course material the respective values are 12% and 13% for sand-clays and gravel-sand-clays. For all types of lateritic gravels for use for unstabilized base-course, maximum linear shrinkage of 5% is not to be exceeded for Nigerian environment. Nigerian Ministry of Transport specified maximum linear shrinkage of 4 to 5% as corresponding to the maximum liquid limit of 25% and plasticity index of 9%. The Zambian Public Works Department also specified a maximum linear shrinkage of 3.3% for lateritic gravels for road-base construction (Newill, 1961).O’Reilly and Millard (1969) and Dreyfus (1962) have recommended linear shrinkage limits together with the liquid limit and plasticity index, etc. for selecting materials for specific climatic conditions (Table 4). Table 4. 1952)

Rating of potential laterite base materials performance under bituminous surfacing (See Dreyfus, Field Performance Rating

Soil Properties

Excellent

Average

Poor

Linear Shrinkage (%)

0-4

4-6

above 6

Plasticity Index (%)

0-6

6-8

above 12

Liquid Limit (%)

14-21

22-30

above 30

Swell in CBR mould after saturation (%)

0-0.2

0.3-0.4

above 0.4

Optimum Moisture Content (%)

-

8-10

-

5.3. Plasticity Modulus as a Specification Requirement Factor An indication of the importance of plasticity modulus as a factor influencing the stability of aggregates was given by Dunn (1966) (see Fig. 6). The plasticity modulus has been variously defined as the product of fines (i.e. passing 0.425mm, 0.075mm sieve sizes, etc.) and such plasticity parameters as the liquid limit, plastic limit, plasticity index, as well as linear shrinkage and optimum moisture content. This parameter has been used for materials selection and in acceptance specifications (e.g. Townsend et al., 1982; Cocks and Hamory, 1988; Bhatia and Yeboa, 1970; USAID/BRRI, 1971; De Graft-Johnson et al., 1972). Typical specifications involving the use of plasticity modulus are summarized in Table 5 Table 5. Recommended Criteria for selection of base course materials (from Townsend et al., 1982) Criteria

Road Classification

Current FMW* specifications

Class I

Class II

Class III

(Heavy Traffic)

(Meduim Traffic)

(Low Traffic)

Design CBR

80 or 100 (Min)

60 or 70 (Min)

50 (Min)

80 (Min) 30 (Sub-base)

Liquid Limit (LL)

35 or 45 (Max)

40 (Max)

-

25 (Max) 45(Sub-base)

(LL)x (% passing 0.075mm)

600 (Max)

900 (Max)

1250 (Max)

125-375 (for Gravel base) 500-625 ( sandy clay base)

Plasticity Index (PI)

10 or15 (Max)

12(Max)

-

20 (Max) 25(Sub-base)

(PI)x(%passing 350 (Min) 350-400 500 (Max) 0.075mm) *Federal Ministry of Works, Nigeria (See Teme et al., 1987)

100-300(for Gravel base) 400-500 ( sandy clay base)

22

M.D. Gidigasu / Towards Developing Paving Materials Acceptance Speciﬁcations

6. SOME PROBLEMS OF TESTING AND RATING OF TROPICAL PAVING AGGREGATES AND SOILS 6.1. General The contributions by Hamrol (1961), Duncan (1970), Clauss (1963) and Weinert (1968) to the subject of evaluating weathering rocks and residual soils for engineering purposes have emphasized the need to undertake detailed studies aimed at identifying significant parameters for evaluating and rating various grades of tropically weathered soils for engineering purposes. For example, the so-called saturation moisture content (Duncan, 1969) and secondary mineral content (Scott, 1955) criteria are key parameters for rating and predicting the engineering behaviour of decomposed rocks and tropically weathered soils in the roadway. As regards the fine-grained soils, the difficulties associated with obtaining consistent particle size and Atterberg limit test results appear to be the main problem. For example, the effect of pretest drying, type of dispersing agent and time of stirring on the laboratory determined compositional and index properties for hydrated and volcanic ash soils has been widely discussed (Townsend et al., 1971; Terzaghi, 1958). It has been found that most tropical soils are amenable to satisfactory cement, lime and chemical stabilization and considerable strength gains have been recorded for typical tropical soils (e.g. Ingles and Metcalf, 1972). However, there are limited studies related to the effect of pretest preparations and testing procedures on the strength gains or strength losses for these soils. For example, significant effect of lapse of time between mixing and compaction on the strength loss was reported for some West African lateritic gravels (Gidigasu and Amankwa, 1975). This would suggest that if cement and lime stabilization of lateritic soils for road construction is to prove useful then intensive studies would be required to establish their usefulness and limitations. There is the real need to evaluate fully the laboratory and field engineering behaviour of cement and lime stabilized lateritic soils both in the laboratory and in the field. 6.2. Problems of Laboratory and Field Compaction and Quality Control Testing Review of pavement engineering practice in some 30 tropical countries (Tanner, 1963) revealed that pavement design based upon the CBR method has been established as most applicable to tropical soils and climatic environments. For example, information available indicate that provided realistic testing conditions are selected, the CBR procedure provides a reasonable basis for estimating pavement thickness. However, to ensure long-term pavement stability, the moisture content of the sub-grade should preferably represent the “stable” moisture condition likely to prevail during the design life of the pavement. Consequently, it is necessary, to define this “equilibrium” moisture condition for most project sites at which to determine the strength of the subgrade in the laboratory as well as during field quality control testing. In some tropical climates the equilibrium sub-grade moisture content under sealed pavements is noted to rarely exceed the plastic limit of the soil and also the standard Proctor compaction optimum moisture content may reasonably represent the “equilibrium” sub-grade moisture content (e.g. O’Reilly and Baker, 1963). However, in some climates high subgrade moisture contents frequently approaching full saturation do occur. For such conditions, we need the long-term mean insitu moisture content on which laboratory

M.D. Gidigasu / Towards Developing Paving Materials Acceptance Speciﬁcations

23

and field compaction and strength control tests would be carried out to ensure longterm stability of the pavement structure. 6.3. Insitu Moisture Content Vrs. Optimum Moisture Content The relation between the insitu moisture content and the Modified AASHO compaction optimum moisture content is illustrated for a group of lateritic gravels from the moist sub-humid zone in Ghana (Fig. 8).

Fig. 8 Relation between the INSITU and Modified Fig. 9 The relation between the optimum moisture ASSHO Compaction Optimum Moisture Contents content and the moisture content corresponding to (from Ghana Highway Authority, 1970) maximum CBR value (from Gidigasu, 1980)

It is noted that for the given lateritic gravels (from the moist sub-humid zone), it would be unrealistic to adopt the Modified AASHO compaction optimum moisture contentrelated placement moisture content. This is because soils with insitu moisture contents wet of optimum moisture contents would tend to absorb more water to attain the “equilibrium” moisture content; this could lead to reduced strength of the pavement structure. Similarly, soils with natural moisture contents generally wet of optimum moisture content of an adopted compactive energy would present construction problem (Gidigasu, 1980a). For example, in the dry sub-humid climatic zone, the insitu moisture content of the gravels is lower than the optimum; in such a case, the strength and stability of the pavement is not likely to undergo significant deterioration since there would not be any additional water absorption to cause strength loss. 6.4. Effect of Compaction Moisture Content on Stability of Lateritic Gravels The danger of specifying the optimum moisture content for pavement placement has been noted for humid environments (e.g. Gidigasu, 1980). This is because the CBR at the optimum moisture content may sometimes be as low as 30% of the peak values obtainable at moisture content dry of optimum. Typical inter-relationships between the moulding dry density and moisture content on the one hand and stability or bearing strength (CBR) for a lateritic gravel on the other have been found by Hammond (1970). It has been observed that at low moisture content, an increase in density improves the stability of the soil; however, at moisture contents of say 10% and above the stability

24

M.D. Gidigasu / Towards Developing Paving Materials Acceptance Speciﬁcations

increases with density only up to a certain point and then further increases in density produces a decrease in stability. Indeed, at moisture content of about 16% or higher, the stability decreased with any density above 1752kg/m3 (110 lb/ft3) (Gidigasu, 1991). The relation between the optimum moisture content and moisture contents corresponding to the maximum CBR for some fine-grained soils is given in Fig. 9. It is noted that the moisture content at which the highest CBR is obtained is dry of the optimum moisture content. The adverse effects of moulding moisture contents on the stability of compacted micaceous sandy loamy soils have also been noted elsewhere (e.g. Gidigasu and Mate-Korley, 1980). For example, it was noted that, for samples compacted at optimum moisture content and at moisture contents dry of optimum the stability is quite high. However, the same soil compacted at moisture content wet of optimum gives very low stability. This phenomenon has been attributed to the over compaction (De Graft-Johnson et al., 1967) involving excessive destruction of the natural soil structure accompanied by considerable loss of strength through mobilization of high pore water pressure at high moisture content, involving high compactive effort. Foster (1955) discussed the reduction in stability with increase in moulding density for fine-grained soils with degrees of saturation generally above that represented by the line joining optimums for a series of moisture density relations. These observations illustrate the probability of loss of stability which may accompany over-compaction to too high a density, as well as the danger of using too high water content if high strength is required. 6.5. Reproducibility of Laboratory Test Results During Roadway Construction A major problem relates to the reproducibility of laboratory test results under field condition during construction. For example, Fig. 10 illustrates the difference between the laboratory and field compaction characteristics. It is also to be noted that laboratory compaction produces more coarse particle breakages than field compaction using, say, 10-12 ton vibratory rollers (Fig. 11).

Fig. 10a Relation between laboratory and field Fig. 10b Relation between field control and compaction curves for lateritic gravels (from laboratory dry densities for lateritic gravels (from Gidigasu, 1991) Gidigasu, 1991)

M.D. Gidigasu / Towards Developing Paving Materials Acceptance Speciﬁcations

25

Fig. 10c

Field control density related to field optimum moisture content (from Gidigasu, 1980)

Fig. 11

Effect of laboratory and field compaction on degree of breakages of lateritic gravels (from Gidigasu, 1991)

6.6. Some Aspects of Tropical Roadway Geotechnical Practice The results of laboratory soaked CBR tests on base and sub-base materials taken both from gravel and bituminous surface dressed road sections at over 2,000 sites in Ghana were analysed at the Ghana Highway Authority (GHA, 1972) and the results are summarized in Fig 12.

Fig. 12 Frequency distribution of pavement thickness in Ghanaian roads (from Ghana Highway Authority, 1972)

Fig. 13 Performance of tropical road pavements in relation to age (from Tanner, 1963)

26

M.D. Gidigasu / Towards Developing Paving Materials Acceptance Speciﬁcations

The test results from the bituminous surface dressed roads were obtained from a combination of roads where failures have already occurred together with roads which are only in the initial stages of deterioration. The results of the insitu CBR studies and examination of the sources from which these materials have been excavated have led to the conclusion that the quality of the existing base materials in failed sections of the bituminous surface dressed and in gravel roads are generally too low for consideration as satisfactory road base material by temperate zone standards. It was also noted that desiccating environments sometimes promote improvement of compacted lateritic soil in terms of CBR at some pavement sections to prevent premature failure of pavements built with sub-standard or inferior materials. It was also found that premature failure of some pavements was caused not by the use of sub-standard materials or misuse of the CBR method of pavement design. For example, it was noted that sites with no failures are those where the actual construction thicknesses are higher than the design thickness based upon CBR method of design. Similarly, in areas of under-design, moderate and severe failures occurred. Studies of failed pavement sections revealed that in fact, failure were due to the fact that instead of the proposed total pavement thickness of 28cm, most of the failure sections had thicknesses below 25cm (Fig. 12). Clearly, the failures were due to either under-design or poor construction, and this was probably true for both bituminous surface dressed and unpaved roads. The causes of failures of tropical low-cost pavements investigated in 30 tropical countries (Tanner, 1963) under low to medium traffic volume, have led to similar conclusions (e.g. Figs. 13/14).

Fig. 14

Performance of low cost tropical road pavements as a function of the relation between design and construction thickness (from Tanner, 1963)

6.7. Performance of Some Lateritic Aggregates and Gravels in the Roadway Figure 15 illustrates how the grading of lateritic gravels affect pavement performance in Central Africa (Remillon, 1955). As could be expected, it is noted that the wellgraded gravels perform satisfactorily while those with excessive gravel content corrugated and those with excessive silt and sand contents became slippery during the rainy season. Results of field studies in Ghana have shown interesting picture; these are shown in Fig. 16. It is noted that some materials that would have been rejected by traditional standard specifications may perform well provided the right testing procedure had been done and the drainage conditions of the roadway are good. On the other hand, some apparently well-graded gravels falling within standard grading

M.D. Gidigasu / Towards Developing Paving Materials Acceptance Speciﬁcations

27

envelopes have produced pavement failures under adverse climactic and/or poor drainage conditions.

Fig. 15

Performance of laterite gravels in pavements (from Remillon, 1955)

Fig. 16a Typical poorly graded material that have performed satisfactorily in well drained road sections (Data from Ghana Highway Authority, 1972

Fig. 16b Typical well-graded gravel that have failed in roads at poorly drained pavement sections (Data from Ghana Highway Authority, 1972

Considering the influence of climatic conditions, the depth of ground water and temperature variations may be critical factors that need serious attention. Seasonal temperature and moisture variations with depth may adversely influence pavement performance. Depending upon the relationships between the insitu placement and longterm stable moisture contents coupled with density variations, pavement foundation performance may be poor. For example, under poor drainage conditions the bearing strength of the pavement foundation may fall to failure condition within a matter of say, 2 years (Fig. 17); for such situations special design and construction precautions should be taken to ensure the safety and long-term stability of the pavement structure. Results of studies in some countries relating to the performance of crushed lateritic stone and other pedogenic rock aggregates in tropical pavements are given in Table 6. Climatic and drainage conditions appear to most influence the pavement performance pattern, and so also is the key factor of the materials constituents which are direct results of the mode of formation of the materials (e.g. USAID/BRRI, 1971).

28

M.D. Gidigasu / Towards Developing Paving Materials Acceptance Speciﬁcations

Fig. 17

The loss of bearing strength (CBR) of a sub-base in existing pavement (from Bhatia, 1968)

Table 6.

Performance of lateritic gravels, crushed lateritic stone, calcrete and coral in pavements

Locality/ Country

Parent Rock Types

Nature of Degradation / Pavement failure

Climate (Environment)

References

South Africa

Calcrete

Degradates during construction and in pavement

Semi-arid zone

Netterberg, 1971

Ghana

Crushed lateritic rock and lateritic gravels

Degradates during construction and under traffic load

Semi-arid environment

Gidigasu, 1975, 1979, 1980; Arulanandan, 1969

Central Africa

Lateritic gravels

Degradates and change in grading for et environment

Wet tropical environment

Remillion, 1955

Various roads in India

Crushed lateritic rock

Strength reduction under wet condition. Los Angeles Abrasion specification proposed

Wet tropical environment

Nanda and Krishnamachari, 1968

East, West & Central Africa

Lateritic gravels (pisoliths)

Polishes fast even under medium traffic

Wet tropical environment

Millard, 1962

Jamaica

Corals

Disintegrates and polishes under light to medium traffic load

?

Hosking and Tubey, 1969

7. SOME CONTRIBUTIONS TO DEVELOPING PAVING MATERIALS SPECIFICATIONS FOR TROPICAL CONDITIONS 7.1. Specifications Requirements for Crushed Stone Aggregates Field and laboratory studies have shown (e.g. Clauss, 1963; Weinert, 1968) that in some cases standard aggregates tests do not give information on the future mineralogical or textural alterations that occur in a road pavement aggregate. For example, the weathering tests (e.g. sodium or magnesium sulphate soundness test or simple alternate wetting and drying test) undoubtedly provide some information on certain physical properties of the material but this may be only part of the complex alteration a rock may undergo in a road foundation. Indeed, Weinert (1968) found that in certain cases the most careful performance of engineering tests and the strictest control during construction may not prevent failure prior to the expiration of the design life of a road pavement. For soft materials some of the aggregate tests are not very reliable. For example, Loubser (1967) observed that the very popular British Standard

M.D. Gidigasu / Towards Developing Paving Materials Acceptance Speciﬁcations

29

Aggregate Crushing Test (British Standard Institution, 1975) is relatively insensitive to differences in the strength of weak materials. A consideration of the test procedure suggests that this difficulty could result from the fact that the weaker materials are compacted to dense mats before the specified 40 tons has been applied, thus preventing any further crushing in the later stages of the test. The Los Angeles Abrasion Test may also give misleading results on the same basis as the Aggregate Crushing Test, because weak materials often form a uniform powdery mass before 500 revolutions have been completed. Loubser (1967) suggested, however, that the crushing value by the 10% fines crushing (British Standard Institution, 1975) provides a good method of evaluating soft aggregates such as shales for road aggregates. Weinert (1964) has proposed a criterion for accepting or rejecting decomposed rock aggregates based on amount of secondary minerals and a climatic factor N (Fig. 18), where E1 is the potential evaporation during the warmest month, Pa is the total annual precipitation, and thus N is a numerical expression for the balance between significant climatic factors (Weinert, 1961). E Equation 3 N = 12 1 Pa

Fig. 18

Rating of natural aggregates for pavement construction (from Weinert, 1980)

7.2. Specification Requirements for Natural Residual and Lateritic Aggregates and Gravels Specifications for many African countries in relation to maximum sizes obtained from results of studies elsewhere in terms of acceptable materials for base and sub-base materials are given in USAID/BRRI, 1971. In terms of the strength and durability of the coarse particles, considerable studies have also been undertaken to define acceptable limits for specific gravity, water absorption, aggregate impact value, Los Angeles Abrasion value as well as 10% fines content for both quartzitic and lateritic gravels (De Graft-Johnson et al., 1972). Specific ranges of values of these parameters were related to the performance of the materials as aggregates in pavement (see Table 2). Perhaps, the first attempt to establish a criterion for selecting paving gravels for surfacings in terms of values of the liquid limit, plasticity index and the linear shrinkage, for the moist temperate and wet tropical, seasonally wet tropical, and semiarid climatic conditions was made by O’Reilly and Millard (1969). This is perhaps the first indication that stringent temperate zone material acceptance specifications (where,

30

M.D. Gidigasu / Towards Developing Paving Materials Acceptance Speciﬁcations

for example, base gravel should have not more than 25% liquid limit and not more than 6% plasticity index irrespective of the climatic conditions) may be relaxed for the rather unpredictable soils and tropical and sub-tropical climatic environments (Table 7). Table 7.

Plasticity characteristics preferred for gravel surfacing (from O’Reilly and Millard, 1969)

Climate

Liquid limit not exceeding (%)

Plasticity index range (%)

Linear shrinkage range (%)

Moist temperate and wet tropical

35

4-9

2.5-5

Seasonally wet tropical

45

6-20

4-10

Semi-arid and arid

55

15-30

8-15

Studies were undertaken in Ghana to investigate the effect of the plasticity and fines content for formulating the local paving gravel specifications and to see how they relate to the bearing strength in terms of the CBR values. Relationships were tried between the CBR on the one hand and passing 63 μm sieve size, liquid limit, the plastic limit and the plasticity index on the other hand for many soil (gravel) systems. It was noted that there are no correlations between the CBR on the one hand, and the other parameters individually. However, some correlations were found for some lateritic gravels formed over phyllite in the moist sub-humid zone between the CBR on the one hand and the liquid limit x passing 63 μm sieve size, plastic limit x passing 63 μm, plasticity index x passing 63 μm, on the other. Based upon these correlations a tentative acceptance specification for gravels from the dry and moist sub-humid climatic zones were proposed (Tables 8/9). Using the allowable 10% passing 63μm BS size, it was also possible to separate good, border line, and poor paving gravel materials. As regards the maximum dry density values it was noted that materials with satisfactory performance have the maximum dry density (West Africa compaction) of 2.16 Mg/m 3 and above. Using the 80% soaked CBR criterion it was also possible to modify the existing imported British standard specification and redefined the maximum liquid limit to around 37% and the plasticity index to 10% for the gravels in the moist sub-humid zone. Studies on similar gravels from the semi-arid climatic zone led to the definition of a similar tentative specification for paving gravels for that climatic zone. It is to be noted, therefore, that specifications at the moment can only be formulated for particular materials and climatic environments. Apparently, regional studies along this line should be encouraged at the moment; generalizations may only be possible when more data from laboratory and field studies have been gathered and more co-operative research activities among Institutions of different countries have materialized Table 8.

Specifications for lateritic gravels in the dry sub-humid zone

Material

Passing No. 200 sieve

Liquid Limit (%)

Plastic Limit (%)

Maximum dry density (kg/m3)

Optimum moisture content (%)

Relative compaction (%)

Base

9±2

24±6

7±4

2130.5±4

8±2

101±1

Sub-base

10±4

32±7

13±4

2098±4

9±1

101±2

Sub-grade

36±23

34±9

15±7

2066.4±9

9±1

94±5

31

M.D. Gidigasu / Towards Developing Paving Materials Acceptance Speciﬁcations

Table 9. Acceptance Specifications for lateritic gravels from the moist sub-humid zone for base course (from Gidigasu, 1982) 1

Maximum passing No. 200 sieve

12.5%

2

Maximum liquid limit

37.5%

3

Maximum plasticity Index

10%

4

Maximum product of liquid limit and passing No 200 sieve size

300%

5

Maximum product of plasticity limit and passing No. 200 sieve size

200%

6

Maximum product of plasticity index and passing No. 200 sieve size

100%

7

Lowest west African compaction dry density

2162kg/m3

These have to be correlated with field performance data before they can be incorporated into acceptance specification. A survey of acceptance specification from various African countries (USAID/BRRI, 1971) revealed different values for the amount passing the 63 μm sieve size, the minimum liquid limit and plasticity index values (Table 10). The CBR values were generally 80% or above for base course gravels except for more arid countries such as Mali where, for example, 50% CBR value was stipulated. Based upon literature studies and field experiences gathered to date, it appears that the development of local highway geomechanics should be related not only to the available materials but also to the environment. The literature on specification requirements for tropically weathered rocks and soils is fairly extensive but is usually not specific. One of the best summaries on the subject is provided by Weinert (1980) and all of the information that follows is from this source. Table 10. Insitu CBR of materials in existing pavements in relation to surfacing type (from GHA, 1972) Percent of test results less than

Road Type

No of Test

40% CBR

60% CBR

Gravel roads

98

51

77

96

Bituminous surface dressed

138

47

64

84

Bituminous surface dressed (failed sections)

40

53

79

88

80% CBR

Fig. 19a Particle size distribution curve of materials that had exhibited good performance under Ghanaian conditions (from PWD Records, Ghana, 1952)

32

Fig. 19b

M.D. Gidigasu / Towards Developing Paving Materials Acceptance Speciﬁcations

Recommended optimum grading specifications for lateritic gravels (from USAID/BRRI, 1971)

7.3. Specifications for Natural Gravels for Base and Sub-base Generally, natural gravel base have the same requirements regarding grading and plasticity for crushed stone bases (Netterberg, 1971). The limits for plasticity index can be relaxed for calcrete, which exhibit self-cementing properties (Netterberg, 1975); they may be as high as 15 for lightly trafficked roads but not more than 8 for highly trafficked ones. Ferricrete with a plasticity index of 8 has performed satisfactorily in Southern Africa main roads, where the host material in the ferricretes is quartz sand. The strength requirements for the coarse aggregate are to accept an ACV of 29 per cent or 110 kN obtained in the 10 per cent FACT on dry material, provided the strength of the wet material is at least 75 per cent of that of the dry material. Such generalized specifications involve the risk that the values may be too low for certain materials and too high for others, and that the values actually used for selection depend on the anticipated traffic. Some South African Authorities accept disintegrating, and certain pedogenic materials, whose dry/wet strengths are as low as 50/40 kN in areas where the climatic N-value is greater than 5, in the bases of lightly trafficked roads carrying less than 1,000 vehicles per day (Mitchell, 1971). The use of weathered rock and gravels as coarse and fine material for the base requires fairly strict control of the stage of weathering which is the principal determinant of the material’s durability. It is known that weathering proceeds at a faster rate under a bituminous surfacing than in the local natural environment (Clauss, 1967). Hence, a weathered material initially of a marginal condition acceptable in regard to grading, plasticity and strength, may continue weathering in the road to reach an unacceptable condition before the end of the structural design life of the pavement. Huge maintenance or early reconstructions are the consequences of lack of durability of paving aggregates. The severity of the effect of continued weathering under a bituminous surfacing depends on the environment and the type of rock. In areas where disintegration is the dominant form of weathering, and the selection and quality control of the material must concentrate on its strength, its resistance to early degradation during hauling, dumping and rolling, and also to some extent under the traffic. This strength can be controlled by crushing tests (the 10 per cent Fines Aggregate Crushing Test (10 per cent FACT) carried out on air-dry and on wet samples), and by specifying

M.D. Gidigasu / Towards Developing Paving Materials Acceptance Speciﬁcations

33

limits based on the strength of the dry material and the acceptable decrease of the strength of the wet material. Decomposition plays the most important role when the quality and durability of crystalline rocks, such as granite or dolerite, is assessed, in areas where N is less than 5. Crushing tests are proposed for assessing the long-term strength of base materials. In addition, crystalline rocks whose composition makes them liable to undergo at least partial decomposition must be checked for the quantity of secondary minerals before being used. The relation between the local N-value and recommended percentages of secondary minerals is shown in Figure 18. Each material which plots left of the curves is suitable for the purpose represented by the curve concerned and that which plots right is unsuitable. 7.4. Specifications for Crushed Stone Base and Sub-base This is the crushed fresh rock used for bases. The quality requirements of the material are similar to those of the surfacing aggregate. However, some relaxations are permissible with regard to strength, freshness and cleanliness. The rock surface may be weathered since an oxidized layer does not affect the strength and durability of the material adversely. These relaxations are permissible because the base material will eventually be fully embedded in the base layer and there are no demands with respect to adhesion and polishing. The aggregate may be continuously or, in the case of a black base, semi-gapgraded, and the maximum size of the stone must not exceed 17.5 mm, to arrive at the required density, impermeability and stability. The fine aggregate is usually the crusher sand derived from the crushing process. This is, however, not always possible because the quantity of fines produced during the crushing of different types of rock varies and, occasionally, a special effort is required from the crusher to produce the necessary fines. Although not acceptable to all road authorities, some of them have used ‘soil’ when it satisfied the design requirements in regard to plasticity and grading. Since the plasticity must be low (liquid limit not more than 25% and plasticity index not more than 6%), the selection of such a fine aggregate is restricted to sandy soils whose –0.425 mm fraction must not contain more than 10% of montmorillonite, while the kaolinite content can be 25% to 30% (Weinert, 1980). The strength of crushed fresh rock, coarse aggregate, of particular importance during construction when the material has to stand up to hauling, dumping and compaction, is largely a function of the intergrowth and size of the minerals in the case of crystalline rocks, and of the nature of the cementing matrix in the case of all other types of rock. In addition, the strength of a rock is determined by latent fissures which easily escape detection by visual inspection. They can of course be seen in microscopic slides and they may be the cause of unexpectedly low crushing values and especially of larger-than-expected differences between the results of the 10 per cent FACT on dry and wet samples. In general, coarse-grained rocks are less resistant to crushing than fine-grained ones, and the more similar the cementing matrix and the predominating mineral in sedimentary rocks are (e.g. quartzitic sandstone) the stronger is the bond. The strength of granular sedimentary rocks is also affected favourably by angular grains which provide for additional interlock. Not all rocks are equally easily crushed to the required shape and certain rocks are notorious for producing elongated or flaky chips. The hornfels is such a rock and there are others especially those which are very hard or schistose. The problem can be overcome, however, by proper setting of the

34

M.D. Gidigasu / Towards Developing Paving Materials Acceptance Speciﬁcations

reduction rate of the crusher (e.g. Shergold and Greysmith, 1947); in which case excellent stone can be produced from the strong rock, while schistose types, although perhaps of the intended shape initially, will tend to part along the planes of schistosity during construction or in service (Weinert, 1980). The problems which have been experienced in the past with cement-stabilized crushed stone bases have not been due to the natural material but to inadequate thickness and the shrinkage of the cement. Cement- or bitumen-stabilisation will often be required for crushed stone bases especially to obtain the required bearing capacity. Certain types of rock contain sulphide minerals such as pyrite, or chalcopyrite, from which sulphuric acid and eventually sulphate salts develop when these minerals decompose. These decomposition products are, of course, deleterious to cement in the same way as organic substances may be. 7.5. Specifications for Gravel Wearing Courses The gravel wearing courses of roads should consist of materials whose quality is comparable to that of the sub-bases of other pavements. Oversize stones should be avoided because of the risk such stones provide to the moving traffic. Similarly, the material must be less clayey than is permissible for sub-bases otherwise the road will become slippery, lose its shape and potholes may develop in wet weather. Too sandy a material, on the other hand, will cause corrugation. The material of a gravel wearing course should be evenly graded (maximum grain-size not more than 50 mm) and the material smaller than 4.75 mm should possibly not comprise more than two-thirds of the total. Nevertheless, material used for gravel wearing courses in dry areas would be rather clayey to prevent corrugation and the development of dust, while less plastic material is preferable in wetter areas. Ideal materials are therefore evenly graded sandy soils or weathered rocks with a low to medium clay content. The dust nuisance, characteristic of gravel roads, is a result of the type of material used for the carriageway and the finer the material it contains or develops under traffic, the worse the nuisance is. Materials which possess self-cementing properties, e.g. certain calcretes and ferricretes, are less dusty than the same type of materials without this property. Salt-containing soils are also less dusty than salt-free materials because provided the air does not contain too much or too little moisture, the salts which migrate to, and accumulate at, the surface hydrate and exhibit slightly binding properties (Weinert, 1980). Since all salts are hygroscopic, these roads become excessively slippery if the atmospheric moisture rises above the level normal for the climatic condition concerned. Most gravel wearing courses corrugate in time, particularly at places or in lanes where vehicles accelerate. Corrugation is particularly severe where a too sandy material has been used in dry areas. Pedogenic materials, if satisfactorily graded and not too strong or too weak, would perform better in gravel roads than most other soils or weathered rocks.

8. SUMMARY OF CONCLUSIONS In recent years, considerable laboratory and field studies have taken place to solve the problem of premature roadway failures arising from apparent misuse of non-traditional tropical paving materials. In spite of the progress made so far, there is need for further

M.D. Gidigasu / Towards Developing Paving Materials Acceptance Speciﬁcations

35

laboratory and experimental field studies to evolve paving materials acceptance specifications and roadway construction methods for specific problematic materials and environmental conditions. Projects to develop sustainable paving construction methods applicable to specific tropical pedogenic aggregates, lateritic and saprolitic soils and other problem soils in diverse geologic, climatic and drainage conditions in many African countries has been ongoing. Surveys of existing roadway pavement conditions and identification of failure patterns related to the nature of the materials, pavement design and construction methods developed for rather stable temperate zone materials and environments are applicable to some tropical materials, climatic and drainage conditions. However, there are many other problem paving aggregates, gravels and fine-grained soils that pose roadway geomechanical problems. Results of studies undertaken on pavement test sections including monitoring of field experimentation is helpful for providing useful data for evolving sustainable paving materials acceptance specifications as well as pavement construction methods for the rather unstable lateritic and saprolitic soils. Experiences have also shown that stage construction involving the exposure of the various pavement layers to adverse effects of traffic and climatic conditions for at least alternate wet and dry season help to identify areas of inadequate pavement thickness, poor compaction as well as high water table. The sections are rectified and improved before the next layers are laid. The ability to reproduce laboratory measured properties under construction conditions has also raised serious concerns. The choice of the most appropriate placement method for some tropical materials in extreme climates are best achieved from existing records of field performance of similar materials under the same traffic, geologic , climatic and drainage conditions. It is clear from the above that tailoring of tropical paving materials acceptance specifications, as well as construction methods and quality control techniques are required for developing sustainable roadway geomechanical practice in the tropics. The need to apply statistical tools for separating, grouping and rating of geotechnical data for variable and unstable soils has also been felt. Insitu testing, field instrumentation and observational methods in tropical highway geomechanics as well use of statistical methods for characterisation and geotechnical rating of test data are considered useful for developing sustainable principles and practices of roadway geomechanics for marginal and problem tropically weathered soils.

REFERENCES Ackroyd, L.W., 1985: Some Nigerian laterites. Proc.1st Intern. Conf. on Geomechanics in Tropical lateritic and Saprolitic soils, Brasillia, Vol. 1pp:29-38 Ackroyd, L.W., and Rhodes, F.G., 1960: Notes on the application of linear shrinkage test to the selection of Western Nigeria gravel soils. Ministry Works. Transp., Ibadan, Western Nigeria. Tech. Pap., 5: 3pp. American Association of State Highway Officials (ASSHO), 1966: Standard specifications for highway materials and method of sampling and testing, Washington, 9th edition. American Society for Testing Materials (ASTM), 1964: Procedures for Testing Soils. Sponsored by ASTM Committee, D-18 on Soils for Engineering Purposes Andrews, J.H. amd Vlasic, Z. I., 1968. Decomposed lithic sandstone as a flexible pavement material. Proc. 4th ARRB Conf., Vol. 4 pt 2, pp: 1083-1099

36

M.D. Gidigasu / Towards Developing Paving Materials Acceptance Speciﬁcations

Arulanandan, K., 1969: Classification, engineering properties and behaviour of laterites. Spec. Sess., Eng. Prop. Lateritic Soils. Proc. Int. Conf. Soil Mech. Found. Eng., 7th, Mexico, Vol.2, pp: 163-179. Arulanandan, K. and Tunbridge, A.D., 1969: The performance of soil-cement roads in Ghana. Spec. Sess. Lateritic Soils. Proc. Intern. Conf. Soil Mech. Found. Eng. 7th, Mexico, Vol. 2, pp: 181-190. Bethune, J. D., 1971: Case studies in roadway failures. Paper presented to Transportation and Highways Branch, Victoria Division, The Institute of Engineers, Australia, 18th August, 1971, 12pp, Appendix. Bhatia, H. S., 1968: Personal communications. Building and Roads Research Institute (BRRI), Kumasi, Ghana. Bhatia, H. S., and Hammond, A. A., 1970: Durability and strength properties of lateritic aggregates of Ghana. Building and Road Res. Inst., Kumasi, Ghana, Proj. Rep., SM.9: 15pp. Bhatia, H. S. and Yeboa, S. L., 1970: A Study on the engineering characteristics of laterite gravels of Ghana. Building and Road Res. Inst., Kumasi, Ghana, Proj. Rep., SM. 10: 20pp. Blight, G.E., 1976: Migration of sub-grade salts damage thin pavements. Proc. Amer. Soc. Civ. Engrs., Vol. 102, Note 4, pp: 779-791. Brand, E.W. and Phillipson, H.B. (Editors), 1985: Sampling and testing of residual soils. A Review on International Practice. Scorpion Press, Hong Kong, 194 pp. British Standards Institution, 1975: Methods of test for soils for civil engineering purposes. Brit. Std. 1377, London, 143 pp. Carpenter, C.A., and Willis, E.A., 1939: A study of sand clay gravel materials for base course construction, Public Roads, Vol. 20. Castro, E. de, 1969: Swelling test for the study of lateritic soil. Proc. Specialty Session. Eng. Properties of Lateritic Soils. Intern. Conf. Soil Mech. Found. Eng. 7th, Mexico, Vol. 1, pp: 97-106 Clauss K.A., 1963: The influence of the pH value on the pore moisture of the dolerites in road foundation. National Inst. for Road Research, Pretoria, Report RS/16/63, 14pp. Clauss, K.A., 1967: The pH of fresh and weathered dolerite as an indicator of decomposition and of stabilization requirements. Proc. 4th Reg. Conf. Africa Soil Mech. Found. Eng., Cape Town, Vol.1, pp: 101-108. Clauss, K.A., and Loudon, P.A., 1971: The influence of initial consumption of lime on the stabilisation of South African road materials. Proc. 5th Reg. Conf. Afr. Soil Mech. Found. Eng., Vol. 1, Luanda, pp: 5/615/68. Clemente, F. M. Jr., 1981: Foreword to Proc. ASCE Geotechnical Engineering Division, Specialty Conference. Engineering and construction in tropical and residual soils. Honolulu, January 11-15, 1982, ASCE, 735pp Cocks, G.C. and Hamory, G., 1988: Road construction using lateritic gravel in Western Australia. Proc. 2nd, Intern. Conf. Geomechanics in tropical soils, Singapore, Vol.1, pp: 369-384. Comerford, L.E., 1986: A review of subdivision road design criteria, ARRB Special Report SR33, Austr. Road Res. Board, Vermont, South Victoria. Curtayne, P.C. and Todres, H.A., 1971: Investigation and correlation of parameters determining structural properties of subgrades. Proc. 5th Reg. Conf. Afr.Soil Mech. Fndn. Engg, Luanda, 3-11 - 3-16. Davidson, W.H., 1972: The influence of constitution on the engineering properties of crushed volcanic breccia. Aust. Road Res. Board. Proc. 6th Conference, Canberra, Vol.6, Part 5, pp: 71-90. Day, H.L., 1962: A progress report on studies of degrading Basalt aggregate bases. Highw. Res. Board, Bull. No. 344; pp: 8-16. De Bruijn, C. M. A. de, 1967: Annual vertical movement and soil moisture redistribution at the Onderstepoort test site. Proc. Reg. Conf. Africa, 4th, SMFE, Cape Town, South Africa, pp: 235-342. De Graft-Johnson, J.W.S., 1975: Laterite soil in road construction. Proc. Reg. Conf. Africa Soil Mech. Found. Eng, Durban, South Africa, pp: 89-98 De Graft-Johnson, J.W.S., Bhatia, H.S., and Gidigasu, M.D., 1967: The consolidation and swell characteristics of Accra mottled clays. Proceedings Asian Regional Conference, Soil Mechanics Foundation Engineering, 3rd, Haifa, Israel, pp: 75-80. De Graft-Johnson, J. W. S., Bhatia, H. S., and Gidigasu, M. D., 1969: The engineering characteristics of lateritic gravel of Ghana. Proc. Specialty Session-Engineering Properties of Lateritic Soils. Intern. Conf. Soil Mech & Foundation Eng. 7th, Mexico, Vol. 4. pp: 117-128. De Graft-Johnson, J.W.S, Bhatia, H. S. and Hammond, A. A., 1972: Lateritic gravels evaluation for road construction. Journal Soil Mech. And found. Division, Amer, Soc. Civil Engrs. Vol. 98, pp. 1245-1265. Dehlen, G. L and Williams A. A.B., 1969: The performance of full-scale base and surfacing experiments on national route 3-1 at Key Ridge. Proc. 1st Conf. Asphalt pavement South Africa, 5B/1-5B/13. Deklotz, I. A., 1940: Effect of varying the quantity of the soil portion of highway aggregates on their stability. Proc. H.R.B., Vol. 20, pp: 787 Dreyfus, J., 1952: Laterites. Revue General des Routes et des Aerodromes, Paris, 22(245): 88-101.

M.D. Gidigasu / Towards Developing Paving Materials Acceptance Speciﬁcations

37

Duncan, N. 1967: Discussion: On Subjective Classification of Hardness of Rocks and Associated Shear strength. Proc. Reg. Conf. Africa Soil Mech. Found. Eng., 4th, Cape Town, Vol.2, pp: 398. Dunn, C.S., 1966a: The stability of aggregates used in road sub-base. Roads and Road Construction. MarchApril, Vol.44, No. 219, pp: 77-81. Dunn, C.S., 1966b: Specification for crushed stone road base and sub-base aggregates. J. Brit. Granite and Whistone Federation, Vol.6, No. 2, pp: 7-22. Easterbrook, P. L., 1961: The preliminary selection of road base material in Western Nigeria by linear shrinkage test. Min. Works Transp. Ibadan, Tech. Paper, 9: 2pp. (Unpublished) Emery, S. T., 1987: Unsoaked CBR design to reduce the cost of roads. Proc. Annual Transportation Convention. Pretoria, Vol. 5B, Paper 11, 15pp Farrah, J. and Thenoz, B., 1965: L’alterabilité des roches, ses faeteur, sa prevision. Ann. Inst. Tech. Bat. Trav. Publ., Paris, Nov, 1965 Fookes, P.G. and French, W. J., 1977: Soluble salt damage to surfaced roads in the Middle East. Journ. Inst. of Highway Engineers, London, Vol. XXIV, No. 12, pp: 10-20. Fossberg, P.E., 1963: Gravel Roads – the performance and testing of materials. Proc. 3rd Reg. Conf. Soil Mech. Found. Eng., Salisbury, Vol.1, pp: 69-72. Foster, C.R., 1955: Reduction of soil strength with increase in density. Trans. Am. Soc. Civil Engr., paper 2763, pp: 803-822. Foster, M.D., 1955: The relation between composition and swelling in clays. Clays and clay minerals, Vol. 3, pp: 205-220 Frost, R.J., 1967: Review of specifications for gravel paving materials for low-cost roads. Proc. Southeast Asian Reg. Conf. Soil Eng., Bangkok, pp: 451-460. Fuller, W.B. and Thompson, S.E., 1907: The laws of proportioning concrete. Transactions Am. Soc. Civ. Engrs., Vol. 59, pp: 67-172 Ghana Highway Authority (Public Works Department), 1970: Detailed engineering design and material report. Contract 2, Ghana Highway Authority, Accra Ghana Highway Authority (Public Works Department), 1972: Ghana highway study, Road Rehabilitation Programme, Soil Assessment, Ghana Highway Authority, Accra, Vol. 2. Ghana Public Works Department (PWD), 1970: Trunk road technical specifications, Central Materials Laboratory, Accra (unpublished sheets). Gidigasu, M.D., 1971: Parameters for classification of fine-grained laterite soils of Ghana. Highway Research Board, Washington D.C., Record 374, pp: 57-79 Gidigasu, M.D., 1974: Identification of problem laterite soils in highway engineering. Highway Research Board, Washington D.C. Record 497, pp: 96-111. Gidigasu, M.D., 1975a: Behaviour of lateritic soils in roads. Ghana Engineer. Journal, Ghana Institution of Engineers, Accra, Vol.7, No.1, pp: 52-78 Gidigasu, M.D., 1975b: Some problems relating to identification of laterite soils for engineering purposes. Ghana Engineer, J. Ghana Inst. Engrs, Vol.5 No. 2, pp: 23-42 Gidigasu, M.D., 1976: An approach to the development of highway geotechnical specifications for Ghana. Ghana Engineer, Journal, Ghana Institution of Engineers, Accra, Vol.7 (2) Accra, pp: 10-22. Gidigasu, M.D., 1980a: Geotechnical evaluation of residual gravels for pavement construction. Engineering Geology, Amsterdam, Vol.15, pp: 173-194 Gidigasu, M.D., 1980b: Some contributions to tropical soils engineering in Ghana “Special Lecture”: Proceedings Regional Conference, Africa Soil Mechanics Foundation Engineering, 7th, Accra, Vol.2, pp: 508-626. Gidigasu, M.D., 1980c: Variability of geotechnical properties of sub-grade soils in a residual profile over phyllite; Proceedings Regional Conference, Africa Soil Mechanics Foundation Engineering, 7th, Accra, Vol.1, pp: 95-104. Gidigasu, M.D., 1982/83: Development of acceptance specifications for tropical gravel paving materials. Engineering Geology, Amsterdam, Vol.19, pp: 213-240. Gidigasu, M.D., 1982: Importance of material selection, construction control and field performance studies in developing acceptance specification for lateritic paving gravels: Solos e Rochas, Rio de Janeiro, Brazil, Vol.3, No.1, pp: 1-12. Gidigasu, M.D. 1985: Construction control of low-cost pavements on residual lateritic gravels over phyllite. Proceedings, Intern. Conf. on Tropical Lateritic and Saprolitic soils. 1st, Brasilia, Vol.4, pp: 261-278. Gidigasu, M.D., 1988: The use of non-traditional tropical and residual materials for pavement construction – A Review. Proc. Intern. Conf. on Geomechanics in Tropical Soils, 2nd, Singapore, 12-14 December, Vol.1, pp: 397-403. Gidigasu, M.D., 1991: Characterisation and use of tropical gravels for pavement construction in West Africa. Geotechnical and Geological Engineering, An Intern. Journal, Chapman and Hall Publishers, London, Vol.9, No.3/4, pp: 219-260.

38

M.D. Gidigasu / Towards Developing Paving Materials Acceptance Speciﬁcations

Gidigasu, M.D., 1991: A contribution to the development of acceptance specifications for lateritic paving materials in relation to the environment. Summary of contributions (1964-1990) to lateritic materials characterisation, rating and use in tropical pavement construction. 389pp, (Unpublished manuscript) Gidigasu, M.D., 2004: Field performance based specifications for landcretes in hot humid regions. Proc. CIB World Building Congress, 2004, Building for the Future. 2-7 May, 2004, Toronto, Canada. Gidigasu, M.D., 2008: Geomechanical characterisation of marginal and problem tropical weathered soils. A review, paper under review for possible publication in an International Journal. 100pp. Gidigasu, M.D., 2011: Lateritic and saprolitic soils in the design and construction of low to medium trafficked roadways. A review, 117pp, (under publication) Gidigasu, M. D., and Bhatia, H. S., 1971: The importance of Soil Profile in the Engineering Studies of Lateritic Soils. Proc. Reg. Conf. Africa. Soil Mech. Found Eng., 5th Luanda, 1: 255-260. Gidigasu, M.D and Amankwa, A., 1975: Road construction studies. BRRI Current Note No.3, Kumasi, Ghana, 41pp. Gidigasu, M.D. and Yeboa, S.L., 1972: The significance of pretesting preparations in evaluating index properties of laterite materials. Highway Research Board, Washington, D.C. Record 405, pp: 105-116. Gidigasu, M.D. and Bani, S.K., 1973: Geotechnical properties of troublesome laterite materials. Proc. Intern. Conf. Soil Mech. Found. Eng. 8th, Moscow, Vol. 2, pp: 89-96 Gidigasu, M.D and Dogbey, J.L.K., 1980a: Importance of strength criterion in selecting some residual gravels for pavement construction. Proceedings, Reg. Conf. Africa Soil Mechanics Foundation Engineering, 7th, Accra, Vol.1, pp: 317-330. Gidigasu, M.D. and Dogbey, J.L.K., 1980b: Geotechnical characterization of lateritised decomposed rocks for pavement construction in a dry sub-humid environment. Proceedings, South-East Asian Conf. On Soil Engineering, 6th, Tapei, Taiwan, pp: 493-506. Gidigasu, M.D. and Appeagyei, E., 1980: Geotechical evaluation of natural and treated black cotton clay for possible use for low-cost pavement. Proceedings, Reg. Conf. Africa Soil Mechanics and Foundation Engineering, 7th, Accra, Vol.1, pp:363-375. Gidigasu, M.D. and Mate-Korley, E., 1980: Highway geotechnical characterization of residual micaceous soils over granite. Proceedings, Reg. Conf., Africa Soil Mechanics and Foundation Engineering. 7th, Accra, Vol.2, pp: 831-844. Gidigasu, M.D. and Mate-Korley, E., 1984: Tropical gravel paving materials specifications in relation to the environment. Proceedings, Reg. Conf. Africa Soil Mechanics and Foundation Engineering, 8th, Harare, Zimbabwe, Vol.1, pp: 267-273. Gidigasu, M.D, Asante, S.P.K. and Dougan, E., 1985: Acceptance specifications for gravel paving materials for a moist sub-humid climatic zone. Proceedings, Intern. Conf. on Tropical Lateritic and Saprolitic Soils, 1st Brasilia, Vol.4, pp: 279-293. Gidigasu, M.D., Asante, S.P.K. and Dougan, E., 1987: Identification of suitable non-traditional tropical and residual paving materials in relation to the environment. Canadian Geotechnical Journal, Vol.24, pp: 58-71. Grey, J. E., 1962: Characteristics of graded base course aggregates determined by Triaxial Tests. Nat. Crushed Stone Assoc., Bulletin 12, July1962. Hammond, A. A., 1970. A study of engineering properties of some lateritic gravels from Kumasi District. Building and Road Res. Institute, Kumasi-Ghana, Project Report SM5, 17pp. Hamory, G. and Ladner, P.A., 1976: Bitumen stabilisation of a limestone (calcareous Aeolinite). Proc. 8th ARRB Conf., Perth, Australian Road Research Board, Vermont South, Victoria. pp: 12-21 Hamrol, A.A., 1961: Quantitative classification of the weathering and weatherability of rocks. Proc. Intern. Conf. Soil Mech Found. Eng., 5th, Paris, Vol.2, pp: 771-774. Hartley, A., 1974: A review of the geological factors influencing the mechanical properties of road surface aggregates. QL. J. Engineering Geology, Vol. 7, pp: 69-100. Hatcher, N.F., 1962: Shale in road pavements. Proc. 4th IRF World Meeting, Madrid. Hosking, J.R. and Tubey, L.W., 1969: Research on low grade aggregates. Road Research Laboratory (UK), Report LR 293, 30pp. Ingeroute Consultancy, Paris, 1970: Ghana Highway Study Accra-Kumasi Section of the Golden Triangle. Materials Report, Contract No. 3, Ghana Ministry of Works and Housing, Accra, 184pp. Intern. Soc. Soil Mech. Found. Eng. (ISSMFE), 1985: Proc. Intern. Conf. on Geomechanics of Tropical Lateritic and Saprolitic soils, Brasillia-Brazil, Vol.1 (475pp); Vol. 2 (449pp); Vol. 3 (399pp); Vol. 4 (435pp) Intern. Soc. Soil Mech. Found. Eng. (ISSMFE), 1982-1985: Peculiarities of geotechnical behaviour of tropical lateritic and saprolitic soils. Progress Report, Committee on Tropical soils of ISSMFE, Sao PauloBrazil, 449pp. Jamieson, I.L., Netterberg, F., Edler, A.C and Mac Carron, C. M., 1979: A small-scale road experiment to compare different types of prime for the curing and protection of a stabilized layer. Proc. 3rd Conf. Asphalt Pavement S. Africa, Durban, 1 pp:58-77

M.D. Gidigasu / Towards Developing Paving Materials Acceptance Speciﬁcations

39

Knight, K. and Dehlen, G., 1963: The Failure of a Road Constructed on Collapsing Soil. Proc. 3rd Reg. Conf. Africa Soil Mech. Found. Eng., Salisbury, Vol. 1 pp: 31-34. Lee, G. and Kennedy, C.K., 1975: Quality, shape and degradation of aggregates. QL. Jl. Engineering Geol., Vol.8, pp: 193-209. Laesk, A., 1963: Recent advances in the selection and testing of materials. Main Roads vol. 29, No. 1, Australia, September, 1963. Little, A.L., 1969: The Engineering Classification of Residual Tropical Soils. Proc. Spec. Sess. Intern. Conf. Soil Mech. Found. Eng. 7th, Mexico, Vol. 1, pp: 1-10. Liu, T.K., 1967: A review of engineering soil classification systems. Highway Research Board Record 156. pp: 1-22 Lohnes, R.A., Demirel, T., and Tuncer, E.R., 1976: Strength and structure of laterite and lateritic soils. Eng. Res. Inst. Iowa, Report ISU-BRI-Ames, 77021, 83pp. Lohnes, R.A., Rush, R.O. and Demirel, T., 1971: Geotechnical properties of selected Puerto Rican soils in relation to climate and parent rock. Geol. Soc., Am. Bull. 82, pp: 2617-2624. Loubser, M.M., 1967: Shale in road foundations. Proc. Reg. Conf. Soil Mech Found. Eng., 4th, Cape Town, Vol. 1, pp: 129-134. McDowell, C., 1966: Comparison of AASHO and Texas Test Methods and specifications for flexible base materials. Texas Highway Dept., Austin, Research report, 48-F.1966 McLerran, J.H., 1954: The engineer and pedology-State of Washington Engineering Soil Manual, pt 1, Washington State Council for Highway Research. Melville, P.L., 1948: Weathering studies of some aggregates. Proc. Highw. Res. Board, Bulletin 28, pp: 238248. Metcalf, J.B., 1991: Use of naturally-occurring but non-standard materials in low-cost road construction. Geotechnical and Geological Engineering, Vol.9, pp: 155-165. Millard, R. S., 1962: Road Building in the Tropics. Journal of Applied Chemistry, Vol.12, pp.342-357. Ministry Of Works and Housing, Ghana and Scott Wilson Kirkpatrick and Partners, 1971: Ghana Highway Study, Road Rehabilitation Programme. Vol. 2. Soil Assessment, 81 pp. Mitchell, R.L., 1971: The strength of the bases for flexible pavements with reference to overlays. Proc. 5th Reg. Conf. Soil Mech. Found. Eng., Luanda, Vol.1, pp: 5.29-5.34. Morin, W.J., 1982: Characteristics of tropical red residual soil. Proc. Specialty Session, Conf. on Eng. And Const. in Tropical and Residual soils, ASCE, Geot. Eng. Div., Jan, 11-14, 1982, Honolulu, Hawaii, pp: 172-198. Morris, P. O. and Cochrane R. H. A., 1967: Road subgrade compaction requirements- A study of density, moisture and strength of subgrades of Victorian Roads Construction prior to 1975. Australia-New Zealand Conf. S.M.F.E., Auckland, 1966, pp: 95-99 Nanda, R. L., and Krishnamachari, R., 1958: Study of Soft Aggregates from different Parts of India with a view to their use in Road Construction, II. Laterites: Central Road Research Institute, Road Research Papers, No.15: 32pp. Nascimento, U., De Castro, E. and Rodrigues, M., 1964: Swelling and petrification of lateritic soils. National Engineering Laboratory, Lisbon, Technical Paper 215, 19pp. United States Agency for International Development (USAID)/Building and Road Research Institute (BRRI), 1971: Laterite and Lateritic Soils and other Problem Soils of Africa. An Engineering Study Report, AID/CSD – 2164, 290 pp.

40

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-40

Use of Geosynthetics to Improve Seismic Performance of Earth Structures a

Junichi KOSEKIa,1 Institute of Industrial Science, the University of Tokyo, Japan

Abstract. After reviewing seismic performance of earth structures based on case histories in Japan and relevant model test results, advantages of using geosynthetics in improving their seismic performance are demonstrated. As one of successful applications, geosynthetics reinforced soil retaining walls are highlighted, focusing on several influential factors such as facing rigidity, arrangement and properties of reinforcements, and backfill and subsoil conditions. In addition, further applications of the reinforcement method using geosynthetics are introduced, which include combination with other reinforcement methods, application to bridge abutments and piers, and application to ballasted railway tracks. Keywords. Geosynthetics, earthquake, retaining wall, case history, model test

Introduction Figure 1 shows global distribution of earthquake epicenters that took place during tenyear period from 1990 to 2000 with magnitudes equal to or exceeding 4.0 and epicentral depths of 50 km or less. As indicated by an arrow in the figure, Japan is located in a very highly active zone of such seismic events.

Figure 1. Distribution of earthquake epicenters with M>4.0 and depth<50 km recorded from 1990 to 2000 (modified after JMA, 2011).

1

Corresponding Author: Junichi KOSEKI, Professor, Institute of Industrial Science, the University of Tokyo, 4-6-1 Komaba, Meguro-ku, Tokyo 153-8505, Japan; E-mail: [email protected].

41

J. Koseki / Use of Geosynthetics to Improve Seismic Performance of Earth Structures

Table 1. List of recent strong motion records in Japan (modified after NIED, 2011a&b) Earthquake㻌

Station㻌

Hyogoken-Nanbu (1995)㻌㻌 Tottoriken-Seibu (2000)㻌 Tokachi-Oki (2003)㻌 Niigataken-Chuetsu (2004)㻌 Noto-Hanto (2007)㻌 Niigataken-Chuetsu-Oki (2007)㻌 Iwate-Miyagi-Nairiku (2008)㻌

JMA Kobe (NS)㻌 JR Takatori (EW)㻌 KiK-net Hino (NS)㻌 K-net Hiroo (EW)㻌 JMA Kawaguchi (EW)㻌 K-net Anamizu (EW)㻌 K-net Kashiwazaki (NS)㻌 Kik-net Ichinoseki-Nishi (EW)㻌

a)

PGA (gal)㻌

Embankment

(Embankment)

Backfill

b) d)

Embankment

97㻌 136㻌 127㻌 47㻌 146㻌 99㻌 110㻌 62

Level ground

Reduction in necessary area

Inclined ground (slope)

818㻌 645㻌 924㻌 970㻌 1676㻌 782㻌 667㻌 1433

Retaining wall (with backfill)

c)

Level ground

PGV (kine)㻌

Retaining wall (for cut or natural slope)

Inclined ground (slope) (Embankment) Backfill

Figure 2. Schematic illustrations of embankments and retaining walls.

Table 1 summarizes the peak values of horizontal ground accelerations (PGAs) and velocities (PGVs) that were recorded during recent major earthquakes in Japan. After the 1995 Hyogoken-nanbu earthquake, the availability of strong motion data recorded near the epicenter was improved significantly. Therefore, some of them approached or exceeded 800 gals and/or 100 kines. On the other hand, earth structures, such as embankments as schematically shown in Figures 2a&b, have been widely employed to construct highways, railways, river dikes and housing lots. In addition, in order to reduce the area to be occupied by the construction of embankments and thus the volume of fill material, retaining walls (RWs) have also been frequently adopted, as schematically shown in Figures 2c&d. If we convert the horizontal seismic inertia into pseudo-static force as schematically shown in Figure 3a, the direction of apparent gravity will be inclined. Then, the driving moment to trigger the sliding failure along a potential failure plane will be increased, as shown in Figure 3b. Under such circumstances, adding reinforcements in the embankment with their tensile forces mobilized effectively will increase the resisting moment, as shown in Figure 3c.

42

J. Koseki / Use of Geosynthetics to Improve Seismic Performance of Earth Structures

a)

Embankment mah (ah=980 gal in this case) mg, g=980 gal

c) Increase in resisting moment by tensile force mobilized in reinforcements

b) Increase in driving moment

Inclined direction of apparent gravity

Figure 3. Schematic illustrations on effects of horizontal inertia of embankment and tensile force mobilized in reinforcements.

In view of the above, by addressing the following questions in this paper, attempts are made to share Japanese experiences on the use of geosynthetic-reinforcement to improve seismic performance of earth structures: Q1: How different are the seismic performances of earth structures with/without geosynthetic-reinforcement? Q2: What are the influential factors in improving effectively the seismic performance using geosynthetics? Q3: How can we extend the application of geosynthetic-reinforcement technologies to other structures? In order to answer the above questions, the paper begins with a review of seismic performance of earth structures in Japan. Next, influential factors in improving seismic performance of retaining walls using geosynthetics are discussed. Some of further applications of geosynthetic-reinforcement are briefly reviewed as well, which are followed by conclusions.

1. Seismic Performance of Earth Structures In order to answer the question 1 raised in INTRODUCTION, the following two subtopics are reviewed in this chapter, while updating the summary made by Koseki et al. (2007): How earth structures with/without geosynthetic-reinforcement behaved in case histories, and model tests?

J. Koseki / Use of Geosynthetics to Improve Seismic Performance of Earth Structures

43

Figure 4. Damage to RWs without reinforcement caused by 1995 Hyogoken-nanbu earthquake: a) and b) cantilever-type RW at Ishiyagawa before and after the earthquake, respectively, and c) gravity-type RW at Ishiyagawa, and d) leaning-type RW at Sumiyoshi (Tatsuoka et al. 1996).

1.1. Case Histories in Japan Herein, case histories from the following three major earthquakes in Japan are reviewed: 1995 Hyogoken-nanbu (Kobe) earthquake 2004 Niigataken-chuetsu earthquake 2007 Noto-hanto earthquake 1.1.1. 1995 Hyogoken-nanbu (Kobe) earthquake Figure 4 shows damage to RWs without reinforcement that were located in severely shaken area by the January 17, 1995 Hyogoken-nanbu earthquake. Conventional type RWs without foundation, such as cantilever, gravity and leaning-type ones, suffered overall tilting and/or failure of the wall body. Most of them had to be removed and reconstructed after the earthquake. In contrast to the above, one geosynthetic-reinforced soil (GRS) RW with fullheight rigid facing, which was also located in the severely shaken area, survived with minor residual lateral displacements of about 10 to 20 cm that are measured relative to the neighboring culvert box structure (Figure 5a). The standard procedures for staged construction of this type of GRS RWs with a full-height rigid facing is illustrated in Figure 5b.

44

J. Koseki / Use of Geosynthetics to Improve Seismic Performance of Earth Structures

d b)

Figure 5. a) Residual displacement of GRS RW at Tanata caused by 1995 Hyogoken-nanbu earthquake (Tatsuoka et al. 1996), and b) staged construction procedures for GRS RW with full-height rigid facing (Tatsuoka et al. 1995).

Figure 6. a) Failure of highway gravity-type RW and railway embankment at Tenno caused by 2004 Niigataken-chuetsu earthquake, and b) reconstruction of highway RW (Koseki et al., 2006a).

On the other side of the culvert structure, a cantilever-type RW with bored-pile foundation suffered similar amounts of residual lateral displacement, suggesting that this wall and the previous GRS RW without foundation exhibited almost the same seismic resistance. Refer to Tatsuoka et al. (1995, 1996, 1997, 1998) and Koseki et al. (1999) for the detailed results of the damage investigation and its back-analysis. 1.1.2. 2004 Niigataken-chuetsu earthquake earthquake Figure 6a shows damage to highway RW and railway embankment by the October 23, 2004 Niigataken-chuetsu earthquake (Tatsuoka et al. 2006, Koseki et al., 2006a). Although the original structures were without reinforcement, the highway was reconstructed using GRS RW with segmental facing panels made of pre-cast concrete (Figure 6b), while the railway on the down slope side was reconstructed using GRS RW with a full-height rigid facing and rock bolts. Such decisions were made

J. Koseki / Use of Geosynthetics to Improve Seismic Performance of Earth Structures

45

considering ground conditions, construction time and available backfill material, while adopting the same concept that the reconstructed earth structures shall exhibit improved seismic performance. Numerical verification of such improvement was made by Shinoda et al. (2009) on the railway embankment. 1.1.3. 2007 Noto-hanto earthquake Figure 7 shows damage to an embankment for Noto toll road by March 25, 2007 Notohanto earthquake. The embankment had been constructed by filling a valley. In this case, the fill material was weathered tuff, which flowed down the valley for a distance exceeding 100 meters. Note also that, based on the survey conducted after the earthquake, the ground water level was found within the fill. As shown in Figure 8, the collapsed embankment was reconstructed using GRS RW, while ensuring the drainage of ground and surface water. The waste soil that had originally been a part of the collapsed embankment was re-used after lime-treatment for the construction of the upper fill (Ishikawa Pref., 2007). On the other hand, embankments of Anamizu road that connects to the north end of the Noto toll road could survive the earthquake with minor damage. Such good performance may be attributed to the use of lime-treatment for the fill material, while ensuring the drainage of ground water by installing non-woven geosynthetic sheets.

Temporary road Original road before earthquake

Before earthquake 1: 1. 5

Temporary excavation

Estimated ground water level

1: 1. 5

Fill

1:

1.8 1:

Failure plane Collapsed fill

1.8

1: 2

.0

After earthquake

Heavily weathered tuff-breccia Old surface soil Estimated level of firm layer (N value>20)

Figure 7. Failure of an embankment of Noto toll road (at site No. 32) caused by 2007 Noto-hanto earthquake.

Figure 8. a) Schematic illustration on reconstruction of failed embankments for Noto toll road (Ishikawa Pref., 2007), and b) typical reconstruction work using GRS RW (at site No. 9).

46

J. Koseki / Use of Geosynthetics to Improve Seismic Performance of Earth Structures

155

140

140

Surchrge 1kPa

Surchrge 1kPa

53

Surchrge 1kPa

53

53

ModelBackfill 㻞㻜

ModelBackfill

(Dr=90%)

㻞㻟㼏㼙

b. Gravity type(G)

a. Cantilever type(C) 140 20

ModelBackfill 18

㻞㻜

㻞㻜

c. Leaning type(L)

140

Surchrge 1kPa

50

45

50

d. Reinforced-soil type 1(R1)

50

Extended Reinforcement

ModelBackfill

ModelBackfill

㻞㻜

Surchrge 1kPa

35

80

ModelBackfill 㻞㻜

140

Surchrge 1kPa

20

㻞㻜

e. Reinforced-soil type 2(R2)

f. Reinforced-soil type 3(R3)

Figure 9. RW models on level ground (Watanabe et al. 2003).

80 70 60 50 40 30 20

Local failure due to loss of bearing capacity

dtop

90

G

d a)

L

Backfill

C

Conventional type C: cantilever G: gravity L: leaning Reinforced-soil R1: type 1 R2: type 2 R3: type 3

35

R1

R3

R2

10 0

Normal stress at bottom of base footing, (kPa)

Wall top displacement, dtop(mm)

100

30

Seismic coefficient kh

4 3 2

20

1-9: Shaking step 7

d b)

1

15

LT7 6 5 4 LT7 (toe) 9 8

LT5

10

a)

LT6

5 0

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4

6 5

25

Location of loadcells

LT4 (heel) 0

10

20 30 40 50 60 70 80 90 100 110 Wall top displacement, dtop (mm)

Figure 10. a) Comparison of residual wall top displacements, and b) change of subgrade reaction of gravitytype RW model (Watanabe et al. 2003).

1.2. Model Tests Herein, focusing on RWs, results from relevant model tests on their different seismic performances with/without geosynthetic-reinforcement are reviewed. 1.2.1. Test conditions and procedures A series of relatively small-scale 1-g model shaking tests was conducted on six different types of retaining walls resting on level ground as shown in Figure 9. The wall models were about 50 cm high and the subsoil and backfill were modeled by very dense dry sand layers. They were subjected to several sequential horizontal excitations in 0.1 g increments. Refer to Watanabe et al. (2003) for the detailed test conditions. 1.2.2. Test results Figure 10a compares the cumulative horizontal displacements near the top of each RW model. The seismic coefficient plotted in the horizontal axis is defined as the peak base acceleration during each shaking step that is normalized with the gravitational acceleration. Up to seismic coefficient of about 0.35, no significant difference could be

J. Koseki / Use of Geosynthetics to Improve Seismic Performance of Earth Structures

47

observed. However, under higher seismic loads, the residual wall displacements accumulated rapidly with the conventional RWs, i.e., cantilever, gravity and leaningtype ones. In contrast, the GRS RWs with a full-height rigid facing exhibited more ductile behavior, in particular with the ones having extended reinforcements (types 2&3, Figures 9e&f). The reason for the less ductile behavior of the conventional RWs can be understood from Figure 10b. The subgrade reaction at the toe of base footing of the gravity-type wall increased sharply with the accumulation of wall top displacement. It suddenly decreased, however, after showing a peak state, suggesting a local failure due to loss of bearing capacity. On the other hand, the subgrade reaction at the heel of the base footing decreased in the beginning, followed by a slight increase with the occurrence of the local failure at the toe. In case of GRS RWs, as shown in Figure 11a, the tensile forces in the reinforcements measured at three different heights increased with the accumulation of the wall top displacement. Such a response of GRS RWs is the key feature for their good performance under high seismic loads. It should be noted that the mobilization of tensile force was concentrated to the uppermost long reinforcement for the type 2 GRS RW, which could effectively resist against the overturning of the facing. Due attentions should be paid on such concentration of tensile force. d U p p e r m o s t r e in fo r c e m e n t

15 10

T ype 3 (L = 3 5 c m )

b) Before shaking

T y p e 2 (L = 8 0 c m )

5 T yp e 1 (L = 2 0 c m )

0 0 15

10

20

T ype 3 (L = 3 5 c m )

3 0

40

50

60

M id d le - h e ig h t r e in f o r c e m e n t

10 5

T y p e 1 (L = 2 0 c m )

Tensile force (N)

Tensile force (N)

a)

c) After shaking

T y p e 2 (L = 4 5 c m )

0 0

10

20

3 0

40

50

60

25 T y p e 2 (L = 2 0 c m )

15

T y p e 1 (L = 2 0 c m )

Location of end of longer reinforcements when failure planes were formed.

Tensile force (N)

L o w e s t r e in f o r c e m e n t

20

10

20cm

T y p e 3 (L = 3 5 c m )

5 80cm

0 0

10 20 30 40 50 60 W a ll t o p d is p la c e m e n t , d to p ( m m )

45cm

70

d

A

d)

41.3 o 40.3 o

B

Location of failure plane observed along the center line of the sand box Backfill soil Subsoil layer

Figure 11. a) Change of reinforcement tensile force of GRS RW models (Watanabe et al. 2003), b) and c) schematic illustration of wall displacement induced by shear deformation of subsoil layer, and d) shear deformation of reinforced backfill and formation of failure planes in unreinforced backfill of type 2 GRS RW model.

48

J. Koseki / Use of Geosynthetics to Improve Seismic Performance of Earth Structures

It should be noted that, as mentioned above, the RW models with/without reinforcement exhibited similar wall displacements up to around 5 mm during the shaking steps at relatively low excitation levels (Figure 10a). This behavior is possibly affected by the shear deformation of subsoil layer. When such deformation occurs, as schematically illustrated in Figures 11b&c, the RW would suffer residual horizontal displacement without any slippage at the interface between the base of the RW and the underlying subsoil layer. Under the same subsoil conditions, as were the cases with the present model tests, the amount of such residual displacement would not depend largely on the difference in the RW types. Note also that, in case of GRS RWs, not only the subsoil deformation but also the shear deformation of reinforced backfill was observed in model tests as typically shown in Figure 11d. In evaluating the residual displacements of GRS RWs, such effects of shear deformation of reinforced backfill should be considered properly. However, in many of the relevant design guidelines, the reinforced backfill has been modeled as a rigid body that would not undergo any shear deformation (Koseki et al., 2006b).

2. Influential factors in Improving Seismic Performance of Retaining Walls Using Geosynthetics In order to answer the question 2 raised in INTRODUCTION, the following four subtopics on the seismic performance of GRS RWs are reviewed in this chapter: How their seismic performance is affected by facing rigidity, arrangement of reinforcement, properties of reinforcement, and backfill and subsoil conditions? 2.1. Facing Rigidity Tatsuoka (1993) investigated and discussed in detail the effects of facing rigidity on the stability of GRS RWs, as summarized briefly below. a) As the facing becomes more rigid, the earth pressure acting on the back face of facing increases. This large earth pressure confines the backfill soil immediately behind the facing, which decreases the deformation and increases the ultimate stability of the wall. b) A large degree of flexibility is not necessarily a preferable property for completed GRS RWs, although this property is required to accommodate possible large deformation of the subsoil so that a deep foundation becomes unnecessary. The above summary a) suggests also that the geosythetic-reinforcement technique is not a method to reduce earth pressure exerted from the backfill soil. Rather, this technique takes advantage of the tensile force mobilized in reinforcements to resist against the earth pressure (and inertia force of facing in case of earthquakes). Regarding the above summary b), one of the possible compromises are that walls are made as much as flexible during construction, while they are made stiff enough before they are open to service. The staged construction procedures to cast-in-place the

J. Koseki / Use of Geosynthetics to Improve Seismic Performance of Earth Structures

49

full-height rigid facing at the last stage, as shown in Figure 5b, enable us to control the flexibility in such a manner, while they require in general longer construction period than GRS RWs with pre-cast segmental types of facing. It should be noted that, in case of GRS RWs with segmental types of facing, local instability of facing due to failure at the connection between the facing and the reinforcement may lead easily to overall failure. Such a failure mode was observed in the 1999 Chi-chi earthquake in Taiwan as typically shown in Figure 12a. Note also that, with the above particular case, the vertical spacing of reinforcements was 80 cm, which exceeded the value recommended by relevant design guidelines. At the time of the earthquake, therefore, insufficient number of reinforcements may have been subjected to excessive tensile force, causing local rupture at their connection with the facing, and/or the overall facing rigidity may have been too small to resist against the earthquake loads, causing excessive deformation of the stacked facings and pull-out of connecting pins, as schematically illustrated in Figure 12b. On the other hand, in case of GRS RWs with full-height rigid facing, even if local rupture or failure takes place, the tensile force mobilized in the reinforcements would be re-distributed more easily, since the rigid facing is supported simultaneously by many layers of reinforcements. This feature would enhance their redundancy against overall failure. GRS RWs with full-height rigid facing that were constructed in Japan following the procedures shown in Figure 5b have performed well under not only working loads but also large earthquake loads (e.g., Figure 5a). Therefore, this particular type of GRS RWs has been adopted for constructing important permanent earth structures to support such as bullet train tracks and highways (Tamura, 2006). Their application in terms of total wall length exceeded 120 km as of June, 2010. Hereafter, focusing on GRS RWs with full-height rigid facing, effects of other influential factors are discussed. d a)

d Precastconcrete blocks

b)

Overall instability (Possible failure plane)

80cm (Collapse) Local instability

Reinforcements (Geogrid)

(Bulging) (Rupture of reinforcement or pull-out of connecting pin)

Figure 12. Damage to GRS RW with segmental facing caused by 1999 Chi-chi earthquake, Taiwan; a) photograph of damaged structure and b) schematic cross-section (Koseki and Hayano 2000).

50

J. Koseki / Use of Geosynthetics to Improve Seismic Performance of Earth Structures

2.2. Arrangement of Reinforcement In the model tests presented previously (Figure 9), three types of reinforcement arrangement were employed: R1 or type1 having relatively short reinforcements with equal length, R2 or type 2 having partially extended reinforcements, and R3 or type 3 having relatively long reinforcements with equal length. Among the above three types, as can be seen from Figure 10a, the type 3 wall exhibited the smallest amounts of residual wall top displacement. In addition, though the total length of reinforcement of the type 2 wall was about 80 % as large as that of the type 3 wall, their seismic performances in terms of the wall top displacement were similar to each other. Such good performance of the type 2 wall confirms that partial extension of upper reinforcements (preferably, connection with another wall on the opposite side) improves the seismic stability significantly, since the upper reinforcements can resist more effectively against the overturning mode of failure. It should be noted that, as shown in Figure 11d, the extended uppermost reinforcement in the type 2 wall prevented full formation of a failure plane (marked as A in the figure) in the backfill as well, which passed through the end of the other extended reinforcement at the middle height. In turn, as mentioned previously on Figure 11a, the tensile force mobilized in the reinforcements concentrated into the uppermost reinforcement of the type 2 wall, implying that this reinforcement was the key to exhibit the above good performance. Note also that, with the type 2 and 3 walls, the calculated values of critical seismic coefficient to induce a factor of safety equal to unity in pseudo-static limit-equilibrium stability analysis against overturning failure were different from each other (0.55 and 0.70, respectively). Refer to Watanabe et al. (2003) for the details of calculation. In spite of such difference, the two walls exhibited similar seismic performances. This is possibly affected by the shear deformation of reinforced backfill (Figure 11d), which is not considered in evaluating the above critical seismic coefficients. 2.3. Properties of Reinforcement As an extension of the model tests presented previously (Figure 9), another series of 1g model shaking tests was conducted by Nakajima et al. (2007a)where two kinds of geosynthetic-reinforcement models, called herein as PB (phosphor bronze) model and PE (polyester) model, were employed. As summarized in Table 2, their tensile stiffness per single strip evaluated in direct tension tests was higher with the PB model. On the other hand, their tensile stiffness per unit width of the grid was higher with the PE model, since this model consisted of much larger number of strips than the PB model. In addition, as shown also in Table 2, their ultimate pull-out resistance per unit width of the grid was higher with the PE model, due possibly to its mesh size (= 3*3 mm) that could confine approximately 10 particles of the sand material (with a mean diameter of 0.23 mm) that was used to prepare the backfill in the model tests. In contrast, the PB model had a much larger mesh size (= 50*95 mm), and it exhibited larger pull-out resistance per unit width of the grid at small levels of pull-out displacement (up to about 1 mm, Fig. 13a) due possibly to the effects of sand particles that were pasted on the surface of the reinforcement to mobilize the frictional resistance.

51

J. Koseki / Use of Geosynthetics to Improve Seismic Performance of Earth Structures

Table 2. Properties of model reinforcements (modified from Nakajima et al., 2007a)

Property

Secant tensile stiffness at T=Tmax/2

Unit

per single strip

per unit width

Ultimate pull-out resistance, Tmax at v=5 kPa per unit width for buried length of 0.5 m

(kN//strip)

(kN//m)

(kN/m)

3.5-5.7 41-66 2.96* 0.31-0.36 105-121 4.48 * Strips in the air ruptured under this tensile force.

PEv=5kPa PB v=5kPa

㼀㼜㼑㼍㼗㻩㻝㻞㻜㻥㻺

㻌

800

Rupture of phosphor bronze 400

d a) 0

㻜㻡㻝㻜㻝㻡 Pullout displacement(at 5cm away from wall facing) (mm)

Figure 13. a) Pull-out test results on model reinforcements and b) & c) comparison of residual tilting angles and base sliding displacements of wall facing (Nakajima et al., 2007a).

㻞㻜

㻿㼘㼕㼐㼕㼚㼓㻌㼐㼕㼟㼜㼘㼍㼏㼑㼙㼑㼚㼠㻘㻌㼐㼟㻔㼙㼙㻕

Pullout resitance(N)

1200

㻤㻢

d b)

㻠

PE

PB

㻌

㻌 1600

㻞㻜

㻌㻌

㻠㻜㻟㻜

c)

㻞㼚㼐

d 㻞㼚㼐

㻞㻜

㻌

㼀㼕㼘㼠㼕㼚㼓㻌㼍㼚㼓㼘㼑㻘㻌䃗㻔㼻㻕㻌㻿㼑

PB PE

㻝㼟㼠

㻝㼟㼠

㻝㻜㻜

㻜㻚㻤

㻝㻚㻜

㻝㻚㻞

㻝㻚㻠

㻝㻚㻢

㻹㼍㼤㼕㼙㼡㼙㻌㼎㼍㼟㼑㻌㼍㼏㼏㼑㼘㼑㼞㼍㼠㼕㼛㼚䚸䃐㼙㼍㼤㻔㻳㻕

Despite the above differences in the reinforcement properties, the observed cumulative tilting angles and base sliding displacements of the GRS RW models (type 2, Fig. 9e) employing the two types of reinforcements, respectively, were in general similar to each other, as shown in Figures 13b and c. Note that, in these model tests, the peak base acceleration was 0.9 g in the first shaking step, which was increased in 0.3 and 0.4 g increments in the second and third shaking steps, respectively. Further, the fourth shaking step was conducted by using the same base acceleration as in the third one. The influential factor to affect the tilting behavior of GRS RWs observed in the present model tests would be the pull-out resistance mobilized at small displacement levels. As mentioned above, it was larger with the PB model than with the PE model, though the ultimate resistance was vice versa (Figure 13a). The slight difference between the two models in terms of the cumulative tilting angles as can be seen in Figure 14b may have been affected by such pull-out properties at small displacement levels. On the other hand, since no rupture or pull-out failure of reinforcements was observed, these properties would not have affected the tilting behavior in the present model tests. Regarding the base sliding of GRS RWs shown in Figure 13c, no significant difference was observed between the two models. This is possibly because the resistance against base sliding under the model configurations employed for the present model tests is not largely affected by the reinforcement properties but

52

J. Koseki / Use of Geosynthetics to Improve Seismic Performance of Earth Structures

predominantly affected by the backfill and subsoil conditions around their interfaces. For small displacement levels, in addition, one needs to recall the effects of shear deformation of subsoil layer as discussed previously (Figures 11b and c), which are also independent from the reinforcement properties. 2.4. Backfill and Subsoil Conditions Not only the conditions of reinforcement arrangement and facing rigidity as discussed above, but also the conditions of backfill and subsoil affect the seismic performance of GRS RWs. For example, backfill soils that are not well-compacted may not mobilize sufficient pull-out resistance of reinforcements, no matter how the reinforcements themselves are stiff and strong enough. In addition, poorly-compacted backfill soils may suffer excessive settlement during their service period prior to earthquakes, resulting into local failure of reinforcements at their connection with the facing. Once pull-out or local failure of reinforcements takes place, it would trigger overall instability of GRS RWs during earthquakes as well as under working load conditions. It should be noted that, as compared to the backfilling work without reinforcement that is employed for conventional type RWs, the backfill soil with reinforcements can be compacted in a more effective manner, since the existence of reinforcement would confine the lateral deformation of the backfill soil during compaction work. Therefore, by taking such advantage, due attentions shall be paid in constructing GRS RWs to compact the backfill soil sufficiently. The above confinement of the lateral deformation of the backfill soil would in turn mobilize initial tensile forces in the reinforcements. Such effective mobilization of tensile forces would contribute to reduce the displacement of the facing as well. Note also that, by placing reinforcements at a specified vertical spacing (equal to 30 cm in case of GRS RWs with full-height rigid facing shown in Figure 5), one may ensure the lifting height of the backfilling work to be equal to or smaller than this vertical spacing. When considering the effects of shear deformation of subsoil layer as discussed previously (Figures 11b and c), one may understand easily that the subsoil conditions also affect the seismic performance of GRS RWs. The effects of subsoil conditions and applicability of large diameter nailing as aseismic measures for GRS RWs are further investigated by Kato et al. (2002) based on model test results.

3. Further Applications of Geosynthetic-Reinforcement In order to answer the question 3 raised in INTRODUCTION, the following three subtopics are reviewed in this chapter: How geosynthetics are used for combination with other reinforcement methods, application to bridge abutments and piers, and application to ballasted railway tracks?

J. Koseki / Use of Geosynthetics to Improve Seismic Performance of Earth Structures

53

3.1. Combination with Other Reinforcement Methods In the model tests conducted by Kato et al. (2002), GRS RWs exhibited substantially high seismic stability when they were further reinforced with large diameter nails (Figure 14a). The nails could effectively prevent formation of full failure plane in backfill and sloped subsoil layers. On the other hand, the full failure plane formation led to less ductile behavior of GRS RW model on sloped subsoil layer without nails (Figure 14b).

Figure 14. a) GRS RW model with large diameter nails on sloped subsoil layer and b) full failure plane formation in backfill and subsoil layers of GRS RW model without nails (Kato et al. 2002).

㼂㼑㼞㼠㼕㼏㼍㼘㻌㼍㼞㼞㼛㼣㼟㻌㼕㼚㼐㼕㼏㼍㼠㼑㼠㼔㼑㻌㼍㼏㼏㼛㼙㼜㼘㼕㼟㼔㼙㼑㼚㼠㻌㼛㼒㻌㼠㼔㼑㻌㼒㼍㼕㼘㼡㼞㼑㻌㼜㼘㼍㼚㼑㻌㼕㼚㻌㼠㼔㼑㻌㼎㼍㼏㼗㼒㼕㼘㼘㻌㼘㼍㼥㼑㼞

㻢㻠

㻌㻾㻾㼃㻿㻼 With SP Without 㻌㻾㻾㼃

㻞

d

㻜

a) 㻌㻌

㻝㻜㻜

Crashed rock (t=300)

ȟ

㻤㻜

(Unit: mm) Cement-mixed fill (150 kg/m3)

㻢㻜

ds

㻠㻜 Sheet pile (SP)

㻞㻜㻜

b)

d b)

㻜

Cement-mixed fill (105 kg/m 3)

d 㻌

㻮㼍㼟㼑㻌㼟㼘㼕㼐㼕㼚㼓䠗㼐㼟㻔㼙㼙㻕

㻌

㼀㼕㼘㼠㼕㼚㼓㻌㼍㼚㼓㼘㼑䠗䃗㻔㼐㼑㼓㼞㼑㼑㻕㻌

d a)

㻠㻜㻜㻤㻜㻜㻝㻞㻜㻜㻮㼍㼟㼑㻌㼍㼏㼏㼑㼘㼑㼞㼍㼠㼕㼛㼚㻧䃐㼎㼍㼟㼑㻔㼓㼍㼘㻕

㻝㻢㻜㻜

Figure 15. Comparison of residual displacements of GRS RW models with/without embedded sheet pile (Nakajima et al. 2006).

Geogrid Gravel mat (secondary (t=500) reinforcement)

Figure 16. Failure of railway embankment at Tsukanoyama caused by 2004 Niigatakenchuetsu earthquake and its reconstruction (Morishima et al. 2005).

54

J. Koseki / Use of Geosynthetics to Improve Seismic Performance of Earth Structures

In addition, as shown in Figure 15, the residual tilting angle of the facing of GRS RW model could be effectively reduced by installing a sheet pile at the foot of the facing and connecting it to the facing (Nakajima et al. 2006). Further, not only soil reinforcement methods but also soil improvement methods have been combined with geosynthetic-reinforcement. For example, Izawa et al. (2009) reported successful development and practical application of a new method that combines geosynthetic-reinforcement with a fibre-mixed soil-cement wall. The benefits of similar approaches to increase the stability by using a combination of geosyntheticreinforcement and cement-treatment will be also described in the next section on bridge abutments. The above combination of geosynthetic-reinforcement with other reinforcement or improvement methods has been adopted in Japanese practice as well. Herein, some relevant case histories on its application to reconstruction works of earth structures damaged by earthquakes are reviewed briefly. 3.1.1. Case histories in 2004 Niigataken-chuetsu earthquake Figure 16a shows the collapse of a railway embankment. It was reconstructed by reusing the collapsed fill material (Morishima et al., 2005). As illustrated in Figure 16b, the fill material was improved by adding a cement-origin stabilizer at a mixing ratio of 150 kg/m3 for the upper fill or 105 kg/m3 for the lower fill. It was further reinforced with geogrid sheets that were placed at a vertical spacing of 1.5 m as secondary reinforcement. In order to ensure the drainage, a gravel mat was placed at the bottom of the embankment. 3.1.2. Case histories in 2007 Noto-hanto earthquake As shown in Figure 8a, the collapsed highway embankments were reconstructed using GRS RWs, where the collapsed fill material was re-used after lime-treatment for the construction of the upper fill (Ishikawa Pref., 2007).

d

45

d b) (Unit: mm)

Ordinary type 30 Conventional (untreated & 25 unreinforced (Model1) 20 backfill 35

15 10

1340 Surcharg (1kPa) e 6

Bridge girder

5

Displacement(mm)

40

a)

4

New type (cement-treated backfill reinforced with Proposed geogrids) Type

5

3

LT2

2

Reinforcement

1

Cement-mixed loam

620

LT1

0 0 200 400 600 800 100012001400

Toyoura sand

200

1030

Gravel 250

Base acceleration(gal) Figure 17. a) Comparison of residual wall top displacements and b) new type bridge abutment model with cement-treated backfill reinforced with geogrids (Watanabe et al. 2002).

J. Koseki / Use of Geosynthetics to Improve Seismic Performance of Earth Structures

55

Figure 18 Bridge abutment using GRS RW with cement-treated backfill gravel (Aoki et al. 2005).

3.2. Application to Bridge Abutments and Piers In past major earthquakes, bridge abutments suffered from several types of damage, including settlement of their backfill soil, extensive residual displacements of the wall body, and structural failure of the wall body. In order to improve the stability of retaining walls supporting bridge girders, Watanabe et al. (2002) and Aoki et al. (2003) conducted a series of 1-g model shaking tests on different types of bridge abutments. As shown in Figure 17a, the wall displacement could be reduced significantly by using cement-treated backfill that was reinforced with geogrids (Figure 17b) as compared to the ordinary type wall model using unreinforced and untreated backfill sand. The backfill settlement could be also reduced extensively. On the other hand, Aoki et al. (2003) revealed as well that using cement-treated backfill without reinforcements is not enough to improve the seismic stability of bridge abutments up to sufficiently high levels. As reported by Aoki et al. (2005), a similar type bridge abutment using cementtreated backfill gravel for the GRS RW with a full-height rigid facing (Figure 18) has been implemented in practice to construct an abutment for new bullet train in Kyushu Island, Japan. By employing this system, the construction cost could be saved by 20 % as compared to the ordinary method. Tatsuoka et al. (2009) and Aizawa et al. (2007) conducted another series of 1-g model shaking tests on bridge abutments. The models included integral types with or without reinforcements in the backfill (Figure 19a), where the bridge girder was firmly connected to both of the walls. They were subjected to several sequential horizontal excitations with 20 cycles of sinusoidal waves at a frequency of 5 Hz. As shown in Figure 19b, the seismic stability could be improved significantly by combining the integral bridge and the GRS RW systems. It should be noted that, with the above GRS integral bridges, in addition to the rupture strength and pull-out resistance of reinforcements, the connection strength between the facing and the reinforcement was found to be one of the key issues to maintain the high seismic stability (Hirakawa et al., 2007).

56

J. Koseki / Use of Geosynthetics to Improve Seismic Performance of Earth Structures

Figure 19 a) Bridge abutment model and b) comparison of residual backfill settlements (Tatsuoka et al. 2007). RC block

4.4m 2.4m

Hydraulic jacks

Girder 0.8m RC facing

2.7m

Gravel bags

Grid (TR =58.8kN/m = 6tonf/m ) Anchored for 4 m-deep

Tie rod

9 - 11m

Columns of cement-treated clay

Soft clay deposit

Figure 20 Bridge pier using GRS RW with pr-eloaded and pre-stressed backfill (Uchimura et al. 2003).

Note also that, for bridge piers to support girders, pre-loaded and pre-stressed gravel backfill for GRS RW with a full-height rigid facing has been also implemented in practice for a railway in Kyushu Island, Japan (Figure 20, Uchimura et al. 2003). Its high seismic stability has been confirmed through model shaking tests (Shinoda et al. 2003), and its applicability to bridge abutments has been verified as well (Aoki et al. 2003). 3.3. Application to Ballasted Railway Tracks In order to reduce deformation of ballasted railway tracks during large earthquakes, a method to reinforce their shoulders using stacked geosynthetic bags that are filled with ballast material, as schematically shown in Figure 21a, was developed by Kachi et al. (2010). The stacked bags were further reinforced with iron bars by driving them through the bags and embedding them to the base layer. This method can be regarded as an extension of conventional soil bag methods, while it employs a mesh-type bag to mobilize better interlocking at the interface between adjacent bags. The opening of the mesh is about 25 mm, which is much larger

J. Koseki / Use of Geosynthetics to Improve Seismic Performance of Earth Structures

57

than those of conventional soil bags. As schematically shown in Figure 22, the tensile force mobilized in the bag would in turn enhance the bearing capacity of the bag by adding an apparent cohesion. Refer to Matsuoka and Liu (2003) for more detailed discussions on the advantages of soil bags. It should be noted that, during large earthquakes, the direction of the major principal stress that is mobilized in the ballasts will rotate from the vertical direction. Therefore, referring to the pioneering work by Matsushima et al. (2008), the method was improved by stacking the bags in an inclined manner as shown in Figure 21b. In order to resist against the overturning moment more effectively, some of the reinforcing iron bars were also inclined. In addition, in order to increase the overall stiffness of the stacked bags, adjacent bags were connected to each other by using a Ushaped iron bar.

Figure 21 a) Original concept and b) improved version of ballasted railway tracks reinforced with geosynthetic bags and iron bars (Kachi et al., 2010).

External force Tension

With bags Without bags

c Figure 22 Schematic illustration on effects of soil bags (modified after Matsuoka and Liu, 2003).

d

Figure 23 a) Full-scale model for 1-g shaking test on improved version of ballasted railway tracks reinforced with geosynthetic bags and iron bars and b) maximum horizontal displacements during shaking (modified after Kachi et al., 2010).

58

J. Koseki / Use of Geosynthetics to Improve Seismic Performance of Earth Structures 㻔㼓㼍㼘䠅

㼕㼚㼜㼡㼠㼍㼏㼏㼑㼘㼑㼞㼍㼠㼕㼛㼚

㻝㻡㻜㻜㻝㻜㻜㻜㻡㻜㻜㻜㻙㻡㻜㻜㻙㻝㻜㻜㻜㻙㻝㻡㻜㻜

㻜

㻡

㻝㻜

㻝㻡

㻞㻜㼠㼕㼙㼑䠄㼟㼑㼏䠅

㻞㻡

㻟㻜

㻟㻡

Figure 24 Excitation time history for full-scale 1-g model shaking test (Kachi et al., 2010).

After confirming better performance of the improved version by horizontal monotonic loading tests, a full-scale 1-g model shaking test was conducted (Figure 23a). The model was subjected to a severe horizontal excitation as shown in Figure 24. The maximum response in terms of horizontal displacements during the excitation is shown in Figure 23b. The horizontal displacement at the tie position (see Figure 21b) was as small as 6 mm, which implies that the improved structure can perform well even under severe earthquake loads in preventing high-speed trains from derailment. Thus, it has been adopted for actual reinforcing works of existing ballasted tracks for Tokaido bullet train (or Tokaido Shinkansen) of Central Japan Railway Company. Kachi et al. (2010) reported as well that the newly developed method has good workability, as compared to an alternative method using a pre-cast concrete block with a mass of about 200 kg, and thus requires heavy equipments. With the new method, on the other hand, each of the geosynthetic bags can be handled easily without heavy equipments, since they are simply filled with about 25 kg of ballast material and can be easily compacted into a specified dimension of 400*400*100 mm with a plate compacter.

4. Conclusions The contents of the present paper on the use of geosynthetic-reinforcement to improve seismic performance of earth structures can be summarized as follows. a) As compared to unreinforced earth structures, geosynthetic-reinforced soil retaining walls (GRS RWs) performed well during past large earthquakes in Japan. Their ductile behavior under large earthquake loads was also confirmed by relevant model tests. Thus, reinforced earth structures have been adopted for new construction of important permanent structures as well as their damage rehabilitation works. b) The seismic performance of GRS RWs is affected by rigidity of facing, pull-out resistance and arrangement of reinforcements, and their connection strength. Due attentions shall be also paid on the effects of shear deformation of subsoil and reinforced backfill layers and formation of full failure plane in these layers. c) Combination with other reinforcement or improvement methods, such as nailing, anchoring, sheet-piling and cement-treatment, enhances further the seismic performance of GRS RWs. By employing such combination, collapsed soil materials can be re-used effectively. d) Due to the above advantages, the application of geosynthetic-reinforcement is extending to wider areas. For example, it has been successfully applied to bridge abutments and ballasted railway tracks in practice.

J. Koseki / Use of Geosynthetics to Improve Seismic Performance of Earth Structures

59

Acknowledgements The author wishes to express his sincere appreciation to Prof. F. Tatsuoka (Tokyo University of Science) and Dr. M. Tateyama (Railway Technical Research Institute) who initiated immediately after the 1995 Hyogoken-nanbu earthquake the long-lasting series of researches on seismic performance of GRS RWs as well as conventional type RWs, which were referred to extensively in this paper. Among many others, contributions of late Dr. Y. Munaf, Dr. K. Watanabe, Mr. N. Kato, Dr. S. Nakajima and Mr. K. Hong (formerly, graduate students at University of Tokyo) in carrying out the researches are also acknowledged.

References [1] Aizawa, H., Nojiri, M., Hirakawa, D., Nishikiori, H., Tatsuoka, F., Tateyama, M. and Watanabe, K. (2007). Validation of high seismic stability of a new type integral bridge consisting of geosyntheticreinforced soil walls, New Horizons in Earth Reinforcement, Taylor and Francis, 819-825. [2] Aoki, H., Watanabe, K., Tateyama, M. and Yonezawa,T. (2003). Shaking table tests on earthquake resistant bridge abutment, Proc. of 12th Asian Regional Conf. on Soil Mechanics and Geotechnical Engineering, 1, 267-270. [3] Aoki, H., Yonezawa,T., Tateyama, M. Shinoda, M. and Watanabe, K. (2005). Development of aseismic abutment with geogrid-reinforced cement-treated backfills, Proc. of 16th International Conf. on Soil Mechanics and Geotechnical Engineering, 3, 1315-1318. [4] Hirakawa, D., Nojiri, M., Aizawa, H., Nishikiori, H., Tatsuoka, F., Watanabe, K. and Tateyama, M. (2007). Effects of the tensile resistance of reinforcement in the backfill on seismic stability of GRS integral bridge, New Horizons in Earth Reinforcement, Taylor and Francis, 811-817. [5] Ishikawa Prefecture (2007). Damage to toll road due to Notohanto earthquake and its rehabilitation, Official Bulletin, 1, 4p. (in Japanese). [6] Izawa, J., Ito, H., Saito, T., Ueno, M. and Kuwano, J. (2009). Development of rational seismic design method for geogrid-reinforced soil wall combined with fibre-mixed soil-cement and its applications, Geosynthetics International, 16(4), 286-300. [7] Japanese Geotechnical Society (2007a). Compiled data on damage to highways due to the 2007 Notohanto Earthquake, 79p. (in Japanese) [8] Japanese Geotechnical Society (2007b). Report of damage survey committee on the Niigata-ken Chuetsu Earthquake, 518p. (in Japanese) [9] Japan Meteorological Agency (2011). http://www.jma.go.jp/jma/kishou/know/whitep/2-1.html (in Japanese, accessed on March 10, 2011) [10] Kachi, T., Kobayashi, M., Seki, M. and Koseki, J. (2010). Evaluation tests of ballasted track reinforced with geosynthetic bags, Proc. of 9th International Conference on Geosynthetics, Brazil, 1499-1502. [11] Kato, N., Huang, C.C., Tateyama, M., Watanabe, K., Koseki, J. and Tatsuoka, F. (2002). Seismic stability of several types of retaining walls on sand slope, Proc. of 7th International Conf. on Geosynthetics, Nice, 1, 237-240. [12] Kitamoto, Y. Abe, H., Shimomura, H., Morishima, H. and Taniguchi, Y. (2006). Rapid and strengthened repair construction for a seriously damaged railway embankment during violent earthquakes, Proc. 8th Int. Conf. on Geosynthetics, Yokohama, 3, 861-864. [13] Koseki, J., Tateyama, M., Horii, K., Munaf, Y. and Kojima, K. (1999). Back analyses of case histories and model tests on seismic stability of retaining walls, Proc. of 11th Asian Regional Conf. on Soil Mechanics and Geotechnical Engineering, 1, 399-402. [14] Koseki, J. and Hayano, K. (2000). Preliminary report on damage to retaining walls caused by the 1999 Chi-Chi earthquake, Bulletin of ERS, Institute of Industrial Science, Univ. of Tokyo, 33, 23-34. [15] Koseki, J., Sasaki, T., Wada, N., Hida, J., Endo, M. and Tsutsumi, Y. (2006a). Damage to earth structures for national highways by the 2004 Niigata-ken Chuetsu earthquake, Soils and Foundations, 46(6), 739-750. [16] Koseki, J., Bathurst, R.J., Guler, E., Kuwano, J. and Maugeri, M. (2006b). Seismic stability of reinforced soil walls, Proc. of 8th International Conference on Geosynthetics, Yokohama, 1, 51-77.

60

J. Koseki / Use of Geosynthetics to Improve Seismic Performance of Earth Structures

[17] Koseki, J., Tateyama, M., Watanabe, K. and Nakajima, S. (2007). Stability of earth structures against high seismic loads, Proc. of 13th Asian Regional Conf. on Soil Mechanics and Geotechnical Engineering, Kolkata, 2, 222-241. [18] Matsuoka, H. and Liu, S.H. (2003). New earth reinforcement method by soilbags (“donow”), Soils and Foundations, 43(6), 173-188. [19] Matsushima, K., Aqil, U., Mohri, Y. and Tatsuoka, F. (2008). Shear strength and deformation characteristics of geosynthetic soil bags stacked horizontal and inclined, Geosynthetics International, 15(2), 119–135. [20] Morishima, H., Saruya, K. and Aizawa, F. (2005). Damage to earth structures for ordinary railways and its rehabilitation, Foundation Engineering and Equipment (Kiso-ko), 33(10), 78-83 (in Japanese). [21] Nakajima, S., Koseki, J., Watanabe, K., Tateyama, M. and Kato, N. (2006). Shaking table model tests on geogrid reinforced soil retaining wall with embedded sheetpile, Proc. of 8th International Conference on Geosynthetics, Yokohama, 4, 1507-1510. [22] Nakajima, S., Hong, K., Mulmi, S., Koseki, J., Watanabe, K. and Tateyama, M. (2007a). Model tests on seismic performance of reinforced soil retaining walls by using different geo-grids, International Workshop on Earthquake Hazards and Mitigations, Guwahati, India, 319-325. [23] Nakajima, S. Koseki, J. Tateyama, M. and Watanabe, K. (2007b). Shaking table model tests on retaining walls reinforced with soil nailings, New Horizons in Earth Reinforcement, Taylor and Francis, 707-712. [24] National Research Institute for Earth Science and Disaster Prevention (2011a). http://www.knet.bosai.go.jp/k-net/topics/chuetsuoki20070716/pgav5v20070716.html (in Japanese, accessed on March 10, 2011) [25] National Research Institute for Earth Science and Disaster Prevention (2011b). http://www.knet.bosai.go.jp/k-net/topics/Iwatemiyaginairiku_080614/IWTH25_NIED.pdf (in Japanese, accessed on March 10, 2011) [26] Shinoda, M., Uchimura, T. and Tatsuoka, F. (2003). Improving the dynamic performance of preloaded and prestressed mechanically reinforced backfill by using a ratchet connection,” Soils and Foundations, 43(2), 33-54. [27] Shinoda, M., Watanabe, K., Kojima, K., Tateyama, M. and Horii, K. (2009). Seismic stability of a reinforced-soil structure constructed after the mid-Niigata prefecture earthquake, Geosynthetics International, 16(4), 274-285. [28] Tamura, Y. (2006). Lessons learnt from the construction of geosynthetic-reinforced soil retaining walls with full-height rigid facing for the last 10 years, Proc. of 8th International Conference on Geosynthetics, Yokohama, 3, 941-944. [29] Tatsuoka, F. (1993). Roles of facing rigidity in soil reinforcing, Earth Reinforcement Practice, Balkema, 2, 831-870. [30] Tatsuoka, F., Tateyama, M., Koseki, J. and Uchimura, T. (1995). Geotextile-reinforced soil retaining wall and their seismic bahaviour, Proc. 10th Asian Regional Conf. on Soil Mechanics and Foundation Engineering, 2, 26-49. [31] Tatsuoka, F., Tateyama M. and Koseki, J. (1996). Performance of soil retaining walls for railway embankments, Soils and Foundations, Special Issue of Soils and Foundations on Geotechnical Aspects of the January 17 1995 Hyogoken-Nambu Earthquake, 311-324. [32] Tatsuoka, F., Koseki, J. and Tateyama, M. (1997), Performance of reinforced soil structures during the 1995 Hyogo-ken Nanbu Earthquake, Earth Reinforcement, Balkema, 2, 973-1008. [33] Tatsuoka, F., Koseki, J., Tateyama, M., Munaf, Y. and Horii, N. (1998). Seismic stability against high seismic loads of geosynthetic-reinforced soil retaining structures, Proc. of 6th Int. Conf. on Geosynthetics, Atlanta, 1, 103-142. [34] Tatsuoka, F., Konagai, K., Kokusho, T., Koseki, J. and Miyajima, M. (2006). Special Session on the 2004 Niigata-ken Chuetsu earthquake, Proc. of 16th International Conf. on Soil Mechanics and Geotechnical Engineering, 5, 3279-3287. [35] Tatsuoka, F., Hirakawa, D., Nojiri, M., Aizawa, H., Nishikiori, H., Soma, R., Tateyama, M. and Watanabe, K. (2009). A new type of integral bridge comprising geosynthetic-reinforced soil walls, Geosynthetics International, 16(4), 301-326. [36] Uchimura, T., Tateyama, M., Koga, T. and Tatsuoka, F. (2003). Performance of a preloadedprestressed geogrid-reinforced soil pier for a railway bridge, Soils and Foundations, 43(6), 33-50. [37] Watanabe, K., Tateyama, M., Yonezawa, T., Aoki, H., Tatsuoka, F. and Koseki, J. (2002). Shaking table tests on a new type bridge abutment with geogrid-reinforced cement treated backfill, Proc. 7th International Conf. on Geosynthetics, Nice, 1, 119-122. [38] Watanabe, K., Munaf, Y., Koseki, J., Tateyama, M. and Kojima, K. (2003). Behaviors of several types of model retaining walls subjected to irregular excitation, Soils and Foundations, 43(5), 13-27.

Section 2 Dams

This page intentionally left blank

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-63

63

Pathology of foundation of Ghezala dam, a Tunisian case history Mounir BOUASSIDAa,1 H. KAROUI and Moncef BELAID b University Tunis El Manar, National Engineering School of Tunis b Ministry of Agriculture and Water Resources, Division of Dams. Tunisia. a

Abstract. Ghezala Earth Dam was built in the North West area of Tunisia, located west of Mateur City. First full filling of reservoir, of storage capacity close to hundred thousand of cubic meters, started in 1985 and lasted two years. The foundation of dam is essentially composed of a consistent clayey marl layer. At lower level of the dike’s body a pipe, made up of reinforced concrete, has been designed for water evacuation from upstream side to downstream side. The pipe structure is composed of eleven sections of 15 m length each. During dam exploration, over twenty six years, a monitoring system permitted, in particular, the measurement of consolidation settlement which occurred in the soil foundation. This follow up made possible the observation of cracks which mainly occurred in central portion of the pipe of evacuation as result of differential settlements. In this context, a 2D numerical simulation of the dam foundation was performed by using the Mohr Coulomb behavior for the dam’s material and the soil foundation. Numerical predictions of the evolution of consolidation settlement were compared to recorded ones. A good agreement between predictions and in situ records confirmed the effectiveness of adopted modeling for the studied case history.

Key words: Settlement, pipe, disorders, elastoplastic, consolidation

Introduction The control of the safety of dams consists in detecting and controlling the mechanisms of degradation which affect their structures and may lead to failure. Based on the study of behavior it is intended to recommend corrective actions aiming to the maintenance of dams in normal conditions of exploitation i.e. by limiting the evolution of mechanisms of degradation. Further a follow up of the behavior of dam during its exploitation is necessary, and then this implies the installation of an in situ set of measurement devices and apparatus: piezometers, inclinometers, settlement gauges, etc. The collection of records during time, their treatment and interpretation will enable to understand the functioning of the actual behavior of dam [1]. This paper aims to the analysis of the behavior of reinforced concrete pipe for water evacuation of Ghezala dam that has been affected by severe cracks and significant opening of joints which were accompanied by piping observed at the down stream side. The comparison between in situ recorded measurements and predicted 1

Corresponding Author: BP 37 Le Belvédère 1002, Tunis. Tunisia. [email protected]

64

M. Bouassida et al. / Pathology of Foundation of Ghezala Dam, a Tunisian Case History

results obtained by finite element computations gives a helpful insight explaining the mechanism which had occurred during the exploitation of Ghezala dam over twenty years [4], [5].

1. Presentation of Ghezala dam Ghezala dam was built from 1981 to 1984 as homogeneous earth dike founded, on compressible marl formation, of 31 m height and 560 m length at crest, the elevation of High Water Level (HWL) is 86.1 m NGT. Ghezala dam provides a storage potential of approximately 10 thousand capacity cubic meters essentially aimed to the irrigation of 900 hectares agriculture lands. The dam structure is crossed by a visiting reinforced concrete pipe, installed in 1982, of rectangular section 2 m x2.5 m and 0.6 m thickness, located at 25 m below the elevation of full reservoir. At the end of dam construction, a settlement of 200 mm has been recorded in the central part of the pipe. In four portions of 60 m length openings of separation joints (J3 to J8) and transverse cracks have been observed as schematized in figure 1. Among the ten main cracks (C1 to C10) six of them have an opening higher than 10 mm (figure 1). After first fill up of reservoir in 1985, repairing works of cracks have been executed in the pipe structure. Later on, the non stopped settlement evolution has declutched piping with estimated amount by 1 kg per week in 2003. The maximum recorded settlement has reached 455 mm under the crest section in 2008.

Figure 1. Cross section of Ghezala dam with localization of cracks in pipe structure.

2. Analysis of settlement records The evolution of settlement of Ghezala dam foundation has been followed up by means of topographic device made up of twenty five ankles installed under the base of pipe at time of construction in 1982. The topographical surveys were only recorded in altimetry. The cumulative settlement of the pipe, from 1982 to 2008, is represented in figure 2. Over twenty six years, the maximum recorded settlement is about 45 cm in the central section of pipe located at crest elevation of the dam. Settlements decrease linearly towards the upstream and downstream sides quasi proportionally with the

M. Bouassida et al. / Pathology of Foundation of Ghezala Dam, a Tunisian Case History

65

reduction of height dike. It is noted that the observed settlement over the three years after pipe installation is about 26.5 cm in 1985, and then approximately reached 32.2 cm in 1987, that is by 71% of the current maximum settlement recorded after 26 years.

Figure 2. Evolution of recorded settlement underneath the pipe of Ghezala Dam.

3. Evolution of pipe settlement in time From figure 3, the evolution of recorded settlements from 1982 to 2008 comprises three periods: first between 1982 and 1985 (1,000 days) during which the construction of dike followed by the first fill up of reservoir had induced, in the central portion of pipe, the highest rate of settlement of about 90 mm/year. The second period was between 1985 and 1987 where full fill up of reservoir had kept maintained, a moderate settlement evolution was recorded by 20 to 25 mm/year. During the third period, between 1987 and 2008, the rate of settlement was from 5 to 6 mm/year. Worth mentioning the recorded settlement, due to primary consolidation of marl compressible layer was initially unexpected. Further, the decreased trend of settlement evolution leads to say the major part of consolidation is finishing and the behavior of soil foundation of Ghezala dam will be stabilized.

66

M. Bouassida et al. / Pathology of Foundation of Ghezala Dam, a Tunisian Case History

Figure 3. Evolution of settlement versus time.

4. Numerical simulation of pipe behavior 4.1. Geotechnical conditions of Ghezala dam’s pipe After recorded settlements, during the exploitation of Ghezala dam, undrained conditions prevail to explain the observed behavior of pipe modeled as a reinforced concrete frame with rectangular section 2×2.5 m2, thickness of 0.6 m and 170 m length (figure 4). In order to simulate the behavior of pipe along the cross section of dike, that involves various loading conditions, a plane strain analysis is undertaken for three sections presented in figure 4: (A-A) at upstream side with equivalent water pressure of height 25 m; (B-B) at crest level of maximum dam elevation of 30 m and (C-C) at downstream side [2], [3].

Figure 4. Pipe cross sections for numerical computations.

M. Bouassida et al. / Pathology of Foundation of Ghezala Dam, a Tunisian Case History

67

Figure 5 shows the typical plane strain model including the rectangular cross section of the pipe overlaid by variable thickness of dike material or height of water. The width of numerical model, compared to that of pipe, was taken big enough so that zero horizontal displacement boundary condition will not significantly affect the behavior of soil around the pipe. The latter is modeled as a reinforced concrete plate element having a linear elastic behavior having normal rigidity EA and bending stiffness EI.

Figure 5. Numerical model adopted for pipe. 4.2. Geotechnical parameters of numerical model The Mohr Coulomb perfect elastoplastic behavior has been adopted both for the compressible marl layer and the clayey material of dike. Geotechnical parameters were adopted as summarized in table 1 showing three scenarios for the degree of saturation, Sr, of the marl layer. The objective is to simulate three different periods according to water infiltration under the pipe and the occasioned evolution of settlement.

Table 1. Geotechnical parameters and materials constitutive of numerical model Material Clayey Marl

Dike’s material

Sr (%) 100 90 50 --

γd(kN/m3) 16 15.4 15.35 16

γh (kN/m3) 20 19.4 17.5 20

C’(kPa) 35 30 25 30

ϕ’ (°) 25 20 20 35

E (MPa) 20 20 20 15

4.3. Mesh and boundary conditions Numerical computations are conducted by the use of half model represented in figure 5 due to the geometrical and mechanical symmetry. The 40 m width of numerical model was taken big enough, compared to pipe dimensions, so that the prediction of soil behavior around the pipe will not be affected by the condition zero horizontal displacement along vertical boundaries (figure 6). Thickness of the marl layer equals 11 m, while the earth dam layer has a variable thickness depending on the studied section. Numerical mesh is composed by 15 nodes triangular elements. The initial at rest state: K0 = 1 – sin ϕ’ is considered for all materials which are supposed normally

68

M. Bouassida et al. / Pathology of Foundation of Ghezala Dam, a Tunisian Case History

consolidated layers. The plane strain hypothesis was used by considering the primary consolidation procedure in clayey marl layer and by assuming anisotropic hydraulic conductivity for all material so that kv = 10-9 m/s and kh = 10-7 m/s.

Figure 6. Finite element mesh of studied model generated by Plaxis software.

4.4. Processed loading The embankment is not primarily activated, and then it is activated by single layers in accordance with the phases of loading. Such optimized numerical procedure reveals efficient to simulate the staged construction as it happens during embankment edification. Indeed, the construction of earth dike was modeled by ten layers built in 1090 days in conformity with the executed project. The water level at upstream side is modeled by uniformly distributed pressure on the upper side of embankment as shown in figure 7 referring to the last stage of applied loading. Two numerical computations were carried out, by using the mechanical model described above, in order to predict settlement of pipe foundation. Each simulation included two loading phases, first is the initialization of stresses, and second is the edification of earth dam.

Figure 7. Boundary condition of the plane strain numerical model.

M. Bouassida et al. / Pathology of Foundation of Ghezala Dam, a Tunisian Case History

69

4.5. Interpretation of results Recorded settlements and predicted ones from numerical computations are compared for the three sections of pipe over time in Figures 8, 9 and 10. First, it is clear the adopted numerical model overall retraces the observed behavior in term of settlement evolution which happened quasi identically in the three sections (A-A), (B-B) and (CC). Second, numerical results well confirmed the evolution of settlement had mainly occurred in the period from 100 and 1,000 days. And third, the magnitude of settlement depends on the elevation of dike material and water pressure if applies. Finally, the remaining settlement of consolidation (after 10,000 days) will not affect the pipe behavior due to the non significant observed evolution of settlement between 1,000 to 10,000 days. It is then concluded the mechanical model and geotechnical parameters used for numerical computations were representative of the observed behavior of Ghezala dam.

Figure 8. Settlement evolution of pipe under Section (A-A)

Figure 9. Settlement evolution of pipe under Section (C-C)

70

M. Bouassida et al. / Pathology of Foundation of Ghezala Dam, a Tunisian Case History

Figure 10. Settlement evolution of pipe under Section (B-B)

5. Conclusion The behavior of the foundation of Ghezala dam was simulated by using software Plaxis to explain the origin of disorders which had seriously affected the pipe of water evacuation. A suitable 2D plane strain model made it possible the simulation of observed settlements over 10,000 days underneath the pipe. The good agreement obtained between predicted and recorded settlements has led to conclude the evolution of consolidation settlement of marl foundation of the pipe tends to be stabilized.

References [1] M. Arfaoui. Préparation des fissures du plot n°4 de la galerie de prise d’eau du barrage Ghezala. Rapport n° 2. Ministère de l’agriculture et des ressources en eaux. Tunis, 2007. [2] M. Al Husein. Etude du comportement différé des sols et ouvrages géotechniques. Thèse de doctorat, Université Joseph-Grenoble; 2001. [3] E. Boidy. Modélisation du comportement différé des cavités souterraines. Thèse de doctorat, Université Joseph-Grenoble I, (2002), 35-47. [4] P. Thao. Etude en place et en laboratoire du comportement en petites déformations des sols argileux naturels. Thèse de doctorat, ENPC Paris; (2008), 7-33. [5] Zhenyu Yin. Modélisation du comportement visqueux de l’argile naturelle. XXIVemes Rencontres Universitaires de Génie Civil, (2006), 1-8.

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-71

71

Injection of Contraction Joints at Pretarouca Dam António Costa VILARa, Duarte CRUZa Teixeira Duarte, S.A, Lisbon, Portugal

a

Abstract. Within the scope of the “Pretarouca Dam Construction” Project the injection of joints’ compartments between the blocks that comprise the dam body was carried out, to ensure its water-tightness and to achieve a monolithic behavior of the entire structure. The works carried out comprised three different stages: washing and saturation of the joint compartments with water, filling up secondary circuits with bentonite (for preservation against deficient operation of non-return valves) and injection of the joints with cement grout, with special emphasis on the use of bentonite.

Keywords. Bentonite, compartment, injection, joint.

Introduction Pretarouca Dam is a concrete gravity dam, approximately 330m long and with 28,5m maximum height (Figure 1) located in the hydrographic basin of the Balsemão River, close to Pretarouca, situated approximately 10 Km from Lamêgo, Portugal.

Figure 1.Upstream view of Pretarouca dam.

The structure of the dam is composed of 20 concrete blocks, separated by 19 contraction joints and is crossed by three galleries. In accordance with the dimension and location of each joint, the joints were divided by means of a “water-stop” blade into one, two or three compartments (lower, middle or upper) which, in turn, were divided into one or two cells (upstream or downstream) (Figure 2). Each cell is fitted with an injection circuit comprising a primary and secondary circuit; the latter being a reserve circuit for future use or in case of malfunction of the primary circuit (Figure 3).

72

A.C. Vilar and D. Cruz / Injection of Contraction Joints at Pretarouca Dam

UPPER COMPARTMENT UPPER GALLERY

MIDDLE GALLERY DOWNSTREAM CELLS

MIDDLE COMPARTMENT UPSTREAM CELLS

LOWER GALLERY

LOWER COMPARTMENT

Figure 2.Contraction joint side view during construction.

Upper Compartment

Primary injection circuit Primary return circuit Secondary injection circuit Water-stop blade Injection slits Collector slits Midle Compartment

Lower Compartment

Figure 3.Contraction joint side view with scheme of the compartment’s injection circuits.

A.C. Vilar and D. Cruz / Injection of Contraction Joints at Pretarouca Dam

73

1. Procedure 1.1. Washing and Saturation of Compartments with Water Injection At the beginning of the works, injection circuits and compartments were washed by means of water injection, in order to confirm respective operating conditions and detect the existence of eventual obstructions in the circuits. The process also serves to verify the existence of possible water leaks in the compartments, to clean any debris, to moisten the compartments’ walls before the grout injection and to measure the volume of each compartment. 1.2. Grout Injection in each Joint Compartment The injection of the joints was done using the primary injection circuits, starting from the central joints to the lateral ones and from the lower compartments to the upper ones, alternating between margins. Each injection operation began with type 1/4 (C/W) grout, progressively changing into a denser grout (1/2, 1/1 and in some cases 2/1 of C/W), where “C/W” is the cement-water ratio. The injection of increasingly denser grouts only took place when it was verified that the density of the injected grout, recovered in the primary return circuits, was identical to that injected in the primary injection circuits. Whenever this situation occurred during the injection of denser grout, injection proceedings for that compartment were considered concluded.

Figure 4.Injection circuits apparatus within one of the galleries.

74

A.C. Vilar and D. Cruz / Injection of Contraction Joints at Pretarouca Dam

2. Principal problems and difficulties 2.1. Leaks in the joints compartments Detected leaks were repaired before the injection of cement grout. In cases where such compensation was not possible, it was necessary to modify the injection procedures in order to ensure proper completion of the joint compartments with cement grout. In these cases the injection was started with a denser cement grout with a water/cement ration of no less than 1/1.5 to 1/2, in order to fill the existing leaks. 2.2. Malfunction of non-return valves of the secondary circuits The deficient operation of a significant number of non-return valves of the secondary circuit, resulted in the search, as quickly as possible, for a technically reliable and economically feasible solution, in order not to endanger the achievement of compartment’s injections through the primary circuits in the appropriate time, as well as preserving the functionality of the secondary circuits for future use. The solution was based on the experience gained by the company in carrying out other work in the geotechnical field and consisted of the use of bentonite, taking advantage of its thixotropic properties, by preparing a water and bentonite solution which, when strongly mixed, presents a fluid behavior, allowing it to be injected into the deficient secondary circuits. Once inside the circuits the mixture acquires a “cake” like texture which blocks the grout entrance during the filling of the joint compartments through primary injections circuits. This process successfully prevented the entrance of cement grout into the secondary circuits, but to ensure the future usability of the secondary circuits, the bentonite still had to be removed from inside the circuits. For that purpose, a deflocculant solution was injected into those circuits.

3. Conclusions In general the procedures involved adjustments in the grouting methodology adopted from compartment to compartment, since the injection conditions are not always the same. The work was done in order to guarantee an efficient injection of all compartments and the circulation of grout through the cells of the upstream and downstream components of each compartment and thus to each joint of the dam. All works were specially aimed at the correct filling of joints with cement grout to ensure water-tightness, and to achieve a monolithic behavior of the dam structure, preserving, whenever possible, the secondary circuits for later use. The preservation of the secondary circuits was only made possible by filling them with bentonite prior to injecting grout into each compartment through the primary circuits. This was achievable thanks to the physicochemical properties of bentonite.

Section 3 Environmental Engineering

This page intentionally left blank

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-77

77

The Challenge of Designing & Constructing Steep Landfill Capping Sealing Systems using Geogrid Veneer Reinforcement a

Jörg Klompmaker a,1 and Burkard Lenze b BBG Bauberatung Geokunststoffe GmbH & Co. KG, Espelkamp-Fiestel, Germany b NAUE GmbH & Co. KG, Espelkamp-Fiestel, Germany

Abstract. Engineered capping systems for landfills are an important tool in the containment strategy for waste disposal. The principal goal of an engineered landfill capping system is to prevent or control infiltration of precipitation, thereby controlling or preventing leachate production and to control the development and emission of landfill gas, thus preventing its emission into the atmosphere. The landfill cap is a composite system of soil layers and geosynthetic components. Steeply sloped sections of capping sealing systems are permanently subjected to shear stresses which require a detailed design of the stability of the sealing system against sliding keeping in mind individual construction stages, operating conditions as well as the long aftercare period or reuse. According to experience a so called "veneer reinforcement" is required in cases where the slope of a landfill cap exceeds an inclination of approx. 20°. The geogrid absorbs the "deficit resistance", which cannot be provided by the sealing system itself due to insufficient interface friction. In this paper the characteristics of the veneer reinforcement design as well as construction details will be addressed based on two landfill case studies from South Africa and Germany. In addition to this results of a monitored landfill veneer reinforcement project will be presented, which allows the comparison of measured deformations in the reinforcement to those predicted by current design standards. Keywords. Geogrid, Steep Slopes, Design, Veneer Stability, Landfill, Capping

Introduction Landfill capping sealing systems are designed to control infiltration of rainwater into the waste and the emission of landfill gas from the waste over very long periods. The landfill cap consists of a layered system made of soil and geosynthetic components. For steep landfill slopes the layered system is permanently subjected to shear forces. During the design the stability of the sealing system needs to be analysed for the construction phase, the operation phase and for the long aftercare period or after use. The failure mechanism "Sliding" can physically be described using basic soil mechanical principles. 1 Corresponding Author: Jörg Klompmaker, BBG Bauberatung Geokunststoffe GmbH & Co. KG, P.O. Box 3025, 32332 EspelkampFiestel, Germany, E-mail: [email protected]

78

J. Klompmaker and B. Lenze / Designing & Constructing Steep Landﬁll Capping Sealing Systems

In the revised version of the GDA-Recommendations [1] the design of the stability against sliding for a landfill cap considers the partial safety factor concept, taking into account DIN 1054 [2] & DIN EN 1997: Eurocode 7 [3] design standard. Proof is provided in accordance to the limit state GEO - design approach 3 (Loss of overall stability) according to DIN EN 1997: Eurocode 7 [3]. With slope inclinations steeper than 20° the use of veneer reinforcement is generally required. In this paper specifics of such veneer reinforcement will be addressed, mainly with regard to details of the design and anchorage using the example of the landfill Duisburg-Sudamin in Germany and Holfontein Hazardous Waste Landfill in South Africa.

1. Capping Sealing System with Geogrids 1.1. Stability Analysis The design approach for the stability of a capping sealing system according to current design standards is sufficiently described in GDA E2-7 [1]. The crucial question of the stability analysis is less the method, but rather the definition of the effective shear parameters in the individual interfaces of the sealing system, which can be mobilized over the design life of the structure. This means that the verification of sufficient stability against sliding is fundamentally depending on the frictional behaviour in the individual shear planes of the sealing system and/or of the soil layers respectively. In case that the calculation of the stability exceeds the allowable degree of utilization μ (μ = ratio of driving vs. resisting forces) with μ > 1.0, veneer reinforcement is required, which absorbs the deficit shear resistance in an anchor trench at the slope crest. The procedure for designing the veneer reinforcement and the load distribution within the anchor trench is described in detail in EBGEO (2010) [4] and will not further be discussed in this paper. 1.2. Design of Veneer Reinforcement Geogrids are made of polymer raw materials, which possess elasto-plastic behaviour. Under constant loading not only elastic (short-term) deformations, but also viscous; time-dependent deformations (creep) takes place. This leads to construction relevant consequences: - Limited load capacity and/or - Larger strains compared to short-term behaviour. The limited load carrying capacity can lead to a failure (creep-rupture failure) and increasing strains (creep-strain) can cause inadmissible deformations in the sealing system. In the design stage both influences can be taken into consideration using product specific results of creep rupture- and tensile creep tests according to EN ISO 13431 [5]. For a design in accordance to Eurocode 7, creep rupture and tensile creep are relevant for the examination of the limit state GEO (loss of stability of structure) and the limit state STR (excessive deformation of the structure or structural elements, like e.g. geogrid). The creep behaviour of a particular geogrid is thereby depending on

J. Klompmaker and B. Lenze / Designing & Constructing Steep Landﬁll Capping Sealing Systems

79

the type of polymer, the manufacturing technique, the amount of tensile load, the loading time and the temperature. 1.3. Design of Anchor Trench To allow for the absorption of shear stresses in the slope area the veneer reinforcement must be anchored at the slope crest. The dimensions of an anchorage trench need to be designed sufficiently big to prevent the veneer reinforcement from being pulled out and secondly to prevent a shear failure along the base of the anchor trench. The limit state equations together with the decisive partial safety factors are described in detail in EBGEO (2010) [4].

2. Case Study – Landfill Duisburg-Sudamin, Germany The landfill Duisburg-Sudamin is an old factory landfill of the smelter Duisburg AG (MHD). The landfill is filled with slag as by product of former zinc production. The slag contained large amounts of heavy metal. The restoration of the landfill was integrated into the development of a public park (Angerpark). The total surface of the landfill to be covered is approx. 140,000 m². The lower slopes of the truncated, pyramid- shaped landfill were constructed with an inclination of 1(V):2(H) and a maximum slope length of approx. 40 m. A restoration soil layer thickness of 1.60 m was installed on top of the sealing system. Figure 1 shows a typical cross section of the sealing system of the landfill cap.

Figure 1. Typical Cross Section of Capping Sealing System (Malakou 2009 [6])

Amongst other aspects, the challenge in the project was the anchorage concept for the veneer reinforcement under consideration of the following aspects: 1. Adherence to slope line during installation of the veneer reinforcement in the inner and outer curves of the slope sections: •

Precise planning of the geogrid installation to reduce potential overlaps, to maintain optimum soil/geogrid interaction.

•

Construction of anchor trenches in the shoulder area between two slope sections to prevent multiple overlaps.

80

J. Klompmaker and B. Lenze / Designing & Constructing Steep Landﬁll Capping Sealing Systems

2. A continuous installation of the geogrid across intermediate berms was prohibited, as well as a continuous installation of the sealing system through anchor trenches is generally not desirable (for an optimum soil/geogrid interaction and negligence of drainage inside the anchor trench). With the installation of the anchor trench in the berm (between two slope sections) connection between the upper and the lower sealing system is to be accomplished. •

In the design stage, the stability of the sealing system in the construction phase (construction equipment on the slope) must take into consideration the filled trench, but not the additional load from the top soil layer [7].

To be able to record the deformations of the sealing system a monitoring system was installed. This system consists of strain gauges which were applied to the installed laid and welded PET geogrid veneer reinforcement in defined distances along the slope. The measurements are taken at regular intervals and are evaluated to allow for conclusions to be drawn from the measured tensile stresses in the veneer reinforcement under installed conditions. Creep monitoring of the geosynthetic soil veneer reinforcement has shown much lower strains in the geogrid (approx. 0.3%) compared to predictions from design. One of the main reason for this is the fact that creep tests according to EN ISO 13431 [4] are carried out as so called "in air" tests, whereas the geogrid under realsitic conditions is confined by the cover soil layer. It can be concluded that creep deformations are very much overestimated in the current design procedures, which results in additional safety reserves.

3. Case Study – Holfontein Hazardous Waste Landfill, South Africa The Enviroserv Holfontein hazardous waste landfill site is situated approximately 80 km east of Johannesburg in South Africa and is one of only three landfills in the country that carries the so called "H:H" hazard rating in terms of South Africa’s „Minimum Requirements“ document [8]. This means that it is capable of accepting all categories of hazardous waste, with the exception of radioactive waste. At the site, specific types of hazardous waste are encapsulated in concrete to isolate them from the environment. The encapsulation facility that required closure and capping consists of a multi-tiered concrete structure in which barrels of highly hazardous liquid and solid waste have been encased in reinforced concrete. However, no standards exist for the closure and rehabilitation of encapsulation cells in the Department of Water Affairs (DWAF) Minimum Requirements (MR) specification. In determining the possible requirements of a closure capping design, Jones & Wagener [9] took an in-depth look at the long-term integrity of typical landfill capping designs and found them to be unsuitable for the specific case. Most capping designs aim to limit the ingress of moisture in the form of rainfall. However in this particular case, exposure to CO 2, O2 and water would result in the deterioration of the encapsulating concrete through depassivation of the reinforcing steel, resulting in corrosion and eventual spalling of the concrete. Hence the selected capping design would have to be not only “waterproof” but “gasproof” as well. The selected capping components are illustrated in Figure 2.

J. Klompmaker and B. Lenze / Designing & Constructing Steep Landﬁll Capping Sealing Systems

81

Figure 2. Typical Cross Section of Capping Sealing System (Johns et al. 2007)

Due to the slope length (20 m) and gradient 1(V):3(H) of the completed capping, concerns arose regarding the possibility of a veneer or infinite slope failure at the interface of the topsoil layer and the geomembrane. Stability analyses were conducted which analyzed the construction phase as well as the long-term "end" state of the slope. The analysis resulted in an insufficient stability of the 500 mm thick cover soil layer on top the HDPE geomembrane in the construction phase because of the additional static and cyclic loads imposed on the system by construction machinery. In the case of the Holfontein landfill project the geogrid together with the geomembrane was not anchored in an anchor trench at the slope crest, as usually done. This was done in a different way as the HDPE geomembrane had been draped over the whole structure of the landfill, which had a symmetrical shape. As a result of this the loads on the membrane on one side were balanced by those on the other, thus eliminating the need for an anchor trench. The same configuration was to be used for the geogrid. However, the geogrid, unlike the geomembrane did require anchorage at the crest of the slope because of the topsoil placement method, in which the covering topsoil was placed from the bottom of the slope up. This would tend to pull an unanchored geogrid down the slope. Therefore the geogrid had to be anchored at the crest of the slope before installation of the soil on the slopes could begin. This was achieved by placing the full depth of topsoil on the structure top surface. This in itself presented a problem: construction machinery could not be allowed to drive on the surface of the HDPE geomembrane because this would result in severe damage to the membrane. Therefore the construction machinery had to build a road in front of itself to ensure that it was always driving on the topsoil and not on the membrane. The development of this is shown in Figure 3.

Figure 3. Placing of Anchoring Soil on top of Cell

82

J. Klompmaker and B. Lenze / Designing & Constructing Steep Landﬁll Capping Sealing Systems

After the topsoil had been placed on the top of the encapsulation cell, the pushing up of soil in the side slopes could begin. Once all the covering soil had been placed, labor intensive methods, using laborers from the local community, were used to install "sausage" drains along each side slope. The purpose of these drains is to intercept and quickly remove rainwater seepage on the geomembrane/soil interface in the slope, thereby reducing the build-up of seepage stresses. Labor intensive methods were required because excavation for the drains by machinery would very likely have resulted in damage to the geomembrane. Finally downchutes were constructed on the slopes to rapidly channel storm water off the capping and preventing erosion.

4. Conclusion Geogrids, which are used as veneer reinforcement, represent an exceptional position within the sealing system of landfills, as they are not initially planned as integral part of the sealing system against infiltration of rainwater into the landfill, but rather integrated as statically required construction element to sustain the stability of the complete system. The design has to be carried out in such a way, that no relative movements and thus tensile stresses can be transferred into the sealing system components, like e.g. the geomembrane or a drainage geocomposite. Therefore geogrids with high creep resistance, high robustness and high tensile modulus should be chosen. For this purpose detailed design has to be provided. A precise planning of the geogrid veneer reinforcement installation is generally recommendable. Prior to the planning of the geogrid installation an anchorage concept on the basis of a detailed stability analysis for all relevant sections and construction phases has to be established. Decisive for a stable landfill capping sealing system with high durability under difficult geometrical boundary conditions is the quality of the construction and the professional handling of the geosynthetic components under consideration of given specifications for installation, installation planning, overlaps, joints, soil coverage, etc.

References [1] DGGT e.V., Empfehlungen des Arbeitskreises „Geotechnik der Deponiebauwerke“, E2-7 „Standsicherheit von Dichtungssystemen“, Entwurf, Bautechnik 9/2008 [2] DIN 1054: Baugrund - Sicherheitsnachweise im Erd- und Grundbau. Normenausschuss Bauwesen (NABau) im DIN. DIN Deutsches Institut für Normung e.V. 01/2005. [3] EN 1997-1:2004: Eurocode 7: Geotechnical Design – Part 1: General Rules [4] EBGEO (2010): Empfehlungen für Bewehrungen aus Geokunststoffen der Deutschen Gesellschaft für Geotechnik e.V., Verlag Ernst & Sohn, Berlin [5] EN ISO 13431: 1999-11 Geotextilies and geotextile- related products: Determination of the tensile creep and creep rupture behaviour [6] Malakou, E. (2009): Endgültiger Abschluss der MHD-Deponie Duisburg-Sudamin. Vortrag und Tagungsband. 6. NAUE Geokunststoffkolloquium, Bad Wildungen, 2009 [7] Werth, K.; Witolla, C. (2008): Standsicherheit von Oberflächendichtungssystemen – geotechnische und hydraulische Interaktion von Rekultivierungsschicht und geosynthetischen Drän- und Filterschichten – Bemessung, Ausführung und Qualitätssicherung. Tagungsband zur 24. SKZ-Tagung „Die sichere Deponie“, Süddeutsches Kunststoffzentrum, Würzburg, 2008 [8] Department of Water Affairs And Forestry, Minimum Requirements for Waste Disposal by Landfill, Second Edition 1998 [9] Johns, D.; Shamrock, J., Improved Slope Stability of the Closure capping System, of a Hazardous Waste Encapsulation Cell, 2nd Young Geotechnical Engineers' Conference (YGEC), Tunisia 2007

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-83

83

Design of Soil Covers in Tropical Africa: A Perspective a

Celestina ALLOTEYa,1 and Nii Kwashie ALLOTEY b Environment and Community Affairs, Anglogold Ashanti Ghana, Accra, Ghana b Civil Analysis Group, Atomic Energy of Canada, Mississauga, ON, Canada

Abstract. Soil cover systems are widely used for the containment of waste with the objective of limiting the ingress of precipitation and oxygen, and also providing a medium for vegetation growth. The positive aspect of soil covers is that they are a natural resource that supports vegetation used in land reclamation. In designing effective soil covers for the African region, it is important that the wide range of varying climatic conditions be taken into account in developing designs that are tailored to fit the local conditions. This paper argues the point that there are benefits for cover design in Africa to be based on a formalized design methodology, rather than based on generic designs Keywords. soil covers, Africa, design methodology, environmental management.

Introduction Soil or dry cover systems are widely used for the containment of waste (i.e., municipal, hazardous and mining), with the main objective of limiting the ingress of water and/or oxygen, and also providing a medium for vegetation growth. The good thing about soil covers is that they prevent groundwater contamination using a natural resource, and restore the land by supporting vegetation; they are, thus, an environmentally-friendly contamination mitigation option. In most African countries, groundwater is, or is becoming a major source of drinking water for the population, it is therefore of utmost importance that the contamination of the groundwater resource be avoided. To ensure this, the disposal of all forms of waste should be in properly engineered waste containment facilities. Unfortunately, this is not the case in most African countries. However, with promotion of the need for good environmental management, there is a growing awareness in these countries of the socio-economic, health and societal benefits of pursuing such programs. To help sustain this drive, it is important that waste containment systems that play the crucial role of being the final “resting place” of waste, be designed to effectively fulfill their function in the most cost-effective and sustainable manner. In this regard, dry covers are known to be an effective waste cover method. This paper presents a brief overview of soil covers, and then discusses best-practice cover design approaches that would benefit African countries, if used. The discussion presented is in regard to all forms of waste, however, reference is made more to that of mine waste. 1

Corresponding Author: Environment and Closure Manager, Anglogold Ashanti Ghana, PO Box 2665, Accra, Ghana; Email: [email protected]. Formerly called Celestina Adu-Wusu.

84

C. Allotey and N.K. Allotey / Design of Soil Covers in Tropical Africa: A Perspective

1. Soil Covers A wide range of covers have been designed and constructed all over the world. For a soil cover to be effective, there must be a good understanding of its objective, and the critical factors that affect its performance. Generally, the design and performance objectives for a cover include: limiting infiltration, controlling oxygen ingress, resisting erosion, supporting vegetation growth, in addition to exhibiting good long-term performance. Different types of covers can be used, however, the specifics of the cover are dictated by the type of waste covered, the prevailing environment conditions (i.e., climate), and the governing jurisdictional regulations. Soil covers may be constructed from a different combination of earthen and other geo-materials. Each material type is chosen based on its unique geotechnical properties to function either as a vegetation support layer, erosion resistant layer, percolation control layer, moisture retention layer, or a foundation layer. Among others, the main types of soil covers are: resistive barriers, capillary barriers, and store and release/water balance covers, also referred to as monolithic or evapo-transpiration barriers. • Resistive barriers: A “resistive barrier” also termed a conventional barrier comprises soil with low hydraulic conductivity that resists the percolation of water to the underlying waste material [1, 2]. Commonly used resistive barriers are compacted clay layers, geosynthetic clay liners (GCL) and geomembranes combined with a soil layer. • Capillary barriers: This cover comprises a fine-grained soil overlying/underlying a coarse-grained soil, thus creating a “capillary barrier” effect [3-5]. Drainage/evaporation from the fine-grained layer is prevented as a result of the lower hydraulic conductivity of the coarse-grained layer under unsaturated conditions. The coarse-grained layers lose water quickly due to an initially higher saturated hydraulic conductivity, however as a result of their large pore sizes, their ability to hold water by capillary action in the unsaturated zone is limited. This reduction in water content results in a decrease in the saturated hydraulic conductivity, which consequently reduces its ability to transmit significant amounts of water, thereby preventing water from draining/evaporating from the compacted fine-grained layer. Capillary barriers can be as simple as two layers of contrasting particle size, or multiple layers of fine-grained and coarse-grained soils. • Monolithic/Evapo-transpiration barriers: These are also referred to as store and release, or water balance covers: In arid and semi-arid climates, maintaining cover saturation might not be feasible and could be expensive. Soil covers are thus designed to prevent moisture movement into the waste material, rather than to act as an oxygen barrier. This is achieved by using a well-graded, homogeneous upper surface layer, having enough storage capacity to retain water during rainfall events. This water is released through evapo-transpiration during dry periods thus preventing percolation into the underlying waste materials. Monolithic barriers require limited compactive effort in comparison to conventional resistive barriers. To prevent percolation it is important to ensure that the soil water capacity of the material used in the cover is more than its water storage. Soil water storage capacity depends on layer thickness, unsaturated hydraulic properties and soil layering. It may therefore be necessary to use a very thick soil layer to increase water storage capacity. Also, the vegetation type for such a cover must be well selected in

C. Allotey and N.K. Allotey / Design of Soil Covers in Tropical Africa: A Perspective

•

•

85

order to ensure optimal transpiration, which is affected by among others: rooting depth and distribution, percent cover and leaf area, water uptake by the roots, and the composition of the plant community [6, 7]. Oxygen barrier: Since Acid Rock Drainage (ARD) develops in the presence of oxygen, most mine waste covers are designed to function as oxygen barriers. The diffusion coefficient of oxygen in water is about 10 4 times that in air, prevention of oxygen ingress is therefore achieved by ensuring that the barrier layer remains permanently saturated. This prevents the diffusive movement of oxygen into the waste rock, and hence limits the development of ARD [8]. Other barrier types: Geochemical covers (e.g., hardpan formed via chemical precipitation), barriers made using atypical materials like asphalt, and barriers constructed with unusual soils (e.g., residual and lateritic soils).

2. Cover Design Methodology 2.1. Established Framework Based on earlier work by Yanful and Lin [9] and Wels and O’Kane [10], Rykaart and Caldwell [11] have recently presented a formalized approach for the design of soil covers, which is presented in Figure 1.

Figure 1. Flow chart for soil cover design methodology.

86

C. Allotey and N.K. Allotey / Design of Soil Covers in Tropical Africa: A Perspective

The methodology comprises twelve steps, starting from the definition of the cover function up to long-term cover maintenance. The approach relies on the use of test covers and numerical predictive modeling to aid in the choice of a suitable cover; it then bases the prediction of long-term performance of the cover, on the results of a calibrated numerical model. Furthermore, the design approach includes performance monitoring with numerical model re-calibration, in combination with long-term cover maintenance. Not many covers, worldwide, have successfully followed the approach. Some that have done so are Barrick Gold’s Les Terrains Aurifères mine in Quebec [12, 13], and Inco’s Whistle Mine in Ontario [14-16]. There are a number of cover designs where steps 1-9 have been implemented, but with no long-term monitoring done [11]. There are also others that totally skip the issue of test plots and base their design solely on past experience, or on the results of previous case studies, i.e., skip steps 5-7 [5]. The literature, on the other hand, includes case studies of pilot experimental projects, that seem to have been research-based, and did not get to the full-scale cover construction stage, e.g. Heath Steele Mine [17]. From the literature, it can be noted that most cover designs that have followed significant parts of the formalized cover design methodology are located in temperate zones. A few exceptions are the Questa coal mine in Indonesia [18], opencast coal mines in South Africa [19], and recently, Santa Catarina coal mine sites in Southern Brazil [20, 21]. The cover design methodology is a comprehensive list of design steps that seeks to encourage the use of locally available materials, ensure proof of concept and, also, provide for good long-term cover performance. Designs that follow the entire plan provide gained experience, which then becomes helpful in the design of other covers in similar areas. For example, performance monitoring of test covers at Whistle Mine by Adu-Wusu et al [14], showed that the GCL test cover worked very well in the first year, but performed poorly in the following years. Post-closure studies showed that GCL degeneration as a result of severe weather conditions and installation errors destroyed the ability of the GCL to self-heal, resulting in a marked increase in its hydraulic conductivity. This would not have been identified if design ended at the cover construction stage, i.e., at step 9. Critical to the cover design methodology is the use of predictive numerical modeling. Many computer codes, both one-dimensional and two-dimensional are currently available (see Bohnhoff et al. [7] for a list of different programs). Among others, Bohnhoff et al. [7] and Adu-Wusu et al. [15] have obtained generally satisfactory water-balance predictions with codes such as VADOSE/W [22]. Appropriate numerical predictive modeling tools are therefore currently available, and this aspect of the design process should not serve as a hindrance to the approach being adopted in practice. 2.2. Implementation in Africa In some Sub-Saharan African countries, where the climate in a particular country could vary from wet to semi-arid, it is important that soil covers are designed to meet specific performance goals, and not just based on generic designs. Not many cover designs in Sub-Saharan Africa have followed the formalized soil cover design methodology. This is unfortunate since the approach encourages design innovation with the use of locally available materials. For example, at the Santa Catarina coal mine site in Southern

C. Allotey and N.K. Allotey / Design of Soil Covers in Tropical Africa: A Perspective

87

Brazil, pilot tests have shown a soil cover comprising compacted clay with upper and lower layers of bottom ash acting as a capillary barrier, to be a suitable cover. Bottom Ash is a local material obtained from thermoelectric plants in the area; its ready availability makes the choice of such a cover a cost-effective option 2. It is the authors’ belief that the established cover design methodology provides a framework within which innovative area-specific designs can be realized in Africa. It provides an opportunity to assess the proof of concept of designs involving locallyavailable materials (e.g., wood chippings, sawdust and palm nut kernels); when proven, this goes to further add value to the worth of these locally available materials. As an example, wood chippings could be used in combination with soil to create a capillary barrier. This will certainly be less costly than using a geosynthetic material, which in most African countries would have to be imported. Other possible innovative cover ideas are the use of a simple vegetated thick layer of coarse soil as a cover over mine waste in arid regions. Such a cover will not prevent oxygen from entering the waste rock, however, if there is no water to carry the contaminants to the groundwater table, this would not be a problem. Such a design concept can be fully assessed within the framework of the cover design methodology. Furthermore, the methodology provides a framework in which the effectiveness of unproven cover designs can be assessed. The authors are aware that regulators in some African countries prescribe the use of laterite as a final cover for mine waste rock. Such prescribed cover designs are mostly unproven, and could be adequately assessed with the methodology. Lastly, under conditions where conventional covers are not suitable due to the lack of sophisticated heavy machinery, the cover design approach can be followed in the development of an alternative cover that requires the use of less compactive effort.

3. Conclusions Soil or dry cover systems are one of the options available for covering waste, be it municipal, hazardous or mine waste. They are designed with the main purpose of controlling percolation, retaining moisture, limiting erosion or to support vegetative growth. Soil cover design has recently been formalized into a twelve-step process. The authors believe that following this prescribed approach allows for design innovation, which results in cover designs that are both cost-effective and reliable. In the African context, this is beneficial and would contribute towards sustainable environmental management.

2

An anonymous reviewer of the draft manuscript noted that in South Africa Bottom Ash is not used in soil covers since they typically have high metal concentrations, leading to ground water contamination. For the Santa Catarina pilot test covers, it is noteworthy to mention that three and a half years of monitoring has shown heavy metal concentrations in collected water to be less than corresponding allowable limits.

88

C. Allotey and N.K. Allotey / Design of Soil Covers in Tropical Africa: A Perspective

References [1] [2] [3] [4] [5] [6] [7]

[8] [9] [10]

[11] [12] [13] [14] [15] [16] [17]

[18] [19]

[20]

[21]

[22]

R. Shulz, R. Robert and E. O’Donnell, Control of water infiltration into near LLW disposal units. US Regulatory Nuclear Commission Report No. NUREG/CR 4918, 3, Washington DC, 1998. C. Benson, P.J. Bosscher, D.T. Lane and R.J. Pliska, Monitoring systems for hydrologic evaluation of landfill covers, Geotechnical Testing Journal, 17(2) (2004), 138-149. S.L. Barbour, Reduction of acid generation in mine tailings through the use of moisture retaining cover layers as oxygen barriers, Discussion, Canadian Geotechnical Journal, 27 (1990), 398-401. M. Fayer, M. Rockhold and M. Campbell, Hydrologic modeling of protective barriers: comparison of field data and simulation results, Soil Science Society of American Journal, 56(3) (1992), 690-700. M. Khire, C. Benson and P.J. Bosscher, Capillary barriers: design variables and water balance, ASCE Journal of Geotechnical and Geoenvironmental Engineering, 126(8) (2000), 695-708. P.E. Mcquire, B.J. Adraski and R.E. Archibald, Case study of a full-scale evapo-transpiration cover, ASCE Journal of Geotechnical and Geoenvironmental Engineering, 135(3) (2009), 316-332. G.L. Bohnhoff, A.S. Orgozalek, C. Benson, C.D Shackelford and P. Apiwatrangoon, Field data and water balance predictions for a monolithic cover in a semi-arid climate, ASCE Journal of Geotechnical and Geoenvironmental Engineering, 135(3) (2009), 333-348. E.K. Yanful, Oxygen diffusion through soil covers on suphidic mine tailings, ASCE Journal of Geotechnical Engineering, 119(8) (1993), 1207-1228. E.K. Yanful and M. Lin, An integrated approach to designing soil covers for reactive mine waste, Proc. of 51st Canadian Geotechnical Conference, Edmonton, Alberta, I, 141-148, 1998. C. Wels and M. O’Kane, Design of mine waste cover systems, linking predicted performance to groundwater and surface water impacts, 2003, Available at http://www.robertsongeoconsultants.com /hydromine/ (accessed 15 June 2009). M. Rykaart and J. Caldwell, Covers: state of the art review, 2006, Available at: http:// technology.infomine.com/covers/ (accessed 24 November 2010). MEnd Project 2.22.4a, Construction and instrumentation of a multi-layer cover; Les Terrains Aurifères, prepared by Golder Associés, December 1998. MEnd Project 2.22.4b, Field Performance of a multi-layer cover; Les Terrains Aurifères, prepared by Golder Associés, March 1999. C. Adu-Wusu and E.K. Yanful, Performance of engineered test covers on acid-generating waste rock at Whistle Mine, Ontario, Canadian Geotechnical Journal, 43 (2006), 1-18. C. Adu-Wusu, E.K. Yanful, L. Lanteigne and M. O’Kane, Prediction of the water balance of two soil cover systems, Geotechnical and Geological Engineering, 25 (2007), 215-237. C. Adu-Wusu and E.K. Yanful, Post-closure investigation of engineered test covers on acid- generating waste rock at Whistle Mine, Ontario, Canadian Geotechnical Journal, 44 (2007), 496-506. E.K. Yanful, A.V. Bell and M.R. Woyshner, Design of a composite soil cover for an experimental waste rock pile near Newcastle, New Brunswick, Canada, Canadian Geotechnical Journal, 30 (1993), 578-587. C. Wels, S. Fortin and S. Loudon, Assessment of store-and-release cover for Questa Tailings Facility, New Mexico, Proc. of Tailings and Mine Waste, Fort Collins, Colorado,459-468, 2002. J.A. Wates, J.J. Vermaak and N. Bezuidenhout, Design of soil covers to reduce infiltration on rehabilitated opencast coal mines, Geotechnics for Developing Africa, eds., Wardle, Blight and Fourie, Balkemaa, Rotterdam, 109-119, 1999. A.B. Soares, M.O. Ubaldo, V.P. de Souza, P.S.M Soares, M.C. Barbosa and R.M.G Mendonca, Design of a dry cover pilot test for acid mine drainage abatement in Southern Brazil Part I: material characterization and numerical modeling, Mine Water and the Environment, 28 (2009), 219-231. A.B. Soares, M.V. Possa, V.P. de Souza, P.S.M Soares, M.C. Barbosa, M.O. Ubaldo, A.V. Bertolino and L.S. Borma, Design of a dry cover pilot test for acid mine drainage abatement in Southern Brazil Part II: pilot unit construction and initial monitoring, Mine Water and the Environment, 29 (2010), 277284. J. Krahn, Vadose zone modeling with VADOSE/W, GEO-SLOPE Int., Calgary, Alberta, Canada, 2004.

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-89

89

Is there a future for GCLs in Waste Barrier Systems? Peter LEGG a and Molly McLENNAN b Peter Legg Consulting, Benoni, South Africa b Golder Associates Africa, Johannesburg, South Africa a

Abstract. Geosynthetic clay liners (GCLs) are commonly used in geocomposite waste barrier systems as replacement for compacted clay liners. In recent years a number of problems associated with GCLs have been identified including panel shrinkage, desiccation cracking, chemical incompatibility, cation exchange and lack of hydration in geomembrane/GCL composite liner systems. The most important factor affecting the performance of a GCL is cation exchange in the bentonite. Leachate compatibility testing of the GCL is therefore necessary, initially by swell index testing, followed by hydraulic conductivity testing if the swell index test indicates promising results. The results of leachate GCL compatibility testing performed for four waste disposal facilities are presented and their influence on the respective barrier designs. It is concluded that there is a future for the use of GCLs in waste barrier systems provided that the compatibility of the GCL is confirmed through testing Keywords. GCL, leachate compatibility, cation exchange, swell index, hydraulic conductivity, barrier

Introduction Geosynthetic clay liners (GCLs) were first developed in the United States in the mid 1980s, and high quality GCLs were introduced into South Africa in the mid 1990s as an alternative for compacted clay liners (CCLs) in waste containment barrier systems. Because of their initial high cost and some scepticism in the industry, the use of GCLs in engineered barriers was very limited initially. However, with increased competition, prices reduced and the use of GCLs as replacements for CCLs has increased significantly in recent years, both in bottom liners and landfill capping systems. The main benefits of a GCL over a CCL include superior performance in terms of hydraulic conductivity, ease of installation, and increased landfill airspace due to the relative thickness of a GCL compared with an equivalent CCL. Typically a GCL consists of a thin layer of sodium bentonite clay sandwiched between two geotextiles which are needle-punched together to form an intact monolithic sheet, usually less than 10 mm thick. The bentonite provides the low permeability barrier property of the GCL while the geotextiles provide the physical strength of the GCL. When the bentonite within a GCL is hydrated, the interlayer regions of the montmorillonite particles expand due to repulsion between the polar water molecules and the charged clay platelets. This interlayer expansion manifests as swelling of the bentonite on the macroscopic scale. The interlayer adsorbed water is tightly bound and very little space is available for flow of free water which results in a

90

P. Legg and M. McLennan / Is There a Future for GCLs in Waste Barrier Systems?

low hydraulic conductivity of the hydrated bentonite. The swelling property of the sodium bentonite when hydrated provides the GCL with its low hydraulic conductivity and self-healing properties. Sodium bentonite typically consists of 50-90% Na +, 5-20% Ca2+, 3-15% Mg2+ and 0.1-0.5% K+ (Egloffstein 2001)[1]. Under ideal conditions with water as a permeant, a GCL can achieve a saturated hydraulic conductivity of less than 10-9 cm/s.

1. South African Landfill Barrier Requirements The regulatory standards for landfill design in South Africa are contained within the Department of Water Affairs & Forestry’s “Minimum Requirements” documents that set out graded minimum standards for landfill design based on the classification of the landfill. In the first edition of the “Minimum Requirements” (1994), GCLs were not even mentioned. In the second edition (1998)[2], GCLs were allowed as replacements for CCLs, based on equivalent performance. In the draft third edition of the “Minimum Requirements” (2005), GCLs have been specified for use in landfill covers. In the case of general waste landfills, the prescribed bottom liner consists of compacted clay liners only without any geomembrane, and the maximum permissible hydraulic conductivity of the compacted clay liner is 1x10-6 cm/s. For hazardous waste landfills, the compacted clay liner acts as the support component to the geomembrane in a geocomposite liner, and the maximum permissible hydraulic conductivity of the compacted clay component reduces to between 1x10-7 cm/s and 3x10-7 cm/s, dependent on the nature of the hazardous wastes contained. In the case of landfill covers, there is no maximum permissible hydraulic conductivity prescribed, however the GCL would be expected to minimise the amount of rainwater percolating through the cover into the waste body to the same degree as a compacted clay cover would.

2. Problems associated with GCLs in Barrier Systems In recent years, a number of problems associated with the use of GCLs in barrier systems have been identified, that adversely affect the hydraulic conductivity performance of the barrier. These include panel shrinkage, desiccation cracking, chemical incompatibility, cation exchange and lack of hydration in geomembrane/GCL composite liner systems. Various papers highlighting these problems have been published in recent years, leading to a reticence by many designers to specify GCLs in landfill barrier systems. Panel shrinkage of GCLs resulting in separation of adjacent GCL panels has been observed in geocomposite liner installations where the GCL was overlain by an HDPE geomembrane which was left exposed for durations ranging from two months to five years with no soil cover (Thiel et al, 2006)[3]. The shrinkage of the GCL panels was the result of cyclic changes of temperature and consequent desiccation and shrinkage of the bentonite in the GCL. Several reported case histories have found that the hydraulic conductivity of GCLs can increase by several orders of magnitude if cation exchange is combined with desiccation and cracking of the bentonite (Melchior 2002; Lin and Benson, 2000)[4, 5]. Typically this phenomenon is more common in landfill covers as a result of seasonal fluctuations of temperature and moisture content (Meer and Benson, 2007)[6], however

P. Legg and M. McLennan / Is There a Future for GCLs in Waste Barrier Systems?

91

the potential for desiccation of the GCL has been identified in the bottom liner of landfills (municipal solid waste and industrial waste) due to the heat produced by waste decomposition or chemical reaction (Southen and Rowe, 2005)[7]. Chemical incompatibility problems have occurred where a GCL has been used beneath a geomembrane in a geocomposite liner for chemical waste impoundments. In particular, where the contained waste (or leachate) is highly acidic or highly alkali, in the event of a leak in the geomembrane, the leachate breaks down the bentonite within the GCL without enabling any degree of hydration. In certain geocomposite lining systems, the GCL is sandwiched between two geomembranes together with a geosynthetic leak drainage layer. Unless the GCL is prehydrated before installation of the top geomembrane, the GCL remains unhydrated until such time as a leak occurs in the overlying geomembrane. Often the quality of the leachate is such that effective hydration of the bentonite cannot occur and the performance of the GCL is compromised. The most important factor affecting the performance of a GCL is cation exchange in the bentonite. Literature (e.g. Jo et al, 2004)[8] shows that when a GCL comes into contact with solutions containing a significant concentration of divalent cations (such as Ca2+ and Mg2+), the hydraulic conductivity of the GCL could increase significantly as a result of cation exchange of the monovalent Na+ cations by the divalent Ca2+ and Mg2+ cations. The physical mechanism that causes these changes is a reduction of the thickness and related absorption capacity of the diffuse double layer of water molecules surrounding the clay minerals. This results in an effective decrease in the volume of the clay, since the water molecules are not attracted to the clay particles. The magnitude of the increase in hydraulic conductivity depends on a number of factors among which the following are the critical ones: • Concentration of the divalent cations in the leachate or surrounding soil • Pre-hydration of the GCL • Confining pressure on the GCL Under ideal conditions, the GCL would normally provide a saturated hydraulic conductivity of less than 10-9 cm/s, however through Ca2+ cation exchange, there can be an increase in hydraulic conductivity of at least two orders of magnitude to 1x10 -7 cm/s. It is therefore vitally important to assess the compatibility of the GCL with the contained leachate or surrounding soil before a decision is made to incorporate a GCL into a landfill barrier system (bottom liner or cover).

3. GCL / Leachate Compatibility Testing There is extensive work reported in the literature on this topic. The most common method of determining the long-term compatibility of a GCL in its complete form is by subjecting it to hydraulic conductivity testing using the expected or actual leachate as the permeant. Factors that have been investigated include the effects of both weak and strong inorganic solutions containing monovalent and divalent ions, the effects of strongly acidic and alkali solutions, and the effect of pre-hydration (Johns and Shamrock, 2009)[9]. Long term hydraulic conductivity testing of a GCL subjected to various solutions containing divalent cations such as Ca2+, performed by Jo et al (2005)[10], indicated that it can take a very long time (years) and many pore volumes of permeant flow to achieve chemical equilibrium. Johns[9] has therefore proposed the

92

P. Legg and M. McLennan / Is There a Future for GCLs in Waste Barrier Systems?

use of swell index testing of the bentonite within the GCL as a far quicker indicator test for leachate / GCL compatibility, prior to conducting long term hydraulic conductivity tests. Based on the literature (Kolstad et al, 2004; Lee et al, 2005)[11, 12], Johns[9] used the correlation between swell index and hydraulic conductivity to rapidly assess the compatibility of several bentonites with various leachates. The results of the swell index tests were used to either reject the use of a GCL in certain applications, to call for further hydraulic conductivity testing in marginal cases, or to proceed with the use of a GCL without further testing.

4. Case Histories The authors have recently been involved with the assessment and/or design of the barrier systems for four industrial waste disposal facilities. These case histories are presented to illustrate the importance and effectiveness of GCL leachate compatibility testing when designing the appropriate barrier system for a waste disposal facility. 4.1. Manganese residue facility The liner at a manganese residue facility near Nelspruit consists of a double geocomposite lining system, with a geosynthetic clay liner (GCL) beneath both the primary and secondary HDPE geomembrane liners. No leachate compatibility testing was carried out at design stage and the GCLs were not prehydrated at the time of installation. During the recent assessment of the barrier system for the disposal of brine and gypsum on the landfill, concern was expressed over the performance of the GCLs in the barrier system when subjected to calcium rich leachate which could result in cation exchange and a resultant reduction in the performance of the GCL. Considering the much higher concentrations of other divalent cations in the manganese larox waste already being deposited on the landfill, it was decided to carry out laboratory hydraulic conductivity testing on samples of the installed GCL recovered from the landfill. A simple “free swell” test on bentonite recovered from the GCL indicated insignificant hydration of the bentonite when subjected to manganese larox filtrate immersion. Flexible wall permeameter testing was carried out in accordance with test method ASTM D 5084 “Standard Test Method for Measurement of Hydraulic Conductivity of Saturated Porous Materials using a Flexible Wall Permeameter”. Five scenarios were tested with the following results (see Table 1): Table 1. Flexible wall permeameter test results – Manganese residue facility GCL GCL hydration, permeant Unhydrated GCL + tap water Unhydrated GCL + larox filtrate Pre-hydrated GCL (100% mc) + larox filtrate Unhydrated GCL + larox filtrate (50% larox diluted) Unhydrated GCL + larox filtrate (25% larox diluted)

Hydraulic conductivity (cm/s) 2.6 x 10-9 9.0 x 10-6 1.2 x 10-7 5.8 x 10-6 4.8 x 10-6

The following conclusions were drawn: • The high concentrations of various salts in the larox filtrate severely affected the ability of the bentonite in the GCL to hydrate and swell.

P. Legg and M. McLennan / Is There a Future for GCLs in Waste Barrier Systems?

93

•

The unhydrated GCL subjected to larox filtrate permeant did not achieve any form of acceptable hydraulic conductivity in terms of landfill liner performance. Dilution down to 25% did not show a significant improvement of performance. • The GCL prehydrated to 100% moisture content subjected to larox filtrate had a hydraulic conductivity of 1.2 x 10-7 cm/s, which is comparable with the hydraulic conductivity requirements for a compacted clay liner within a geo-composite hazardous waste landfill liner, as required by the “Minimum Requirements”. The performance of the installed barrier system was therefore reassessed using the following recommended values for hydraulic conductivity of the GCL: • For the primary liner GCL, a hydraulic conductivity value of not less than 4.8 x 10-6 cm/s, as it was believed that the real landfill leachate would be significantly more dilute than pure larox filtrate. • For the secondary liner GCL, a hydraulic conductivity value of not less than 1.2 x 10-7 cm/s, as it could be safely assumed that this secondary GCL had hydrated fully from moisture in the underlying subgrade soil. 4.2. Brine disposal pond A water reclamation plant treats polluted groundwater from the coal mines in the Witbank area to a potable water standard and, in the process, generates brine and sludge waste streams that have to be disposed of in a lined evaporation pond. The brine is classified as hazardous due to its high concentration of salts (especially Na, S, Ca, K, Fe, Li, Mn and Sr). According to the “Minimum Requirements”, such a pond should have a liner system equivalent to that of a hazardous waste lagoon - a double composite liner system with a leakage detection and collection layer separating the primary and secondary composite liners. This prescriptive liner system recommends the use of either compacted clay liner (CCL) or geosynthetic clay liner (GCL) as part of each composite liner. Due to the unavailability of suitable clay, a GCL was initially considered as a possible replacement for the CCL. However, in the case of the brine, the Ca2+ concentration was almost 1 000 mg/ℓ, which could result in cation exchange and compromise the performance of the GCL. It was therefore necessary to assess the chemical compatibility and performance of a GCL prior to its use in a geocomposite barrier system (Habte and Legg, 2009). A series of hydraulic conductivity tests were carried out on samples of GCL subjected to the brine as a permeant using fixed wall permeameters. Four scenarios were tested for 2 000 hours with the following results (see Table 2): Table 2. Fixed wall permeameter test results – Brine disposal pond GCL hydration, permeant Unhydrated GCL + tap water Unhydrated GCL + brine Pre-hydrated GCL (100% mc) + brine Fully hydrated GCL (219% mc) + brine

Hydraulic conductivity (cm/s) 1.0 x 10-8 6.5 x 10-7 1.0 x 10-7 1.1 x 10-7

The following conclusions were drawn: • The permeability of the GCL with brine as permeant varied between 1.0 x 10-7 and 6.5 x 10-7 cm/s with a slightly increasing trend. The fully hydrated GCL yielded the lowest permeability values. • In the brine tests, the permeability was very low initially, however with time the permeability increased probably as a result of the Na+ cation exchange with Ca2+.

94

P. Legg and M. McLennan / Is There a Future for GCLs in Waste Barrier Systems?

Based on the permeability test results, it was decided that it would be inappropriate to replace the CCL component of the primary liner with a GCL, as the GCL would not have access to moisture for hydration being sandwiched between two geomembranes, and because the composite barrier system would not have a confining pressure for much of its lifespan. The first time the GCL would be subjected to moisture would be in the event of a leak where the moisture would be brine rather than water. It was however felt that the GCL could be used as a replacement for the two compacted clay layers in the secondary liner, as it would be placed a soil subgrade and would therefore draw moisture from the underlying soil for hydration. An alternative triple geomembrane liner system was proposed for the lining of the brine evaporation pond. The concept of this liner system was to provide an additional geomembrane liner and a drainage layer instead of the 600 mm CCL. This provides an additional defence system whilst reducing the head on the secondary geomembrane liner. The potential problem of panel separation of the GCL seams was addressed by increasing the overlap width. 4.3. Vanadium calcine waste disposal facility A new lined waste disposal facility for calcine waste from a vanadium plant was recently constructed. The leachate from the existing calcine dump was analysed and found to have high concentrations of mainly Na, V, Ca and Cr, amongst many other chemicals. The barrier system had to comply with the Minimum Requirements for a hazardous waste facility. To evaluate the proposed barrier design, swell index testing was carried out on the bentonite from two locally available GCLs, one containing granulated bentonite, and the other containing powdered bentonite. The swell index tests were performed using leachate obtained from the existing calcine waste dump, and compared with tests using distilled water and tap water. The results of the testing indicated that both bentonites showed a swell of only 9 mℓ/2g, compared with the industry standard of 24 mℓ/2g, suggesting that the high inorganic salt content of the leachate inhibits hydration of the sodium bentonite. There was therefore no need to carry out hydraulic conductivity testing. It was decided not to use a GCL as a replacement for the compacted clay layers in the primary liner. In the case of the secondary liner, a GCL placed directly on a soil subgrade beneath a geomembrane liner will hydrate from natural moisture in the underlying soil, and will be subjected to a confining pressure from the overlying pioneering waste layer. It will therefore be fully hydrated if/when it is subjected to contact with leachate that might leak through the primary and secondary geomembrane liners. The installed barrier therefore comprised of a triple geomembrane liner system with two leakage detection and drainage layers, with a GCL beneath the tertiary geomembrane liner. 4.4. Pulp mill waste disposal facility The landfill at a pulp and paper mill is being extended and, due to the classification of the waste stream as hazardous, the barrier for the proposed extension must comply with the Minimum Requirements for a hazardous liner design. The clayey soils underlying the site do not meet the permeability specification of less than 10 -7 cm/s, so it was necessary to consider replacing the CCL component of the primary liner with a single GCL.

P. Legg and M. McLennan / Is There a Future for GCLs in Waste Barrier Systems?

95

Swell Index testing was conducted to obtain a first indicator of the effect of the leachate on the swelling potential of the bentonite within the GCL, as compared to tap water as the baseline standard. Two makes of GCL were tested, one containing granular sodium bentonite and the other containing powdered sodium bentonite. The landfill leachate has high concentrations of sodium, while most of the other cations have fairly low concentrations. The results of the Swell Index tests are shown in Table 3 below. Table 3. Swell Index test results – Pulp mill waste disposal facility GCL / Bentonite source Water 21 17

GCL type 1 – granulated bentonite GCL type 2 – powdered bentonite

Swell Index (mℓ/2g) Leachate 15 13

Although the swell indices of both bentonites in water were lower than the industry standard of 24 mℓ/2g, the leachate reduced the swelling index of both bentonites by 28.6% and 23.5%. This reduction in swelling of the bentonite was regarded as significant and it was therefore decided to perform hydraulic conductivity testing by flexible wall permeameter to determine the actual hydraulic conductivity of the two GCLs when subjected to leachate as a permeant. Four scenarios were tested for each GCL type with the following results (see Table 4): Table 4. Flexible wall permeameter test results – Pulp mill waste disposal facility GCL hydration, permeant

Unhydrated + tap water Unhydrated + leachate Pre-hydrated (100% mc) + leachate Saturated (+300% mc) + leachate

Hydraulic conductivity (cm/s) GCL 1 – granulated GCL 2 – powdered bentonite bentonite 2.2 x 10-9 1.9 x 10-9 2.6 x 10-9 3.0 x 10-9 2.6 x 10 1.9 x 10-9 2.6 x 10 1.8 x 10-9

Based on the hydraulic conductivity test results, it was evident that the sodium rich leachate has little or no adverse effect on the hydraulic conductivity performance of both GCL types tested. It was therefore recommended that a GCL be used beneath the geomembrane in the primary liner of the landfill’s proposed barrier system.

5. Conclusions and recommendations The case studies presented in this paper have emphasized the importance of leachate compatibility testing of a GCL before considering its use in a geocomposite barrier system. The swell index test should be used as a first indicator of potential suitability. If the swell index of the GCL’s bentonite is very low, a GCL should not be considered further. If the swell index test shows marginal swelling, hydraulic conductivity testing of the GCL with leachate as permeant should be performed to confirm compatibility. If the swelling of the bentonite is hardly affected by the leachate, there should be no need to carry out hydraulic conductivity testing. There is still a future for the use of GCLs in barrier systems, provided the necessary leachate compatibility testing is done, and the GCL is installed with adequate hydration and confining pressure.

96

P. Legg and M. McLennan / Is There a Future for GCLs in Waste Barrier Systems?

References [1] Egloffstein, T.A. Natural bentonites – influence of the ion exchange and partial desiccation on permeability and self-healing capacity of bentonites used in GCLs. Geotextiles and Geomembranes, 19(7) (2001), 427-444. [2] Department of Water Affairs & Forestry, Second Edition, 1998. Waste Management Series. Minimum Requirements for Waste Disposal by Landfill. [3] Thiel, R., Giroud, J.P., Erickson, R., Criley, K. and Bryk, J. Laboratory measurements of GCL shrinkage under cyclic changes in temperature and hydration conditions. Proceedings of 8th International Conference on Geosynthetics, Yokohama, 2006. [4] Melchior, S. Field studies and excavations of geosynthetic clay barriers in landfill covers. Clay Geosynthetic Barriers, (2002), 321-330. [5] Lin, L.C. and Benson, C.H. Effect of wet-dry cycling on swelling and hydraulic conductivity of GCLs. Journal of Geotechnical and Geoenvironmental Engineering, 126(1) (2000), 40-49. [6] Meer, S.R. and Benson, C.H. Hydraulic conductivity of geosynthetic clay liners exhumed from landfill final covers. Journal of Geotechnical and Geoenvironmental Engineering, 133(5) (2007), 550-562. [7] Southern, J.M. and Rowe, R.K. Modeling of thermally induced desiccation of geosynthetic clay liners. Geotextiles and Geomembranes, 23(5) (2005), 425-442. [8] Jo, H.Y., Benson, C.H. and Edil, T.B. Hydraulic conductivity and cation exchange in non-prehydrated and prehydrated bentonite permeated with weak inorganic salt solutions. Clay and Clay Minerals, 52(6) 2004), 661-679. [9] Johns, D.G. and Shamrock, J.S. Swell index testing of GCL bentonites with general and hazardous waste leachates. Proceedings of African Regional Conference on Geosynthetics, Cape Town, 2009. [10] Jo, H.Y., Benson, C.H., Shackelford, C.D., Lee, J. and Edil, T.B. Long term hydraulic conductivity of a GCL permeated with inorganic salt solutions. Journal of Geotechnical and Geoenvironmental Engineering, 131(4) (2005), 405-417. [11] Kolstad, D.C., Benson, C.H., and Edil, T.B. Hydraulic conductivity and swell of non prehydrated geosynthetic clay liners permeated with multispecies inorganic solutions. Journal of Geotechnical and Geoenvironmental Engineering, 130(12) (2004), 1236-1249. [12] Lee, J., Shackelford, C.D., Benson, C.H., Jo, H. and Edil, T.B. Correlating index properties and hydraulic conductivity of geosynthetic clay liners. Journal of Geotechnical and Geoenvironmental Engineering, 131(11) (2005), 1319-1329.

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-97

97

Design of Hazardous Waste Landfill Liners: Current Practice in South Africa Riva NORTJÉ, Danie BRINK, Jonathan SHAMROCK, David JOHNS and Jabulile MSIZA Jones & Wagener Consulting Civil Engineers, South Africa

Abstract. A hazardous waste landfill can be regarded a complex system, with many components that contribute to the overall environmental performance of the facility. These include access control, waste containment, drainage systems, operations, gas management systems, and so forth. Of these components, the landfill liner is the first line of defense against long-term environmental contamination. While the requirements for and behavior of compacted clay liners (CCLs) are generally known, the geosynthetic components of hazardous landfill liners also require understanding and informed design. This paper considers issues currently considered in hazardous waste landfill liner design in South Africa, including sub-soil drainage requirements, desiccation, chemical compatibility issues, slope stability issues, and leachate collection requirements. It is noted that this paper cannot be considered to be exhaustive, given its brevity. Keywords. Hazardous waste, landfill liner design, geosynthetics, South Africa

Introduction A hazardous waste landfill can be regarded a complex system, with many components that contribute to the overall environmental performance of the facility [1]. These include access control, waste containment, drainage systems, operations, gas management systems, and so forth. Of these components, the landfill liner is the first line of defense against long-term environmental contamination. The siting of a landfill impacts on the complexity of its liner designs: the slope of the site; type, quantities and properties of soils available for lining; depth to groundwater and proximity to surface water bodies are key inputs. Conducting suitably detailed geotechnical and geohydrological investigations and surveys on candidate landfills and considering the results in determining the preferred site are critical. Hazardous waste landfills in South Africa are currently lined in accordance with specifications included in the Minimum Requirements for Waste Disposal by Landfill (MRs)[2]. South Africa currently has two classes of hazardous waste landfills: H:H landfills can accept Hazard Class 1 to 4 wastes, while H:h landfills can accept Hazard Class 3 and 4 wastes[3]. Both liner designs include a secondary compacted clay liner (CCL), a leakage detection layer, a primary composite liner comprising a 600mm CCL overlain by a geomembrane (GM), and a leachate collection layer. Requirements for each layer are specified in the MRs[2]. A new waste classification system is currently being developed by the Department of Water and Environmental Affairs (DWEA) in South Africa, and is expected to be gazetted by early 2012. Existing hazardous waste landfill sites will become Class A

98

R. Nortjé et al. / Design of Hazardous Waste Landﬁll Liners: Current Practice in South Africa

landfills (for high risk wastes) under these regulations [4], and a new landfill liner specification has been proposed. The proposed liner differs from the MRs liner in that an additional GM (1.5mm thick) plus protection layer has been included over the lower CCL, a geosynthetic leakage detection layer is deemed allowable, the GMs are now specified as high density polyethylene (HDPE), and the stone leachate collection layer has been increased from 150mm to 300mm. The inclusion of a secondary composite liner will offer a higher degree of environmental protection, while a thicker leachate collection layer is more likely to be effective. While the requirements for and behavior of compacted clay liners (CCLs) are generally known, the geosynthetic components of hazardous landfill liners also require understanding and informed design. The proposed Class A landfill liner design includes four required geosynthetics layers (two HDPE GMs and two separation geotextiles). Where in situ materials are not suitable for the required layers, it is possible to use geotextiles as protection layers over the GMs, a geosynthetic cuspated drainage product for the leakage detection system, as well as geosynthetic clay liners (GCLs) as a partial substitution for the CCLs. For landfill liner design, the designer must be aware of issues such as sub-soil drainage requirements, desiccation, chemical compatibility issues, slope stability issues, and leachate collection requirements, which are considered in this paper.

1. Consideration of subsoil drainage Landfills should be located so as not to impact on existing or future groundwater resources. Hazardous waste landfills are, however, frequently excavated to significant depth, either to gain airspace, or for the excavation of cover material. In such a case, most of the unsaturated zone may be removed, making the groundwater more vulnerable to pollution. When landfills are constructed below the groundwater in South Africa, they are typically engineered to draw down the water table. This increases the depth from the liner to the water table, while relieving pressure build-up beneath the liner. Constructing a subsoil drainage layer as the first layer of the lining system achieves these objectives. The layer should be designed to intercept water flowing through the pores of the base soil, allowing the passage of water while preventing movement of the base soil particles. Sand and gravels can be used for this purpose [5], although geotextiles and geocomposites are increasingly being used. Ideally, the outlet of the subsoil drainage system needs to be designed so that the water can be freely drained in future. A ‘walk away solution’ for deep cells, equipped with pumped systems, is therefore not attainable for sites in deep excavations.

2. Desiccation of clay liners Desiccation (i.e. drying out) of clay liners leads to cracking, and these cracks frequently do not self-heal when the clay is rehydrated. Both CCLs and GCLs are susceptible to desiccation, and desiccation can occur during construction of the liner layers themselves; during placement of overlying liner, protection and drainage layers; as well as after placement of the waste[6]. There are a number of factors influencing the risk of desiccation of any particular clay liner, including the properties of the foundation layer,

R. Nortjé et al. / Design of Hazardous Waste Landﬁll Liners: Current Practice in South Africa

99

the properties and thickness of the clay layer, the overburden stress on the liner at the time of hydration, the temperature gradient across the liner, and depth to the underlying water table[6]. In South Africa, it is necessary that the design of any landfill liner incorporating clay consider the risk of desiccation[2], and construction and operational planning incorporate measures to reduce the risk of desiccation.

3. Compatibility considerations 3.1. Geosynthetic clay liners (GCLs) GCLs achieve their low permeabilities through swelling of the bentonite clay when wetted, which restricts pore spaces and limits the flow of water. The swelling capacity of the bentonite clay may be reduced by contact with the leachate intended to be retained. Leachates with high concentrations of cations, particularly multi-valent cations such as Ca2+, Mg2+ and Fe3+ may, by the process of cation exchange, result in clay particles with significantly reduced hydrated radii. The pH of the leachate may also affect swell and, therefore, the hydraulic conductivity of the bentonite. It is therefore imperative to check compatibility of the bentonite component of any GCL with site specific leachate. This can be done by several methods: the most obvious of which is a falling-head permeability test (ASTM methods D 5887 and D 6766). Jo et al have shown that such tests are often terminated too early, resulting in un-conservative estimates of long-term hydraulic conductivity being made [7]. However, tests conducted with appropriate termination criteria can take lengthy periods of time to complete, which may be impractical for the project programme. More rapid, although only qualitative, methods to assess compatibility of GCL bentonites with leachates are by index tests, for example the swell index test and the fluid loss test, as prescribed by ASTM D 5890 and ASTM D 5891 respectively. The test results may be compared to correlations of the index tested and long-term hydraulic conductivity. Such correlations are established in technical literature: see references 8, 9 and 10. Site specific leachates will not necessarily be available at design stage. For a new site where leachate has not yet been generated, designers are cautioned that bentonite compatibility testing with synthetic leachates may be either unconservative or overly conservative[11]. In addition, judging compatibility purely on the chemical composition of the leachates should be avoided, as experience has shown that apparently similar leachates produce significantly different swell indices [12]. Recent research has indicated that, when occurring together with cation exchange, desiccation of GCLs increases the hydraulic conductivity of the GCL by orders of magnitude[10, 13]. This emphasizes the need to prevent desiccation of bentonite in a GCL. 3.2. Compacted clay liners (CCLs) All clays have some potential for change in hydraulic conductivity when permeated. Therefore, in common with GCLs, CCLs may undergo detrimental changes from interaction with leachates and permeants. The only way to assess whether there could be a problem is to examine the effect of the particular contaminant of interest on the proposed clay liner material, which may include direct permeation tests or index tests. The sensitivity of a CCL to a leachate varies depending on the natural mineralogy of the clay. Rowe states that most inactive soils whose minerals consist of illites and

100

R. Nortjé et al. / Design of Hazardous Waste Landﬁll Liners: Current Practice in South Africa

chlorites are relatively insensitive to typical municipal solid waste leachate, and their hydraulic conductivities may in fact decrease due to Na+ adsorption[14]. In contrast, soils containing significant amounts of swelling minerals (e.g. vermiculite, montmorillonite) may experience contraction with consequent increases in hydraulic conductivity. The large chemical adsorptive capacity of a thick CCL is an inherent advantage, which may aid in offsetting any detrimental effects of chemical attack. 3.3. Geotextiles There are several types of non-woven geotextile, which have unique advantages and disadvantages. A little known compatibility issue is that of the chemical compatibility of polyester geotextiles with leachate. Note that the term “polyester” itself most commonly refers to polyethylene terephthalate (PET) (not to be confused with polyethylene). PET is susceptible to degradation by hydrolysis in both alkaline and acidic environments, and is the only polymer used in geosynthetics to be degraded in this way. Significant research has been performed to assess the hydrolytic susceptibility of PET geosynthetic products, indicating an order of magnitude drop in the halflifetime of the PET yarns being evaluated in one instance[14]. Hsuan et al define the pH boundary for hydrolysis, recommending that a pH of 9 is the upper limit for the use of PET geosynthetics for critical applications [15]. The lower limit is a pH of approximately 4, but this value is less certain. 3.4. Geomembranes (GMs) While there is an extensive database of chemical resistance guidelines for a number of GM polymers, designers should always be aware of the possibility of encountering chemicals in landfills about which little is known. The landfill designer should always use the existing database to check the chemical compatibility of the proposed GM polymer against the known chemicals in the site leachate. If exposure to potentially aggressive chemicals is anticipated, it is necessary to perform chemical resistance testing on the polymer. The immersion standards are prescribed by ASTM D 5322 and ASTM D 5496, while the physical and mechanical testing requirements are described in ASTM D 5747. Designers should also be aware of the effects of elevated temperatures on polymers. If exposure to high temperatures is expected, designers should consider conducting aging tests on the proposed GM, or specifying a product with enhanced anti-oxidant packages.

4. Liner shear strength considerations A number of landfill failures have occurred, in South Africa and elsewhere, where designers have neglected liner shear strength considerations. Careful consideration of liner stability is therefore an essential component of the liner design process, particularly where geosynthetics are included. The use of GMs in lining systems for landfills, whether in direct contact with a geotextile or a soil, almost invariably introduces a weak shear strength interface along which failure can occur. For example, the residual interface friction angle between a smooth GM and a geotextile can be less than 10 degrees, whereas the friction angle between a smooth GM and a CCL can typically vary between about 15 and 20 degrees.

R. Nortjé et al. / Design of Hazardous Waste Landﬁll Liners: Current Practice in South Africa

101

The interface friction angle between a smooth GM and a GCL can be very low (less than 10 degrees), particularly when the bentonite powder on the interface is hydrated. If the GCL is placed on a sloped surface and a significant load is applied (in the form of waste deposition), the internal shear strength of the GCL, which is dependant on the tensile strength of the reinforcing fibres, could be exceeded. When that happens, the shear strength of the hydrated bentonite, which is in the order of 6 degrees in a drained condition, controls the strength of that particular interface. Almost all slopes encountered are steeper than this and failure will be unavoidable. A number of practical guidelines related to the use of geosynthetics and stability considerations are given below: • When designing for slope stability, it is essential to consider all the interfaces within a lining system and to carry out interface shear strength tests, using the actual geosynthetic materials and soils to be used during construction. • Interface shear strength test results often show a significant difference between peak and residual strength values. Due to the consolidation of waste, fairly large strains can occur on top of a lining system. It is therefore considered appropriate to use residual shear strength values in stability analyses. Standard shear box tests typically do not allow sufficient strain to induce the residual shear strength value, so that high strain equipment, such as a ring shear box apparatus, should be used. • The interface shear strength of a geomembrane in contact with either a geotextile or a soil can be improved by using a textured GM. • In order to prevent tensile stresses from developing within a GM, it is always important to ensure that the frictional resistance on top of a GM liner is less than the frictional resistance below the liner. In practical terms, this means that a consolidating (and moving) waste body will move on top of the liner system without introducing strains into the GM. • The stress-strain relationships of different geosynthetic materials can vary significantly. Some materials, for example geotextiles, develop their peak strength at very high strains, whilst other stiffer materials mobilize their strength at much lower strains. This aspect needs to be taken in consideration when analysing the behaviour of composite lining systems.

5. Leachate collection system design Leachate drainage is arguably one of the most important layers in a liner system, but often receives the least attention in landfill design and specification. If the leachate collection/extraction system in a cell is functioning, it will minimise the hydraulic head acting on the liner, and hence minimise the flow through the liner. International best practice for leachate drainage was summarised in a report compiled for the UK Environment Agency in 2002[16]. From this, the main functional requirements of leachate collection systems are that they must: • Be able to drain leachate within the landfill; • Facilitate the control of leachate levels within the design parameters of the lining system; • Be strong enough to withstand physical damage from the loading imposed by the waste and any equipment working over the system; • Be resistant to chemical attack in the corrosive environment of the landfill;

102

R. Nortjé et al. / Design of Hazardous Waste Landﬁll Liners: Current Practice in South Africa

•

Be able to function without clogging during and after the operational life of the landfill; and • Be capable of inspection and maintenance until such time as the system is no longer required to function. In order to achieve these requirements, the properties of the drainage layer that must be considered include layer thickness, physical strength, particle size distribution, hydraulic conductivity, physical clogging and filter stability, chemical stability and clogging, biological clogging, and redundancy. The pipework to collect and convey the collected leachate to an extraction point in the landfill is as important as the drainage media. The important characteristics of the collection/conveyance pipework are the material chosen, diameter, pipe perforations, pipe layout and spacing, physical strength, pipe surround, pipe jointing, connection with the extraction point(s) and, again, redundancy.

6. Discussion The landfill liner provides the first line of defense against long-term contamination of the environment, which is particularly critical for hazardous waste landfills. South Africa’s current hazardous waste landfill liner specifications are twelve years old, but are due to be upgraded by early 2012. The implementation of the proposed new specification will see an increase in the use of geosynthetics, which offer particular challenges to the landfill designer. This paper has considered issues currently considered in hazardous waste landfill liner design in South Africa, including sub-soil drainage requirements, desiccation, chemical compatibility issues, slope stability issues, and leachate collection requirements. It is noted that this paper cannot be considered to be exhaustive, given its brevity.

References [1] Rowe, R.K. and Hosney, M.S., A systems engineering approach to minimizing leachate leakage from landfills, 9th International Conference on Geosynthetics Proceedings, 2010, 501-510. [2] Department of Water Affairs and Forestry, Waste Management Series: Minimum Requirements for Waste Disposal by Landfill, Second Edition, 1998. [3] Department of Water Affairs and Forestry, Waste Management Series: Minimum Requirements for the Handling, Classification and Disposal of Hazardous Waste, Second Edition, 1998. [4] Department of Environmental Affairs, Draft Standard for Disposal of Waste to Landfill, September 2010. [5] United States Department of Agriculture, Soil Conservation Service, National Engineering Handbook, Chapter 26: Gradation Design of Sand and Gravel Filters, 1994. [6] Rowe, R.K. Long-term performance of containment barrier systems, Géotechnique 55 No 9 (2005), 631648. [7] Jo H., Benson C.H., Shackelford C.D., Lee M., & Edil T.B., Long-Term Hydraulic Conductivity of a Geosynthetic Clay Liner Permeated with Inorganic Salt Solutions, Journal of Geotechnical and Geoenvironmental Engineering 131 No.4 (2005) 405 – 417. [8] Kolstad D.C., Benson C.H. & Edil T., Hydraulic Conductivity and Swell of Nonprehydrated Geosynthetic Clay Liners Permeated with Multispecies Inorganic Solutions, Journal of Geotechnical and Geoenvironmental Engineering 130 No.12 (2004) 1236 – 1249. [9] Lee J., Shackelford C.D., Benson C.H., Jo H. & Edil T.B., Correlating Index Properties and Hydraulic Conductivity of Geosynthetic Clay Liners, Journal of Geotechnical and Geoenvironmental Engineering 131 No.11 (2005) 1319 – 1329.

R. Nortjé et al. / Design of Hazardous Waste Landﬁll Liners: Current Practice in South Africa

103

[10] Benson, C.H. and Meer, S.R., Relative Abundance of Monovalent and Divalent Cations and the Impact of Desiccation on Geosynthetic Clay Liners, Journal of Geotechnical and Geoenvironmental Engineering 135 No.3 (2009) 349 – 358. [11] Ruhl J.L. & Daniel D.E., Geosynthetic Clay Liners Permeated With Chemical Solutions and Leachates. Journal of Geotechnical and Geoenvironmental Engineering 123 No.4 (1997) 369 – 381. [12] Johns D.G. & Shamrock, J.R. Swell Index Testing of GCL Bentonites with General and Hazardous Waste Leachates, GeoAfrica 2009 Proceedings, 2009. [13] Touze-Foltz, N., State of the art and durability insights regarding the use of geosynthetics for lining in hydraulic and environmental applications, 9th International Conference on Geosynthetics Proceedings, 2010, 511-530. [14] Rowe R.K. (Editor), Geotechnical & Geoenvironmental Engineering Handbook, Kluwer Academic Publishers, 2001. [15] Hsuan Y.G, Schroeder H. F., Rowe K., Müller W., Greenwood J, Cazzuffi D, & Koerner R.M., Longterm Performance and Lifetime Prediction of Geosynthetics, EuroGeo4 Proceedings, 2008. [16] Environment Agency R&D Dissemination Centre, R&D Technical Report P1-397/TR, Landfill Engineering: Leachate Drainage, Collection and Extraction Systems, 2002.

104 Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-104

Geosynthetic Clay Liners: A Useful New Tool for Environmental Protection in the Engineer’s Toolbox Peter DAVIES1 Kaytech Engineered Fabrics, Durban, South Africa

Abstract. In the history of construction materials used by humankind, Geosynthetic Clay Liners (GCLs) are relatively new, having been in use internationally since the late 1980s (Heerten 2002) [1]. However in the author’s (and many others’) experience the way they work, and their advantages and limitations are not well understood by designers who have not studied them in depth, or who do not work with them often (Heerten & Koerner 2009)[1]. Their international success over the last 30 years however, has recently led to a modern South African manufacturing plant, and this paper adopts a practical approach and examines the construction, applications and performance of these useful water and gas tight lining materials. The paper highlights the equivalence of GCLs to compacted clay linings (CCLs) and the substantial cost savings that their use can show over ‘conventional’ materials such as CCLs in structures including landfill liners and containment structures of many kinds. Their constraints are also noted. In addition, an independent generic specification (GIGSA 1200W) [2] for GCLs is presented as a model non-commercial example of how to specify and construct linings incorporating these products. Keywords. Geosynthetic Clay Liners, GCLs, Government Regulations, Compacted Clay Linings, CCLs, Equivalence, Constraints, Limitations.

Introduction A GCL is a low permeability (typically ±10-11 m/s) lining, with the low permeability typically being provided by sodium bentonite clay contained within or on geosynthetic materials such as geotextiles and geomembranes to form a liner on a roll. Calcium bentonite is also used by some manufacturers, but as this has a lesser sealing performance than Sodium bentonite, nearly twice the mass of bentonite is required for equivalent performance (von Maubeuge 2002) [3], (Zanzinger & Touze-Folz 2009)[4] which has negative commercial implications in terms of the cost of transporting these products, and there are few manufacturers of such products. GCLs (Note: not GCL’s as is often written) are typically used to prevent the escape of liquids or gasses from containment structures. Applications include basal linings for many types of containment structure, closure cappings for landfills etc, instead of (or as an adjunct to) compacted clay linings (CCLs), geomembranes, and other low-permeability materials.

1

International Marketing Manager, Kaytech Engineered Fabrics, 11 Livingstone Road, Pinetown, Durban, South Africa. E-mail [email protected]

P. Davies / GCLs: A Useful New Tool for Environmental Protection in the Engineer’s Toolbox

105

Typically, a GCL usually less than 10mm thick can be equivalent to a CCL up to a metre thick (Giroud et al 1994 [5], Koerner & Daniel, 1995 [6], Rowe 1998) [7]. GCLs used instead of CCLs can show many advantages including inter alia technical superiority (Heerten & Koerner 2009)[1], ease of installation by relatively unskilled labour, lower construction costs and given their relative thinness, more airspace (and thus more material stored) over a given containment footprint. Their use instead of or as an adjunct to CCLs in pollution control works (and their superior performance in many instances), has increasingly led to their approval by environmental authorities around the world including South Africa (DWAF 1998) [8] and they are now manufactured in many countries around the world. Despite the foregoing, in a study conducted in 2007, where the landfill regulations of 52 countries were analysed, in 73% of those countries the authorities still considered a classical CCL lining system to be adequate (Heerten & Koerner 2009) [1]. In the words of these internationally respected experts: “From the standpoint of the authors and many expert colleagues, this is a surprising situation …. This situation calls for global, rigorous corrective action.” This author says Amen to that!

1. The Components of a Typical Needlepunched GCL GCL products can take a number of forms, including where the bentonite is glued onto a geomembrane, contained between a cover and carrier geotextile which is stitched together, or contained within a cover and carrier geotextile which are needlepunched together, all to form a liner on a roll. It is not the intention of this paper to cover all forms of construction in detail, as this has been done in numerous previously-published references, most recently in GRI-GCL5 (2011)[9] which contains many useful definitions and descriptions.

Figure 1. Long-edge section through a typical Needlepunched GCL. The vertical lines between cover and carrier geotextiles represent around 2.5 million fibre bridges per m2, imparting hydrated internal shear strength to the bentonite. “Heat burnished” is a technique used to lock the fibre bridges into the Carrier Geotextile by heat-sealing

The quality of bentonite used is critical, but often ignored in specifications: “The consensus now is that to simply specify the clay component of the GCL by a generic classification of “sodium bentonite” ignores subtle differences such as impurities and

106

P. Davies / GCLs: A Useful New Tool for Environmental Protection in the Engineer’s Toolbox

particle siize. The key message is th hat all benton nites are not created c equall and small changes within w benton nite mineralog gy, clay chem mistry and pa article size caan have a significant effect on both the short and a long-term performancee of the GCL iin the liner system.” (Gates et al 2009) [10]. dlepunched tyype has emerrged to be th he most wideely-used form m of GCL The need construction around thee world and in nter alia they are now madee in Germanyy, the USA, ussia and India, to name but ut a few countrries. China, Caanada, Australlia, Poland, Ru CE-C Certified [11] South S African n manufactureed needlepuncched GCLs arre typically supplied in n roll form ± 5.3 m wide byy 40 m long, and a ± 960 kg per p roll. A tyypical longedge crosss section throu ugh a South African A producct is shown in n Figure 1. Geeosynthetic componen nt layout and bentonite b quan ntity shown vary v with makke / grade of prroduct, but generally, wherever th hey are made,, they all exh hibit a manuffactured perm meability of around 10 0-11 m/s (10-9 cm/s). c In the type of GCL shown in Fig gure 1, the lonng edges of the GCL roll r are self-sealing to the next roll, mak king site jointting of these materials m a non-comp plicated exercise that can be b done by an ny competentt contractor w without the usual of sp pecialised equ uipment. 2. Simplee Illustration of some of th he benefits of a GCL over a CCL

Picture 1: Transport of GCLs

Picture 2: Installation of a GCL

In Picturee 1 a 7-tonnnee flatbed truck k is shown traansporting 1 055 m² of needdlepunched GCL, with h a loaded mass m of ±5 ton nnes. This on ne truckload of o GCL is equ uivalent to ninety 7-tonne truckloaads of CCL co ompacted to a 600 mm thiickness at 95% % standard a of lining,, the benefits of reduced trruckloads, maaintenance, Proctor. For a given area ur etc. are sellf-evident. No ot included in n these “self-eevident” benefits are the fuel, labou greenhousse gas emissio ons as a result of GCL vs. CCL C use: ±4 kg of CO2 per m m² of GCL installed vs. v ±10 kg per p 600 mm thick CCL in nstalled (Eglo offstein et al 2002) [12]. Obviouslyy, these figurees will vary with w transporttation distances, and these variations are examiined in the reference given n. For Greenh house Gas red duction reasons alone it would seeem more atten ntion should be paid to the use u of GCLs in i place of CC Cls, but the author hass never heard of such a perttinent factor used u in lining choice consid derations in southern Africa. A

P. Davies / GCLs: A Useful New Tool for Environmental Protection in the Engineer’s Toolbox

107

In Picture 2, a civil engineering contractor is seen laying a GCL (a product they had not worked with before). Kaytech technical staff spent a morning with the laying team and they then proceeded to lay over 200 000 m² of our product without requiring further instruction. This illustrates the ease with which GCLs can be used.

3. Some Constraints on the use of GCLs The substantial benefits to be found in the use of GCLs have in some instances, led to their use in projects where the material was found not to be performing to expectations. As with all construction materials, there are some constraints on the use of GCLs: •

Chemical Compatibility: Sodium bentonite as used in GCLs is a natural montmorillonite clay and is subject to the same constraints vis-à-vis chemical compatibility that compacted clay linings are. If a chemical cocktail such as can be found in some industrial effluents will damage a CCL, then it is most likely that a GCL will also be affected, and often more severely and in a shorter time due to the thin nature of a GCL compared to a comparatively massive compacted clay lining. Designers should be aware of this and insist on compatibility testing wherever possible. This takes time and should be planned well in advance. This includes ion exchange considerations as mentioned below.

•

Ion Exchange: This forms part of the chemical compatibility question, and is one that has been receiving much publicity in recent years. This is the phenomenon under certain circumstances (particularly where unhydrated or desiccated Sodium bentonite is hydrated with solutions containing divalent cations such as Calcium or Magnesium) of ionic exchange where the Sodium bentonite in a GCL can become converted to Calcium bentonite in service. This can result in an increase in the permeability of the GCL by orders of magnitude. It is a complex reaction which has resulted in volumes of research. The whole situation is probably best presented concisely in a paper by Benson & Scalia (2010) [13] but there are many other publications on the subject.

4. GIGSA GCL 1200W[2]: An African Independent Generic GCL Specification When GCLs were first introduced to South Africa around 1995, a number of producers and their distributors and agents began making claims of superiority for their products, and the competition became so heated that some specifiers became first confused and then annoyed by the clamour. The need for an independent specification became clear – one that all players in the market could buy into. The Geosynthetics Interest Group of South Africa (GIGSA), as the local chapter of the International Geosynthetics Society (the IGS)[3] was the perfect body to draw this up, as it represents all the players in the geosynthetics market. Environmental law was also being tightened up, and the promulgation of the National Environmental Management Waste Act (NEMWA 2008) [14] demonstrated the RSA Government’s determination to minimise the effects of waste management on society. For the reasons given, GCLs will play an increasingly

108

P. Davies / GCLs: A Useful New Tool for Environmental Protection in the Engineer’s Toolbox

important part in waste management facilities, and the 1200W document provides a non-commercial comprehensive guide on how to work with these products. • For ease of integration, the document is laid out in accordance with the South African National Standard (SANS) 1200 suite of specifications for civil engineering construction works. Any competent contractor can lay GCLs, as specialist joining or welding equipment, as is the case for geomembranes, is not required. • Accordingly, the greater part of the twenty pages of GIGSA 1200W goes around the proper handling and installation supervision of GCLs, rather than the actual specification of their properties. • For product specifications, Section 3 of 1200W inter alia simply refers the specifier to the 11-page GRI-GL3 specification provided by the Geosynthetic Research Institute of the USA (2009) [15]. This is a lowest common-denominator type of specification and was designed to allow for most of the GCLs produced around the world. Some first-class GCLs display properties that are superior to the specifications to be found in the GRI-GCL3 document, and local specifiers should decide whether the GRI-GCL3 GCL specified minimum qualities are suited to their particular requirements. 4.1 What Practical Information is to be Found in this Document? • How to pack, transport and handle GCLs. At around 900 kg each, GCL rolls are heavy and need to be handled and stored in the correct manner. Section 4 of GIGSA 1200W gives precise instructions on how this is to be done. As construction materials, GCLs are relatively fragile and it is much easier to damage a thin geosynthetic layer than it is to damage a 600 mm thick CCL during installation! • How to install GCLs. Contractors who have never installed a GCL before may struggle to price their supply and installation bill items properly, as they will be unsure of what is involved in laying the material to specification. The inclusion of GIGSA 1200W in contract specifications will remove most of such uncertainty, as section 5.3 of this document includes a comprehensive laying and jointing procedure that will enable any quantity surveyor to formulate an accurate costing for tender purposes. This includes the correct preparation of receiving and covering earthworks. In the author’s experience this is the aspect of GCL work that is frequently neglected in project specifications, and it has been the cause of much dissension in the past when the engineer realises that the project specification was lacking in this regard, and tried to force the contractor into doing the ‘right thing’ without adequate compensation. • How to repair GCLs. It is probably inevitable (given that these materials are less than 10 mm thick and that heavy plant can be involved in their placing and covering), that some damage will be done to the GCL during installation, and section 5.3.6 of GIGSA 1200W provides concise and clear instructions on how damage to GCLs is to be repaired. • How to measure and pay. Misunderstanding of how measurement and payment will be made (particularly with regard to overlaps, joints and wastage) has frequently caused argument between contractor and engineer in

P. Davies / GCLs: A Useful New Tool for Environmental Protection in the Engineer’s Toolbox

109

the past. Section 9 of GIGSA 1200W gives clear guidelines on how measurement and payment is to be made for the various aspects of GCL supply and installation. 4.2 Summary on GIGSA 1200W • GIGSA 1200W is a well-formulated document, compiled by a team representative of all needlepunched GCL suppliers active in South Africa at the time of compilation. All regional specifiers of GCLs are urged to integrate it into their project documentation. By doing so, clear information is given to contractors, who are thus able to price projects on an ‘apples vs. apples’ basis. It clearly sets out who is responsible for what, and how payment will be made. It is sincerely hoped that regional waste management and civil engineering communities will accept it as a standard specification, for use on all projects incorporating Geosynthetic Clay Liners. • GIGSA 1200W is specifically formulated around needlepunched GCLs. Other types are not catered for, but should there be sufficient demand in future, GIGSA may consider producing a similar document for them, or adding them to this document. • Other countries which do not yet have a national specification are encouraged to use this document as a resource when compiling their own. • GIGSA 1200W may be downloaded free (196 kb file size) off the GIGSA web site at www.gigsa.org. There is no copyright on this document and all may use it within their own specifications, with the only proviso being that an acknowledgement is made to GIGSA as the source.

5. Conclusion GCLs are here to stay, and it is incumbent on the professional designer to become aware of these materials, and acquainted with their strengths and constraints. Failure to do so can (and has) led to some very expensive mistakes being made where a GCL did not perform as hoped by an uninformed designer. Having said that, the flood of published material on the proven performance (Heerten 2002)[16] and usefulness of GCLs is overwhelming in its implication – which is that these products should sit in the engineer’s toolbox along with concrete, clay, brick and steel and should be understood as well as they are.

References [1] Heerten, G. & Koerner, R.M. (2009): “Clay Sealing Layer in Landfills and Brownfields – Lessons to Learn”. Proceedings of GIGSA GeoAfrica 2009 Conference, Cape Town, 2-9 September 2009. Geosynthetics Interest Group of South Africa, ISBN 978-0-620-43722-6 [2] GIGSA 1200W (GCL) (2008): “Pro-Forma Standardised Specification for Reinforced Needlepunched Geosynthetic Clay liners (GCLs)”. Geosynthetics Interest Group of South Africa, Edenglen, South Africa. This document may be downloaded free of charge at www.gigsa.org (196 kb). [3] International Geosynthetics Society (IGS). 605 Belvedere Rd., Suite #13 , West Palm Beach, Palm Beach USA. www.geosyntheticssociety.org/

110

P. Davies / GCLs: A Useful New Tool for Environmental Protection in the Engineer’s Toolbox

[4] Von Maubeuge, K.P. (2002) “Investigation of bentonite requirements for geosynthetic clay barriers”. Proceedings of the International Symposium IS Nuremberg 2002, Nuremberg, Germany 1-17 April 2002. AA Balkema Publishers, the Netherlands. ISBN 90 5809 380 8 [5] Zanzinger, H. & Touze-Folz, N. (2009): “Clay geosynthetic barriers performance in landfill covers”. Proceedings of GIGSA GeoAfrica 2009 Conference, Cape Town, 2-9 September 2009. Geosynthetics Interest Group of South Africa, ISBN 978-0-620-43722-6 [6] Giroud J.P., Badu-Tweneboah K., & Soderman K.L (1994), "Evaluation of Landfill Liners", Fifth International Conference on Geotextiles, Geomembranes and Related Products, Vol. 3, Pages 981-986, Singapore 1994. SEAC-IGS. ISBN 981-00-5823-3 [7] Koerner R.M. & Daniel D.E (1995), “A suggested methodology for assessing the technical equivalency of GCLs to CCLs”, Geosynthetic Clay Liners, Pages 73-98, A.A. Balkema, Rotterdam, 1995. ISBN 905410-5194. [8] Rowe, R. K. (1998), “Geosynthetics and the Minimization of Contaminant Migration through Barrier Systems beneath Solid Waste”. Sixth International Conference on Geosynthetics, Atlanta, Georgia, USA. International Geosynthetics Society. ISBN0-93583-07-6 [9] DWAF (South African Department of Water Affairs & Forestry) (1998). Minimum Requirements for Waste Disposal by Landfill. Second edition, September 1998. ISBN 0620-22993-4. [10] GRI-GCL5 (2011):"Design Considerations for Geosynthetic Clay Liners (GCLs) in Various Applications”. Standard Guide published January 26, 2011. Geosynthetic Institute, Folsom, Pennsylvania, USA. May be downloaded free at http://www.geosynthetic-institute.org/specs.htm (570kb). [11] Gates, W.P., Hornsey, W.P., Buckley, J.L. (2009): “Geosynthetic Clay Liners – Is the key component being overlooked?” Proceedings of GIGSA GeoAfrica 2009 Conference, Cape Town, 2-9 September 2009. Geosynthetics Interest Group of South Africa, ISBN 978-0-620-43722-6 [12] SKZ – TeConA GmbH (2010): Certification of Factory Production Control 1213-CPD-4772 awarded to Kaytech Engineered Fabrics, Atlantis, South Africa, April 2010 [13] Egloffstein, T.A., Heerten, G., von Maubeuge, K.P. (2010): “Comparative life cycle assessment (LCA) for clay geosynthetic barriers (GBR-C) versus clay liners and other sealing systems used in river dykes, canals, storm water retention ponds and landfills”. Pages 317-323. Proceedings of GBR-C 2k10: 3rd international Symposium on Geosynthetic Clay Liners, Würzburg, 15 & 16 September, 2010. ISBN 978-3-00-029863-9 [14] Benson, C.H. & Scalia, J. (2010):“Hydraulic conductivity of exhumed geosynthetic clay liners from composite barriers”. Pages 73-82. Proceedings of GBR-C 2k10: 3rd international Symposium on Geosynthetic Clay Liners, Würzburg, 15 & 16 September, 2010. ISBN 978-3-00-029863-9 [15] Department of Environmental Affairs, (2008): National Environmental Management: Waste Act 2008 (Act No. 59 of 2008) (NEMWA), promulgated 10 March 2009, Government Gazette No. 32000, Notice No. 278. Government Printer, Pretoria. [16] GRI-GCL3 Rev. # 1 (2009):"Standard Specification for Test Methods, Required Properties, and Testing Frequencies of Geosynthetic Clay Liners (GCLs)”. Geosynthetic Institute, Folsom, Pennsylvania, USA. May be downloaded free at http://www.geosynthetic-institute.org/specs.htm (105kb). [17] Heerten, G. (2002):“Geosynthetic clay liner performance in geotechnical applications”. Proceedings of International Symposium IS Nuremberg 2002, Nuremberg, Germany 1-17 April 2002. AA Balkema Publishers. ISBN 90 5809 380 8

Section 4 Foundations

This page intentionally left blank

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 113 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-113

Numerical modeling of skirted foundation subjected to earthquake loading W. R. Azzam Tanta University, Tanta, Egypt

Abstract. This paper explores the possibilities of using skirted foundation system to mitigate the liquefaction hazards, control the horizontal soil movement and decrease the pore water pressure underneath the foundation during the earthquake. This technique is investigated numerically using finite element analysis. Four stories reinforced concrete building resting on raft foundation is idealized as twodimensional model with and without skirts. In the present study, two dimensional plane strain program PLAXIS, dynamic version is used. A series of models for the problem under investigation was run under different skirts depth below the foundation level. The results showed improved effectiveness in liquefaction mitigation due to confinement effect and decreasing horizontal soil movement. The presence of such skirts can modify and decrease the induced excess pore water pressure. In addition, the skirts can significantly reduce the foundation acceleration and the ground motion amplification is reversed. Keywords. Liquefaction mitigation, Foundation, Skirts, Lateral confinement

Introduction It's recognized that the liquefaction is process involving energy dissipation due to frictional loss along grain contacts during cyclic loading as a result of increases in pore water pressure, that leading to collapse the soil structure and this information can be applied to develop methods for liquefaction mitigation. The energy required to cause liquefaction depends on the density of packing grains. Extensive damage to foundation and structures in areas of liquefaction and lateral spreading has been observed in many earthquakes around the world as presented by Chu et al., (2004). The factors that lead to liquefaction in the subsoil are important as reported by different authors e.g. Hatanaka et al. (1987) and Ishihara, (1993). Liquefaction mitigation can be done using a variety of soil improvement technique as dynamic compaction, vibro stone columns and drainage wicks (Dise et al., (1994) and Luehring et al., (2001)). Improving liquefaction potential strength by using micropiles and inclined reinforcement is also investigated by McManus et al. (2005); Naein and Moaye (2006). These techniques of soil improvement against liquefaction aim to increase the soil density and develop the ideal densifications that avoid large increases in pore water pressure. Based on this approach of densification, lateral confinement technique of soil underneath the foundation was adopted using structural skirts that fixed rigidly to the foundation edges. This technique was used in improving the bearing capacity and the settlement characteristics (Martinez et al. (2008); Azzam and Nazer (2010). Nevertheless, their effects in liquefaction mitigation cannot be scrupulously investigated. Consequently, a new alternative technique is suggested for liquefaction mitigation by lateral

114

W.R. Azzam / Numerical Modeling of Skirted Foundation Subjected to Earthquake Loading

confinement method compared with other technique of soil improvement to reduce liquefaction under building foundation. This adopted confinement was done using structural steel skirts with the conjunction of foundation.

1. Numerical modeling and selection of parameters The plane strain model was used with the 6 node element. The mesh was generated by the program and refined in the area around the footing. The subsoil is consisted of a deposit of sand layer of 20 m thickness. It assumed to be linear elastic in dynamic analysis. The properties of the adopted sand (γ =17 kN/m3, ν = 0.3, Eref = 40000 kPa, the Raleigh damping is considered at vertical boundaries and taken α , β = 0.01). The ground water table is assumed at 2 m below the ground surface to consider the excess pore water pressure so the soil material is assumed to be undraind. The adopted building consists of 4 floor and basement. It is 6 m wide as simulated in program PLAXIS (2002). The building and foundation are simulated as plate properties, Beam elements as elastic material. The floor and wall plate properties are (EA = 5 x 106 kN/m, EI = 9000 kN/m2/m with weight of 5 kN/m/m and Poisson’s ratio ν = 0). The building foundation is assumed as a reinforced concrete raft, it simulated as an elastic beam element, the raft thickness is 0.5 m thickness its plate properties are (EA = 105 kN/m and EI = 21.875 kN/m2/m). The skirts are simulated as beam elements, elastic material. The steel skirts are adopted with thickness of 4 mm and varied in depth, the studied skirts depth (L) as a function in foundation width (B). L/B ratio were modeled in the program in values (0.5, 1 , 1.5 and 2) . The skirts properties are axial stiffness and bending stiffness where, EA = 63000 kN/m and EI = 8.4 kN/m2/m. The earthquake is modeled by imposing a prescribed horizontal displacement at the bottom of boundary in contrast to standard unit length (Ux = 0.01m and Uy = 0). The geometry of finite element model adopted for the analysis is shown in Figure 1. 38 310 3 12 3 14 3

39

Soil model

8

4

3 6 4 5 2 1

36 3 11

3 15 2

37 7

17

20m

Building model

3 13

5 16

Absorbent boundary in the vertical sides

skirts

Horizontal prescribed displacement

y

0

~

1

x

~

100 m

Figure 1: The element of the adopted finite element model.

1.1 Analysis procedure The calculation procedure involves two phase. The first one is a normal plastic calculation in which the building is constructed. The second is a dynamic analysis in which the earthquake is simulated. In this phase the displacement are reset to zero and the time interval 22 sec, the sub step is set to 1. The acceleration of the input earthquake is chosen from the default acceleration data file in program (225smc) (SMC, Standard Earthquake Boundary).

W.R. Azzam / Numerical Modeling of Skirted Foundation Subjected to Earthquake Loading

115

2. Results and analysis 2.1 Results of foundation response output The lateral spreading of soil associated to liquefaction for foundation without skirts is obtained in the finite element outputs, which show that the maximum horizontal soil movement is obtained directly under the foundation. The contours convergence is densely founded under the foundation because at the bottom of the foundation the liquefaction is induced due to significant increase in excess of pore water pressure. The contour of excess pore water pressures is indicated that the extreme values were extended to depth equal to half the foundation width (0.5B) due to liquefaction effect. While, the results of finite element analysis of the model with skirts which presented the horizontal movement showed that, the existence of the skirts with sufficient anchorage length decreases both the flow and the horizontal movement of subgrade particles. The extreme horizontal displacements were also developed beyond the confined region. The horizontal deformations of subgrade particles between skirts were also virtually nonexsisted. The skirts were prevented the enclosed foundation soil from flowing into the free field. Essentially this retrofit technique results a perfect containment of the foundation subgrade particles. The skirts were reduced the liquefaction potential of soil due to lateral confinement. It also, increased the cyclic resistance and decreased the excess pore water pressure. It has a considerable effect on modifying the migration of the pore water pressure as shown in Figure (3). Where, the skirts decreased the induced pore water pressure between the confined subgrade and the pore water pressure migration is dissipated beyond the confined zone.

Figure 2: The contours of the excess pore water pressure for foundation with skirts under the earthquake effect.

2.2 Effect of skirts on the foundation horizontal settlement The rate of the building horizontal movement as well as ground movement can be observed in the horizontal settlement history plot in Figure 3. This plot shows the horizontal displacement due to the horizontal component of acting earthquake with time for foundation with and without skirts at different skirts depth. The existence of skirts were modified the horizontal displacement time curves and decreased the horizontal foundation movement compared with case of normal foundation. The skirted

116

W.R. Azzam / Numerical Modeling of Skirted Foundation Subjected to Earthquake Loading

foundation able to restore and resist the dynamic response due to confinement effect. It also provided significant interlocking for subgrade particles. The skirts can significantly increase the subgrade stiffness so the horizontal movement is limited. It can be concluded that, the skirts that installed with sufficient anchorage depth, the induced passive resistance along the skirts depth can play important role in decreasing the horizontal movement underneath the skirted foundation soil system. 2.3 Effect of skirts on the excess pore water pressure The computed excess pore-pressure history with and without skirts under the foundation at depth equal to 0.25B below the foundation level and inside the confined zone at different skirts depth is plotted in Figure 4. The existence of skirts extensively decreases the induced excess pore water pressure. The subgrade particles inside the skirts were interlocked and compacted. The skirts were made the pore water pressure migration downward to dissipate below the confined block. The skirts can also decrease the distortion rate in the sheared zone and reduce the ultimate shear stress mobilized in the shear zone. Therefore, the skirts were reduced the liquefaction potential and increase the cyclic resistance of the subgrade. It noticed that, as the skirts depth is increased the excess pore water pressure is decreased. In foundation without skirts, when the Earthquake induces pore water pressure in the surrounding soil below the foundation; the pore water pressure is migrated randomly along each side of the foundation. Whereas, in skirted system, the confined zone and the foundation acted as a one unite, and the pore water pressure is highly induced along each side of the skirts and outside the confined block. Therefore a hydraulic head difference is created between the confined zone (low pore water pressure zone, treated zone) and the soil under the confined block. As a result, the pore water flows from the confined zone into the bottom below the confined region. Where the pore water pressure does not developed sufficiently to induce liquefaction, but liquefaction will take place at the soil under the confined zone. The excess pore water pressure is gradually increased with the increase of shaking duration and the soil shear failure associated with liquefaction is developed under the treated confined zone due to a potential decrease in strength. This indicated that the excess pore water pressure is increased with the increase of depth below the confined zone. It can be confirmed by Figure 5, where, the excess pore water pressures at different depth (d) below the foundation level are extracted at different skirts depth. The figure shows that, the pore water pressure is increased at different depth below the confined block. At the same time sharp decrease in excess pore water pressure is observed with the increase of the skirted depth. It can be confirmed the effectiveness of the skirts to relief the excess pore water pressure and the soil shear failure associated to liquefaction is obtained outside the confined zone as justified before by the finite element output. 2.4 Effect of skirts on the foundation peak acceleration Figure 6 shows the lateral acceleration history of foundation with and without skirts directly at the foundation level. It can be demonstrated the effectiveness of the skirts in subgrade interlocking, the foundation and the confined block within the liquefiable soil behaves as if stiff one unit that resist and absorb the ground motion. It is known that, the ground motion amplification show that the acceleration at the top layer of improved zone is increased and decreased in lower part as stated by Liu and Dorby (1997). In

W.R. Azzam / Numerical Modeling of Skirted Foundation Subjected to Earthquake Loading

117

current case of adopting such skirts, the ground motion amplification is reversed. But it can be seen that, increasing the confined zone below the foundation and between the skirts relatively to overall liquefiable sand thickness produced obviously decreasing in the horizontal acceleration as shown in Figure 6.

Figure 3: Horizontal settlement during Earthquake for foundation with and without skirts for different skirts depth.

Figure 4: Computed excess pore-pressure history with and without skirts under the foundation at depth 0.25B.

Figure 5: Variation of the maximum excess pore-pressure with skirts depth at different depth below the foundation level.

The lateral acceleration is reduced with the increase of skirts depth. The installation of skirts with sufficient depth are prevented the ground motion amplification from incidence. The foundation peak acceleration is sharply decreased with the increase of skirted depth as clarified in Figure 7. This again justified that, the confined block and foundation behaves as if a coherent mass absorb the dynamic response and acted as a damper due to significant increase in confined stiffness. Also, it can be consider a good technique to increase the stability of building during earthquakes. 3. Conclusions 1.

2.

The installation of skirts with sufficient anchorage length is a good method of liquefaction mitigation that increasing the subgrade interlocking, controlling the soil movement and decreasing the excess pore water pressure. The foundation and soil between skirts behaves as a coherent mass and the soil shear failure associated with liquefaction is took place under this treated confined zone.

118

3.

4. 5.

W.R. Azzam / Numerical Modeling of Skirted Foundation Subjected to Earthquake Loading

The skirts can cause a significantly reduction in pore water pressure as the skirts depth increased. The direction of pore water migration is modified to migrate downward to dissipate below the confined block. The skirts not only decreasing the horizontal but also decreasing the lateral foundation peak acceleration which increase the building stability. The ground motion amplification is modified due to confinement effect and the resulting peak foundation acceleration is decreased with the increase of skirts depth and sharply increased under the confined block.

Figure 6: Computed lateral acceleration history with and without skirts directly at the foundation level.

Figure 7: Variation of maximum lateral foundation acceleration with skirts depth.

References [1] B.V. Plaxis 8.2 Professional version, Finite element analysis in geotechnical engineering. Theory and Application Thomas Telford, London, (2002). [2] D.B. Chu, J.P. Stewart, S. Lee, J.S., Tsai, Lin, P.S., Chu, B.L., Seed, R.B., Hsu, S.C., Yu, M.S. and Wang, M.C.H., Documentation of soil conditions at liquefaction and non-liquefaction sites from 1999 Chi-Chi (Taiwan) earthquake, Soil Dynamics and Earthquake Engineering, 24 (2004) , 647- 657 [3] H. Martinez, S. Gourvenec and M. Randolph, An experimental investigation of shallow skirted foundation under compression and tension, J. of Soil and Found., Vol., 48(2)(2008), 24-254 [4] J. K. McManus, J. P. Turner and G. Charton “Inclined reinforcement to prevent soil liquefaction, NZSEE Conference, (2005), 11pages [5] K. Dise, M. Steven and J. L., Von, Dynamic liquefaction to mitigate liquafable embankment foundation soils”. ASCE, Geotch. Eng. J., 45(1994),1-5. [6] K. Ishihara, Liquefaction and Flow Failure during earthquakes (Rankine Lecture)". Geotechnique,43 (3) (1993), 351-415, [7] M. Hatanaka, S. Yosho and M. Miake, Mitigation of Earthquake Liquefaction Hazards, Proceedings 8th Australia New Zealand, (1987) 237-243. [8] R. Dobry, V. Taboada and L. Liu, Centrifuge modeling of liquefaction effects during earthquakes, Keynote Lecture, Proc. 1st Int. Conf. Earthq. Geotech. Engrg., Balkema, Rotterdam, 3(1997),1291–1324. [9] R. Luehring, L. Snoteland and M. Stevens, Liquefaction mitigation of a silty dam foundation using vibro-stone columns. Proc. 21st annual metting and lectuer, Denver, (2001), 767-778. [10] S. Naein and R. Moayed, Improving liquefaction potential strength by using Micropiles. IAEG the Geological Scisty of London , (2006), 333-338 [11] W. R. Azzam, and A. Nazer, Improving the bearing capacity of footing on soft clay with and pile with/without skirts, International Review of Civil Engineering Journal. 1 (2010), 32-38.

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 119 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-119

Prediction of the Axial Capacity of Bored Piles Using Methods Based on CPT and Static Analysis Approaches Abdul Karim M. ZEINa and Samah B. MOHAMMADb Associate Professor, BRRI, University of Khartoum, Sudan b Postgraduate student, BRRI, University of Khartoum, Sudan a

Abstract This study aims at evaluating the performance of five methods based on CPT and static analysis in predicting the axial capacity of bored piles installed in some Sudanese soils. These methods were used for estimating pile capacity at six study sites located in Khartoum and White Nile States. Static pile load tests and CPT soundings were performed at adjacent points in each site to enable comparison of measured and estimated data. Pile capacities were determined using the Van der Veen method of load test results interpretation. Statistical analysis was conducted to evaluate the performance of the five methods according to established ranking criteria. Based on the evaluation criteria adopted, Touma and Reese method showed the best performance by achieving the closest agreement between estimated and measured pile capacities whereas the Dutch method showed the lowest overall performance. From comparison of predicted and measured values of the pile end bearing (Qp) and skin friction (Qs) capacity components obtained for an instrumented pile, the Touma and Reese method gave the best comparisons between measured Qp and Qs whereas the Schmertman method showed a quite reasonable estimate of Qs. The other three methods revealed unrealistic and gross overprediction or under-prediction of the two components of bored pile capacity. Keywords: Bored piles, Pile load test, CPT, bearing capacity, Statistical analysis.

1.

Introduction

Deep foundations are often used for supporting heavy and large structures at sites where the competitively economical shallow foundations would lead to excessive settlements. The use of bored piles deep foundation system has received acceptance and application during the two last decades among foundation designers and contractors in Sudan. This study aims at examining the applicability of some methods developed in other countries for the prediction of the bearing capacity of bored pile foundations installed in Sudanese soils. The determination of the carrying capacity of deep foundations is required for preparation of an adequate and safe design. An investigation was undertaken to determine and estimate pile capacities at six sites located in Khartoum and White Nile States. Five methods were selected from those published in literature and used for the estimation and comparison of predicted and measured bearing capacities of bored piles. The methods selected included two based on static analysis (Meyerhof [1] and Touma and Reese [2]) and three based on the CPT (Schmertmann [3], Dutch [4] and De Ruiter and Beringen [5]. The method developed by

120

A.K.M. Zein and S.B. Mohammad / Prediction of the Axial Capacity of Bored Piles

Van der Veen [6] was adopted for the determination of pile capacities through interpretation of pile load test results. Comparisons were made between the ultimate pile capacities estimated by the five predictive methods and the pile capacities determined from pile load test results. Statistical analysis was made on the predicted and measured pile capacities to assess which of the prediction methods compares favourably with measured pile capacities.

2.

Evaluation of the Bearing Capacity of Bored Piles

The ultimate bearing capacity (Qult) of a pile is the sum of the end bearing resistance (Qp) of the pile tip and the pile skin frictional resistance (Q s) acting along the pile shaft. An evaluation of the bearing capacity of piles is affected by many factors e.g. the soil-pile interaction, the behaviour of the various types of soils, types of loads and pile installation methods. Several methods have been developed for estimating the axial pile capacity which can broadly be classified as direct and indirect methods. Indirect methods require evaluation of the soil characteristic parameters whereas the direct methods utilize in-situ test results directly in the estimation of pile capacity. Direct methods rely mainly on the standard penetration test (SPT) e.g. Meyerhof [1], and O’Neill and Reese [7]) and cone penetration test (CPT). The latter reflects the vertical pile loading mechanism better than the SPT. The most popular CPT based methods include the Dutch [4], DeRuiter and Beringen [5], Schmertmann [3], and the LCPC method (Bustamante and Gianeselli [8]). Most CPT methods relate the base and shaft resistance to the cone resistance (q c) or sleeve friction (fs) using empirical parameters relating pile resistance to soil and pile types. Since every method has been developed under different conditions, including soil and pile type, such factors must be considered in the selection of pile design methods.

3.

Study Sites and Research Methodology

3.1. Characteristics of Study Sites Five of the sites studied are located in Khartoum city namely; (i) New Blue Nile Bridge (NBNB) site (ii) Fly-Over Bridge near railway station (iii) Electricity Load Dispatch Centre building, West Soba district, (vi) Friendship Conference Hall and (v) Byblos Bank Africa building in the city centre. The sixth is located in a sugar factory compound in the White Nile State at about 150km south of Khartoum. Generally, the profiles revealed from boreholes drilled showed that the soils in Khartoum State are alluvial deposits in the upper zone consisting of silty and sandy clays of low to high plasticity or clayey silts underlain by clayey and silty and occasionally gravely sands of different patterns of gradation and variable natural states of compactness in the lower zones. Weathered sandstone and mudstone formation deposits are normally encountered at greater depths (20-30m below ground level). The soils in the White Nile State site consists of an upper layer of very dense clayey sand underlain with a layer of a hard silty clay of high plasticity. The investigation works carried out at these sites included drilling of deep boreholes and laboratory testing of soil samples, performing cone penetration test (CPT) soundings and carrying out pile load tests for determination of the ultimate capacity of bored piles.

A.K.M. Zein and S.B. Mohammad / Prediction of the Axial Capacity of Bored Piles

121

3.2. Field Testing Methodology 3.2.1

Static Cone Penetration Tests

To examine the possibility of using the CPT data for the estimation of the bearing capacity of bored pile, CPT soundings were performed adjacent to the tested piles locations within a distance varying from 5 to 10m. The type of CPT used in this study was a 200kN capacity mechanical machine equipped with an adhesion friction jacket cone and the cone resistance (qc) and sleeve friction (fs) soil parameters were measured at 20cm depth intervals at each test point. 3.2.2

Static Pile Load Tests

Pile load tests were performed using the slow maintained load methods for individual piles and the tests set-up and procedures were in general conformance to the procedure by ASTM D1143. The loading of test piles was applied by using calibrated hydraulic jacks pushed against loaded kentledge platforms in five sites a beam-reaction piles counter force system in one (NBNB) site in Khartoum. The loads were applied in increments of 10-15% of the design load and maintained until the pile settlement rate becomes 0.15mm per hour. During testing, the piles were subjected to maximum loads of about 150 to 200% of the design load. The bored piles considered in the present study ranged in length from 8m to 22m and 0.45 to 1.20m in diameters. The bored pile tested at the NBNB site (22m long and 1.2m diameter) was fully instrumented with electrical pressure sensors consisting of strain gauges installed at different levels distributed along the pile shaft to measure the skin friction resistance in the various soil strata and load cells at the pile base to measure the pile tip resistance. Figure 1 shows the load versus settlement curves for total, end bearing and skin friction resistances of the pile tested at the NBNB site. Different methods can be used for determination of the ultimate pile capacity from the load-settlement curve which provides an important indication of pile load-carrying capacity. In general, there is no unique criterion that can clearly define a “failure load” of a pile. The approach selected for interpreting a load-settlement curve should account for the characteristics of the load-settlement curve and the soil condition. Based on a previous local experience [9], the Chin’s method did not prove successful when applied for interpretation of load tests performed on bored piles. Also, according to Abdelrahman et al [10] the Davisson and De Beer methods need the pile to be loaded to failure to be applicable. In this study, the Van der Veen [6] method was selected as it seems to be the most appropriate for the type of pile loading procedure followed and due the fact that all the piles considered had not reached failure during load testing a case limiting the application of some other interpretation methods. Figure 2 depicts a graphical illustration of how this method can be applied for the determination of pile capacity from pile load test results as described below: • Choose an arbitrary value for the failure load, say (Qv)ult. • Plot {ln (1-Qva/(Qv)ult.)}for different Eva values against pile movement • When the plot becomes a straight line, then the corresponding (Qv)ult. represents the correct failure load as shown in Figure 2.

122

A.K.M. Zein and S.B. Mohammad / Prediction of the Axial Capacity of Bored Piles

Figure 1 Load-settlement curves of pile Test at the NBNB site

4.

Figure 2 Graphical illustration of Van der Veen method

Analysis and Discussion of Study Results

4.1. Predicted and Measured Ultimate Bearing Capacity of Tested Bored Piles For each of the bored pile sites, the soil profile was obtained from borings drilled close to the test pile and the soil properties required for static analysis of bearing capacity such as soil cohesion and angle of internal friction were estimated from the appropriate laboratory shear strength test results for cohesive soils and from field standard penetration test (SPT) for cohesionless soils. The capacity for each pile was estimated using these strength parameters by the Meyerhof [1] and Touma and Reese [2] methods. Similarly, the cone resistance (qc) and sleeve friction (fs) parameters deduced from CPT soundings performed adjacent to pile locations were used for estimation of pile capacities using the Schmertmann [3], Dutch [4] and DeRuiter and Beringen [5] methods. The predicted pile capacities (Qp) obtained as described above were then compared with pile capacities (Qm) measured from pile load test results according to the Van der Veen method. From a comparison of the predicted and measured pile capacities summarized in Table 1 for the six investigated sites, the following general comments may be noted: a) In five out of six sites covered in this study the different prediction methods tend to give comparable performances in estimating pile capacities with an overall discrepancy range of -26 to +12% from the values determined from pile load test results. However, at the FHP site all methods grossly underestimated the pile capacity of the tested pile by indicating pile capacities ranging from 25 to 75% of the corresponding value determined from pile test results. b) The Meyerhof and Touma and Reese methods indicated opposite predictions of pile capacities such that they tend to underestimate and overestimate respectively the capacities determined from the Van der Veen method. c) Schmertmann method based on the CPT data indicated both overestimation and underestimation of pile capacities in the six sites.

A.K.M. Zein and S.B. Mohammad / Prediction of the Axial Capacity of Bored Piles

123

Table 1 Summary of predicted and measured bored pile capacities Method/ Site Location

Bored Pile Capacity (kN)

New Blue Nile Bridge Fly Over Bridge

Qp

Load Dispatch Centre Byblos Bank

Friendshi p Hall Palace White Nile Sugar

Predicted Pile Capacities According to Meyerhof Touma Schmert- Dutch De Ruiter & mann and Reese Beringen 13345

5655

9329

10577

9328

Measured Pile Capacity According to Van der Veen Method 4500

Qs

1493

7779

9805

2229

3949

8000

Qult YƉ

14839 ϰϳϮϬ

13434 ϭϲϵϲ

19133 Ϯϵϳϳ

12807 ϮϵϮϰ

13278 Ϯϵϳϳ

12500

Qs

1150

5384

2684

691

927

Qult Qp

5870 2128

7081 2356

5661 2356

3614 2503

3904 2356

Qs

520

1910

1418

480

856

Qult

2648

4266

3775

2984

3212

Qp

7198

3770

4247

5278

4247

Qs

1891

7860

6587

2070

3009

Qult Qp

9089 91

11630 488

10834 908

7348 1120

7256 234

Qs

765

1921

1650

375

1289

Qult

855

2408

2558

1495

1523

Qp

179

829

1551

2584

429

Qs

983

3396

2294

882

2530

Qult

1162

4225

3845

3466

2959

6000

3500

10000

3500

3000

d) The Dutch and De Ruiter and Beringen CPT based methods compared well with each other but both underestimated the measured pile capacities. 4.2.

Evaluation Criteria for Considered Pile Capacity Predictive Methods

Statistical analysis was used previously to provide a measure of ranking and evaluate the performance of the static and CPT based pile capacity methods for bored and driven piles (Abu-Farsakh et al [11] and El Sakhawy et al [12]). This approach was adopted and followed in this study for the comparison of predicted and measured pile capacities which can be expressed as a (Qp/Qm) ratio. A pile capacity shall be underestimated when Qp/Qm is less than 1 and overestimated when Qp/Qm is greater than 1. Briaud and Tucker [9] proposed that in such analyses one should consider the plots of estimated versus measured ultimate pile capacity together with statistical analysis of the Q p/Qm data. Therefore in order to evaluate the performance of the five different methods for predicting the axial capacity of bored piles the criteria adopted in the present study included: (a) The best-fit lines of Qp and Qm correlation equations with the coefficient of determination (R2). Regression analysis was conducted to obtain the best fit line and R 2 for the relationship between Qp and Qm determined from interpretation of pile load test results. The index of this criterion is denoted as (R1).

124

A.K.M. Zein and S.B. Mohammad / Prediction of the Axial Capacity of Bored Piles

(b)The arithmetic mean and the standard deviation for the total ultimate Q p/Qm values obtained by different methods and criterion index is denoted as (R2), (c)The predicted to measured ratios of the end bearing (Qpp/Qpm) and skin friction (Qsp/Qsm) pile capacity components obtained by different methods from the results of the instrumented pile load test performed at the NBNB site. The magnitudes of Q pp and Qsp of pile tested at the NBNB site were estimated through extrapolation of the loadsettlement curves plotted from pile test results as shown in Figure 1 and the corresponding criterion indices are denoted as(R3) and (R4). (d) The 50 and 90% cumulative probabilities (P50 and P90) of the total bearing capacity (Qp/Qm ) ratio for all tested piles and predictive methods. Based on this criterion, a pile capacity prediction method with a P50 value closer to one and with a smaller P90-P50 difference is considered as the best one. Data for this criterion was obtained from plotting the cumulative probabilities against the Qp/Qm ratio the five predictive methods considered and obtaining the P50 and P90 values. The criterion index is denoted as (R5). An overall ranking index (R), defined as the sum of the ranks from the different criteria shall be used to quantify the performance of each pile capacity predictive method such that the lower the value of this index the better would be the performance of the method and vice versa. 4.3. Discussion of Evaluation Results for Pile Capacity Prediction Methods Based on the performance evaluation results summarised in Table 2 for ranking the five methods applied for bored pile capacity prediction, it may be noted that: a) The five prediction methods indicated different overall performances with overall index (R) values ranging from 6 to 22, noting that the best and the worst scores would be 5 and 25 respectively according to the adopted evaluation criteria. b) The Touma and Reese method gave the overall best performance (R= 6) by achieving a very close agreement between predicted and measured pile capacities. Table 2 Evaluation of performance of different capacity prediction methods for bored piles Evaluation criterion

(a) Best fit line and correlation coefficient

Prediction method

Meyerhof Touma and Reese Schmertman Dutch DeRuiter and Beringen

R2 Qp/Qm 1.00 0.88

(b) Mean ( μ ) and standard deviation (σ ) of Qpt/ Qmt

Data of instrumented pile at NBNB site (c) end (c) skin bearing friction

R1 1

μ σ 0.74 0.36

R2 5

Qpp /Qpm 2.9

Qsp R3 /Qsm 5 0.19

(d) Cumulative probability at P50 P90

R4 4

0.8

1.1

Overall Ranking Results R5 R 3 18

Final Rank 4

1.11

0.96

2

1.12 0.24

1

1.25

1

0.97

1

1.1

1.5

1

6

1

1.32

0.87

5

1.11 0.28

2

2.1

2

1.22

2

1.1

1.5

1

12

2

0.86

0.87

4

0.8 0.27

4

2.4

4

0.28

5

0.8

1.2

5

22

5

0.86

0.87

3

0.8 0.24

3

2.1

2

0.49

3

0.8

1.1

3

14

3

c) The two methods proposed by Schmertmann, DeRuiter and Beringen based on the CPT data yielded similar overall evaluation ratings in predicting pile capacities and ranked as numbers 2 (R=12)and 3 (R=14)respectively.

A.K.M. Zein and S.B. Mohammad / Prediction of the Axial Capacity of Bored Piles

125

d) The Meyerhof method revealed different ranks pertaining to the five evaluation criteria but all other methods showed consistency in relation to their performance toward individual criteria. e) The Dutch method showed the lowest overall performance based on the evaluation criteria used in this study, possibly due to the small skin friction contributions (16 to 28%) to the total pile capacity obtained by this method in all sites. This may be compared to 45% and 80% skin friction contributions obtained from load test at the NBNB site and predicted by Touma and Reese method respectively. f) The Touma and Reese method showed a reasonable overestimation of the pile end bearing capacity at the NBNB site with a (Qpp/Qpm) ratio equals to 1.25. On the other hand, all other methods indicated gross overestimation of the measured pile end bearing capacity value. g) Touma and Reese method gave an excellent prediction of the pile skin friction capacity (Qsp/Qsm = 0.97) followed by Schmertmann method which indicated a reasonable comparison between Qsp and Qsm at the NBNB site. However, the other three prediction methods revealed extremely low (20-50% of total pile capacity) and unrealistic estimates of Qsp compared to the corresponding Qsm.

5

Conclusions

The following findings may be drawn from analyses of the study results and the evaluation of the performance of five pile capacity prediction methods in estimating axial capacities of bored piles constructed in Sudanese soils: a) Generally, the prediction methods showed comparable overall performances in estimating the total capacity of bored piles in five sites with a small discrepancy between measured and predicted capacities. b) The methods by Meyerhof, Dutch engineers and DeRuiter and Beringen tend to underestimate the total bearing capacity of piles whereas the Touma and Reese method tends to slightly overestimate it. The Schmertman’s method showed overestimation of pile capacity in some sites and underestimation in others. c) Based on the criteria adopted for method performance evaluation, the Touma and Reese method ranked best overall by indicating a very close agreement between estimated and measured total pile capacities. The Schmertman and DeRuiter and Beringen methods revealed similar evaluation ratings and ranked in the second place. d) The Dutch method showed the poorest overall performance in predicting total pile capacity by indicating gross under-prediction of the skin friction component. e) From comparison of the end bearing and skin friction capacity components determined from results of the instrumented pile tested at the NBNB site and the corresponding values estimated from the five prediction methods, it was found that: • The best estimate of the pile end capacity component (Qp) was obtained by Touma and Reese whereas all other methods revealed gross over-prediction of Q p values. • For the pile skin friction (Qs), an excellent prediction was obtained by the Touma and Reese method while Schmertman method gave a reasonable agreement with measured Qs. However, the other three prediction methods gave extremely low estimates of the skin friction component of pile capacity. On the basis of the findings of this study, it may be suggested to use the Touma and Reese method for estimating the bearing capacity of bored pile installed in Sudanese

126

A.K.M. Zein and S.B. Mohammad / Prediction of the Axial Capacity of Bored Piles

soils. However, in view of the small number of piles tested and considered in analysis, these findings should be considered as indicative at this stage. In fact, a more comprehensive research work on this topic is currently in progress at BRRI to verify or modify, if necessary, the findings presented in this paper.

References

[1] Meyerhof, G.G. “Bearing Capacity and Settlement of Pile Foundations” The 11th Terzaghi Lecture, Journal of Geotechnical Engineering Division, ASCE, Vol. 102, No. GT3, 1976, pp. 195-228 [2] Touma, F.T.,and Reese, L.C.(1974) “Behavior of bored piles in sand” J. Geotechnical Eng. Div., ASCE, Vol. 100, GT7, pp. 749-761. [3] Schmertmann, J. H. (1978). ‘‘Guidelines for cone penetration test, performance and design.’’ Report No. FHWA-TS-78-209, US Department of Transportation, Washington, D.C. [4] De Ruiter, J. and Beringen, F. L. (1979) ‘‘Pile foundations for large North Sea structures’’, Mar. Geotech., Vol 3, No. 3, pp. 267–314. [5] Sanglerat, G., (1972). “The Penetrometer and Soil Exploration” , Elsevier Publishing Co., Amsterdam. [6] Van der Veen, C. “The bearing capacity of a pile” Proc Of 3rd Int. Conf. on Soil Mechanics and Foundation Eng., Vol. 2, Zurich,1953, pp. 84-90. [7] O’Neill, M., and Reese, L. (1999) “Drilled Shafts: Construction Procedures and Design Methods”, Volume II Chapters 9–19. U.S. DOT, FHWA, Washington, D.C [8] Bustamante, M. and Gianeselli, L. (1982) ‘‘Pile bearing capacity predictions by means of static penetrometer CPT.’’ Proc., 2nd European Symposium on Penetration Testing, ESOPT-II, Amsterdam, The Netherlands, Vol. 2, 493–500. [9] Ali H.A.B. (2007) “Use of CPT data for prediction of bearing capacity of bored piles in Khartoum City”, Unpublished M.Sc. thesis, BRRI, Univ. of Khartoum. [10] Abdelrahman G.E., Shaarawi E.M. and Abouzaid K.S. (2003) “Interpretation of axial pile load test results for continuous flight auger piles” Proc. 9th Arab Structural Eng. Conf. Abu Dhabi, pp. 791-801. [11] Abu-Farsakh, M.Y., Titi, H.H., (2004) “Assessment of direct cone penetration test methods for predicting the ultimate capacity of friction driven piles” J. of Geotech.l and Geo-environmental Engineering, ASCE 130 (9), 935–944. [12] El-Shkhawy N.R., Youssef K.M. and Badawy R.A.E. (2008). “Prediction of the axial bearing capacity by five cone penetration test based design methods” Proc. 12th Int. Conf. on Computer Methods and Advances in Geomechanics, Goa (India), pp. 3415-3423.

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 127 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-127

Case History On The Design Of Foundation For Oil Storage Tank on Coastal Plain Sands a

E. A. J. Georgea,1, T. J. Atuboyediab,2 and M. Ojua,3 Enoch George Associates Limited, Port Harcourt, Nigeria b University of Port Harcourt, Port Harcourt, Nigeria

Abstract. A tank farm to contain four oil storage tanks was proposed for an area lying close to the Abonnema Wharf in Port Harcourt, Nigeria. The intended tanks would be 25.0m diameter and 13.5m high. For a tank diameter of 25.0m, two boreholes to depth of 30.0m and three cone penetration tests with pore pressure measurements were considered adequate for the subsurface investigation. The subsurface investigation in the tank site disclosed that the upper portion of the soil deposits in the area is composed essentially of Coastal Plain Sands sediments with few intrusions of Mangrove Swamp soils. Beneath the upper clayey deposits is a prevalent deposit of sands. The Coastal Plain Sands formation is composed of sandy, clayey sediments which have been laterized through the leaching of silica and the concentration of alumina and iron. The laterization is generally enhanced by tropical weathering and where the sediments are sufficiently dry, the weathering process produces iron oxide as cementing agent in the soil profile causing the soil to be brown or reddish brown. The project area is characterized by a thick layer of this soil. From the analyses of the results of the laboratory tests, high total and differential settlement of the tanks would be expected if the existing ground was not treated. Ground improvement techniques such as replacement of the weak soils with competent soils, preloading of the tank area with a surcharge and controlled loading of the tank foundation during water testing were considered. Evaluation of the feasibility of adopting pad or concrete ring or concrete raft foundation for the tanks was also carried out. In this paper, is presented the ground improvement technique adopted, the foundation type selected, the bearing capacity and settlement analyses and the results of the water testing of the tanks. Keywords. Ground Improvement, Preloading, Water testing, Bearing Capacity, Settlement.

Introduction Oil storage tanks as the name implies are tanks used for the storage of petroleum products. There are different environmental regulations applied in the design and operation of these tanks. These regulations depend on the nature of the liquid to be stored in the tank. The tanks can be elevated or placed underground. Tanks could have different shapes and sizes but are essentially cylindrical in shape, erected perpendicular 1

Enoch George Associates Limited, Email: [email protected]. University of Port Harcourt, Email: [email protected]. 3 Enoch George Associates Limited, Email: [email protected]. 2

128

E.A.J. George et al. / Case History on The Design of Foundation for Oil Storage Tank

to the ground with flat bottom and fixed or floating roofs. Tanks are mostly grouped together in a formation which gives them the name of tank farms and are often in close proximity to water ways used for the transportation and delivery of the products stored. A tank farm to contain four oil storage tanks was planned for an area close to a derelict railway line behind the Nigerian Port Authority in Port Harcourt and also near to the bank of the Bonny River. The tanks will be 25.0m diameter and 13.5m high and to be enclosed with a concrete wall. The geology of the area revealed that project site falls wholly within the Coastal Plain Sands formation. No storage tank has been constructed in this vicinity as to provide some knowledge of the behavior of tanks founded in this geological formation. Ground investigation was carried out with the brief to assess the prevailing subsoil conditions at site and to provide pertinent soil parameters for the design of foundations for the tanks. The investigation disclosed the near-surface soils are relatively weak and highly compressible and cannot support the tank load without some treatment of the soil. This paper discusses the results of the ground investigation, the foundation type selected and the performance of the tanks during the water testing.

1. Site Description and Geology The project site lies close to a disused railway line which previously linked Port Harcourt a sea port with the hinterland. As usual, the site is on close proximity to a river, the Bonny River. The geological records of the area indicated that the intended tank farm will lie within the depositional environment of the Coastal Plain Sands with the mangrove swamps which rim the north bank of Bonny River nearby. The Coastal Plain Sands is also known as Benin Formation. The surface sediments of this formation consist of sands, clays and sandy clays (Whiteman,1982). The clayey sediments occupy the upper portion of the surface deposits followed by sands of substantial thickness. The clays have been laterized through the leaching of silica and the concentration of alumina and iron. The laterization is often enhanced by tropical weathering. From the x-ray diffraction analysis carried out on samples collected from the clayey sediments of the Coastal Plain Sands in the neighborhood of the project area it was found that the soil contained 45% of quartz, 29% of smectitic clay, 23% kaolinite and 3% of goethite (Skipper et al.2004) The site is characterized by relatively thick layer of brown to reddish brown laterized sandy clays resting on sand deposit.

2. Ground Investigation For the tank diameter under consideration, it was necessary to explore the soil under the tank to a depth of at least 30m. Two boreholes were sunk to depths of 30m using the boring method involving shell and auger. Two cone penetration tests were also carried out using piezocone with the soundings achieving depth of 30m. The investigation confirmed that the site is underlain by comparatively thick layer of laterized sandy clay about 16m thick with an intrusion of a band of sand about 2m thick found in both boreholes. Beneath the clay is sand deposit which was continuous to the end of the borings.

129

E.A.J. George et al. / Case History on The Design of Foundation for Oil Storage Tank

Samples retrieved from the borings were tested in the laboratory. Index property tests , strength tests including unconfined compression (UC), unconsolidated undrained triaxial (UU) and compressibility tests were performed on the undisturbed cohesive samples. The description of the strata and the summary of the results of the laboratory tests are presented in Table 1. Table 1. Soil Profile and Summary of Laboratory Test Results Depth From

Summary of Laboratory Tests Results

Description Moisture of Strata Content To (%)

Liquid Limit (%)

Plastic Limit (%)

Soft clay 22 42 18 Firm 21 44 20 clay Stiff 15 35 15 9.00 11.00 clay Clayey 11.00 13.00 sand Stiff 13.00 16.00 23 55 23 clay Medium dense to 16.00 30.00 dense sand Coefficient of Volume Compressibility mv (m2/MN) Coefficient of Consolidation Cv (m2/yr)

Applied Pressure (kPa)

Cu (kPa)

0.00 2.00

-

2.00 9.00

60 87

mv cv mv cv

2550 0.24 25.7 0.14 30.3

50100 0.16 19.0 0.12 25.6

100200 0.11 18.7 0.10 21.7

200400 0.08 20.9 0.09 24.5

400800 0.06 20.3 0.05 17.7

137

3. Bearing Capacity and Settlement Estimates The soil strata encountered during the ground investigation are described in Table 1 together with the summary of the results of the laboratory tests. The profiles from the cone penetration tests provide continuous information on the thickness of each stratum and the variation in levels which were observed to be small. Correlations in the information from the borings and the cone soundings allowed a design soil stratigraphy to be established. This is described in Table 2. Table 2. Design Soil Stratigraphy. Depths (m)

Stratum 1 2 3 4 5 6 7

From 2.00 9.00 11.00 13.00 16.00 21.00 21.00

To 9.00 11.00 13.00 16.00 21.00 26.00 30.00

Description of Soil Firm clay Stiff clay Clayey sand Stiff clay Medium dense sand Dense sand Dense sand

3.1. Estimation of Bearing Capacity The ultimate bearing capacity of the near surface soil was estimated using the equation proposed by Skempton (1951) where the undrained strength was taken as 50kPa. ⎛ 0.2 D ⎞⎛ 0.2 B ⎞ qu = 5.14C ⎜1 + ⎟ ⎟⎜1 + B ⎠⎝ L ⎠ ⎝

130

E.A.J. George et al. / Case History on The Design of Foundation for Oil Storage Tank

Where qu is ultimate bearing capacity, C the soil shear strength, B and L are diameter of tank and D the foundation depth The results of the calculation showed the safe bearing capacity to be less than the tank load.

3.2. Estimation of Settlement The equation shown below was used to predict the settlement of the tank. S=

Δp.H EC

Where S is consolidation settlement, Δp the vertical increase in stress due to the tank load, H the thickness of loaded soil layer and E c the constrained modulus The details of the calculations are shown in Tables 3 and 4. Table 3. Estimated Settlements for tank (Center of tank).

Location

Layer No.

Soil Description

Depth of layer (m) From

Centre of Tank

Thickness of layer, Z (m)

To

Constrained Modulus Ec (kPa)

Increase in Vertical Stress, ∆p (kPa)

Settlement (mm) Δp.H

S=

Ec

1

Clay

2.0

9.0

7.0

3,927

166.6

2

Clay

9.0

11.0

2.0

5,709

151.3

297.0 53.0

3 4 5 6 7

Sand Clay Sand Sand Sand

11.0 13.0 16.0 21.0 26.0

13.0 16.0 21.0 26.0 30.0

2.0 3.0 5.0 5.0 4.0

17,400 6,944 29,800 54,000 52,000

127.5 115.6 95.2 57.8 44.2 Settlement

14.7 49.9 16.0 5.4 3.4 439

Thickness of layer, Z (m)

Constrained Modulus Ec (kPa)

Table 4. Estimated Settlement for tank (Edge of tank).

Location

Layer No.

Soil Description

Depth of layer (m) From

Edge of Tank

To

Increase in Vertical Stress, ∆p (kPa)

Settlement (mm) Δp.H

S=

Ec

1

Clay

2.0

9.0

7.0

3,927

79.9

142.4

2

Clay

9.0

11.0

2.0

5,709

64.6

22.6

3 4 5 6 7

Sand Clay Sand Sand Sand

11.0 13.0 16.0 21.0 26.0

13.0 16.0 21.0 26.0 30.0

2.0 3.0 5.0 5.0 4.0

17,400 6,944 29,800 54,000 52,000

61.2 57.8 51 39.1 25.5 Settlement

7.0 25.0 8.6 3.6 2.0 211

The constrained moduli used in the calculations were obtained from the correlations between CPT and the drained constrained modulus recommended by Lunne and Christopherson (1983). The vertical stress distribution were obtained from the chart proposed by Foster and Ahlvin (1954)

131

E.A.J. George et al. / Case History on The Design of Foundation for Oil Storage Tank

4. Water Testing 4.1. Procedure for the Test It is usual to test a storage tank after the erection to ensure that it is water tight and that the foundation provided is adequate to carry the tank loadings without distressful settlements taking place. This is a form of preloading of the foundation soils to reduce the post construction settlements. However, the filling of the tank should be carried out under controlled conditions to assess the stability of the tanks. The testing of the first tank is critical because subsequent testing arrangements for the other tanks are adjusted on the light of the first test results. For a tank of 25m diameter eight monitoring points were selected along the circumferential length. Before water was added to the tank, the level at each reference point was taken. Permanent reference levels were established in locations near the tank but unaffected by the tank loading. The test was carried out in steps and each step allowed the filling of the tank to a height of 1.5m the maximum filling rate was not allowed to exceed 200mm per hour. Between two steps the filling was stopped for 24 hours and the settlements measured every 6 hours. After filling the tank to the full water load, the load was maintained for 4 days and measurements of settlements were carried out daily.

4.2. Settlement Records In the water testing carried out, only peripheral settlements were recorded. The measurements were made using simple survey level and were carried out during initial filling and at regular intervals thereafter. The settlements readings are plotted on radial bases around the tanks perimeter and shown in Table 5 and Fig.1. This plotting has the advantage that a visual impression of the peripheral settlement is immediately available. For an example, the settlement rings plotted for Tank 1 show that the north east and south sides of the tank settled more compared with the settlements obtained in the east and west sides of the tank. Table 5. Results of the Water Testing. Tank

1

2

3

Filling Level (%) 0 25 50 75 100 100 0 25 50 75 100 100 0 25 50 75 100 100

Settlement Readings (mm) 1

2

3

4

5

6

7

8

0 15 64 118 152 152 0 12 82 161 260 260 0 10 83 157 200 200

0 15 88 144 183 183 0 15 84 161 244 244 0 10 79 153 196 196

0 13 58 100 124 130 0 21 96 175 229 229 0 14 72 151 193 193

0 9 75 123 151 151 0 13 91 159 231 231 0 12 77 157 199 199

0 10 88 140 183 183 0 13 81 159 230 230 0 15 74 156 197 197

0 14 74 119 145 145 0 15 88 167 244 244 0 15 73 152 199 199

0 8 76 121 150 150 0 10 77 164 238 238 0 12 72 156 200 200

0 11 90 141 185 185 0 12 89 168 232 232 0 12 72 147 202 202

132

E.A.J. George et al. / Case History on The Design of Foundation for Oil Storage Tank

4

Settlement Ring 25% Full

0 0 25 15 50 94 75 169 100 230 100 230 Settlement Ring-50% Full

0 8 88 164 229 229

0 9 87 165 235 235

0 11 100 179 241 241

0 15 105 185 263 263

0 14 103 185 251 251

Settlement Ring-75% Full

0 18 105 189 259 259

0 12 97 183 254 254

N

Settlement Ring-100% Full

TANK 1

TANK 3

TANK 2

TANK 4

Figure 1. Typical settlement rings around tank periphery

References [1] BS 1377, 1990-British Standard Methods of Test for Soils for Engineering Purpose. [2] BS 5930, 1999-Code of Practice for Site Investigation. [3] BS 2654, 1989-Specification for Manufacture of Vertical Steel Welded Non-Refrigerated Storage Tanks With Butt-Welded Shells for Petroleum Industry. [4] G. Sanglerat, The Penetration and Soil Exploration, Elsevier, Amsterdam, 1972. [5] J. Skipper, G. Cressey and J. Hugget , Report on Four Soil Samples From Nigeria, Engineering Geology Group, Natural History Museum, London, 2004. [6] A. W. Skempton, Bearing Capacity of Clays, Building Research Congress, London Div. 1, (1951), 180189 [7] T. Lunne, P. K. Robertson and J. J. M. Powell, Cone Penetration Testing in Geotechnical Practice, Spon Press, 2001 [8] T. Lunne and H. P. Christopherson, Interpretation of Cone Penetration Data for Offshore Sands, Proceedings of the Offshore Technology Conference, Richardson, Texas, Paper No. 446, 1983. [9] A. Whiteman, Nigeria: Its Petroleum Geology Resources and Potential. Vol. 1, Graham and Trotman, 1982.

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 133 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-133

Moment-Induced Displacement of Offshore Foundation in the Niger Delta EJEZIE, S.U a,1 and S.B. AKPILAb a

Department of Civil Engineering/Environmental Engineering University of Port Harcourt b Department of Civil Engineering, Rivers State University of Science and Technology, P.M.B. 5080 Port Harcourt.

Astract.

The rotational displacement of offshore shallow foundations on clay due to moment loading has been studied in the Niger Delta Environment. Wave characteristics were deduced from available meteorological and oceanographic data while moments were evaluated from horizontal forces which impact on circular piles of 1.0-2.0 m diameter. The rotational displacement on an equivalent square foundation breadth B ranging from 9.9 m to 17.73 m, typical of circular foundation diameters of 10-20 m, was subsequently evaluated. Undrained shear strength su, of the sub-seabed was obtained from both field and laboratory tests. It is observed that rotational displacement θm1, reduces with increase in foundation breadth B, and Poisson ratio for a given applied moment M. It also reduces as M/B ratio reduces with increasing μ. A dimensionless plot of the ratio of moments to undrained shear strength, foundation breadth and rotational displacement gave values of 18.66 and 37.33 at μ = 0 and 0.5 respectively. Keywords. Rotational displacement, Poisson ratio, foundation breadth.

Introduction The oil and gas exploration and production activities are increasingly venturing into the offshore environment. These activities are carried out on offshore structures whose foundations are subjected to a combination of environmental and gravity loads transferred to them in the form of vertical, horizontal and moment loading. Environmental loading is wave-induced and results in lateral (uh) and rotational (θm) displacements, while gravity loading produces vertical displacement (uv). Moment application on foundation results in rotational displacement, although a pure applied moment additionally could result in some horizontal displacement because of cross coupling effect [1]. Moment loads on the foundation were evaluated from horizontal loads impacting on circular piles of 1.0 to 2.0 m diameters. This paper attempts to present preliminary predictive models on moment-induced displacement of offshore foundations in the Niger Delta. 1

S.U .Ejezie: Department of Civil and Environmental Engineering, University of PortHarcourt, , Nigeria

134

S.U. Ejezie and S.B. Akpila / Moment-Induced Displacement of Offshore Foundation

1. Materials and Methods 1.1. Wave Characteristics The wave characteristics, namely wave height and wave period were deduced from relevant meteorological and oceanographic reports of reputable firms [2, 3, 4], while wave celerity, c, and wave length, L, were evaluated for conditions of shallow water waves [5]. Breaking wave height, Hb, and the fractional values were subsequently evaluated . 1.2. Hydrodynamic Coefficients The Inertia coefficients, CM, and drag coefficient, CD, which depends on Reynold’s number, Re, generally lie in the range of 0.8 to 2.0 [6] and are obtained from standard charts. Reynold’s number can be obtained from the expression;

(1) Where U = stream velocity, D= cylinder diameter, and ν = fluid dynamic viscosity. The dimensionless parameters for maximum drag force, KDM, the inertia force, Kim, maximum drag moment, SDM, and maximum inertia moment, Sim, were obtained from charts.

1.3. Hydrodynamic Forces and Moments The total instantaneous hydrodynamic force F on a submerged structure per elemental length, ds of the cylinder is obtained from;

(2)

The maximum horizontal force is obtained by summing both drag force and inertia force as follows:

(3) Horizontal forces are computed for varying pile diameters (1.0 - 2.0m) and wave heights. The maximum moment at mud line is then evaluated from the expression;

S.U. Ejezie and S.B. Akpila / Moment-Induced Displacement of Offshore Foundation

135

(4) Where d = water depth below still water level and

= unit weight of water.

1.4. Subsea Shear Strength Shear strength parameter, su of soil samples from two representative sites located offshore of the Niger Delta were obtained and the shear strength profile with depth generated. Similar studies on other offshore environments have also been reported [7, 8, 9]. 1.5. Moment loading on foundation The follow expression has been given for moment loading [10];

(5) where θm = rotational displacement, M = moment loading, G = shear modulus of soil, R = radius of foundation and μ = Poisson ratio of soil. In this study, Eq. (5) is rewritten for convenience, giving that G = 39su [11] and R = B̸ √π to obtain:

(6) Graphs of µ versus M/suB3θm1 and θm1 versus μ for various M/B ratios were then generated. 2. Results and Discussion 2.1. Wave Characteristics The meteorological and oceanographic data from the offshore Niger Delta gave maximum directional wave height, Hmax of approximately 7.0m, mean wave period of 17 sec and average wind speed of 14.1 m/s. The wave speed and wave length were subsequently obtained as 12 m/s and 206 m respectively. The breaking wave height Hb and its subsequent quarter values are 12 m, 9 m, 6 m and 3 m. 2.2. Hydrodynamic Coefficients The relationship between drag coefficient, CD and pile diameters (1.0-2.0 m), wind speed, u and kinematic viscosity, ν gave a Cd value of 0.7 and inertia coefficient, Cm, of 1.5.

136

S.U. Ejezie and S.B. Akpila / Moment-Induced Displacement of Offshore Foundation

Evaluated values of KDM, Kim, SDM and Sim for different wave heights are shown in Table 1.The dimensionless parameters decreases with wave height and generally, inertia coefficients are smaller than drag coefficients for any given wave height.

2.3. Hydrodynamic Moment In figure 1, it is observed that moments transferred to the foundation increased with pile diameter and wave height. Table 1. Hydrodynamic Coefficients Wave height (m) 12 9 6 3

Dimensionless coefficients Kim KDm 0.72 0.4 0.64 0.39 0.54 0.35 0.42 0.29

SDM 1.12 0.92 0.75 0.62

SIm 0.86 0.76 0.64 0.56

2.4. Subsea Shear Strength The representative soil samples of the studied sites located offshore Niger Delta, generally consist of very soft, grey, normally consolidated (NC) clay exhibiting very small undrained shear strength from mud line to about 9m. Beyond this point, the undrained shear strength starts increasing in value (i.e., su of 2 kN/m2 for z < 9 m and 21 kN/m2 for z > 11 m) as shown in Figure 2.

Figure 1: Variation of moment, wave height and pile diameter.

S.U. Ejezie and S.B. Akpila / Moment-Induced Displacement of Offshore Foundation

137

2.5. Moment and Rotational Displacement of Footing A typical variation of rotational displacement versus μ for various Moment /Breadth (M/B) ratios is presented in Figure 3. It shows that θm1 reduces with increasing foundation breadth, B, decreased applied moment, M, and decreased M/B ratio. A dimensionless plot (Figure 4) of moment to undrained shear strength su, foundation breadth, B, and rotational displacement, θm1, gave M/suB3θm1 of 18.66 and 37.33 at μ = 0 and 0.5 respectively. Hence, rotational displacement reduces with increasing Poisson ratio and for saturated clays (μ = 0.5) its value reduces by about 50% compared to when μ = 0. Rotational displacement at ђ=0 is included for analytical purpose only to start the origin at zero. The generated predictive model has a perfect positive correlation (R 2=1) and is given by: M/suB3θm1 = 68.58μ3-1.175μ2+20.77μ +18.65

(7)

Figure 2: Variation of undrained shear strength, su with Depth

Figure 3: Typical moment load and rotational displacement.

138

S.U. Ejezie and S.B. Akpila / Moment-Induced Displacement of Offshore Foundation

Figure 4 : Moment loading of Foundation 3. Conclusion On the basis of moment induced displacement of foundation on clays offshore Niger Delta, the following conclusions can be drawn; i) For a given M/B ratio, θm1 reduces with increasing Poisson ratio. ii) With increasing μ, θm1 reduces as M/B ratio reduces. iii) Dimensionless plot of M/suB3θm1 versus μ gives values of 18.66 and 37.33 at μ = 0 and 0.5 respectively; for saturated clays (μ = 0.5) its value reduces by about 50% compared to when μ = 0 (NB: μ = 0 is introduced for analytical purpose only in order to have the origin at zero. Otherwise, μ = 0 has no significance). iv) The generated graphs can be used as preliminary predictive tools in assessing the performance of offshore foundations on clays under wave loading in the Niger Delta. References [1]

Bell, R.W., The Analysis of Offshore Foundations subjected to Combined loading, M.Sc thesis, University of Oxford, U.K, 1991. [2] Exxon Mobil, Eastern Nigerian Shallow Water Metocean Criteria: Version 1.2, 2002. [3] Chevron Texaco, Metocean and Hydrodynamic Criteria for Shallow fixed Structures and Pipelines off W.Africa; Revision 11, 2004. [4] Noble Denton: Adanga Field Development Metocean Study, Report, No. L21208/NDE/RS, 2005.. [5] Sorenson, R.M.: Basic Coastal Engineering, 3th edition, Spring science and Business media, New York.Spencer, 2006. [6] Haritos, N., Introduction to the analysis and design of offshore structures – An overview, EJSE, Special Issue: Loading in Structures pp.55-65, 2007. [7] Bradshaw, A.S., Silva, A. J., and Bryant, W. R., Stress-Strain and Strength behavior of Marine clays from Continental slope, Gulf of Mexico, Engineering Mechanics Division, ASCE (EMD 2000) Conference,Austin, TX. [8] Aubeny, C. P., Moon, S.K., and Murff, J. D., Lateral Undrained Resistance of Suction Caisson Anchors, International Journal of Polar Engineering, 11(2), 95-103,2001b. [9] Sharma, P.P.: Ultimate Capacity of Suction caisson in normally and lightly Overconsolidated clays, Msc. Thesis, Texas A and M University, 2004. [10] Poulos, H.G. and Davis, E.H., Elastic Solutions for Soil and Rock Mechanics, John Wiley, New York, 1974. [11] SNAME : Guidelines for Site Specific Assessment of mobile Jack-up Units”, Society of Naval Architects and Marine Engineers Technical and Research Bullitin 5-5A, New Jersey,1994.

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 139 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-139

Lateral Response of Suction Caissons in Deep Water Floating Structures off Niger Delta Coast Samuel U. EJEZIE and Baribeop KABARI Department of Civil & Environmental Engineering University of Port Harcourt, Nigeria

Abstract. The lateral response of suction caissons used as anchors for floating structures in the offshore Niger Delta has been investigated using the “Lumped Parameter Systems” model. In this process, the dynamic stability (horizontal vibration) of suction caissons used to anchor floating production facilities located deep offshore of the Niger Delta has been examined. Geotechnical conditions prevalent at Niger Delta Deep offshore have been used to determine dynamic soil parameters needed for analyses. Also, dynamic wave properties of the offshore environment which correspond to 100 years return period have served as inputs into the analyses. Results of analyses show that for a given wave condition, an increase in the mass of caisson whose height to diameter ratio is 2:1 causes a decrease in the horizontal amplitudes of vibration of the caissons. Results also reveal that continuous increase in the mass of caisson beyond certain limits does not significantly reduce vibrating amplitude. This is important because it provides information on the limiting mass and hence the size of caisson required in any particular situation. A most striking observation made is the fact that for a given wave steepness, several smaller units of suction caissons can be used rather than a single massive unit. Cases considered show reduction in amplitude by 67%, 41%, 32% and 23% respectively for increase in the number of caissons to 2, 4, 6, and 10. Keywords. Suction caisson, deep water, floating structure, dynamic, Niger Delta

Introduction Renewed interest in the use of suction caissons as an alternative to the conventional pile foundations and gravity structures in deep waters has increased in recent times. The reason for this is due, mainly to the ease with which this type of foundation system can be installed in deep waters compared to driven piles and the stability derived from its adoption. Suction caissons have been used in most of the deep water locations such as the Gulf of Mexico, North Sea and the Gulf of Guinea. In most cases, the design of the caissons has been carried out following the methods used in pile design for onshore conditions. In determining the loads on floating structures and hence forces exerted on suction caissons, consideration has been given to deep water waves which are known to be erratic in nature. This research work studies the lateral response of suction caissons with emphasis on dynamic stability, using the “Lumped Parameter Systems” approach. ________________________________

1 Samuel U. EJEZIE: Department of Civil & Environmental Engineering, University of Port Harcourt, Nigeria; E-mail: samejezie @yahoo.co.uk.

140

S.U. Ejezie and B. Kabari / Lateral Response of Suction Caissons

1. Study Location The study area is located in the offshore of the Niger Delta region of Nigeria. The geotechnical and wave loading conditions here are the same as in other adjoining deep water sites like the Gulf of Mexico and the Gulf of Guinea. The Niger Delta is the oilrich region of south-Eastern Nigeria situated between latitudes 4o N and 12o N and longitudes 5o E and 8o E.

Niger Delta Niger Delta offshore Gulf of Guinea

Figure 1: Location Map showing Niger Delta offshore

2. Nature of Niger Delta Sea Bed Soils It is generally well-known that the deep ocean floor is basically flat and covered by very fine grained sediments. These sediments consist primarily of brown clays as well as calcareous and siliceous ooze with thicknesses ranging from about 300m to 600m. Their accumulation rates range from less than a millimetre per thousand years in the deep sea to a few tens of centimetres per year in the near-shore areas [1]. Specific geotechnical investigations have been carried out by individuals and bodies to determine the properties of the sea bed soils for purposes of the design of suction caissons. Typical results obtained from laboratory tests conducted by researchers [2] on soil samples collected below water depths of 150m to 250m revealed the seabed conditions at the Nkossa oil and gas field in the Gulf of Guinea adjourning the Niger Delta offshore. Within this depth range, several geotechnical borings revealed the presence and predominance of soft, normally consolidated clays [2]. Similar soil conditions are present at the Marlin site in the Gulf of Mexico [3].

3. Wave Forces Computation of wave forces on an offshore structure is more accurately done using the diffraction theory when the structure is large compared to the wave-length [4]. The development of wave forces on a half-submerged cylinder utilizes the two-Dimensional Source-Sink method which is a two-dimensional diffraction theory applicable to structures whose dimensions in the direction normal to the wave approach are large compared to other dimensions. The source-sink method solves problems of wave

141

S.U. Ejezie and B. Kabari / Lateral Response of Suction Caissons

forces within the framework of the linear theory with the usual assumptions of incompressible and irrotational flow. The maximum horizontal and vertical forces are shown in Figures 2 and expressed as: Fx

=

F x ( max ρ gaH

) / 2

(1)

Where F x is force coefficients read from Figure 2, ρ g is unit weight of water, a is cross sectional radius of two dimensional floating structures, H is the double amplitude of water wave.

Figure 2: Normalized maximum horizontal force on a half-submerged cylinder 3.1. Deepwater Wave Forces Deepwater waves exhibit the following characteristics: amplitude, H ranges between 18.3 – 24.4m, Steepness, S ranges between 1/11 and 1/15 [5]. The magnitude of wave forces depends primarily on the wave amplitude, frequency and the relative dimension of the structure to the size of the wave.

4. Use of Suction Caissons as Anchors Suction caissons also called “pneumatic caissons”, “buckets”, “skirted foundations” or “suction anchors” are steel cylinders with thin walls, closed at the top which are driven initially into the seabed by self weight and subsequently by suction pressure created in the interior of the caissons by water pumps. Available report of previous works shows that larger penetration to diameter ratios are required in soft clay deposits for sufficient mobilization of holding capacity since shaft friction, though relatively small, generally tends to improve with depth. Sufficient suction pressure is however needed to overcome the high resistance posed by stiff clays and sands [6]. To overcome the

142

S.U. Ejezie and B. Kabari / Lateral Response of Suction Caissons

resistance offered by stiff soils requires high suction pressure which is accommodated by larger diameter suction anchor.

5. Dynamic Soil Parameters These parameters have been Computed based on the geotechnical properties of subbottom soils offshore Niger Delta. The upper value of specific gravity for organic clay of 2.65 shall be adopted for purposes of computing void ratio and hence the dynamic shear modulus [7]. The masses of caissons for different sizes are also determined. Dynamic parameters such as mass ratio, damping ratio, spring constant and modified spring constant which take into account the embedment depth of caisson are also computed for different geometric sizes of caisson. Table 1 shows the computed combined mass of the caisson and soil plug for increasing caissons size and constant height to diameter ratio (Hc:D) of 2:1. The combined weight of the caisson is the sum of the weights of the soil plug and the steel material. The table also shows the computed dynamic parameters for horizontal oscillation. The modified spring constants as calculated take into account the depth of embedment of caisson. Table 1: Dynamic Parameters Hc (m)

D (m)

Combined mass (kg) x103

2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 11.0 12.0

1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0

1.08 4.18 8.54 15.07 24.16 36.24 52.39 71.40 94.93 122.51 155.61

Dynamic shear modulus G, kPa 2294.2 2294.2 2294.2 2294.2 2294.2 2294.2 2294.2 2294.2 2294.2 2294.2 2294.2

Mass ratio Bx

Damping coefficient Dx

4.67 5.36 4.62 4.17 3.87 3.66 3.54 3.39 3.27 3.19 3.12

0.13 0.12 0.13 0.14 0.15 0.15 0.15 0.16 0.16 0.16 0.16

Spring constant Kx, (KN/m) 6117.8 9176.8 12235.7 15294.7 18353.6 21412.5 24471.5 27350.4 30589.3 33648.3 36707.2

Modified spring constant K(x)d kN/m 15906.4 23859.7 31812.9 39766.1 47719.4 55672.6 63625.8 71579.0 79532.3 87485.5 95438.7

6. Dynamic Stability Analyses Figure 3 shows a system representing the equivalent Mass-Spring-Dashpot Model for suction Caisson. Horizontal X

Fx

Kx/z

Kx/z M Cx/z

Cx/z

Figure 3: Mass-spring-dashpot System

143

S.U. Ejezie and B. Kabari / Lateral Response of Suction Caissons

The model, as depicted, simulates a vibrating soil-caisson-system. When the driving force, Fx causes a sinusoidal motion of the mass, the motion at any time, t, is resisted by a combination of the following forces: Retarding force, m &x& , Damping force, cx x& and Restoring force, k x x . For conditions of equilibrium the expression for horizontal oscillation becomes:

m&x& + c x x& + k x x = Fx sin ω t

,

(2)

where m = mass of caisson, Cx = damping coefficient, Kx = spring constant, Fx = maximum horizontal force, ω = circular frequency of exciting force and x is horizontal displacement. Results of analyses are presented in Figure 4, which clearly shows the trend of the variation of amplitude with increasing mass of caisson.

Figure 4: Amplitude vs. Mass of caisson 6.1 Discussion of Results The values of the amplitude of vibration computed for different caisson sizes (masses) and for different wave loading conditions indicate that amplitude of vibration decreases with increasing mass of caisson. This is in line with what obtains in surface foundations subjected to machine vibrations. It is also observed that the natural frequency of the caisson decreases with increasing foundation mass even though the spring constant increases. The decrease in the natural frequency generally increases the

144

S.U. Ejezie and B. Kabari / Lateral Response of Suction Caissons

frequency ratio, ω / ω n with consequent amplitude reduction. Thus the entire system may be construed as representing a high frequency machine. Another important observation made is the fact that the amplitude of vibration of the caisson can be greatly reduced if several smaller units of suction caissons are used instead of a single massive unit whose weight equals the combined weight of the smaller units. An increase in the number of caissons from 1(single massive unit) to 2, 4, 6, 10 and 20 (smaller units) reduces the amplitude of vibration by 23, 59, 68, 77 and 77% respectively. Selection of an appropriate number of caissons which represents optimum condition can therefore be made bearing in mind the maximum allowable vibration amplitude as well as the cost implications.

7.

Conclusion

Considering the high level of economic risk associated with failure of deepwater floating facilities there is need to thoroughly examine and analyze dynamic wave loading conditions and their effects on anchoring systems to ensure safety. Ocean waves exert enormous hydrodynamic forces on offshore structures which are transmitted to the foundation via mooring lines. The stability of the foundation therefore determines the overall stability of the entire floating system. The “Lumped Parameter System” adopted in this work constitutes an appropriate model for analysis and design of suction caissons since the obtained results are in agreement with surface foundations subjected to dynamic loads. These results show that for a given wave condition, an increase in the mass of caisson whose height to diameter ratio is 2:1 causes a decrease in the horizontal amplitudes of vibration of the caisson. They also reveal that continuous increase in the mass of caisson beyond certain limits does not significantly reduce vibration amplitude implying that there is a limiting mass and hence size of caisson required in any particular situation. Another striking inference is that for a given wave steepness, several smaller units of suction caissons can be used rather than a single massive unit.

References [1] [2] [3] [4] [5] [6] [7] [8]

Mitchell, J. K. (1993): “Fundamentals of Soil Behaviour” 2nd edition, John Wiley and Sons, New York, USA. Colliat, J. L., Boisard, P., Sparrevik, P., and Gramet, J. C. (1998): “Design and Installation of Suction Anchor Piles at Soft Clay Sites” ASCE, 7 Geotech Engrg. 114(5). Jeanjean, P., Knutt, A. H., and Kalsnes, B. (1998): “Soil Parameters for Design of Suction Caissons for Gulf of Mexico Deepwater Clays” Proc. Offshore Technology Conference 8830, Houston, Texas. Chakrabarti, S. K. (2003): “Hydrodynamics of Offshore Structures” Reprint edition, WIT, UK API (2000): “Recommended Practice for Planning, Designing and constructing fixed Offshore Platforms” 2nd Edition, API Publishing, Washington, D. C. Sukumaran, B. (2004): “Suction Caisson Anchors – A Better Option for Deepwater Applications” Rowan University. Bowles, J. E. (1997): “Foundations Analysis and Design” 5th edition, McGraw Hill, New York, USA. Sharma, P. P. (2004): “Ultimate Capacity of Suction Caisson in Normally and Lightly Over Consolidated Clays”. Ms Thesis, Civil Engineering, Texas A & M University.

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 145 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-145

Foundation design and construction for an LPG terminal in a difficult geology and constrained waterfront in coastal Lagos Olaposi FATOKUN and Gianguido MAGNANI TREVI Foundations Nigeria Ltd, Lagos, Nigeria

Abstract. LPG (liquefied petroleum gas or cooking gas) domestic consumption, despite the abundance of natural gas in Nigeria is still very low. Nigeria is the second most endowed nation in Africa with proven reserve of over 260 trillion cubic feet of natural gas either associated with petroleum or occurring on its own gas field, even as up to 2 million cubic feet is flared off daily as crude oil is drilled making Nigeria the largest gas flaring country in the world. The low gas consumption is a result of lack of adequate downstream LPG handling infrastructures to take bulk LPG gas from mother vessels to onshore locations where gas can be easily transported for domestic use.

In 2009, TREVI Foundations Nigeria was commissioned to design and construct all waterfront facilities for a major LPG terminal in coastal Lagos. Subsoil investigation data proved the site is underlain with recent organic clays and young sediments in shallow water ranging between 1.2m to 6.0m. The requirement was to design berthing and mooring facility capable of handling up to 60,000 metric ton LPG gas vessel with water intake channel, fire fighting facility and (water reservoir) pit. In all, the following were installed at this site: 28No. 1.5m bored piles seated at between 35m and 40m below the river bed to support the berthing and mooring dolphins, 10No. 600mm diameter open bottom steel cased piles to support the intake pit, 247No. mono fluid jet grouted (JG) columns acting as bottom plugs to between 11m and 13m below the river bed to seal the bottom of the pit and make excavation works easy in marine environment and 184 elements of precast 600mm wide by 300mm thick by 10-13.0m long reinforced concrete plates with joints grouted installed by vibro jetting to form the intake channel wall. TREVI proprietary software and PLAXIS 8.6 code were utilized for the design and simulation of the various foundation and structural elements. The project was completed in a record time of 5½ months, handed over to the client and is already in use. Keywords. LPG terminal, coastal sediments of Lagos, berthing and mooring facility, fire fighting facility, intake channel and pit, Plaxis 8.6 code.

Introduction The Apapa LPG terminal is constructed to discharge up to 60,000 DWT mother vessels laden with LPG from the oil and gas rich Niger Delta region of Nigeria to the main commercial city centre of Lagos where population growth and urbanization has created an increase in demand for domestic cooking gas coupled with government plan to grow cooking gas consumption by encouraging expansion of LPG infrastructures in the

146

O. Fatokun and G. Magnani / Foundation Design and Construction for an LPG Terminal

country as it would help combat deforestation and minimize emission of green house gases. The client saw this potential and decided to invest in the construction of this facility, thus becoming the largest privately owned LPG terminal in Nigeria today. Apart from the jetty landing facility, the onshore facility consists of spherical gas srorage tanks with connecting pipes linking the jetty head to the loading bay. About 200m shoreline is available to berth discharging LPG vessels.

1. LPG technical requirement LPG is classified as a dangerous material because of its combustive property, causing catastrophic damage to nearby properties if ignited or when it explodes. This hazard therefore makes fire fighting device with ample water supply an essential part of the LPG facility. To save cost, it was decided that the adjoining creek would be ideal to supply the required water for this purpose. Also, there is the need to ensure all-season water for fire fighting in sufficient quantity, even during low tides as a minimum of about 190m3 should be guaranteed during such worst case scenario. Based on this specification, the intake channel and pit were dimensioned. The design of the water intake channel and pit reservoir is detailed in this paper.

2. Geotechnical site characterization and project definition The subsoil geotechnical exploration by Earth surveys[1] revealed the site is underlain directly below by about 7.25m very soft to soft recent silt and organic peat material. To avert equipment sinking, working platform made from free draining sand fill had to be placed prior to commencement of all site works, this fill varied between 600mm and 1.0m thick. The approximate geological stratification detail is given below in Table 1 and Figure 1, with low and high tide river levels at +97.31m and +99.11m respectively: Table 1. Geological stratification (with elevations) Layer (m) 99.047 – 93.547 93.547 – 91.797 91.797 – 82.547 82.547 – 79.547 79.547 – 78.547 78.547 – 76.047 76.047 – 72.797 72.797 – 60.047

Description Very soft SILT Very soft PEAT Loose to medium dense SAND Medium dense SAND Firm CLAY Loose SAND Medium dense SAND Firm to stiff CLAY

Sat. bulk density (kN/m3) 13.0 12.0 17.0 17.5 18.0 17.0 17.5 19.0

The intake channel and pit walls were constructed by vibrating (and jetting) into place precast concrete plate elements to form the confined pathway for water inlet from adjoining creek to the fire fighting facility. The joints between the precast elements were sealed with cement grout. Wall stability was guaranteed using 13m deep by 300mm thick reinforced concrete elements designed as free standing member but braced at the top with concrete beams spaced at 6.5m centers - for the intake channel.

O. Fatokun and G. Magnani / Foundation Design and Construction for an LPG Terminal

147

As stated earlier, the pit is required to guaranty a minimum water reservoir of 190m3 at all times in the chamber, for this to be achieved, the pit was dimensioned 15.3m x 6.9m x 4.2m deep with minimum water depth of 1.863m at low tide. At this depth, excavation for and construction of the 325mm thick bottom slab will generate some ground heave and very difficult under the surrounding hydrostatic pressure respectively. With this anticipated scenario, consideration had to given on the best way to handle the groundwork during the short term phase. Also, there will be need to empty the pit of water for regular post-construction desiltation to remove silts that would be deposited with the lagoon water. The pit slab was constructed from a combination of 175mm thick precast and 200mm insitu structural concrete elements. Figure 1. Site geological stratification.

3. Intake channel pit construction and design To overcome the challenge of floating phenomenon during excavation phase for the pit base construction, 2m to 2.6m deep overlapping single fluid (cement-water) jet grouting treatment was carried out between +87.2m/+87.5m and +89.5m/89.7m within the loose to medium dense sand proved between +91.797m and +79.547m to act as bottom plug. The primary columns were 2.0m to 2.3m deep while the secondary columns were 2.6m. 247 no. jet grouted treatment columns were carried out over an area of about 109m2 under the footprint of the pit. This procedure assisted during excavation for the pit slab base construction works, as it drastically reduced the hydrostatic pressure during the short term with total seepage cutoff by the bottom plug with reduced permeability from the sand layer below. During excavation, limited water infiltrating through the diaphragm wall joints was observed, as the pit base slab was constructed with no seepage from the bottom with the help of the bottom jet grout plug. The minor side leakage from the wall will not compromise the serviceability of the structure. The total load estimated from the pit structural elements and water when the pit is filled is 3,450kN. This load is supported using 10 no. 600mm open-ended steel cased

148

O. Fatokun and G. Magnani / Foundation Design and Construction for an LPG Terminal

piles installed seated in the medium dense SAND at +80.347m with cut-off at +95.347m, each mobilizing an allowable bearing capacity of 345kN with predicted settlement of about 3mm under this working load. However, 2 no. of these piles numbered PD5 and PD6 in Figure 2 below had their cut off at +100.347m in order to be used as conduit for collection of water from the pit whenever it is to be cleaned or de-silted. The upper part of these 2no pipe piles was perforated and filled with selected filter materials to reduce the drainage path. The filter media filled pipe pile will guaranty long term stability of the structure as it reduces the water head under the slab. In addition, lifting pump will be switched on to evacuate water for pit de-siltation. 1650

OPEN END PIPE PILE

PIPE PILE SECTION WITH DRAIN

Ø600

PIPE FILTER Ø600

Ø600

+95.347 m

+95.347 m

3300

perfored pipe from 89.50 to 92.00 m

+92.000 m

+89.500 m

+92.500 m Filter from 89.36 to 92.5 m

3300

empty pipe

+95.347 m

6000

3600

empty pipe

JET GROUTING BOTTOM PLUG 1650

+89.360 m

Detail A

soil

9000

3300

soil

3300

15000

Steel plate 25 mm th.

+80.347 m

+80.347 m

+80.347 m

Figure 2. General arrangement of pit structure.

The pit global stability was modeled under the following five phases using the Plaxis code 8.6[2]: • Geostatic: shoreline in natural condition • Diaphragm wall: DW and capping beam elements activated • Jet grouting treatment: treated bottom plug cluster activated • Dewatering and excavation: drain activated and soil inside the DW removed • Final filling: filling material upstream the DW activated

O. Fatokun and G. Magnani / Foundation Design and Construction for an LPG Terminal

149

The parameters assigned to the various soil layers, jet grout and other structural elements are: Table 2. Materials parameters for Plaxis modeling. Parameter Type γunsat.. (kN/m3) γsat. (kN/m3) kx (m/day) ky (m/day) einit (-) emin. (-) emax. (-) ck (-) Eref (kN/m2) E50 ref (kN/m2) ν (-) Gref (kN/m2) Eoed. (kN/m2) Einc. (kN/m2) Tstr. (kN/m2) cref. (kN/m2) cinc.. (kN/m2/m) φ (0) ψ (0) yref. Rint.

Conc. Drained 25.0 25.0 0.0 0.0 0.5 1.0 E15 2.5E07 0.2 1.04E07 2.78E07 0.0 0.0 1.0

JG col. Drained 17.5 18.0 0.0 0.0 0.5 1.0 E15 1.0 E05 0.25 4.0 E04 1.2 E05 0.0 0.0 120 0.0 34.0 0.0 0.0 0.1

Firm clay Drained 17.5 18.0 0.0 0.0 0.5 1.0 E15 7.5 E03 0.32 2841 10732.3 0.0 0.0 15 0.0 23.0 0.0 0.0 0.67

Silt Drained 13.0 13.0 0.001 0.001 0.5 1.0 E15 2.0 E03 0.35 740.74 3210.0 0.0 0.0 5.0 0.0 21.0 0.0 0.0 0.67

Peat Drained 12.0 12.0 0.432 0.432 0.5 1.0 E15 4.0 E03 0.34 1492.5 6156.7 0.0 0.0 7.5 0.0 17.0 0.0 0.0 0.67

L. sand Drained 16.5 17.0 0.086 0.086 0.65 0.25 0.92 1.0 E15 3.5 E04 0.2 2.8 E04 0.0 0.0 0.1 0.0 30.0 0.5 0.0 0.67

MD sand Drained 17.0 17.5 0.864 0.864 0.57 0.30 0.90 1.0 E15 3.75 E04 0.2 4.2 E04 0.0 0.0 0.1 0.0 33.0 3.0 0.0 0.67

Mohr Coulomb and hardening soil models were used for the cohesive soils/JG columns and cohesionless sandy materials respectively. At long term, the floating verification of the jet grouting plug results was satisfied with a safety factor significantly larger than 1.0. Summary and conclusion The adopted model and parameters proved reliable, this was confirmed during the short term excavation phase with lateral deformation and seepage into the pit falling below the predicted quantities. The following output results were derived from the modeling: Max diaphragm wall deform, dmax = 65.04 mm, 2.60 m from the top of capping beam. Max shear force, Tmax = 115.86 kN/m at 0.85 m from the top of capping beam. Max bending moment, Mmax = 144.95 kNm/m at 4.07 m from top of capping beam The water discharge at the drain to depress the GWL during pit cleaning, Q tot/m = 47.41 liter/day/m considering a 15 m length pit thus Qtot ≅ 700 liter/day. In addition to these output results, Figs 3 to 5 below are some of the output graphic results, and plan view of the completed water intake channel and pit.

O. Fatokun and G. Magnani / Foundation Design and Construction for an LPG Terminal JET GROUTING COLUMN SEQUENCE - PHASE 1

PHASE 2a EXECUTION OF PRIMARY COLUMNS

PHASE 2b EXECUTION OF SECONDARY COLUMNS

Thickness 2.0 m

Thickness 2.3 m

B

B

A Thickness 2.3 m

600 625

525 700

175

Figure 3. Jet grouting treatment scheme.

2600

2000 300

Ø1 00

0

JET GROUTING COLUMNS Ø 1000 mm

2300

2300

2000

300

300

Thickness 2.6 m

7

A

79

150

O. Fatokun and G. Magnani / Foundation Design and Construction for an LPG Terminal

151

Figure 4. Pit dewatering phase with ground water head.

Figure 5. Plan view of completed intake channel and pit.

References [1] Earth surveys and designs Ltd., Geotechnical engineering report for LPG facility at Creek Road, Apapa, Lagos, Nov. 2008. [2] Brinkgreve R. B. J. & Broere W., Plaxis V8 professional version manual, Delft Netherlands, 2004.

152 Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-152

Variation of Hydrodynamic Forces and Moments on Offshore Piles in the Niger Delta a

AKPILA, S.B and S.U.EJEZIEb

a

Department of Civil Engineering, Rivers State University of Science and Technology, P.M.B. 5080, Port Harcourt. b Department of Civil Engineering/Environmental Engineering, University of Port Harcourt

Abstract. The variation of hydrodynamic forces and moments on different circular pile sections of 1.0 to 2.0 m diameters in the offshore Niger Delta Environment has been investigated. Available meteorological and oceanographic data were used to obtain wave characteristics. Inertia coefficient, cm and drag coefficient, cd were obtained from Reynolds number, Re, while maximum inertia force, kim, maximum drag force, kDm, maximum inertia moment, Sim and maximum drag moment, SDm were evaluated from standard charts. The revealed variation of hydrodynamic forces and moments with pile diameter showed that the horizontal forces and moments increase as pile diameter and wave height increase. Drag dominated forces and moments were obtained for wave heights greater than 3m. Keywords. Amplitude, model, inertia, drag

Introduction The offshore environment is notable for the complex gravity and environmental loading which impact on offshore structures. Gravity loads act vertically and are mostly from self weight of the structure while environmental loadings are generally waves and wind induced. The wave force impacted on an offshore structural element depends upon the geometry (the size of these elements relative to the wavelength and their orientation to the wave propagation), the hydrodynamic conditions and whether the structural system is compliant or rigid. Structural elements that are large enough to deflect the impinging wave (diameter to wavelength ratio, D/L > 0.2) undergo loading in the diffraction regime, whereas smaller, more slender, structural elements are subject to loading in the Morison regime [1]. This paper attempts to present the feature of the variation of hydrodynamic forces and moments on offshore piles in the Niger Delta. _____________________________ 1 S.B. Akpila: Department of Civil Engineering, Rivers State University of Science & Technology, Nigeria; E-mail: [email protected].

S.B. Akpila and S.U. Ejezie / Variation of Hydrodynamic Forces and Moments on Offshore Piles

153

1. Materials and Method 1.1. Meteorological and Oceanographic Data Relevant Meteorological and Oceanographic data from offshore Niger Delta environment were obtained from reputable firms [2, 3, 4]. These have been used to deduce wave characteristics namely, wave height and wave period. 1.2. Wave Classification In small amplitude wave theory [5], the surfaces profile is expressed by;

(1) and from the speed of wave propagation [6], the conditions for shallow water waves [7] yield the following; (2) (3) 1.3. Hydrodynamic Coefficients Inertia coefficients, CM and drag coefficient, CD which are dependent upon Reynold’s number, Re, are evaluated from standard charts. Both coefficients generally lie in the range of 0.8 to 2.0 [1].The breaking wave height and fractional values of the offshore environment were determined from which the dimensionless parameters of maximum drag force, KDM, inertia force, Kim, maximum drag moment, SDM and maximum inertia moment, Sim were also evaluated using standard charts for known relative depth. 1.4. Hydrodynamic Forces and Moments The maximum horizontal force F, can also be expressed as follows:

(4) = unit weight of water, D= pile diameter, FDm = drag force and Fim = inertia force. where Hence, for a known wave height, pile diameter, unit weight of water, C m, CD, KDm and Kim, wave forces were computed using Eq. (4).The relationship between drag force, F Dm, inertia force Fim, total force F, and pile diameter for varying wave heights were also examined and

154

S.B. Akpila and S.U. Ejezie / Variation of Hydrodynamic Forces and Moments on Offshore Piles

relevant graphs generated. The maximum moment on the foundation is evaluated from the expression;

(5) Moments generated at mudline were evaluated and the relationships between wave heights and the moments – drag moment, MDm, inertia moment, Mim and total moment, M – for different pile diameters, D were examined and relevant graphs generated for the offshore environment. 2. Results and Discussion 2.1. Wave Characteristics The meteorological and oceanographic data from the offshore Niger Delta were analysed and a maximum directional wave height, Hmax of approximately 7.0 m, mean wave period of 17 sec and average wind speed of 14.1m/s were obtained. Based on breaking wave height, Hb, and the respective fractions, wave heights of 12 m, 9 m, 6 m, and 3 m were obtained. Also wave speed and wave length were obtained as 12m/s and 206 m respectively. 2.2. Hydrodynamic Coefficients Hydrodynamic coefficients, CD and CM assume a constant value of 0.7 and 1.5 respectively for wave heights varying from 6.0 – 12.0m, pile diameter of 1.0 – 2.0m, wind speed, u of 12 m/s and kinematic viscosity, υ of 9.5x10-7 m2/s. The dimensionless parameters of inertia and drag forces (Kim and KDM) assume constant values for a given wave height on pile diameter range of 1.0 – 2.0m. These parameters also reduce with wave height. Generally, for a given wave height, Kim assumes lower values compared to KDM. A similar trend is also observed on Sim and SDM . 2.3. Hydrodynamic Forces The variation of wave force, F, wave height, H and pile diameter, ʔ is presented in figure 1, where wave forces increase with wave height and pile diameter. For example, a 9.0m wave height generated about 236% increase in hydrodynamic force on a 2 m diameter pile as against the hydrodynamic force on a 1m diameter pile. Some of the generated wave force models are presented in Equations (6) – (8). F (ʔ=2.0) = 0.032H3 + 6.266H2 – 3.655H + 20

(6)

F (ʔ=1.6) = 0.041H3 + 4.644H2 – 3.572H + 14.3

(7)

3

2

F (ʔ=1.0) = 0.040H + 2.57H – 2.943H + 7.52

(8)

S.B. Akpila and S.U. Ejezie / Variation of Hydrodynamic Forces and Moments on Offshore Piles

155

In figure 2, the trend for maximum inertia force, maximum drag force, total force and pile diameter for a 3m wave height shows that both inertia force, Fim and drag force, FDm are equal at pile diameter of 1.3m. These are represented by the models expressed in Equations. (9) – (11). Drag dominated forces are however obtained for higher wave heights and for different pile diameters. Ft = 10.17ʔ2 + 12.72ʔ + 0.128

(9)

Fim = 10ʔ2 + 0.128ʔ- 0.042

(10)

FDm = 13.07ʔ- 0.123

(11)

Figure 1: Variation of wave force, wave height and pile diameter

Figure 2: Variation of wave force and pile diameter on 3m wave height.

156

S.B. Akpila and S.U. Ejezie / Variation of Hydrodynamic Forces and Moments on Offshore Piles

2.4. Hydrodynamic Moments The relationship between moment M, wave height H, and pile diameter ʔ, is depicted in figure 3, where a progressive increase in moment is observed for higher wave height and pile diameter. Some of the generated moment models are as follows: M (ʔ=2.0) = 0.077H3 – 0.022H2 + 1.466H – 0.5

(12)

M (ʔ=1.6) = 0.065H3 – 0.111H2 + 1.511H – 1.2

(13)

M (ʔ =1.0) = 0.042H 3– 0.111H 2 + 0.955H – 0.1

(14)

The trend of maximum inertia moment, maximum drag moment, total moment and pile diameter for 3m wave height is presented in figure 4, where both maximum inertia and drag moments are equal for a 1.4 m pile diameter. Their predictive models for a 3 m wave height are as follows; M = 0.812ʔ2 + 1.311ʔ - 0.082

(15)

Mim = 0.834ʔ2 + 0.037ʔ - 0.032

(16)

MDm = -0.022ʔ2+1.274ʔ–0.05

(17)

For higher wave heights, drag moment exceeds inertia moments for all pile diameters.

Figure 3: Variation of moment, wave height and pile diameter.

S.B. Akpila and S.U. Ejezie / Variation of Hydrodynamic Forces and Moments on Offshore Piles

157

Figure 4: Variation of moment and pile diameter on 3m wave height

3. Conclusion The following conclusions are drawn from the study. (i) Horizontal forces and moments increase with pile diameter and wave height (ii) A 3m wave height produced equal inertia force and drag force on a 1.3m diameter pile, while both maximum inertia moments and drag moments are equal on 1.4m diameter pile. (iii) For higher wave heights, drag dominated forces and moments are obtained for varying pile diameters. (iv) The generated graphs and models may be used as predictive tools in assessing the magnitude of forces and moments on offshore piles due to wave loading in the Niger Delta.

References [1] [2] [3] [4] [5] [6] [7]

Haritos, N., Introduction to the analysis and design of offshore structures – An overview”, EJSE, Special Issue: Loading in Structures, 2007, pp.55-65 Exxon Mobil :Eastern Nigerian Shallow Water Metocean Criteria, Version 1.2, 2002. Chevron Texaco: Met Ocean and Hydrodynamic Criteria for Shallow fixed Structures and Pipelines off W.Africa, Revision 11(2004). Noble Denton: Adanga Field Development Metocean Study, Report, No. L21208/NDE/RS, 2005. Airy, G.B.: Tides and Waves Encyclopadia Metropolitian, Vol 5(1845), Art. 192, pp. 241-396. Dake, J.N.K.: Essentials of Engineering Hydraulics, 2nd edition, Macmillan Press Ltd, London, 1983. Sorenson, R.M.: Basic Coastal Engineering, 3th edition, spring science and Business media, New York, 2006.

158 Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-158

Contribution à l'Analyse du Comportement des Pieux sous Chargement Vertical Analyse d'Une Base de Données Locale Ali BOUAFIAa,1 and Abderrahmane HENNICHEb a Université Saâd Dahleb de Blida, Algérie b Ecole Nationale Polytechnique d'Alger, Algérie

Abstract. La communication présente les résultats d'analyse d'une base de données locale relative au nord Algérien, constituée de 39 projets de fondations sur pieux dans lesquels 54 essais de chargement vertical statique ont été menés. Le premier volet comporte une interprétation des essais de chargement des pieux, à la base des courbes normalisées de chargement. Il a été constaté que la charge critique correspond à un tassement assez proche de celui du tassement de référence, défini comme étant le rapport de la charge limite et la pente initiale de la courbe de chargement. En deuxième partie, une étude comparative des méthodes courantes de dimensionnement à la base des essais in-situ a été menée, en l'occurrence les méthodes pressiométrique et pénétrostatique. Keywords. Pieu, capacité portante, essai de chargement, charge critique.

Introduction Le développement croissant en matière d'infrastructures en Algérie a souvent conduit à l'implantation des ouvrages fondés sur pieux. La difficulté des conditions géotechniques de certains sites est une source d'incertitudes quant à la conception et au dimensionnement des pieux. S'inscrivant dans une logique pragmatique, l’essai de chargement d'un pieu en vraie grandeur est un outil pratique d’analyse in-situ de la capacité portante et du tassement d’un pieu isolé qui permet de s’affranchir des incertitudes liées au calcul. Il est systématiquement mené dans le cadre d’un important projet afin de confirmer les prévisions du comportement du pieu, ou chaque fois que les méthodes de calcul mènent à des résultats entachés d’incertitudes. Lorsque le pieu d’essai est seulement instrumenté par des comparateurs pour la mesure du tassement, l’évaluation séparée de la résistance en pointe et du frottement latéral n’est pas possible, mais on peut néanmoins interpréter la courbe de chargement obtenue pour une estimation globale de la capacité portante. La littérature des pieux jalonne d'une diversité d'approches empiriques ou semiempiriques d'interprétation des résultats d'essais sur de tels pieux, en vue de l'évaluation de la capacité portante [1], [2].

1

Corresponding Author: Ali Bouafia, E-mail: [email protected]

A. Bouaﬁa and A. Henniche / L’analyse du Comportement des Pieux sous Chargement Vertical

159

La base de données analysée dans cette communication comporte 54 essais de chargement de pieux simplement instrumentés, réalisés dans le cadre de 39 projets de fondations sur pieux localisés au nord Algérien. Dans l'ensemble, on note que 50% des sites sont de nature argileuse, 30% de nature marneuse et 20% sableuse. En outre, 52 pieux sont en béton armé installés par procédé de forage simple au tube plongeur, alors que 2 pieux sont installés par battage. Notons que 53% de pieux ont un diamètre de 1.2 m, 30% un diamètre de 1.0 m et le reste des diamètres variant entre 0.17 et 0.8 m. Les élancements de pieux varient entre 9 et 43, avec une valeur exceptionnelle de 91 pour un pieu battu. Tous les essais de chargement ont été menés conformément à la norme française NF P 94-150. La figure 1 illustre un dispositif typique d'essai de chargement où le massif de réaction est constitué d'un lest de dalles en béton [3]. Dans ce qui suit, on focalise lors de l'interprétation des courbes de chargement sur les concepts de charge verticale critique et du tassement de référence, en vue d'en dégager des conclusions pratiques.

1. Analyse des Courbes de Chargement 1.1. Ajustement des Courbes de Chargement Dans l'ensemble, 87% essais ont été menés jusqu'à des tassements en tête du pieu égaux à 1% du diamètre, alors que dans 13% des cas, le tassement maximal mesuré varie entre 1 et 9% du diamètre, ce qui ne permet pas une appréciation rigoureuse du comportement du pieu en grands déplacements et l'évaluation de la capacité portante. Un tel constat pousse à analyser la courbe de chargement par ajustement et extrapolation. Les courbes de chargement, illustrant la variation de la charge verticale appliquée Q en fonction du tassement mesuré en tête du pieu, soit v0, ont une allure typiquement hyperbolique décrite par l'équation suivante:

Q=

v0

1 v0 + α Ql

(1)

dans laquelle α et Ql correspondent respectivement à la pente initiale de la courbe (ou raideur verticale du système pieu/sol) et à la capacité portante verticale du pieu. L'ajustement au sens des moindre carrés des courbes par l'équation (1) a permis de déterminer les paramètres précédents, le coefficient de régression étant dans presque tous les cas supérieur à 98%, ce qui est le signe d'une excellente qualité d'ajustement.

1.2. Analyse de la capacité portante La capacité portante Ql, déduite de l'ajustement hyperbolique des courbes de chargement et correspondant théoriquement à des tassements infinis, a été comparée à la charge verticale Q correspondant à un tassement égal à 10% du diamètre B du pieu.

160

A. Bouaﬁa and A. Henniche / L’analyse du Comportement des Pieux sous Chargement Vertical

Figure 1. Dispositif expérimental typique d'essai de chargement vertical de pieu.

Il est remarquable, d'après l'analyse statistique illustrée à la Figure 2, que le rapport Q/Ql est en moyenne égal à 0.98, ce qui laisse à conclure que la capacité portante ainsi déduite par la formulation hyperbolique est confondue avec la capacité portante conventionnelle correspondant à un tassement de 10% de B. Un tel constat justifie le recours à l'ajustement par la formulation hyperbolique de la courbe de chargement en vue de l'extrapoler, particulièrement lorsque les grands tassements n'ont pas été atteints, ce qui est le cas des essais analysés ici. 1.3. Analyse de la Charge Critique Qc L'étude de l'évolution des tassements dans le temps pour une charge donnée, montre qu'ils varient linéairement avec le logarithme du temps, selon une pente β pour chaque effort Q. Au niveau de la courbe de la variation de la pente β en fonction de Q, la charge critique (ou de fluage) correspond au point de brisure de cette courbe. Mécaniquement parlant, au-delà de la charge critique, la vitesse des tassements augmente brusquement, ce qui correspond au domaine d’instabilité des tassements dans le temps et la convergence vers la rupture du sol. Comme le montre la 3, le rapport Qc/Ql est pratiquement constant et fluctue autour de 0.45, alors que selon les recommandations de dimensionnement des pieux, telles que le fascicule 62, ce rapport dépend plutôt de la contribution des efforts limites en pointe et en frottement latéral. 1.4. Concept de Tassement de Référence L'équation précédente peut se réécrire comme suit, en introduisant la notion du tassement de référence vr:

v0 Q v = r Ql 1+ v0 vr

(2)

A. Bouaﬁa and A. Henniche / L’analyse du Comportement des Pieux sous Chargement Vertical

161

35

Probabilité (%)

30

N=45 Marge=0,930-0,998 Moyenne=0,980 Coefficient de variation=1,8%

25 20 15 10 5 0 0,92

0,93

0,94

0,95

0,96

0,97

0,98

0,99

1,00

Q/Ql Figure 2. Histogramme de l'analyse statistique du rapport Q(B/10)/Ql

35

N=43 Marge=0,146-0,691 Moyenne=0,45 Coefficient de variation=31%

30

Probabilité (%)

25 20 15 10 5 0 0,0

0,1

0,2

0,3

0,4

0,5

0,6

Qc/Ql

0,7

Figure 3. Histogramme d'analyse du rapport Qc/Ql.

Les termes Q/Ql et v0/vr sont appelés respectivement niveau de chargement du pieu et niveau de tassement du pieu, et le paramètre vr, défini par l'équation suivante est appelé tassement de référence:

vr =

Ql α

(3)

162

A. Bouaﬁa and A. Henniche / L’analyse du Comportement des Pieux sous Chargement Vertical

D'après l'équation (2), le tassement de référence correspond à la moitié de la capacité portante, soit un niveau de chargement de 50%. Il est remarquable que ce niveau de chargement corresponde approximativement, d'après le paragraphe précédent, à la charge critique, ce qui laisse à conclure que le tassement de référence peut être interprété comme étant le seuil d'instabilité des tassements dans le temps. Le tassement d'un pieu sous charges de service ne doit pas ainsi dépasser cette valeur admissible. En outre, dans le cadre d'une formulation élastoplastique de la courbe de chargement, comme le montre la figure 4-a, on remarque que vr correspond plutôt à la limite du comportement linéaire du système pieu/sol et donc au seuil de la capacité portante, soit Ql. L'étude statistique de ce paramètre montre qu'il varie dans une marge de 0.01 à 1.00% de B, avec une moyenne de 0.24% de B. Le caractère dispersé des valeurs analysées montre que le tassement de référence dépend d'autres paramètres de l'interaction pieu/sol tels que la compressibilité relative pieu/sol et éventuellement de l'élancement du pieu. Reprenons l'équation (2) et écrivons que pour un tassement de 10% de B, la charge verticale est en moyenne de 0.98Ql. On en déduit que:

vr = 0.22% B

(4)

Cette valeur correspond d'ailleurs à la valeur moyenne de ce paramètre, et peut être utilisée à titre indicatif comme un ordre de grandeur du tassement de référence. 1.5. Courbe de chargement normalisée En vue de recommander une approche simplifiée d'estimation du tassement d'un pieu isolé sous les charges de service, en phase préliminaire d'un projet de fondations sur pieux, la courbe de chargement a été normalisée et simplifiée, comme le schématise la Figure 4-b, en une courbe tri-linéaire dont la première portion correspond au domaine des petits déplacements du pieu, limité au tassement de référence, soit de 0.2%B.

Figure 4. Courbes normalisées de chargement (a: schéma élastoplastique, b: courbe trilinéaire simplifiée)

A. Bouaﬁa and A. Henniche / L’analyse du Comportement des Pieux sous Chargement Vertical

163

La deuxième portion correspond au domaine des moyens déplacements du pieu, variant entre 0.2% à 10 % de B et qui correspond au dépassement de la charge critique, et enfin la dernière portion correspondant au domaine des grands déplacements du pieu. L'utilité d'une telle courbe est d'estimer rapidement et manuellement le tassement d'un pieu en phase préliminaire des calculs sans avoir recours aux méthodes traditionnelles de calcul des pieux.

2. Étude comparative des méthodes d'évaluation de la capacité portante Un calcul prévisionnel de la capacité portante a été mené à la base des méthodes de calcul à partir des essais pressiométrique (PMT) et pénétrostatique (CPT), conformément au règlement français CCTG-93, fascicule 62, titre 5 [4], [5]. L'application de la méthode pressiométrique sur 15 pieux d'essai a permis de calculer un rapport λ défini par:

λ=

Ql ( PMT ) Ql ( Hyper )

(5)

Comme le montre la Figure 5, ce rapport varie entre 0.54 et 1.40 et prend une valeur moyenne de 1.10. Selon l'histogramme, la probabilité que λ soit entre 0.9 et 1.3 est de 80%.

Ql(Hyper) Ajustement hyperbolique

50

Ql(PMT) A partir du PMT

Moyenne=1,10 Marge=0,54-1,40 Coefficient de variation=25%

Probabilité (%)

40

30

20

10

0 0,4

0,6

0,8

1,0

1,2

1,4

1,6

1,8

Ql(PMT)/Ql(Hyper)

Figure 5. Histogramme de prévision de la capacité portante par le PMT

Les résultats d'application de la méthode pénétrostatique sont résumés à la Figure 6 qui montre que le rapport μ défini par:

μ=

Ql (CPT ) Ql ( Hyper )

(6)

164

A. Bouaﬁa and A. Henniche / L’analyse du Comportement des Pieux sous Chargement Vertical

varie entre 0.7 et 1.14 et prend une valeur moyenne de 0.89. En outre, la probabilité que ce rapport soit situé entre 0.85 et 1.15 est de 64.0%. Il se dégage de cette étude comparative la bonne qualité de prévision de la capacité portante de ces deux méthodes, comparées à celle déduite de l'ajustement hyperbolique des courbes de chargement. Ces deux méthodes sont d'ailleurs largement utilisées en Algérie dans les projets de fondations sur pieux.

Ql(Hyper) Ajustement hyperbolique Ql(CPT) A partir du CPT 25

Moyenne=0,89 Marge=0,70-1,14 Coefficient de variation=17%

Probabilité (%)

20

15

10

5

0 0,6

0,7

0,8

0,9

1,0

1,1

1,2

Ql(CPT)/Ql(Hyper)

Figure 6. Histogramme de prévision de la capacité portante par le CPT

Conclusions Les résultats d'analyse d'une base de données locale, relative aux essais de chargement de pieux, ont été présentés. L'interprétation des essais de chargement des pieux a focalisé sur la capacité portante, la charge critique et le tassement de référence. En deuxième partie, une étude comparative des méthodes courantes de dimensionnement à la base des essais in-situ a été menée, en l'occurrence les méthodes pressiométrique et pénétrostatique. Il a été constaté que ces deux méthodes permettent une prévision de la capacité portante en bonne concordance avec celle de l'ajustement hyperbolique des courbes de chargement.

References [1] M. Cassan, Les essais in-situ en mécanique des sols, Tome 1: Réalisation et interprétation, Eyrolles, Paris, 1988. [2] US Army Corps of Engineers, Bearing capacity of soils, Technical Engineering and Design Guides as adapted from the US Army corps of Engineers; No. 7, ASCE Press, Reston, Virginia, 1993. [3] A. Henniche, Contribution à l'analyse du comportement des pieux sous chargement vertical- Analyse d'une base de données locale, Thèse de Post-graduation à l'Ecole Nationale Polytechnique d'Alger, 2010, 148 p. [4] Ministère de l'équipement, du logement et du transport, Règles techniques de conception et de calcul des fondations d'ouvrages de génie civil, Fascicule 62, titre V, Eyrolles, Paris, France, 1999, 188 p. [5] A. Bouafia, Les essais in-situ dans les projets de fondations, Office des Publications Universitaires OPU, Alger, 2009, ISBN 9961.0.0692.5, 299 p.

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 165 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-165

The Use of Micropiles as Settlement Reducing Elements H.N. CHANGa and T.E.B. VORSTER b Design Engineer, Esorfranki, South Africa b Technical Director, Aurecon, South Africa a

Abstract. The Gautrain Rapid Rail Link (GRRL) is one of ten Spatial Development Initiative (SDI) projects implemented by the Gauteng Department of Public Transport, Roads and Works (Gautrans) to stimulate economic growth, development and employment opportunities in the Gauteng Province. As part of this development initiative, Viaduct 5c (a 3.2 km long viaduct) crosses highly variable and sinkhole prone dolomitic ground conditions in Centurion, Pretoria. The original design for piers 39 and 40 of Viaduct 5c consisted of four large diameter (1.2m) bored piles designed to be end-bearing onto the dolomite bedrock. Due to severe difficulties encountered with the installation of large diameter endbearing piles and the highly variable rock head and ground conditions encountered, an alternative founding solution at piers 39 and 40 was proposed. The alternative solution consisted of a grid of small diameter piles (micropiles) installed as soil reinforcement elements overlain by a compacted soil mattress. As bearing capacity failure of the footing is unlikely, settlement or differential settlement of the footing was the main concern. Due to the complex and variable nature of dolomitic profiles in general, settlement of the foundation was estimated using a 3D finite element model. The analysis indicated that the micropile/soil mattress system was effective in reducing the settlement of the foundation to an acceptable value. This paper presents the design methodology and analysis that was carried out for the two Piers. Keywords. micropile, settlement-reducing elements, Gautrain, dolomite, finite elements.

Introduction The Gautrain Rapid Rail Link (GRRL) is one of ten Spatial Development Initiative (SDI) projects implemented by the Gauteng Department of Public Transport, Roads and Works (Gautrans) to stimulate economic growth, development and employment opportunities in the Gauteng Province. As part of this development initiative, Viaduct 5c (a 3.2 km long viaduct) crosses highly variable and sinkhole prone dolomitic ground conditions in Centurion, Pretoria. The foundations for piers 39 and 40 of Viaduct 5c were originally designed to comprise four 1.2m diameter oscillator piles installed into the dolomite bedrock. Detailed investigation carried out for these two piers revealed differential ground and rock hardness, numerous hard rock dolomite ‘floater’ (boulders) and highly variable rock head and ground conditions. This would make the installation of large diameter piles extremely difficult, and as a result, alternative solutions were required for Pier 39 and 40 to meet programme. A piled raft solution with percussion bored piles which was being carried out on other piers, was not

166

H.N. Chang and T.E.B. Vorster / The Use of Micropiles as Settlement Reducing Elements

considered an optimal solution in the cases of piers 39 and 40 due to difficulties expected in drilling through numerous and hard dolomite ‘floaters’ (boulders). An alternative founding solution consisting of a compacted gravel-soil mattress and small diameter piles, or micropiles, was proposed. Hollow bar systems with a rock bit would easily penetrate hard dolomite rock boulders and the construction of the compacted gravel-soil mattress could easily be undertaken as part of current construction methods on the GRRL project. The foundation system could also be completed in a short time frame with far less unforeseen difficulties or complications. The design concept of the alternative foundation system Pier 39 and 40 considers the micropiles as soil reinforcing elements for settlement control rather than primary foundation elements. Analysis of the foundation was carried out using the commercially available 3D finite element software, Plaxis 3D Foundation and the results of the analysis are discussed in this paper.

1. Description of Geology A large portion of the Centurion area is underlain by dolomitic terrain of the Chuniespoort Group. A typical dolomitic profile has been described by Wagener [1] as consisting a blanket of transported material, overlying unweathered dolomite rock pinnacles and cavities (which are sometimes filled with soft weathered dolomite residuum in the form of clay, clayey sand or silty sand or wad; an acronym for Weathered Altered Dolomite). The presence of soft wad, highly irregular dolomite pinnacles, floaters and disseminated voids (sometimes referred to as cavities) make construction in dolomitic profiles extremely challenging. In addition, dolomitic soils are often sinkhole prone; an issue, which must be addressed in the design. In the case of the GRRL project the risk of sinkhole formation was limited through a process of controlled void filling through compaction grouting [2]. Four rotary percussion boreholes were drilled for Pier 39 and five for Pier 40 using the Symmetrix method with Jean Lutz recording. The various material types and corresponding stiffness values were established using correlations developed based on full-scale load testing at various positions elsewhere along Viaduct 5c. The 9 boreholes are illustrated in Figure 1. The boreholes indicate that bedrock varied between 20m and 30m in depth, but can be as high as 5m below natural ground level as seen in BH39-5. Above the bedrock, the profiles are extremely variable, ranging from a residual chert/clay matrix in BH40-4 to alternating layers of large dolomite floaters and wad as seen in BH39-2. The combination of hard dolomite rock and softer dolomite residuum makes estimation of footing settlements extremely difficult. This is further complicated by the large variability between profiles across the same pier foundation.

2. Alternative Solution The proposed foundation alternative comprised of 49 micropiles installed in a 7 × 7 grid at 2.25m spacing to a depth of 12m, above which a 1.5m thick compacted G6 quality gravel-soil mattress (with 0.5MPa < UCS < 1.5MPa) was constructed. The base of the pier, a 12m by 12m by 2m thick reinforced concrete raft, was founded on the compacted gravel-soil mattress. The proposed foundation system is shown in Figure 2.

H.N. Chang and T.E.B. Vorster / The Use of Micropiles as Settlement Reducing Elements

167

Figure 1. Design profiles for Pier 39 and 40.

Figure 2. Plan and section view of proposed foundation system.

The micropiles increase the stiffness of the founding soil and as a result reduces total and differential settlements on the piers. The compacted soil raft distributes load from the pier base to the underlying soil/micropiles system and prevents direct contact between the pier base and the micropiles which may cause buckling or crushing failure of the micropiles and reduce the effectiveness of the system. Micropiles were installed using a Titan hollow bar system where drilling and pressure grouting occurs simultaneously. The Titan bar acts as the drill stem and permanent reinforcement for the micropiles.

168

H.N. Chang and T.E.B. Vorster / The Use of Micropiles as Settlement Reducing Elements

3. Analysis of Foundation System With bearing failure of the footing found to be unlikely and the possibility of cavities below or near the foundation addressed through void filling, the design of the pier foundations is performance based with total settlement and differential settlement being the main design criteria. 3.1. Design Information Two load cases were identified as critical load cases on each pier. Unfactored values of the two load cases are summarized in Table 1.

P39 P40

N (kN) 34351 33981

Table 1. Critical load cases C15 and C16. Load Case C15 Load Case C16 HX HY MXX MYY N HX HY MXX (kN) (kN) (kNm) (kNm) (kN) (kN) (kN) (kNm) 1060 904 20406 11265 34351 530 1247 25791 1060 805 14859 10241 33981 530 1050 18490

MYY (kNm) 5634 4874

Settlement criteria under the two critical load cases are specified as: • •

A minimum spring stiffness of 1600kN/mm (or a maximum settlement of 21mm), A maximum differential settlement of 75% of the total settlement (or 16mm) measured across the width of the pier.

Young’s modulus for identified soil layers were allocated using relationships between Jean Lutz monitoring and soil profile descriptors established based on full-scale load testing at various piers along Viaduct 5c: Weathered dolomite Chert/clay residuum Wad Dolomite rock

30-60MPa 30-60MPa 8-60MPa 225-300MPa

3.2. Finite Element Model Modelling of foundation settlement was carried out using Plaxis 3D Foundation finite element (FE) software. The schematic model is illustrated in Figure 3. The 2m reinforced concrete footing was modelled using an infinitely thin plate element with the equivalent stiffness of a 2m thick concrete slab. The pier application area on top of the footing was modelled using a 3m x 4m plate element with infinite stiffness. Vertical point loads were applied on the 4 edges of the ‘rigid’ plate to model the moment from the pier and a point load in the centre of the ‘rigid’ plate was used to model the vertical and horizontal loads. To simplify the model, the 3.5m excavation was modelled using a surcharge rather than an actual excavation. The soil was modelled using a hardening soil model where stress dependency of the stiffness modulus can be taken into account. Drained conditions were used to consider long term behaviour of the foundation.

H.N. Chang and T.E.B. Vorster / The Use of Micropiles as Settlement Reducing Elements

169

Figure 3. Plan and section view of proposed foundation system.

Micropiles were modelled using embedded pile elements, which are line elements with zero volume and interaction between pile and the subsoil takes place through special interface elements. The limiting friction value of the interface element, q s, was determined using (1) stiffness, E’, and SPT ‘N’ correlations of E’ = 1500N [3] and (2) SPT ‘N’ and shaft friction correlation of qs = 2.5N [4]. By substituting (1) into (2), a final correlation of qs = E/500 was used for the interface friction of the embedded pile elements. As micropiles are small diameter piles, the end bearing (modelled as a force at the pile toe) is assumed to be zero. It should be noted that unlike some other FE software which use P-Y curves or equivalent spring constants to model load settlement behaviour of piles, the stiffness response of the embedded pile element subjected to loading is a function of the relative stiffness of the pile and the surrounding soil and not an input parameter. Stiffness of the micropiles was initially based on a 150mm diameter grout body with a 52/26 hollow bar. During the initial stages of modelling, however, it was found that the bending stresses in the embedded pile element exceeded the bending capacity of the micropile, which would result in cracking of the grout body. The contribution of the grout body was therefore ignored and the stiffness of the embedded pile element was based only on the properties of the 52/26 hollow bar (minus 4mm diametrical reduction for corrosion over a 100 years design life under aggressive soil conditions). 3.3. Design strategy Due to the complexity of the geotechnical conditions, every borehole was analysed individually to investigate the range of results that may be encountered. An initial attempt was made to model all boreholes in the same finite element model. This was however problematic as the soil horizons in Plaxis 3D are generated automatically from input borehole data and four or five complex profiles in close proximity often resulted in numerical errors during interpolation. The advantage of using embedded pile elements (line elements) to model piles is simplicity and reduction in computing time. The limitation, however, is that pile head

170

H.N. Chang and T.E.B. Vorster / The Use of Micropiles as Settlement Reducing Elements

bearing forces cannot be taken into account. This is generally acceptable for normal pile foundations where forces can be transferred directly from pile caps to piles. In this case, load is transferred from the concrete base to the soil mattress and then into the underlying soil/micropile system. Since the cross-sectional area of the line element is effectively zero, no axial force is generated and the resulting axial force in the micropile head will be under-estimated. To solve this problem, a model with a single micropile as shown in Figure 4 was setup for every borehole profile. A uniformly distributed unit pressure of 100kPa was applied to a 3m x 3m footing and the axial force in the micropiles was determined for models with and without a 150mm square rigid plate (to cause load transfer to the micropile head). The relative increase in axial force for conditions with the square plate (compared to that without) was calculated as a percentage and applied to the axial force obtained from the main model. The increase in axial force ranged between 0% and 62%. The increase in axial force is proportional to the average stiffness of the profile, provided that no stiff material is present at the surface, in which case the effect will be negligible.

Figure 4. Single pile model to investigate pile head bearing.

The effect of dolomite floaters on the micropile axial forces was estimated in a similar manner as is used for the original piled raft design. The strategy involves setting up an axisymmetric model of a single 150 mm diameter pile in Plaxis 2D. The soil profile in the model was allocated the average soil mass stiffness of the least stiff profile determined using Fraser & Wardle’s solution [5]. The micropile therefore ‘floats’ within this uniform soil body. Each floater (or combination of floaters) identified at a particular pile position (from predrilling at the specific pile position) was then modelled as a separate design case. The single pile settlement for the pile and floater combination in each case was then compared to the settlement of a single pile without floaters. By setting up a curve for different single pile lengths the ‘equivalent’ pile length at a particular pile position which would give similar settlement to the micropile with floaters is then established. Equivalent micropile lengths to account for the effect of floaters ranged from 7.2m to 20.8m compared to the design micropile length of 12m; i.e. floaters were found to cause both softening and stiffening depending on the combination, size and depth location of the floater/s compared to the 12 m single pile stiffness without floaters. The increase in axial force due to floaters was then assessed by setting up full 3D FE models of the pier, including all 49 micropiles in the uniform soil model used to determine the equivalent lengths of the micropiles described above. The first model consisted of the original 12m micropiles and the second model consisted of variable equivalent micropile lengths to simulate floater effect. The increase (or decrease) in micropile axial force due to floaters was then determined as a percentage for each micropile was found to range between +18% (increase) and -87% (reduction).

H.N. Chang and T.E.B. Vorster / The Use of Micropiles as Settlement Reducing Elements

171

The final individual micropile axial force would then be that determined from the main 3D finite element model, plus the effects of pile head bearing and floaters.

4. Results and Discussions During the design stages of the project, it was found that some micropiles at Pier 39 would become overstressed under P39-BH2 conditions. To resolve this problem, the top meter of the compacted gravel-soil mattress was replaced by a weakly cemented mattress (with an UCS of ±5Mpa) to stiffen the system, improve load transfer across the foundation footprint and thereby reduce forces on the individual micropiles. The bottom half meter of the compacted gravel-soil mattress was kept to prevent direct contact between the micropiles and the cemented mattress which may result in localized crushing of the micropile head. The results given below are therefore for Pier 39 with composite (5MPa UCS/C4) mattress, and for Pier 40 with compacted gravelsoil (C4, 0.5 < UCS < 1.5 MPa) mattress. 4.1 Pier Settlement An initial analysis was carried out to investigate the necessity for micropiles to be installed. Total and differential settlement of the piers was estimated using the 3D finite element model for conditions without micropiles and results are summarized in Table 2.

Borehole BH1 BH2 BH3 BH4 BH5 Ave

Table 2. Settlement of piers without micropiles. Pier 39 Pier 40 without micropiles with micropiles without micropiles with micropiles Ave Diff. Ave Diff. Ave Diff. Ave Diff. 21.7 10.3 14.8 6.1 21.9 9.6 17.0 9.0 28.0 12.0 14.0 6.0 21.1 12.9 16.8 12.5 14.1 3.9 8.0 3.2 14.4 5.2 13.0 5.0 37.7 18.8 30.1 15.9 25.3 22.3 16.8 14.6 8.9 5.1 6.6 4.9 22.4 12.4 14.6 7.9 20.7 10.1 15.7 9.1

Settlement of the piers were relatively high, with results for three of the borehole conditions in each pier exceeding the maximum allowable settlement of 21mm. The average spring stiffness of the piers was 1536kN/mm for Pier 39 and 1640kN/mm for Pier 40. The average differential settlement under combined loading is 12mm for Pier 39 and 10mm for Pier 40, but values as high as 22mm and 19mm are possible. Since the total and differential settlements for the piers borderline the maximum allowable limits it was decided that micropiles need to be installed to reduce the risk of excessive settlement. Total and differential settlement for the piers with a grid of 49 micropiles are also given in Table 2. The addition of micropiles significantly improved the performance of the foundation system, with a reduction in average settlement of 35% for Pier 39 and 25% for Pier 40. The addition of micropiles also resulted in a reduction in differential settlement of 35% and 10% for Piers 39 and 40 respectively. The average spring stiffness of the foundation is 2353kN/mm for Pier 39 and 2164kN/mm for Pier 40. Average differential settlement is reduced to an acceptable value of less than 10mm.

172

H.N. Chang and T.E.B. Vorster / The Use of Micropiles as Settlement Reducing Elements

4.2 Micropile axial force Unlike conventional piled foundations, load from the structure is transferred via the compacted soil mattress into the underlying soil/micropile system. Development of stresses (and forces) along the pile shaft is a function of the ground conditions in which the micropile is installed. A plot of axial force against depth is given for three distinctly different subsoil types in Figure 7, namely: Subsoil type I: Soft layer throughout (BH40-4) Subsoil type II: Soft layer overlying rock with micropiles socketed into rock (BH39-5) Subsoil type III: Rock at surface overlying soft layer (BH40-3)

Figure 5. Axial force generation under various soil conditions.

Under subsoil type I, downdrag of the surrounding soil in the upper parts of the micropile results in an increase in axial force up to a maximum value where relative movement between the micropile and surrounding soil is zero. Below this point, the micropile moves down relative to the soil (as with a conventional friction pile) and axial force is dissipated through the shaft friction. Where micropiles are socketed into rock at depth (subsoil type II), there is a steep increase in the axial force due to compression and downdrag of the overlying soft layer. The axial force is then dissipated by the section socketed in the rock. Subsoil type III is analogous to the conventional pile/cap system where load from the ‘cap’ (rock) is transferred directly into the piles. In this case, axial force at the base of the rock layer (‘fixed pile head’) is a maximum and dissipates into the soft layer as depth increases. The maximum axial force was 409kN (including cap and floater effects) and the maximum compressive stress due to combined axial force and bending is 403MPa. This is approximately 75 % of the yield strength of the hollow bars (550MPa) and is deemed acceptable.

H.N. Chang and T.E.B. Vorster / The Use of Micropiles as Settlement Reducing Elements

173

4.3 Buckling Considerations Since micropiles are slender elements, design of micropiles generally includes a check on buckling. Unlike conventional piled foundations, however, load from the structure is transferred to the micropiles through downdrag of the surrounding soil as observed in Figure 5 This eliminates the possibility of buckling, as an increase in axial force is accompanied by a decrease in effective length. 4.4 Local failure at micropile-soil mattress interface The use of line elements in the model with an infinitely small pile tip area results in high localized stresses which suggest punching of the micropiles through the compacted gravel-soil layer in the finite element model. Under these conditions, stress redistribution occurs with additional settlement of the foundation system. Punching of the micropiles would therefore reduce the effectiveness of the system, but will not constitute failure of the foundation system. 4.5 Effectiveness of micropiles as soil reinforcing elements The effectiveness of the foundation system can be assessed by comparing the normal distributions of pier settlement with and without micropiles. Normal distributions for average total and differential settlements are given in Figures 6a and 6b respectively.

Figure 6. Normal distributions of average total and differential pier settlements.

Besides the reduction in average total and differential settlements, the micropiles were also effective in reducing the range of possible settlement values (i.e. variability across the foundation footprint) and the risks associated with the variability of the ground. The benefit in relation to settlement of having a stiffer mattress is also evident in Figure 6, where the reduction in mean and standard deviation of settlement values in Pier 39 is significantly greater than in Pier 40. The percentage of load carried by the micropiles is plotted against average profile stiffness in Figure 7. The percentage of load carried in the micropiles is indirectly proportional to the average stiffness of the profile, with average values in the order of 20% to 30%. The effectiveness of the micropiles acting as settlement reducers (‘stiffners’) is evident in the significantly smaller proportion of the load carried compared to conventional piled (100%) or piled raft foundations (60% to 90%).

174

H.N. Chang and T.E.B. Vorster / The Use of Micropiles as Settlement Reducing Elements

Figure 7. Plot of percentage load carried by micropiles against profile stiffness.

5. Conclusions The design and analysis of a micropile foundation solution for the Piers 39 and 40 of Viaduct 5c of the GRRL project is presented. The foundation system consists of a compacted cement stabilised soil mattress and 49No 52/26 Titan hollow bar micropiles designed as soil reinforcing elements. Due to the complexity of the foundation system and the dolomitic profile, analyses were carried out using 3D finite element software, Plaxis 3D Foundation v2.2. The installation of micropiles results in reduction in total settlement of between 25% and 35% and a reduction of differential settlements of between 10% and 35%. The proposed system also reduces the risks associated with the stiffness variability of the dolomitic profile. Forces in the micropiles are primarily generated from downdrag of the surrounding soil, and are highly dependent on the local profile in which the micropile is installed. Load carried in the micropiles was found to be on average approximately 20% to 30% of the total load, and is indirectly proportional to the stiffness of the profile. The proposed micropile system is a cost effective alternative to end-bearing piled foundations in dolomitic areas with fewer risk of delays and complications associated with installation.

References [1] R.A. Fraser and L.J. Wardle, Raft foundation – case study and sensitivity analysis, Proc. 2nd International Conference on Applications of Statistical & Probability in Soil & Structural Engineering, 15th – 18th September, 1975, Aachen, Germany. [2] R. Tosen, R.B. Storry and M. Baribault, Grouting for dolomitic soil and rock, Gautrain Rapid Rail Link, Proc. International Symposium on Ground Improvement Technologies and Case Histories, 9th – 11th December, 2009, Singapore. [3] G.G. Meyerhof, Penetration tests and bearing capacity of cohesionless soils, J. of Soil Mechanics and Foundation Division 85 (1956), 1-19. [4] M.A. Stroud, The standard penetration test – its applications and interpretations, Proc. I.C.E. Conference on Penetration Testing in the U.K., Birmingham, U.K, 1989. [5] F. von M. Wagener, Engineering Construction on dolomites, PhD Thesis, University of Natal, 1982.

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 175 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-175

Rigid inclusions in sand formation resting on compressible clay Mounir BOUASSIDA,1 Simpro, Geotechnical consulting bureau, Tunis, Tunisia.

Abstract. A typical soil profile composed by relatively thick sand layer resting on very thick compressible clay layer is considered. In such geotechnical condition the construction of very high embankments induces unallowable consolidation settlement. Then a novel foundation system is suggested by installing floating rigid inclusions in the upper sand layer so that skin friction and tip components will develop along the inclusion as reaction to the concentrated vertical load applied on capped mini slabs. For the design it will be of interest to determine the optimum length of rigid inclusions and spacing to come up with negligible settlement at surface of the sub clayey layers. The analytical prediction of bearing of bearing capacity of rigid inclusions in sand formation is presented. The consolidation settlement of clayey layers is estimated.

Keywords. Settlement, rigid inclusion, bearing capacity, sand, clay

Résumé. L’édification de remblais à grande hauteur est étudiée dans le cas d’une coupe géotechnique compose d’une couche de sable dense, d’épaisseur 8 m, reposant sur une couche d’argile compressible très épaisse. Afin d’éviter un tassement de consolidation primaire inadmissible dans l’horizon d’argile on propose l’exécution d’inclusions rigides verticals flottantes dans la couche de sable dense. Une concentration de la charge transmise par le remblai est assure par le biais de dalettes places en tête des inclusions dont la reaction se produit par frottement latéral et la résistance de pointe. Pour optimizer le dimensionnement d’un tel système de fondation on est amené à determiner la longueur des inclusions rigides et leur espacement, en function de leur diameter, afin de réduire au maximum le tassement en surface de la couche d’argile. Mots clefs. Tassement, inclusions rigides, capacité portante, sable, argile

Introduction In North Area of Tunis Lake the soil profile is essentially composed of an upper fine sand layer of 6 to 8 m thickness resting on two compressible clay layers which extend up to 30 to 35 m depth. When high buildings or high embankments are projected in such geotechnical condition a serious problem of consolidation settlement is posed due to significant vertical stress induced in clayey sub layers. Therefore, the question is to 1

Corresponding Author: ENIT, [email protected]

BP 37 Le Belvédère 1002, Tunis. Tunisia; E-mail:

176

M. Bouassida / Rigid Inclusions in Sand Formation Resting on Compressible Clay

prevent vertical stress propagation in clayey layers. Rigid inclusions (RI) are a very recent technique of soft ground improvement. The principle of rigid inclusions is based on the use of end bearing vertical reinforcing elements, made up of steel bars or reinforced concrete, head capped by mini slabs overlaid by a thin mattress of granular layer. The inclusions are generally located in squared regular pattern. When such a reinforced medium is subjected to vertical loading a concentration of vertical stress is induced on the surface of mini slabs. Therefore the vertical stress induced at surface of soft soil is quite reduced because of the concentration of vertical stress on the inclusions. By the latter the load transfer is essentially ensured by the skin friction component, while the tip resistance at stratum level might be neglected. In the present contribution a novel reinforcement scheme is suggested by which rigid inclusions are embedded in relatively thick sand formation overlying very compressible clay layer(s). The idea came out from the study of a Tunisian case history where the use of vertical geodrains has revealed totally inefficient due to the presence of upper thick sand layer. This latter, in fact, makes lengthy geodrains installation time consuming and not cost effective as well. Therefore a new improvement technique should be thought to minimize, in allowable limit, the primary consolidation settlement in very thick compressible clay layers, overlain by the upper sand layer of 8 m thickness. From the studied case history the upper sand layer has a good penetration resistance characterized by a friction angle of 32°. In this view it has been judged to reinforce the upper sand layer by rigid inclusions (RI) of length less or equal to 6 m so that tip effect does not extend to the sub compressible clay layer H2.2 and, an allowable settlement in sand should occur. The efficiency of rigid inclusions (RI) is enhanced when covering its upper part by a small cap raft to more concentrate the applied load on the inclusion. By consequence, the vertical stress induced at surface of soil between inclusions will be reduced and the corresponding settlement is too. Similar to classic pile foundation, when subjected to embankment loading the soil reacts both with tip and lateral skin resistances (Figure 1).

Figure 1. Unit cell model and soil characteristics used for prediction.

M. Bouassida / Rigid Inclusions in Sand Formation Resting on Compressible Clay

177

In this paper, an axisymmetric unit cell model is adopted to calculate the allowable bearing capacity of purely frictional sand reinforced by floating rigid inclusion. Then, the behavior of reinforced soil is investigated by means of finite element computation to determine the settlement distribution as a function of embankment loading. Finally a synthesis of results is given and practical recommendations are proposed..

1. Design criteria Under loading embankment modeled by vertical surcharge q = γr Hr two verifications should be addressed for stability purposes of reinforced soil by RI: first is punching of sand layer and second is allowable settlement under the tip of RI. 1.1. Punching stability The tip of RI is considered as circular shallow foundation of diameter 2a and length L i embedded at depth equal to the length of RI added to the thickness of mattress H m. The resultant force at inclusion head writes:

Q real = 4qa '2

(1)

2a’ = the side of squared mini raft that represents the cover of inclusions of diameter 2a. As first approximation it is considered: a ' = a + 5cm . Along the interface between RI and sand one can adopt the friction angle (2ϕ/3), then, after static formula of pile foundation, the skin friction component writes (Fig. 1)

Q f = ( 0.5γ ' H 2 + σ soil H ) 2π aKtg (

2ϕ ) 3

(2)

γ’ = effective unit weight; H = Li + H m is the total height where skin friction develops; K = coefficient governing the proportionality between horizontal and vertical effective stresses due to gravity; the more likely pressure at rest prevails, then for sands it can be approximated K = 0.5. The tip component is Q p = π a 2 ( γ ' D + σ soil ) N q + 0.6aN γ

(3)

Nq and Nγ = depth and surface bearing capacity factors; D = embedded depth of tip of RI. The total ultimate bearing capacity, Qult, of RI is the sum of components given by Eqs (2) and (3). The safety factor, F, against punching is F=

Qult Q real

(4)

178

M. Bouassida / Rigid Inclusions in Sand Formation Resting on Compressible Clay

For embankment projects it can be adopted F = 1.5. The stress concentration factor is defined by n=

σ RI σ soil

(5)

σsoil (resp σRI) = vertical stress acting at surface of soil (resp. head of RI). The substitution factor is defined by

η=

4a '2 De2

(6)

De = diameter of unit cell model. Consider the average vertical stress at surface of reinforced soil as q = ησ RI + (1 − η )σ soil

(7)

The allowable stress q is calculated from Eq. (4) and then, from Eqs (5) and (7), it comes σ soil =

q 1 + η (n − 1)

(8)

The distribution of vertical stress on reinforced soil is identified for given η, q and n values. Then, on the basis of given allowable stress distribution the settlement prediction can be carried out to analyze the behavior of reinforced soil. 1.2. Punching stability RI constitutes the strong point of embankment foundation; hence the settlement under the tip of inclusion should be limited in allowable range compatible with embankment stability. Obviously, the allowable settlement depends on thickness of embankment. As example, for 12 m embankment thickness the allowable settlement does not exceed 10 cm.

2. Numerical prediction of reinforced soil behavior 2.1. Mechanical model Numerical computations were carried out using Plaxis software version 8.0 by adopting axisymmetric condition as represented in Figure 1. The equivalent diameter of unit cell model is 3 m, and the effective spacing between RI is s = 2.65 m. All materials are described by Mohr Coulomb behavior. The corresponding mechanical characteristics adopted for the numerical calculation are given in Table 1 (E = Young modulus; ν = Poisson’s ratio; C = cohesion and ϕ = internal friction angle).

M. Bouassida / Rigid Inclusions in Sand Formation Resting on Compressible Clay

179

Table 1. Mechanical characteristics for numerical predictions Material Clayey Marl Sand Compressible clay

ϕ (°) -32

C’(kPa) -1

ν’ (°) 0.3 0.3

E (kPa) 107 1.493 104

27

3

0.33

15

2.2. Numerical results Recorded numerical results by Plaxis software are summarized in Tables 2 and 3 where Hr = Height of embankment; Li = Length of rigid inclusion; sp = immediate settlement under tip of rigid inclusion; sc: settlement at bottom of sand layer; and sv = settlement of high compressible clay layer (horizon H2.1). Table 2. Predicted settlements of reinforced soil by rigid inclusions (2a = 40 cm) Hr (m) 12 11 10 9

Li (m) 6 6 6 6

sp (cm) 10 9 8 7

sc (cm) 8 7 6 5

sv (cm) 5 4.8 4.3 3.8

Table 3. Predicted settlements of reinforced soil by rigid inclusions (2a = 30 cm) Hr (m) 8 7 6 5 4 3 2 1.5

Li (m) 6 6 6 5 5 4 4 4

sp (cm) 6 5 4 4 3 2 1 1

sc (cm) 5 4 4 3 2 2 1 1

sv (cm) 5 4.8 2.6 2.0 1.7 1.2 0.8 0.6

2.3. Synthesis of results For embankment heights ranging from 1.5 m to 8 m a 0.3 m diameter of inclusions is compatible with bearing capacity and settlement verifications above mentioned. Meanwhile in the range Hr = 6 to 9 m the length of RI should be extended up to 6 m associated with an increase of diameter to 0.4 m. In this range of reinforcement predicted settlements are lower than allowable value of 10 cm, as predicted for embankment height of 12 m. Note that predicted settlements in sand layer should occur at short term behavior which corresponds to the end of embankment construction. The latter will be continuous, i.e. without the need of use of staged construction. Recorded settlements in compressible clay layer correspond to the long term behavior, i.e. as consequence of primary consolidation process taken into account in the numerical option provided by Plaxis software. Since the magnitude of these settlements is less than 5 cm, even if the time of consolidation will take some years, the long term stability of embankments will not be affected. Fig. 2 illustrates the evolution of length of RI in function of embankment height. Fig. 3 shows the evolution of predicted

180

M. Bouassida / Rigid Inclusions in Sand Formation Resting on Compressible Clay

settlements at surface of upper sand layer and under the tip of rigid inclusion. These settlements will be rapidly stabilized due to the drained behavior of sand.

Figure 2. Length of rigid inclusion versus height of embankment.

Figure 3. Evolution of settlement at top of compressible clay layer vs height of embankment.

As example, for Hr = 10 m, the coefficient of vertical stress concentration is: n = 2.46. Table 4 summarizes the distribution of the vertical effective stress from the surface of reinforced sand layer to the top of compressible clay layer, where σp = effective stress at tip of rigid inclusion; σI = effective stress at interface mattress-reinforced-soil and σH2,1 = effective stress at surface of compressible clay layer. It is noticed that the concentration effect of vertical stress at the head of rigid inclusion (RI) depends on the embankment height. The vertical stress decreases at the tip of RI with allowable value of bearing capacity of sand layer. Elsewhere the induced vertical stress at surface of clay layer H2.1 well explains the predicted settlement of primary consolidation. As example, for embankment height of 1.5 m the excess of vertical stress (due to such loading) is: 82 – 77 = 5 kPa that corresponds to predicted settlement of 0.6 cm (Table 3). Indeed, the initial vertical effective stress at surface of clay layer is 77 kPa. Illustrated outputs of Plaxis software are presented in Figs 4 and 5 showing the mesh of mechanical model respectively prior and post a loading equivalent to 5m height of embankment applied on a unit cell of 3 m diameter.

M. Bouassida / Rigid Inclusions in Sand Formation Resting on Compressible Clay

Figure. 4 Initial mesh of mechanical model

181

Figure 5. Deformed mesh of mechanical mode

Table 4. Distribution of vertical effective stress in reinforced soil mass by rigid inclusions Hr (m) 12 11 10 9 8 7 6 5 4 3 2 1.5

σp (kPa) 574 532 366 348 366 355 297 261 210 177 98 78

σI (kPa) 4000 3780 3500 3000 3000 3130 2440 2100 1890 1320 877 342

σH2,1 (kPa) 197 184 172 177 150 152 139 155 104 140 89 82

3. Conclusions A novel reinforcement scheme has been suggested for a sand layer resting on very compressible clay horizons. Floating rigid inclusions installed in regular pattern, rafted by mini slabs, make it possible a significant reduction of settlement at surface of reinforced soil and at the top of clay layers. The design has included, first, the determination of bearing capacity of rigid inclusions assumed as end bearing and frictional pile. From this verification the allowable loading has been identified in function of spacing between inclusions and the ratio between the vertical stress applied on the capped inclusions and on the surrounding soil. The second step of design has addressed the prediction of settlement distribution in depth which resulted from numerical computation carried out by Plaxis software by considering a unit cell mechanical model. From numerical results, the evolution of height of embankment, as loading parameter, has been predicted as function of the length of rigid inclusions. Further, the settlement evolution has been predicted in function of embankment height and spacing between inclusions.

182

M. Bouassida / Rigid Inclusions in Sand Formation Resting on Compressible Clay

Due to drainage properties of sand layer, the suggested reinforcement solution has the main advantage to limit the settlement evolution at the end of continuous embankment construction (short term behavior). While settlements in the under compressible clay layer will be insignificant. Consequently the stability of foundation built on such reinforced soil will not be compromised.

References [1] Simpro. Embankment Foundation of Sports Tunis Golf Area. Geotechnical report.04/10. Tunis, 2009.

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 183 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-183

Dynamically Loaded Foundations André ARCHER Aurecon South Africa (Pty) Ltd, Pretoria, South Africa

Abstract. The purpose of this study was to determine whether the closed form solutions used for the calculations of dynamically loaded foundations are sufficient to predict the actual behaviour of the foundation. The main limitation is that only impact loads are considered and this will limit the closed form solutions used for the study. A foundation was constructed and the actual behaviour of the foundation while exposed to an impact load was measured during an experiment. The experiment consisted of placing accelerometers and a LVDT on the foundation and then the foundation was hit with an impulse force hammer. Three parameters were tested namely the natural frequency of the system, the radiation damping of the soil and the displacement of the foundation. Closed form solutions to predict the behaviour of the foundation were obtained through numerous literature sources and these predictions was tested against the experimental values. From the data obtained through the experiment and the closed form solutions it was established that the prediction of the closed form solutions is over conservative. The natural frequency was the most accurately predicted with 5% difference between the predicted value and the experimental value. The damping was the worst parameter and because the damping was too high, the displacement was influenced. Nonetheless, there was some correlation between the value and conclusions can be made. There was a correlation between the predicted values and the experimental values. The overall prediction of the response of the foundation is over estimated, which will lead to a more conservative design. Keywords. Dynamically loaded foundation, Impact load, Closed form solutions, Frequency, Damping, Displacement.

Introduction When designing a foundation both ultimate limit states and serviceability limit states must be satisfied. Serviceability limit state includes vibration resulting in unacceptable effects such as settlement.[1] The aforementioned settlement effects can only be reduced with a proper design and analysis of dynamically loaded foundations. Although computer programs have made it easier for engineers to analyse dynamic foundations, hand calculations are used as a preliminary assessment. It is therefore necessary that reliable hand calculations are used which will render resolute results. The best way to do this is through comparing the predictions from theory with the actual behaviour of foundations. The aim of this paper is to compare the theory of the design of dynamic foundations with the behaviour observed in the experiment. A foundation was constructed, a dynamic load induced and the results were compared to the theoretical answers given by closed form solutions.[2] The focus in this paper will be on vertical

184

A. Archer / Dynamically Loaded Foundations

impact type loads and the parameters considered are natural frequency, damping and vertical displacement.

1. Literature Study The outcome of a dynamic foundation design is to reduce the dynamic response of the foundation and to avoid resonant frequency. The response is expressed as the amplitude at a specific frequency, which is the largest at the resonant frequency. The latter is true for all types of dynamic loads (impact and harmonic), but in this paper the focus will be on impact loads. 1.1. Design Properties Designing a dynamically loaded foundation requires certain properties to be obtained in order to do the design or analysis. There are three key properties to be considered namely: • • •

Machine properties[3], including the weight of the machine, the vertical and horizontal centre of gravity, the frequency, magnitude and applied direction of the unbalanced forces. Foundation properties, including the size, the shape and the weight of the foundation. Soil properties[4], including the density of the soil (ρ), the Poisson’s ratio (ʆ) and the dynamic shear modulus (G). The soil parameters are the most unpredictable factors because they differ from site to site.

1.2. Closed Form Solutions Extensive work was done by Richart et al.[5] on the design of dynamically loaded foundations, based on the ‘lumped parameter system’. The system uses a mass, spring and dashpot as reference to the stiffness and damping of the system. A normal foundation has six degrees of freedom and each mode is analysed independently to simplify the design and to avoid confusion. The equations used in this paper for the damping and the natural frequency subjected to impact and harmonic loading are published by Richart et al.[5]. For the displacement due to impact loads, more specific equations are required since there are three different pulse types namely rectangular, half-sine and triangular. A complex procedure and formulas for the calculation of displacements for impact loads are discussed by El Naggar et al.[2]. The procedure discussed by El Naggar et al.[2] was applied to obtain the displacements for the closed form solution calculations. It should be noted that the effect of impact loads are not often governed by frequencies, but rather by the force applied to the foundation. The closed form solutions in the aforementioned papers are used to compare with the experimental results.

185

A. Archer / Dynamically Loaded Foundations

2. Experiment and Results 2.1. Experiment For the experiment the following instruments were used and Figure 2 is an image of the setup on the foundation. • • •

Accelerometers – which measures the acceleration of the foundation. Linear Variable Differential Transducer (LVDT) – which measures the displacement of the foundation. Force Hammer – which induces the vertical impact load at the centroid of the foundation.

The experiment was carried out by placing the instruments on the foundation as shown in Figure 1. The hammer was used to apply the dynamic load at the centroid of the foundation. A data logger was used to capture the results which were compared to the closed form solutions. The dimension of the model foundation is given in the next section.

Figure 1. Photo of the experimental setup

2.2. Results Figures 2 to 3 show the measurements of acceleration and force respectively. ϲ͘Ϭ

ĐĐĞůĞƌĂƚŝŽŶ;ŵͬƐϮͿ

ϰ͘Ϭ Ϯ͘Ϭ Ϭ͘Ϭ ͲϮ͘Ϭ Ͳϰ͘Ϭ Ͳϲ͘Ϭ Ϭ͘ϬϬϬ

Ϭ͘ϬϬϱ

Ϭ͘ϬϭϬ

Ϭ͘Ϭϭϱ

Ϭ͘ϬϮϬ

Ϭ͘ϬϮϱ

Ϭ͘ϬϯϬ

Ϭ͘Ϭϯϱ

dŝŵĞ;ƐͿ

Figure 2. Close -up acceleration curve for the accelerometer

Ϭ͘ϬϰϬ

186

A. Archer / Dynamically Loaded Foundations

ϲϬ ϱϬ

&ŽƌĐĞ;ŬEͿ

ϰϬ

^ƚĂƌƚ

ŶĚ

ϯϬ ϮϬ ϭϬ Ϭ ͲϭϬ Ϭ͘ϬϬϬ

Ϭ͘ϬϬϱ

Ϭ͘ϬϭϬ

Ϭ͘Ϭϭϱ

Ϭ͘ϬϮϬ

Ϭ͘ϬϮϱ

Ϭ͘ϬϯϬ

Ϭ͘Ϭϯϱ

Ϭ͘ϬϰϬ

dŝŵĞ;ƐͿ

Figure 3. Response curve for the hammer

Different hammer tips (low frequency (soft) to high frequency (hard)) were used during the experiment and it was found that the tip type does not influence the overall conclusion. Only the result for the low frequency tip is given in the paper. The hammer pulse time is taken from the first zero force point to the point where the force is zero again.

3. Data Analysis For analysing the closed form solutions, the design properties in Table 1 were used. The soil shear modulus values was obtained with CWS testing and the density and poisons ratio was taken from previous work done at the experimental site. Table 1. Summary of the design properties used. Soil Properties

Model Foundation Properties

Soil Shear Modulus (G) – MPa

79.8

Length (L) – m

2.5

Density of the soil (ρ) - kg/m3

2000

Width (B) – m

2

Poissons Ratio (ν)

0.35

Height (H) – m

0.4

Mass (M) – kg

5000

Effective radius (r0) - m

1.261

3.1. Natural Frequency From the procedure discussed in Richart et al.[5] the natural frequency of the soil and foundation system was calculated as fn = 51.47 Hz. The accelerometer data as shown in Figure 3 was used to calculate the experimental natural frequency. In order to obtain the experimental natural frequency of the system, a Fast Fourier Transform (FFT) was applied to the accelerometer data. Figure 4 is the FFT analysis results and Table 2 indicates the magnitudes and frequencies of the FFT analysis. From the results the dominant frequency is 48.82 Hz.

187

A. Archer / Dynamically Loaded Foundations

Taking the FFT value as ‘truth’, it follows that the CFS predicted a value approximately 5% higher than the value obtained from the experiment. From this it is fair to assume that the equation for the stiffness of the system can be used with confidence. Table 2. Magnitude and Frequencies of the FFT analysis for the accelerometer Peak Frequencies Magnitude

Frequency (Hz)

Magnitude

Frequency (Hz)

0.761

48.82

0.429

439.45

0.368

292.96

0.157

830.07

Figure 4. FFT analysis for the accelerometer data

3.2. Damping The damping of the soil consists of radiation and material damping. Since the foundation was placed on the surface, material damping is ignored and only radiation damping is taken into account.[5] Following the procedure in Richart et al.[5] the radiation damping calculated as Cb = 3.176 x 106 N/m/s and the critical damping cc = 3.234 x 106 N/m/s. Using these values the damping ratio is D = 0.982 = 98.2%. The experimental damping ratio was calculated using the accelerometer data and the logarithmic decrement procedure as described by Rao[6]. The experimental damping ratio was found to be D = 8.4% and using the critical damping ratio the radiation damping is Cb = 0.29 x 106 N/m/s. The experimental values are about ten times smaller than the closed form solution values, indicating that the predicted damping is un-conservative. A parametric study concluded that the foundation height does not influence the critical damping enough to increase the value which will render a lower damping ratio value. In contrast, the effective radius has too much influence on the radiation damping which renders a high damping ratio value. The combination of these effects is the cause of the high predicted damping ratio and the difference in the results. It is concluded that the closed form solutions for the radiation damping prediction greatly overestimate the actual value.

188

A. Archer / Dynamically Loaded Foundations

3.3. Displacement Since the pulse type is undefined, both the half-sine and triangular pulse were applied to the closed form solutions as given by El Nagar[2]. For the displacement from the experimental results, the accelerometer data and the LVDT data was used. The accelerometer data was numerically double integrated to obtain the displacement and the LVDT data taken directly from the calibrated experiment data. The data for the closed form solutions as well as the experimental results were plotted against time for comparison. Figure 5 is the graph of the displacement against time for the aforementioned data sets. ,ĂůĨͲ^ŝŶĞ

dƌŝĂŶŐƵůĂƌ

>sd

ĐĐĞůĞƌŽŵĞƚĞƌ

,ĂŵŵĞƌ/ŵƉĂĐƚŶĚ

Ϭ͘Ϭϲ

sĞƌƚŝĐĂůŝƐƉůĂĐĞŵĞŶƚ;ŵŵͿ

Ϭ͘Ϭϱ Ϭ͘Ϭϰ Ϭ͘Ϭϯ Ϭ͘ϬϮ Ϭ͘Ϭϭ Ϭ ͲϬ͘Ϭϭ

Ϭ

Ϭ͘ϬϬϱ

Ϭ͘Ϭϭ

Ϭ͘Ϭϭϱ

Ϭ͘ϬϮ

Ϭ͘ϬϮϱ

dŝŵĞ;ƐͿ

Figure 5. Displacement curve with the experimental and calculated data.

From Figure 5 it is noticed that the maximum displacements of all the graphs are at approximately the same time. All the curves tend to follow the same form. The closed form solution values significantly over-predicted the experimental values. The reason for that is attributed to the over estimation of the closed form solution damping values. Despite the differences, there is still a correlation between the predicted values and the experimental values with reference to the time for maximum displacement. Because of the overestimation of the damping, the displacement prediction is conservatively high. It is therefore necessary to obtain a more accurate prediction for a more economic design.

4. Conclusion Tests were conducted on a model foundation that was subjected to a dynamic impact load in order to compare the behaviour of the model foundation with behaviour predicted using closed form solutions. The closed form solutions predicted a natural frequency that was within 5% of the experimental value. The damping calculated from the closed form solutions was ten times higher than the experimental value. The displacement from the closed form solutions was larger than the experimental values, but the time at which the maximum amplitude occurred correlates well with the experiment. This is attributed to the over estimation of damping in the closed form solutions. By using the closed form solutions, the foundation will be designed conservatively, which may have an impact on economics. If more research on damping is done leading to a more accurate damping estimate, a more accurate foundation design may follow.

A. Archer / Dynamically Loaded Foundations

189

References [1] Craig, R. F., Craig's Soil Mechanics, Seventh ed. New York, Spon Press, 2004. [2] El Naggar, M., & Chehab, A., Response of block foundations to impact loads, Journal of Sound and Vibration (2004), 276(276), 293-310. [3] Arya, S., O'Niell, M., & Pincus, G., Design of Structures and Foundations for Vibrating Machine, Houston, Gulf Publishing Company, 1979. [4] Sienkiewicz, Z., & Wilczynski, B., Minimum-Weight Design of Machine Foundation under Vertical Load, Journal of Engineering Mechanics(1993), 119(9), 1781-1797. [5] Richart, F. E., & Whitman, R. V., Design Procedures for Dynamically Loaded Foundations, University of Michigan, 1967. [6] Rao, S, Mechanical Vibrations, Vol. 3, Addison-Wesley Publishing Company, 1995.

190 Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-190

Construction of a Bridge over the Kwanza River at Cabala in Angola Duarte NOBREa, Francisco CAIMOTOa and Baldomiro XAVIERb ªTeixeira Duarte S.A. Luanda Delegation b Teixeira Duarte S.A. Abstract. The crossing of the Kwanza River alluvial valley along approximately 1500m is a reference in terms of the complexity of the foundation technologies used, as well as the auxiliary works. The aim of this paper is to describe the main construction processes adopted for the execution of 1.50m diameter piles, at depths that reached 75m. The environmental constraints underlying the Kwanza River involve the construction of a temporary steel pier, 225m long, a jetty, and the use of a special cofferdam system for the execution of the pile cap blocks for bridge piers grounded in the river without resorting to embankment. Keywords. Piles, jetty, and cofferdam.

Introduction The Bridge over the Kwanza River, located near the town of Cabala, on the road connecting Catete to Muxima, Province of Bengo, Republic of Angola, appears along with the country's economic development. Its strategic location has as main objective to provide safe circulation conditions for crossing the main river bed of the Kwanza River, as well as the entire length of the left river’s flood bank (Figure 1).

dd

Dhy/D Figure 1. Aerial photo of the bridge layout area.

The aim of this paper is to describe the foundation technologies applied in the construction of the bridge and its respective access viaducts over the alluvial valley that stretches over 1500m. The solution proposed in the preliminary design foresaw the execution of a 350m bridge over the river and its respective access embankments. However, the results

D. Nobre et al. / Construction of a Bridge over the Kwanza River at Cabala in Angola

191

obtained in the first site investigation revealed incompetent geological conditions to carry out the access bridge embankments. Thus, it became clear that the construction of an embankment up to 10m high over the left river’s flood bank, which was the originally planned solution, would not be viable as it could cause settlements up to 2.0m. The study of the technically alternatives made the auxiliary works in the river area extremely difficult to implement due to the actual hydrological and geologicalgeotechnical conditions. Instead of the execution of embankment in the river area it was built a temporary 225m long steel pier, a jetty, in order to handle the lack of floating devices with enough capacity to sustain the equipment used for execution of the piles and ensure a safe and effective working platform against the Kwanza river floods. The high cost, as well as the deadline, associated to the execution and removal of independent steel platforms to support the execution of the river piles has forced a variant solution that used the definitive piles casing, previously driven, as the basis of the platform where the drilling equipment subsequently circulated. The constructive difficulties associated to the execution of temporary cofferdams using embankment and cased piles led to the design of a prefabricated cofferdam positioning system applied in the construction of the pile cap blocks.

1. General Conditions The superstructure, with a simple 14.60m wide deck, is formed by a 204m long access viaduct on the North side through the main bridge over the river, with two 68m spans and a 120m central span, and a 1074m access viaduct on the South side, all of which totals 1534m in length (Figure 2).

Figure 2. Panoramic views of the bridge.

The main bridge structure is in reinforced and pre-stressed concrete and was built through incremental launching method using travelling formwork with symmetrical cantilevered beams from two central piers. For the approach viaducts, there are T-beam decks, in reinforced and pre-stressed concrete, with continuous 30m spans. The access

192

D. Nobre et al. / Construction of a Bridge over the Kwanza River at Cabala in Angola

viaduct decks are sustained by 1.50m diameter piles, while the central piers of the Bridge were grounded on 1.50m diameter nine pile cap blocks.

2. Geology The site investigation carried out on the bridge's implantation area revealed the presence of various litho-geotechnical complexes. On the surface, there are thick alluvial deposits, made up mainly by soft mud formations, interbedded with levels of clay and fine sand, extending up to around 8 to 12m deep on the banks of the valley and around 60 to 63m in the alluvial valley. Below, in the central valley, there are sometimes muddy or clayey sandstones that are moderately compact to compact and about between 1.5 and 31m thick. At around between 46 and 70m deep they become very compact sands. In the valley's side areas and under the alluvial deposits, there are clay formations of a variable consistency at the top, which are very solid at the bottom, with foundation capacity (Figure 3).

Figure 3. Geological Profile.

3. Foundation Solutions The structure is grounded on special foundations in the form of bored piles in the soil, 1.5m in diameter, arranged in pile/column pairs on the viaducts and in groups of piles on the main bridge’s piers. The piles for the viaducts were executed from platforms prepared along the route and above the ground-water table, to ensure both the stabilization of the heavy equipment and the actual construction processes. The piers and the access viaducts’ abutments were grounded on 89 1.50m diameter piles and 28 1.0m diameter piles. This involved a total of over 5000m of executed piles, with average depths of 54m and maximum depth of 75m. The piles were executed with a rotation unit with a telescopic rod and using fluid stabilizers for the temporary support of the hole walls (Figure 4).

Figure 4. Execution of the access viaduct piles.

D. Nobre et al. / Construction of a Bridge over the Kwanza River at Cabala in Angola

193

The bridge based on 2 piers grounded on the lower river bed, each with cape blocks of 9 1.50m diameter bored piles. The piles include a total of 18 units, more than 1000m long and with a maximum depth that reaches 57.50m. Due to the water level of the river, the bridge's piles have permanent steel casing which have been applied using a hydraulic vibrodriver suspended from a crawler crane which operates from a temporary steel platform built through consecutive advances. In this case, the future casings of the structure piles, nearly 640m long and with 1.50m diameter served as basis for the drilling equipment's work platform. The works’ foundations obeyed a strict control of the physical integrity of the concreted elements through cross-hole sonic-log tests on the 1.50m diameter piles and stress-wave sonic tests on the 1.0m diameter piles.

4. Auxiliary Works The construction of the new bridge over the Kwanza River is a reference with respect to the complexity of the auxiliary works involved in a project of this scope, carried out in a location where the diversity of the available resources is scarce. As such, it was necessary to create a temporary steel pier, a jetty, allowing the movement of the equipment between the river banks and the access way to the necessary work platforms for the execution of the piles for the bridge’s central piers, as well as the construction of the cofferdams to execute the pile cap blocks (Figure 5).

Figure 5. Overview of the temporary steel bridges - Jetty 1 and 2.

The jetty can be divided into two distinct parts: jetty 1, which consisted on a platform connecting the two shores of the river, 225m long and 5 m wide (25 - 9m x 5m platforms); and jetty 2 (6 - 10m x 5m platforms), which consisted of a platform through Jetty 1 which facilitated the works that were directly related to the river's central piers. The jetty's modules were built through advances from the river bank, sustained over steel vibrodriver casings. These casings were filled with sand and sealed at the top with a concrete plug, in order to contain the sand and mobilize toe resistance, when necessary. When its execution finished, more than 3000 meters of 720mm diameter steel pipes had been applied. In certain cases they reached 44m deep (Figure 6).

194

D. Nobre et al. / Construction of a Bridge over the Kwanza River at Cabala in Angola

Figure 6. Construction details of the temporary steel structures.

The driving of the casings of the structural river piles, as well as future pile drilling and concreting forced the design of a steel platform grounded on its own casing, with the capacity to sustain loads associated to the drilling equipment. Due to the equipments' constraints, this platform was also carried out through advances, starting at Jetty 2 (Figure 7).

Figure 7. Platforms to support the execution of the piles located on the riverbed: (I) vibrodriver of the cases (ii) complete structure (iii) pile drilling.

After the drilling and concreting of the piles, the steel platform over the casings was removed and placed over welded support on the cases, prefabricated in reinforced concrete elements (impacted by the weight to be transported, using a crane). These served as basis for the construction of a reinforced concrete structure, serving simultaneously as a cofferdam and formwork for concreting of the cap blocks, partially set below the river's water level. After the positioning of the prefabricated elements over the supports, above the water level, additional concrete was applied connecting the elements. The remaining height of the wall foreseen for the cofferdam was also concreted. With the cofferdam completed, the structure was lifted using a hydraulic system, consisting of hydraulic jacks and beams and placed in its final elevation, partially submerged in the river. The connection between the cofferdam and the casings was executed through underwater concrete sealing, highly resistant and with low shrinkage. With the cofferdam properly connected to the piles, it was possible to begin the works to clean up the pipes and concrete piles and place the pile cap block's framework and concrete it, with the massif being ready for the construction of the pillar to be started (Figure 8).

D. Nobre et al. / Construction of a Bridge over the Kwanza River at Cabala in Angola

195

Figure 8. Construction phases of the cofferdam: (I) placement of the precast slab panels (ii) in-situ concreting of the wall panels (iii) positioning of the formwork cofferdam at its final elevation.

5. Final Considerations A proper site investigation carried out during the design phase is crucial to define the appropriate solution for the works. The solution of executing the river's piers through embankment would place at risk the safety, the functional and structural requirements of the construction and the contract job’s actual deadline. Thus, the choice of resorting to temporary steel structures, a jetty and a cofferdam system, although conceptually more expensive, proved to be the most effective due to the actual very adverse working conditions. Thus, this solution turned out to be more economical. The bridge and access viaducts are the result of all the skill and art used in the studies that led to the execution of this contract job, with all the guarantees of success (Figure 9).

Figure 9. Construction phases of the bridge deck with advance form travellers.

The first piling works began in November 2008 and the last concreting of the main bridge deck was completed in July 2010. After 20 months as of the beginning of the contract job's deadline, 6km of piles, approximately 33000m3 of concrete, 2700 tons of passive steel and 355 tons of pre-stressed steel had already been used. Thus, the longest bridge - 1534m long - ever built on Angolan soil was concluded.

196 Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-196

Case studies to support recent advances in geogrid technology Clifford D. HALL Tensar International Ltd, Blackburn, UK Abstract. Since the introduction of monolithic biaxial geogrids in the early 1980’s, geogrid developments have been characterised by the creation of geogrids from differing methods of manufacturing to produce the ribs, create the junctions and hence offer the aperture for some form of interaction between the fill and the geogrid aperture. In the period 2002-07, significant development in geogrid technology occurred when the geogrid structure was reviewed and revised so that the aperture shape became triangular rather than rectangular. Subsequent full scale testing in the laboratory and the field showed that for monolithic geogrids, this was a more efficient structure yielding better performance for each kilogram of polymer processed in manufacture. Experience and feedback from the field can be seen with respect to some of the principle stabilisation applications: • • • •

Reduced aggregate thickness Increased pavement life Controlling differential settlement Increasing bearing capacity

The paper draws from experiences and describes four case studies by highlighting both the observations made and the indications of meeting the performance expectations. Keywords. Geogrid, stabilisation, mechanically stabilised layer

Introduction Following the pioneering use of extruded meshes in Japan in the 1970’s, biaxial geogrids were specifically developed for stabilisation and incorporated into road projects as early as 1981[1]. The approach of using a mechanically stabilised layer (msl) was used primarily over very weak or variable formations, Figure 1, and as an alternative to a piled platform in this case. More commonly, the use of msl’s is compared with either excavation and granular replacement or the need to install excessively thick fill layers. Thereafter, msl’s applications were developed for base layers for the extension of pavement life in road construction.

C.D. Hall / Case Studies to Support Recent Advances in Geogrid Technology

197

Figure 1. Early use of biaxial geogrids in ground stabilisation

These first biaxial geogrids were manufactured by the punching of extruded sheets of polypropylene which underwent controlled stretching in the longitudinal, (machine), direction and then the transverse direction. This oriented the long chain molecules of the polymer in order to enhance tensile properties and create a structural framework of orthogonal sets of ribs to form square or rectangular apertures. As a consequence these stiff, planar geogrids had integral rib junctions and the resulting geogrid was monolithic. Other manufacturers subsequently developed different forms of geogrid and in the main these are formed from a series of separate polymer elements which are variously woven, knitted or welded at their rib junctions to create biaxial grids.

1. The case for a different geogrid structure In ground stabilisation, the wheel loading applied to the geogrid is multi-directional. From a rolling wheel passage over a single interlocking geogrid aperture, the force vectors through the msl vary in time, direction and magnitude. It follows that a geogrid which can offer near-uniform radial restraint, by possessing near uniform tensile stiffness in all radial directions, should produce even better performance in its reaction to those force vectors. In 2002, development work began on such a new type of product. The project resulted in a punched and drawn geogrid with stable triangular apertures with six ribs emanating radially from each node. This resulted in a variety of geogrids which possess much more uniform tensile stiffness in the radial sense. This is characterised by a polar diagram showing stiffness at low strain value commensurate with the strains experienced by geogrids in msl’s. The two characteristic forms of polar diagram are shown in Fig. 2. The biaxial geogrid has a cruciform shape, (dotted), whereas the multi-axial geogrid has a rose shape, (solid).

Multi-axial

Biaxial

Figure 2. The characteristic polar plot for multi- axial and biaxial geogrids

198

C.D. Hall / Case Studies to Support Recent Advances in Geogrid Technology

2. Case studies 2.1 Reduced Aggregate Thickness – Carroll Canyon Road, San Diego USA 2009 The road widening project anticipates traffic growth on the local network and comprised the construction of two extra lanes. The conforming design required the incorporation of 560 mm of unbound aggregate base. Plate loading tests were carried out[2] on the conforming and alternative design for a mechanically stabilised layer. The results of the plate loading tests, Figure 3, show that the surface modulus of the unreinforced construction was matched by the designs for the msl, with values of 87 and 99 MPa respectively. The msl designs were on the conservative side, based on methodologies developed for biaxial geogrids.

Figure 3. Plate loading tests, field values

According to flexible pavement design methods in the State of California, the designer can apply the geogrid benefit by connecting it with, or distributing it between, the three design input valuables: • Traffic Index, TI, a scalar property representing traffic load and related to equivalent standard axle passes. • Pavement support, R-value, the quality of support to the msl • Gravel factor, Gf, gravel factor, the increased quality of the aggregate. All factors combine to determine the thickness of the unbound aggregate layer in the empirical Imperial equation: T = 0.0032.TI (100 − R ) / Gf

Eq. (1)

where T is the thickness of the unbound aggregate layer, (ft). In this case study, field studies have demonstrated that T may be reduced from 560mm to 350 mm for the msl. Placing the geogrid effect on Gf, the designer may reflect the improved performance of the aggregate by ‘lumping’ the geogrid effect on the gravel factor. Accordingly, in this example, Gf was increased from 1.36 to 2.02 and a rule of thumb for this project emerges that: Gf ( msl ) ≥ 1.6Gf

Eq. (2)

It is therefore possible, using local evaluation methods, to arrive at a reduced pavement thickness for a mechanically stabilised base layer.

C.D. Hall / Case Studies to Support Recent Advances in Geogrid Technology

199

2.2 Increased Pavement Life – Khalifa Port Industrial Zone, UAE, 2010 In this desk study, a ‘value engineering’ exercise was carried out to preserve the pavement design life in converting the aggregate base layer of a pavement to a msl. The local requirements are based on AASHTO 1993 with the caveat that a minimum structure number be associated with the class or duty of the road. In this case, the Structure Number, SN, is required to achieve 7.9. A design traffic life of 250 million standard axles was the traffic load target that also needed to be achieved. The conforming design met these requirements and the impact of substituting the msl into the sub-base caused the structure number and the design traffic load to increase to 8.81. The effect of the geogrid was to increase the effective thickness of the sub-base layer in accordance with the confinement model, Figure 4, where the magnitude of the lateral confinement of the aggregate varies with depth. This showed that, rather than stabilisation over weak ground, an increase in traffic life was possible and a life cycle cost analysis would show up the benefit. In this particular exercise, as design targets were now exceeded, it was possible to reduce the sub-base thickness. The construction cost benefits of using multi-axial and biaxial geogrids were examined, as shown in Figure 4. The cost index of 1.00 refers to the conforming design.

Figure 4. Confinement model and a comparison of msl options in a value engineering exercise

200

C.D. Hall / Case Studies to Support Recent Advances in Geogrid Technology

Figure 4 indicates that the local requirement for a structure number, for the category of road in question, was preserved and as a result of which, the target traffic continues to be comfortably exceeded. All alternatives are similar in terms of their structural constitution. The effect of the geogrid is less dramatic than that which is found in some geogrid uses – such as stabilising weak ground. The reason is that the sub-base is relatively thick and rigorously evaluated stabilisation factors will be thickness-sensitive, according to the adopted confinement model and the geogridspecific behaviour. To complete the value engineering desk study, the environmental benefit of the multi-axial geogrid alternative showed that some 20kg of CO 2e greenhouse gas emissions would be saved per square metre of road construction. Applied to the whole project, this would amount to approximately 10,000 tons. 2.3 Controlling Differential Settlement – A66 Scotch Corner – Carkins Moor Improvement, UK 2007 The improvement work to a major highway included the remodelling of a junction with a side road. The layout required that the side road be re-aligned to pass over a portion of a former quarry which had been used as a landfill for wastes disposal. There was a concern about differential settlement in the side road along the irregular alignment of the former quarry walls. Here, the road straddled the firm ground and the wastes deposits. In this design and build contract, the contractor and his designers conducted value engineering workshops to meet the incentives that are part of the UK Highways Agency’s innovations that have been introduced into contracts of this nature. For the part of the site where the potential differential settlement problems were expected, the traditional solution would be to remove the wastes material and replace with engineering fill. Environmental legislation and taxes on both landfill disposal and imported aggregate mean that the traditional approach is not favourable. The value engineering exercise used a reinforced granular mattress with two layers of geogrid. The Engineer has devised a transition slab, Figure 5, using compacted layers of aggregate and the reinforced granular mattress both capped the wastes and enhanced the function of the transition slab. A multi-axial geogrid was selected as its more uniform radial distribution of geogrid stiffness (load carrying capacity at small strain) meant that whatever the orientation of the differential settlement that might occur at the rough and irregular quarry walls, the geogrid had a stiffness alignment to respond and smooth out the settlement.

Figure 5. Transition detail for the control of differential settlement

C.D. Hall / Case Studies to Support Recent Advances in Geogrid Technology

201

Regular inspection and monitoring by the client, over the last three years, indicate that whatever differential settlement may have occurred, there is no evidence of strain or cracking in the asphalt surface. 2.4 Increasing Bearing Capacity -Footbridge foundations in Workington, UK, 2009 In 2009, the north west of England experienced widespread flooding. In the town of Workington, one public highway masonry arch bridge collapsed and another was deemed unsafe. The British Army Royal Engineers was commissioned to build a temporary footbridge as soon as possible. The river banks of the selected site comprised sandy clay with low bearing capacity. By constructing a multi-layer msl as the bank seat foundation, its effective load spread angle could be deduced from bearing capacity tests[3]. The tests included pressure meters which measure vertical stress distribution. The envelope of the derived pressure bulbs indicates a vertical load distribution angle, Figure 6.

Figure 6. Bank seat support: load distribution angle based on bearing capacity testing

Summary Some of the pioneering projects using the newly developed multi-axial geogrid are presented indicating four of the primary mechanical stabilisation applications. They show satisfactory performances for base layer thickness reduction, bearing capacity and differential settlement along with the potential to increase the life of unbound aggregate in pavement construction. References [1] F.B. Mercer, Critical aspects of industrial and academic collaboration, The Philips Lecture, Royal Society, 1986 [2] SCS & T inc, Reinforced Pavement Section Study, Carroll Canyon, September 2009, Report 0911070 [3] Watts, K & Jenner, C.G., Large-scale laboratory assessment of geogrids to reinforce granular working platforms. EuroGeo4, 2008

202 Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-202

! " #

$ ' $ # # "

#$

#

+ "

$

$

# # !

X Zσ \δ^Z`σ \δ^ σ

δ # `

σ

{

# "

| `

! ! ! # \}^ $

! #

!

M.M. Shahin / Shaft Resistance of Model Pile in Wet Soil

203

+ \}^ # !

# ! #

# # \}^ ~ ~ ~ + # \~^ \~^ \~^ #

# \~^ \^

#

# ! $ \}^\~^

\^\^\^\^

\}^ \~^ # ## # \}^ \^ `

204

M.M. Shahin / Shaft Resistance of Model Pile in Wet Soil

# ! \~ ~^ \ ^

+ #

## #

+ # ! ! ! \^ # \}^

# } ⁄

} }~ } ~ ~ } } }~} } } }}""} }""}

# \~^ \^

!

# # !

\ # Z^ | \}~^ # \~^ \^ \`Z ~^ | \~^ \^ "

\}~^ " X ! ! # !

205

M.M. Shahin / Shaft Resistance of Model Pile in Wet Soil

\}^

}

\}^ } \}^ \}^

~

\}^

\}^

~

}

\^

\}^

~

\}^

~

\}^

}

\}^

| ~ " }"}

}

\}^ \}^

~

}

\^

| " }"}

\^

}

\^ }

\}^ \}}^

~

\^

\}~^

~

}

\^

| " "}~ ~

\^

~

\^

}

\}^

}

\}}^

\^

\}~^

~

}

\^

| " "}~

206

M.M. Shahin / Shaft Resistance of Model Pile in Wet Soil

#$ !

# \~^ \^{ \}"}^ # \"^

' # \~^ \^

\}^ \`Z ~^ { # "

\}"~^ \^ \^ # ~ σ σ

\^ }}~}

!

~~}}~ }} }}

"

~}}

}}

} ~

!

}}~

}}

}}~

"

}~}

}}~

! }~ } }}

" }~ }}~}~

\~^ } \^ \^ \^

}

\}^

\^

~

}

\^

| " }"

207

M.M. Shahin / Shaft Resistance of Model Pile in Wet Soil

# #

σ σ$

&

&'

*+,

/:

} }

~~}~

!

} }

"

~}

~

}}

!

~}~

}~

}}~ }}~}~

"

}~}

} }}

! }~}~ }} }} " }~ }}

" X ! ! # ! # !

# \~^ \^{ \"^ # \~^ \^

' ~ \}}^ # \~^ \\`Z~ ~^| \^ \^ " \} ^ "

! ! # !

M.M. Shahin / Shaft Resistance of Model Pile in Wet Soil ~

\}^ \~^

~

\^

}

\^

}

\^

208

\^

~

}

\^

| " }"

#$ ! '\"^

# \~^ \^ # \~^ \^ '

' !"

+ # } ` | \^ \^ #

|

# $ #

! " # \~^\^\^\^\^ \^

$

{ ! ! # ` ' !! { !

209

M.M. Shahin / Shaft Resistance of Model Pile in Wet Soil ~

~ }

^ \

} ^ \ ^ \

^ \

~

~

}

}

$

`

| $

}"} }"} ~

}

^ \

} ^ \ ^ \

~

^ \

~

~

}

}

$

`

| $

"}~ "}~

~

}

^ \

}

^ \ ^ \

~

^ \

~

~

}

}

$

`

| } $

}"} }"}

%&

'

" # \ "

^ \ δ^ #

# \^\^ \}^ + \}}^ # \^ ` δ #

δ # $

210

M.M. Shahin / Shaft Resistance of Model Pile in Wet Soil

; #

#

# # { # " # <

&

> ?&

δ δ>

'

'

~ !

"

~} !

}~

~ "

}~ }} ! ~~~~ }~~~ " ~~

"# }

" "

# ! "

M.M. Shahin / Shaft Resistance of Model Pile in Wet Soil

211

'

{ ! ` "

@ <

# " " # X }"{ " \}~^ \}^ #

#}} ~"

"

"+ #$

" \} " }^

\"^ " \^ \"^ " \ ^

" "

H '

}" +# + } | # + }} { | + + ` ` ~"~ ~" } #

{ " {} + }}}} # }~~"}~~ " }

#"~ " ~ ! + " }"

This page intentionally left blank

Section 5 Lateral Support and Retaining Structures

This page intentionally left blank

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 215 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-215

The effect of anchor post-tensioning on the behaviour of a double anchored diaphragm wall embedded in clay a

Amr ELHAKIMa,1 and Abdelwahab TAHSINb Department of Public Works, Cairo University, Giza, Egypt b Arab Consulting Engineers, Giza, Egypt

Abstract. A proper lateral support system is an important part of any deep excavation to lower the risk of damaging any adjacent structures, utilities or roads. For deeper excavations, the use of either strutted or anchored walls becomes necessary to decrease the amount of wall movement. The use of post-tensioned anchors further limits the amount of ground movement for both temporary and permanent structures. This paper investigates the effects of varying the posttensioning ratio on the behavior of a double anchored diaphragm wall retaining clay of different consistencies (medium stiff, stiff and very stiff). A parametric study on the behavior of a 10-m deep excavation with two levels of anchors is considered. The analyses are conducted using the two-dimensional finite element program PLAXIS Version 8.2. The study examines the variation of wall horizontal displacement, and bending moment and ground surface movement to provide guidance for different values of the post tensioning ratio. Keywords. Post tensioning, anchored walls, anchor, deep excavation, numerical modelling

Introduction A deep excavation unloads the surrounding ground because of the large amount of soil removed. Even with the stiffest available lateral support system, some ground movement is inevitable. The use of multi-propped walls has become increasingly widespread for temporary and permanent support of vertical excavations, especially when horizontal displacements must be limited. A wide range of technologies is currently available for braced or tieback walls [1]. Anchored walls have become popular in deep excavations because of the substantial progress in technology and availability of high capacity anchor systems. Additionally, anchored walls provide a working area free of obstructions thus improving the construction conditions in the underground portion of the building. The objective of this research is to numerically investigate the effects of soil relative density/consistency on the behaviour of anchored diaphragm walls embedded in clay under undrained conditions at different post-tensioning ratios. A parametric study was designed to achieve the research goals. 1

Assistant Professor, Soil Mechanics and Foundations Research Laboratory, Department of Public Works, Faculty of Engineering, Cairo University, Giza, Egypt; Email: [email protected].

216

A. Elhakim and A. Tahsin / The Effect of Anchor Post-Tensioning

1. Modelled Wall The problem considered in the analysis is a 10 m deep excavation with two levels of anchors and the total width of excavation is 40 m. The groundwater table is located 5 m below the natural ground surface. The wall is embedded 12-m into the clay layer to ensure its stability. The finite element mesh is extended a distance of 60 meters behind the wall and the total depth of soil to the boundary limit is 40 m to minimize boundary effects. The wall configuration is presented in Figure 1. Two rows of grouted posttensioned ground anchors are installed at levels (-2.00 m) and (-6.00 m). The important characteristics for the geometry of the model are summarized in Table 1. The anchors are inclined at 30 degrees with the horizontal plane and are modelled as link members with properties listed on Table 2. The stages of construction adopted in the current study are presented in Table 3.

Figure 1. Modeled double anchored diaphragm wall

Table 1. Characteristics of model geometry Item

Dimension (m)

Total excavation width

40.00

Excavation depth

10.00

Distance between wall and boundary

60.00

Wall embedment depth below excavation level

12.00

Total depth of soil below ground surface

40.00

Level of upper row of ground anchors

(-2.00)

Level of lower row of ground anchors

(-6.00)

217

A. Elhakim and A. Tahsin / The Effect of Anchor Post-Tensioning

Table 2. Anchor characteristics Free(bonded) length/ fixed (unbonded) length for fine-grained soil

12 m / 6 m

Young's modulus (E) for strands

2.1E8 KN/m2

Axial stiffness (EA) of strands and fixed length

1.15E5 KN/m/m

Table 3. Phases of construction of anchored diaphragm wall Phase

Description

0

Initial condition

1

Installation of diaphragm wall.

2

Excavation to level (-2.50)

3

Installation and post tensioning of upper row for anchors at level (-2.00)

4

Lowering ground water table to level (-7.00)

5

Excavation to level (-6.50)

6

Installation and post tensioning of lower row for anchors at level (-6.00)

7

Lowering ground water table to level (-10.50)

8

Excavation to final level (-10.00)

2. Investigated Parameters The effect of varying clay consistency (medium stiff, stiff, very stiff) on the behavior of the wall is investigated. The adopted clay properties (summarized in Table 4) are based on typical values provided by the Egyptian Code for Soil Mechanics, Design, and Construction of Foundations [2] based on clay consistency. The wall comprises of 60 cm thick reinforced concrete (EA= 1.455E7 kN/m, EI = 0.436E6 kPa). The analysis includes eight levels of post tensioning (PT) ratios; 0% (base model), 60%, 70%, 80%, 90%, 100%, 110% and 120%. Table 4. Clay parameters Clay type

cu

ν

(kPa)

E

γwet

γd 3

(kPa)

(kN/m )

(kN/m3)

Medium Stiff

50

0.45

6000

17

18

Stiff

100

0.45

10000

17.5

18

Very stiff

200

0.45

20000

17.5

18

3. Numerical Model A series of two dimensional (2D) finite element analyses are performed using PLAXIS Version 8.2 to model the wall as a plane strain problem. Soil is modelled using a 15node triangular element, which produces more accurate results in modelling twodimensional problems. The diaphragm wall is modelled as 5-node beam elements which are slender structures in the ground with significant flexural rigidity and normal

218

A. Elhakim and A. Tahsin / The Effect of Anchor Post-Tensioning

stiffness. The unbonded anchor length is modeled using a node-to-node anchor while the bonded length (grouted body) is represented as a geotextile which guarantees a continuous load transfer to the soil and avoids a concentrated point load at the end of the unbonded anchor length [3]. At the vertical boundaries, horizontal fixity is applied (u x = 0), while both lateral and vertical displacements (ux and uy) are restrained at the lower horizontal boundary. The model dimensions of 80m width x 40m depth are selected to minimize the boundary effects on the accuracy of results. These chosen model dimensions relative to the excavation width and depth are in good agreement with values used by Briaud and Lim [4]; and Schweiger [5]. A hardening soil model is used for representing soil behavior as it is more suited for simulating the soil deformations behind earth retaining structures [5].

4. Initial State of Stresses At any site, initial soil stresses exist before any construction work is started. It is important to replicate the in-situ stress conditions because they could influence any subsequent analysis. The vertical overburden stress at any depth z within the soil mass is computed as the summation of vertical stress σvo = (Σ ρig Δzi), where g is the gravitational acceleration, ρi is the soil mass density, and Δz i is the soil layer thickness. Horizontal stresses are not as easily computed. In many cases, the horizontal stress σho’ is determined by the at rest coefficient Ko, where Ko = σho’/ σvo’ [6]. In the current model, initial stresses are computed based on the at rest earth pressure coefficient K o which is evaluated using Jaky’s formula [7].

5. Parametric Study Simulations of the wall response were performed for medium stiff, stiff and very stiff clay. Figure 3 shows the wall displacements for different post tensioning ratios (0%, 60%, 70%, 80%, 90%, 100%, 110% and 120%). As expected, the wall movement decreases as clay stiffness increases. The top wall movements are reduced by 6.6%, 16.1% and 13.5% using 120% PT compared to 0% for medium stiff, stiff and very stiff clays, respectively. Figure 3 shows the bending moment versus depth diagram for different clay consistencies. The figure shows that the maximum bending moments increase by 7.8%, 2.8% and 1.9% using 120% PT compared to 0% PT for medium stiff, stiff and very stiff clays, respectively. The ground surface vertical movement is presented in Figure 4 for the different soil consistencies and anchor post-tensioning ratios. The ground movement decreases as the clay gets stiffer. The anchor posttensioning ratio reduces the ground movement by 9.2%, 7.9% and 11.8% using 120% PT compared to 0% for medium stiff, stiff and very stiff clays, respectively.

6. Conclusions Numerical modeling is used to investigate the effect of varying the anchor post tensioning ratio on the response of a double anchored diaphragm wall embedded in

219

A. Elhakim and A. Tahsin / The Effect of Anchor Post-Tensioning

clay. It is shown that increasing the post-tensioning ratio reduces the wall and ground surface movements. On the other hand, the maximum wall bending moment increases with the increase in post tensioning ratio. The level of influence is affected by the clay consistency as illustrated in the paper.

Wall horizontal displacement (m)

Wall elevation (m)

0.08 0

0.1

0.12

0.14

0.16

Wall horizontal displacement (m)

Wall horizontal displacement (m) 0.18

0.04 0

0.05

0.06

0.07

0.02 0

-2

-2

-2

-4

-4

-4

-6

-6

-6

-8

-8

-8

-10

-10

-10

-12

-12

-12

-14 -16 -18 -20 -22

0.03

0.035

-14

-14

0 PT 60 PT 70 PT 80 PT 90 PT 100 PT 110 PT 120 PT

0.025

0 PT 60 PT 70 PT 80 PT 90 PT 100 PT 110 PT 120 PT

-16 -18 -20

-18 -20 -22

-22

(a)

0 PT 60 PT 70 PT 80 PT 90 PT 100 PT 110 PT 120 PT

-16

(b)

(c)

Figure 2. Wall movement versus depth for (a) medium stiff clay, (b) stiff clay, and (c) very stiff clay

Bending moment (kN.m/m')

Bending moment (kN.m/m')

Wall elevation (m)

-400

-200

0

200

400

600

-200

-100

0

100

200

300

Bending moment (kN.m/m') 400

-100

0

0

0

0

-2

-2

-2

-4

-4

-4

-6

-6

-6

-8

-8

-8

-10

-10

-10

0 PT 60 PT 70 PT 80 PT 90 PT 100 PT 110 PT 120 PT

-14 -16 -18 -20

-14

0 PT 60 PT

-14

-16

70 PT 80 PT

-16

90 PT 100 PT 110 PT

-18 -20

120 PT

-22

-22

(a)

200

300

-12

-12

-12

100

0 PT 60 PT 70 PT 80 PT 90 PT 100 PT 110 PT 120 PT

-18 -20 -22

(b)

(c)

Figure 3. Wall bending moment versus depth for (a) medium stiff clay, (b) stiff clay, and (c) very stiff clay

220

A. Elhakim and A. Tahsin / The Effect of Anchor Post-Tensioning Distance from wall (m)

Vertical Displacement (m)

0 0

-5

-10

-15

-20

-25

-30

-35

-40

-45

-50

-55

-60

;ĂͿ

-0.02 -0.04

0PT 60PT 70PT 80PT 90PT 100PT 110PT 120PT

-0.06 -0.08 -0.1 -0.12 -0.14 Distance from wall (m) 0

-5

-10

-15

-20

-25

-30

-35

-40

-45

-50

-55

-60

-55

-60

Vertical Displacement (m)

0

0PT 60PT 70PT 80PT 90PT

;ďͿ -0.01

-0.02

-0.03

-0.04

Distance from wall (m) 0

-5

Vertical Displacement (m)

0 ;ĐͿ -0.005

-10

-15

-20

-25

-30

-35

-40

-45

0PT 70PT 90PT 110PT

-50

60PT 80PT 100PT 120PT

-0.01 -0.015 -0.02

Figure 4. Vertical ground surface movement for (a) medium stiff clay, (b) stiff clay, and (c) very stiff clay

References [1] Carrubb, P. and Colonna, P. (2000). A comparison of numerical methods for multi-tied walls. Computers and Geotechnics,Vol. 27, 117-140. [2] Egyptian Code for Soil Mechanics, Design, and Construction of Foundations, Ministry of Housing, Cairo, Egypt, 2001. [3] Brinkgreve, R. B. J. and Vermeer, P. Finite element code of soil and rock analyses Version 7, Plaxis B.V., Netherlands, 1998. [4] Briaud, J. L. and Lim, Y., Tieback walls in sand, numerical simulations and design implications, Journal of Geotechnical and Geo-environmental Engineering , ASCE, Vol. 125, No. 2 (1999), 101-110. [5] Schweiger, F., Benchmarking in geotechnics_1, Computational Geotechnics Group, Institute for Soil Mechanics and Foundation Engineering, Graz University of Technology, Graz, Austria (2002). [6] Mayne, P.W., and Kulhawy, F.H., Ko-OCR relationships in soil. Journal of Geotechnical Engineering, 108 (GT6) (1982), 851-872. [7] Jaky, J., Earth pressure in soils. Proceedings of the Second International Conference on Soil Mechanics and Foundation Engineering, Volume I (1948), 103-107.

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 221 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-221

Observed axial loads in soil nails S.W. JACOBSZa,1 and T.S. PHALANNDWAb a University of Pretoria, Pretoria, South Africa b Esor Franki, Johannesburg, South Africa

Abstract. Three instrumented soil nails were installed along a 10m high soil nail and sprayed concrete supported vertical face in a residual andesite profile. The axial load was monitored at several positions along the nails’ lengths as the height of the retained face increased. The mobilised loads were compared with average tensile loads calculated using a simple wedge analysis commonly used in soil nail design. As the excavation deepened, the mobilised loads were initially less than the calculated load, but soon corresponded well with the predicted values. After the soil nail wall had reached about 60% of its final height, the loads in the top row of soil nails stabilised and did not increase further. The ingress of water after rainfall significantly affected the observed loads, illustrating the need of effective drainage. Temperature effects from daily temperature fluctuations were found to be negligible. This paper does not focus on the friction on the nail shaft but only on the axial loads. Keywords. soil nail, axial load, load mobilisation

Introduction Soil nail retaining systems are widely used in South Africa for the support of excavation faces. A substantial amount of work has been done both internationally and in South Africa on the pull-out capacity of soil nails, but less information is available on the mobilisation of load in soil nails as excavation work progresses. Studies are currently been carried out at the University of Pretoria to investigate the mobilisation of axial load in soil nails as the height of the retained face increases. A first trial was carried out on a soil nail retaining wall supporting an excavation for the Gautrain railway line in Pretoria. Three soil nails were instrumented and installed to measure the axial load at several locations along their lengths as the adjacent excavation was deepened. This paper describes the instrumentation of the soil nails and the observed axial loads as the excavation was deepened. It discusses the effects of daily temperature variations and rainfall on the axial forces predicted. Due to data limitations, the paper does not focus on the axial load distribution on nails or the friction generated along the length of the grouted nails.

1

SW Jacobsz, Department of Civil Engineering, University of Pretoria, Pretoria, 0002, South Africa, [email protected].

222

S.W. Jacobsz and T.S. Phalanndwa / Observed Axial Loads in Soil Nails

1. Project background A significant portion of the alignment of the Gautrain Rapid Rail Link between the Pretoria and Hatfield stations follows an existing railway line. In order to accommodate the new railway line next to the existing line, widening of the available railway corridor was required. Space constraints necessitated excavation into adjacent material, requiring several kilometres of excavation faces to be supported. The total length requiring support was 4,2km, with excavation depths varying between 3,5m to 12m. 1.1. Regional geology and soil profile The soil profile along the retaining wall comprises moist, yellow-brown, silty sand from residual andesite of a dense consistency. The structure is widely jointed. The consistency of the excavated material tended to increase with depth to medium dense at the top to very dense at the base of the excavation. On the other side of the railway line, steeply dipping shale occurs. The shale does not occur in the face which required support. 1.2. Soil nail retaining wall The retaining wall comprises 125mm diameter soldier micropiles installed at 2m centres along the excavation face. They were reinforced with four galvanised Y12 bars and grouted with 25MPa grout. Six rows of soil nails were installed between the soldier piles at 2,0m x 1,5m centres (horizontal:vertical spacing). The 175mm thick shotcrete wall between the soldier micropiles was reinforced with two layers of galvanised 395Ref. mesh. The total retained height was 10m. The soil nail lengths of the upper three rows were 12m, followed by two rows of 9m and a bottom row of 6m. A permanent vertical geofabric drain was placed behind the shotcrete skin at 2,0m spacing, together with 50mm diameter weepholes installed at 2.0m x 2.0m centres to ensure full drainage behind the excavation face.

2. Instrumentation system 2.1. Instrumented soil nails The soil nails used were 25mm diameter threaded hollow bar with a yield stress of 549MPa, offering an ultimate tensile load capacity of approximately 210kN per bar. The bar lengths of 3m were joined using screw-in couplings. The soil nails were instrumented by fixing strain gauges to the couplings. Flat surfaces were machined onto the couplings to accommodate the gauges. Two gauges, assembled into half-Wheatstone bridges, were used per coupling. One gauge was orientated parallel to the soil nail and the other perpendicular to the nail, implying a temperature compensated system. The strain gauges used were the CEA-06-W250A350 weldable type with a gauge resistance of 350Ω, manufactured by the MicroMeasurements Group. After completion of the wiring, the strain gauges were covered using silicone sealant and the instrumented couplings were surrounded by a heat shrink sleeve to

S.W. Jacobsz and T.S. Phalanndwa / Observed Axial Loads in Soil Nails

223

provide protection when installed in the ground. Figure 1 illustrates two instrumented couplings, one with and one without a heat shrink sleeve. All load cells were calibrated by applying known tensile loads using a hydraulic testing apparatus in the laboratory before assembly into a soil nail. A Datataker DT615 logger was used for the load cell calibration and the recording of data on site. The logger supplies a constant current excitation of 2,5mA which eliminates lead wire effects. The calibration factor in terms of the output voltage of the instrumented couplings varied slightly, but was approximately 75N/μV. Heat shrink sleeve protection

Strain gauge

Figure 1. Soil nail couplings instrumented to measure tensile load.

2.2. Installation on site Three instrumented soil nails were installed, two with a length of 12m and one with a length of 9m. The spacing of the load cells is illustrated in Figure 2 showing the wall cross section. The outer 1,5m of the soil nail bars were debonded from the grout by means of a debonding tape. The 105mm predrilled holes were filled with a watercement grout, after which the soil nails, fitted with centralisers, were installed. The end-plates and nuts were installed after the grout had gained sufficient strength approximately 3 to 4days later. All strain gauged couplings were connected to a multi-core cable passing along the inside of the hollow soil nail bars. The soil nails were pre-assembled, but with the instrumented couplings only joined to one end of the bars before installation so that the nails could be transported to site in manageable lengths. On site, the nails were straightened out, joined together and inserted into a predrilled hole filled with a 25MPa cementations grout. 2.3. Success rate It proved problematic for the instrumentation to survive the installation process on site and many instrumented couplings did not function correctly. Assembly of the soil nails on site often resulted in damage to the instrumentation cables. As the soil nail segments were screwed together, the wires from each coupling were twisted around the multicore cable inside the hollow rods, breaking connections. Wires also got trapped in the screw thread, damaging connections and causing short circuits.

224

S.W. Jacobsz and T.S. Phalanndwa / Observed Axial Loads in Soil Nails

When assembling the last instrumented nail, the multi-core cable along the inside of the nail was stripped from its sheathing and only the wires required to service each load cell passed along the inside of the nail. This resulted in the wires being able to twist together during assembly of the nails without damage. Also, wires to a specific instrumented coupling could be tensioned from the outside end as the nail was assembled, preventing them from being damaged in the screw thread. Only couplings 1 and 2 in the top soil nail operated correctly, as did all couplings in the third soil nail. However, in the case of the third nail, the signals became noisy towards the end of the monitoring period so that the load could not be determined with certainty. C1 C2

C3

C4

1,5m 4m

12m

6m 7m

12m 9m 9m

10m

Coupling No Instrumented coupling Nail length 12m

10°

Nails not monitored

6m

Figure 2. A cross section of the soil nail retaining wall showing the positions of instrumented couplings.

3. Selected monitoring results 3.1. Axial loads Figure 3 presents the axial loads from the first and second couplings of the upper soil nail. The average axial loads per soil nail as a function of the wall height, estimated using a simplified single wedge analysis, are also presented in Figure 3. Despite the limited resolution of the excavation depth record (recorded once a week), axial load does generally appear to be correlated with excavation depth during the first approximately 7 weeks. At wall heights below 4m, the estimated load significantly exceeded the measured load, probably because the retained face was temporarily supported by suctions generated in the soil in response to the excavation. Load had at this stage not yet been mobilised in the soil nails. As the excavation deepened and movement of the retained face occurred, tensile loads mobilised in the soil nails. The magnitude correlated well with estimates from the wedge analysis. As the height of the retained face reached 6m to 7m, the loads in the upper nail stabilised at levels of between 60% and 70% of the estimated values and did not increase appreciably with further increasing wall height. Over time, the load at Coupling 1 slowly increased, while that in Coupling 2 reduced gradually. Byrne (1992) mentions that this is probably attributable to load redistribution within the nail from the grout annulus to the steel bar, rather than from soil creep. As the height of the face increased beyond 7m, equilibrium was maintained by mobilising load in the soil nails installed at greater depth (data not presented). Despite doubt regarding the accuracy of the measurements from the third instrumented soil nail,

225

S.W. Jacobsz and T.S. Phalanndwa / Observed Axial Loads in Soil Nails

it appears that the magnitude of loads generated here might have exceeded the estimated values. The consistency and joint spacing in the residual andesite increase with depth, resulting in a stronger soil mass at depth. This offers a possible explanation for the stabilisation of the load in the upper soil nail. The stabilisation of the load can possibly also be ascribed to the particular joint configuration at the location of the instrumented nail, so that an increase in wall height did not result in the mobilisation of a larger failure wedge at this particular location. Of interest are the “spikes” in the load records during the first 11 weeks. This was initially ascribed to measurement errors, but when compared with the rainfall record, an excellent correlation was found (refer to the rainfall record in Figure 3(c)). A ditch located immediately behind the retained face allowed water to infiltrate behind the wall, dramatically increasing the soil nail loads. As drainage occurred, the loads rapidly returned to their former values. ϭϬϬ

Coupling C1

(a)

Axial load (kN)

ϳϱ ϱϬ Ϯϱ Ϭ ϬϮͬϮϲ

ϬϯͬϮϲ

ϬϰͬϮϯ

ϬϱͬϮϭ

Ϭϲͬϭϴ

Ϭϳͬϭϲ

Ϭϴͬϭϯ

ϬϵͬϭϬ

ϭϬͬϬϴ

ϭϭͬϬϱ

ĂƚĞ

Axial load (kN)

ϭϬϬ

Coupling C2

(b)

ϳϱ ϱϬ Ϯϱ Ϭ ϬϮͬϮϲ

ϬϯͬϮϲ

ϬϰͬϮϯ

ϬϱͬϮϭ

Ϭϲͬϭϴ

Ϭϳͬϭϲ

Ϭϴͬϭϯ

ϬϵͬϭϬ

ϭϬͬϬϴ

ϭϭͬϬϱ

Date

ϰϬ

ϭϬ

(c)

ǆĐĂǀĂƚŝŽŶĚĞƉƚŚ;ŵͿ

Daily rainfall & wall movement (mm)

ϱϬ

ϯϬ tĂůůŵŽǀĞŵĞŶƚ

ϮϬ

ZĂŝŶĨĂůů

ϭϬ Ϭ ϬϮͬϮϲ

ǆĐĂǀĂƚŝŽŶĚĞƉƚŚ

Ϭ ϬϯͬϮϲ

ϬϰͬϮϯ

ϬϱͬϮϭ

Ϭϲͬϭϴ

Ϭϳͬϭϲ

Ϭϴͬϭϯ

ϬϵͬϭϬ

ĂƚĞ

Figure 3. Axial load variation in top soil nail (a & b) and influencing factors (c).

ϭϬͬϬϴ

ϭϭͬϬϱ

226

S.W. Jacobsz and T.S. Phalanndwa / Observed Axial Loads in Soil Nails

3.2. Temperature effects The soil nail instrumentation comprised half Wheatstone bridges, implying temperature compensation. However, due to daily thermal expansion and contraction of the soil and the section of the soil nail near the supported face, changes in the axial load were possible. In addition, imperfections in the instrumentation could result in some thermal effects being measured. Figure 4 illustrates a six day axial load record from the first two couplings of the upper soil nail commencing 36 hours after installation. The end plate had not yet been installed, but the grout would have set by this time. A small load variation of about 2kN is evident on Coupling 1, with an even smaller variation on Coupling 2 where a smaller temperature variation could be expected. Once the soil nail had reached a steady load, the daily load variation doubled to about 4 kN. Tensile loads increased as temperature increased, suggesting that the expansion of the sprayed concrete wall and soil immediately behind it exceeded the expansion the soil nail. Maximum loads lagged the time of occurrence of the maximum daily temperatures by approximately 2,5 hours. Measured axial load (kN)

4

2

0

-2

Nail 1 Coupling 1 Nail 1 Coupling 2

-4 28-Feb

02-M ar

04-M ar

06-M ar

Date Figure4. Axial load variation due to thermal effects.

4. Conclusions The following conclusions are presented: • Although generally obeying the trend of the increasing excavation depth with time, the mobilisation of loads in the soil nails exhibited a step-like, non-gradual fashion in the residual andesite soil profile investigated. • After a wall height of approximately 60% of the final value had been reached, the mobilised load in the upper soil nail stabilised at values less than that estimated from a wedge analysis and did not increase further with increasing wall height. This is thought to be a consequence of the increase consistency with depth and the jointed structure of the residual andesite. During this time, load mobilisation occurred in the lower rows of nails. • Water ingress after rainfall significantly increased mobilised soil nail loads, illustrating the need for effective drainage behind and above the wall. • Temperature effects on soil nail loads were found to be small, with loads increasing slightly with rising temperatures.

S.W. Jacobsz and T.S. Phalanndwa / Observed Axial Loads in Soil Nails

227

Acknowledgements The authors acknowledge the permission of the Gauteng Provincial Government and the Bombela Concession Company to publish this paper. The views and opinions expressed are those of the authors and not those of the Province or Bombela. The authors also acknowledge the contribution of Mr J.U.H. Beyers who assembled and calibrated the soil nails and assisted with installation, data collection and interpretation.

References [1] R.J. Bryne, Soil Nailing: A simplified kinematic analysis. Proc Grouting, Soil Improvement and Geosynthetics Conference, Vol 2, ASCE Geotechnical Special Publication (1992), pp751-764.

228 Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-228

Deformations of Soil Reinforced Walls in Relationship with Reinforcement Used Edoardo ZANNONI a , Marco VICARI b and Moreno SCOTTO b a Maccaferri Southern Africa, Durban, South Africa b Officine Maccaferri

Deformations become the main design parameter when serviceability limit state analysis becomes more restrictive than ultimate limit state analysis. In this paper, deformation of soil reinforced walls will be related to the properties of the reinforcement material. A 5m high soil reinforced wall will be analysed to relate deformation using a double twisted hexagonal wire mesh and a geosynthetics bonded geogrid as reinforcements. The performances of the double twisted hexagonal wire mesh will be evaluated using test in air and confined in soil. The maximum deformation of the wall is 75% higher in air than in confined soil condition. The confined soil condition represents a more realistic behaviour of the reinforcement. Short-term deformation of a geosynthetics bonded geogrid happen during construction and cannot be highlighted due to the soil consolidation process. Therefore, only a long-term analysis is conducted where only viscous deformation due to creep are accountable. Keywords. Soil reinforced wall, deformations, reinforcement, geosynthetics, double twisted wire mesh,

Introduction Walls are designed taking into account the most hazardous situations, which are commonly sliding, overturning, external and internal stability. The design is usually completed once the wall passes these checks. Most of the codes of practice and guidelines put lot of emphasis on these “main failures”, giving lesser importance to wall deformations, which are very important for the serviceability of the structures. Restrain conditions develop more challenging engineering solutions when new buildings, fly overs, wide bridges and higher wall are required to be erected close to each other where any movements could cause failure. The economic affordability of stabilized earth walls is replacing most of the classic concrete wall solutions. In few situations classic concrete wall is still preferred to stabilize earth walls when restrictive deformation is required. In this paper only deformations of reinforced walls will be considered, without taking into account soil displacement as settlement, swelling or influence of water pressure and local displacement in the facing. The range of soil reinforcement products increased substantially due to the development of geosynthetic’s performances. The design approach for steel as a soil reinforcing is different than geosynthetics because steel does not have creep phenomena. Designing with steel as soil reinforcing is therefore based on the elastic theory as its properties are not affected by ageing. The design of geosynthetics as soil

E. Zannoni et al. / Deformations of Soil Reinforced Walls in Relationship with Reinforcement Used 229

reinforcement must take into account the ageing of the material through to creep phenomena. In the next paragraphs, deformation of reinforced wall will be discussed based on the types and behavior of reinforcement. FEM analysis will be used to demonstrate the relationship between the modulus of the reinforcement and the horizontal displacement of the wall.

1. Wall deformations In this paper, only horizontal wall deformations will be considered. Vertical movements are mainly due to soil behavior and not to the reinforcement, whilst for horizontal movements, especially in long-term, the horizontal deformations are due to the wall itself. Horizontal deformations can take place in the short-term when the structure is still in construction or just completed when the overburden tensions are still not dissipate in the soil and there are still filtration movements in the soil due to consolidation process. Long-term can be defined when all aforementioned situations are over and the horizontal movement can be attributed to the reinforcement. Soil reinforced walls consist of a facing and of a structural backfill. The facing can usually vary between 120mm to 1m in thickness which can be constructed using concrete cladding or gabion boxes filled with rocks. Structural fill consist of selected material compacted and reinforced with steel and geosynthetics. Wall deformations can occur due to the facing or the reinforcement. Generally facing deformations are due to an inappropriate construction methodology which results for instance in bulging of gabions boxes (Figure 1) or movement of concrete panels during installation phase (Figure 2). These facing deformations are usually independent to the behavior of the wall, but they can simply be local displacements due to external factors.

Figure 1. Gabion boxes facing deformation.

230 E. Zannoni et al. / Deformations of Soil Reinforced Walls in Relationship with Reinforcement Used

Figure 2. Concrete facing deformations.

Wall deformations due to the reinforcement are difficult to be understood because the soil consolidation processes take place in the short-term and usually develop higher deformations than those accounted by the reinforcement. Therefore is only possible to analyse the deformation due to the influence of the reinforcement in the long-term.

2. Assessment of reinforcement Most of the design methods for soil reinforced walls are based on either variations of the classical limit equilibrium which developed the wedge method and the coherent gravity method. Both these methods only calculate the tension in the reinforcement. Main points of discussion are how these tensions are developed in a reinforced soil structures and how they behave with time. The tension in reinforcement is activated by the displacement of the reinforcement (strain) in the soil during construction. The result of the research from Bathurst and Allen [6] indicate that it is difficult to accurately predict the total initial strain in the reinforcement, therefore it is difficult to estimate the total lateral displacements of the wall during construction. However they indicate that the general maximum strain in the reinforcement is typically of the order of 2-3% over the first 20-30% length of the reinforcement. The initial short-term strain in the reinforcement develops during construction (generally weeks or months) and only creep strain must be considered over the design life of the structure. Some reinforcements vary their performances if they are confined when placed in soil. Most of stress-strain tests to understand the behavior of the reinforcement are performed in air, but this does not correspond to the reality because the reinforcement is confined in soil. Deformations analyses take into account its elastic modulus J (or stiffness) and not the tensile strength of the reinforcement. In the following paragraphs a FEM analysis will be used on a 5m high wall by vary the elastic modulus of the steel and the geosynthetics reinforcement. The steel reinforcing will consist of a double twisted hexagonal steel wire mesh which will be assessed as a non-creep material (considering the steel in its elastic range). The geosynthetics reinforcement will consist in polyester bonded geogrid which will be assessed in its short-term and long-term elastic modulus values.

E. Zannoni et al. / Deformations of Soil Reinforced Walls in Relationship with Reinforcement Used 231

3. Double twisted hexagonal steel wire mesh The double twisted hexagonal steel mesh has an internal mesh opening of 80 mm with a wire diameter of 2.7 mm wire galvanized and PVC coated. It has been used as reinforcement since the early 1990’s and since then it has been used widely in soil reinforcement applications on both walls and slopes. The assessment of the double twisted mesh is based on its deformation modulus behavior that varies if the double twisted mesh is in air or confined. This behavior is due to its structure, because in air the hexagonal mesh stretches to close transversally and becomes longer longitudinally registering high deformation (or low deformation modulus) whilst in soil it does not occur because the stretching of the hexagon is avoided by the soil. The tensile strength in air (“nominal breaking load”, NBL) when tested in accordance with ASTM A 975-67 and linear deformation modulus (J) at failure strain were as follows: NBL = 50.4

kN ; m

J ≈ 500

kN m

(1)

Confined tests run on the wire mesh (Ismes Geo Internal report, 2003) demonstrated that this reinforcement behaves in a different way if confined in soil as it is in the soil reinforced wall. The reinforcement breaks at point a (Figure 3) developing a J modulus of 2947 kN/m.

Figure 3. Confined test results at σ v′ = 50 kPa Table 1. Behavior of linear deformation modulus (J) In Air

Confining Pressure σ v′ = 50 kPa

500 kN/m

2947 kN/m

232 E. Zannoni et al. / Deformations of Soil Reinforced Walls in Relationship with Reinforcement Used

The effects of the J value on the wall deformation were conducted using the GEOSLOPE programs Sigma/W [5]. This analysis was based on a 5m high wall with two different J values of 500kN/m and 2947 kN/m. 10 steps of 0.5m thick vertical layers were considered in the theoretical construction of the wall. The facing interaction was not considered in this analysis. The influence of J from the analysis is reported in Figure 4.

Figure 4. Horizontal displacement of the wall using double twisted hexagonal steel wire mesh

The J value in air of 500kN/m is not representative of the real J value in a soil reinforced wall where the J value is 2947kN/m. This in air J value in accordance with ASTM A975-67 gives unrealistic high deformation compares with calculate deformation when using confined J as it is in practice. In other words by assuming the “in air” J value there is an overestimation of the calculated displacement than compare with the actual displacement of the real structure; which can be actually measured on real structures [1]. Double twisted hexagonal steel wire mesh being made in steel is not affected by long-term deformations because steel does not presented creep phenomena as does geosynthetics. For this reason displacements in the long-term can be considered negligible with respect to soil deformations.

4. Geosynthetics reinforcement Many types of geosynthetics can be used as reinforcement. Geosynthetic reinforcement must be defined by a tensile strength available at the design life of the structure. The short-term strength of geosynthetic reinforcement has been verify, documented and accepted in the engineering circle; however the long-term strength still need to be assessed in every design.

E. Zannoni et al. / Deformations of Soil Reinforced Walls in Relationship with Reinforcement Used 233

Double twisted hexagonal wire mesh behaves elastically in air and in confined environment, whilst for geosynthetics reinforcement this is not the case because by nature, all geosynthetics are influenced by viscous deformations. All structures are characterized by a design life span, which is typically less than 120 years. There are no data regarding geosynthetics reinforcement at 120 years due to the recent development of geosynthetics. The oldest field test available is about 20 – 30 years old (Naughton et al. 2009) which is not applicable to the typical design life. The main issue to design a soil retained wall with geosynthetics is to understand deformations in the long-term design when the consolidation effects have already dissipated. General geosynthetic behavior (FHWA, 2009) is reported in Figure 5 where the strength experience a reduction due to creep and ageing. Geosynthetics can be assessed considering their short term or long-term behavior.

Figure 5. Geosynthetics strength behavior (FHWA, 2009)

4.1. Short-term behavior Geosynthetics are assessed using standard such as ISO, EN and ASTM; of which the preferred test is ISO expressing the main characteristic of that product. The stress-strain curve characterized all type of geosynthetics used as reinforcement in a short-term analysis (Figure 6). The short term behavior is usually not used in deformation analysis because these tests are representative of the ultimate values and are not design values.

234 E. Zannoni et al. / Deformations of Soil Reinforced Walls in Relationship with Reinforcement Used

Figure 6. Strength – Strain behavior for main geosynthetic reinforcements

If the values reported in Figure 6 are used in a deformation analysis, the results will show high deformation considering common geosynthetics reinforcement as PET Bonded geogrid and PET woven geogrid. What is missing in this statement is the soil – reinforcement interaction because the soil has to deform to activate the bond of the reinforcement and to develop the tensile strength required. Therefore, the reinforcement will deform with the soil during the construction process [3] until the equilibrium between soil and reinforcement is achieved. 4.2. Long-term behaviour Reinforcement material characterized by the design parameters at the end of the design life of the structure. A typical design life for a civil structure is 120 years, which can drop to 50 years for mining structures. Therefore, geosynthetics reinforcement has to be assessed in their long-term performances. Reinforcement deformations have been influenced by viscous behaviour that happens with time due to the intrinsic structure of the raw materials. Viscous deformation or creep gives the effective performance of geosynthetics reinforcement during its life. The result of a creep analysis is a stress-strain graph that reports all isochronous curves as in Figure 7.

E. Zannoni et al. / Deformations of Soil Reinforced Walls in Relationship with Reinforcement Used 235

Figure 7. Stress-strain isochronous curves

The effect of creep is visible in the upper part of the graph where the curves bend to the right. The strain at 24 hours has already happened due to compaction phases, therefore only the difference in abscissa between the 10 6 hours and the 24 hours correspond to the deformation is due to creep in the reinforcement. By analysing the same structure of paragraph 3 for a design life of 104 (10 years) and 106 hours (114 years) using reinforcement with a deformation modulus of 1 and 2% as reported in the Figure 7, the calculated horizontal displacements of the wall are maximum 11mm at 10 years and 20 mm at 114 years (Figure 8).

Figure 8. Horizontal deformation of 5m high wall using GSY reinforcement at 10 and 114 years.

5. Conclusions The choice of the reinforcement under a deformation point of view is of primary importance because it can compromise not only the serviceability limit of the structure but also the ultimate limit if the retained wall is close to other structures than can cause a failure if high deformations occur.

236 E. Zannoni et al. / Deformations of Soil Reinforced Walls in Relationship with Reinforcement Used

Deformation analysis is based on the choice of the elastic modulus J where an incorrect interpretation on the behavior of the reinforcement can compromise an effective design. Although many standards and guidance have been written in the past and are still in process of updating, none of these do not provide to the designer the sufficient knowledge about the reinforcement. Only the right assessment of performances, safety factors and behaviors can provide the difference between an effective design and an overdesign.

References [1] V.N. Ghionna, M. Olivetta M. Vicari, Interpretation of pullout and direct sliding tests on double twisted steel wire mesh reinforcements, Eurogeo 3 2004, 683-688 [2] PJ Naughton. GT Kempton. R Lozano. M Scotto, J Meadows, Assessment of Hydrolysis in Historic Polyester Yarn Recovered From 20 - 30 Year Old Reinforced Soil Structures, GIGSA GeoAfrica 2009 Conference, 2009 [3] Lackner C & Semprich S, Prestressed geosynthetic reinforced soil by compaction, 9th International Conference on Geosynthetics, Brazil, 2010 [4] BBA Agrément Certificate 03/4065, Linear Composites’ soil reinforcement products, 2010 [5] Geo-Slope International Ltd., SIGMA/W for finite element stress/deformation analysis, Version 4. User's guide. Calgary, Alberta, Canada, 1997 [6] Bathurst R.J, Allen T.M. Huang B.Q, Current issues for the internal stability design of geosynthetics reinforced soil, 9th International Conference on Geosynthetics, Brazil, 2010 [7] Ghionna V.N., Fioravante V., Vicari M.. Full scale test on a retaining wall with non-uniform reinforcements. 7th ICG, Nice 2002.

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 237 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-237

Performance Comparison of VerticalHorizontal with Conventional Reinforced Soil Walls Using Numerical Modelling Binod SHRESTHA, Hadi KHABBAZ and Behzad FATAHI University of Technology, Sydney, Australia

Abstract. Conventional reinforced soil walls have demonstrated acceptable performance during earthquake occurrences. Nevertheless, there is still some potential for enhancing the performance of these walls without increasing the cost significantly. This paper presents an overview on the application of vertical components to the reinforced soil in addition to the horizontal reinforcement. The performance of conventional and the modified reinforced soil walls are evaluated and compared to each other. In this study, a series of 2D models is carried out using PLAXIS, finite element software, to investigate behaviour of these walls. The performance of reinforced walls is evaluated under the seismic loads of Kobe earthquake. The results indicate that the proposed wall with vertical reinforcement has superior performance compared with the conventional method and can reduce the risk of failure during earthquakes. Keywords. Vertical reinforcement, numerical modelling, seismic load, reinforced soil wall

Introduction The first modern-day design approach for reinforced earth structures was developed in the 1960s by a French engineer, Henry Vidal, using metal strips as reinforcement. He published his investigations in 1966. In the seventies in Britain, the production of geogrids by the extrusion under controlled heating of high-density polypropylene was started. The use of geosynthetic soil reinforcement has increased exponentially since last three decades as a result of high performance and low cost of construction. Many researchers have examined the stability of reinforced soil walls. A reinforcement embedded perpendicularly or at an inclination to the shear zone in a shear box to study the behaviour of a dry sand reinforced with different types of fibers was evaluated by Gray and Ohashi [1]. Arenicz and Choudhury [2] carried out a series of laboratory investigations to study the effects of different types of random reinforcements on soil strength. Contributions related to the new arrangement of reinforcement have played an active role in the development of reinforced soil technology. Furthermore, the study on conventional reinforced soil, where the reinforcements are implemented horizontally, some new configurations of inclusions were developed. A few studies were carried out to investigate the strength of soil reinforced with multi-layer horizontal-vertical orthogonal elements [3, 4] considering the strength in the case of static loading. In most reinforced soil walls, the reinforcement components are applied purely horizontal. In this study, vertical reinforcing components (connecting each layer of

238

B. Shrestha et al. / Performance Comparison of Vertical-Horizontal

horizontal reinforcement together) are also employed. The combined verticalhorizontal reinforced soil performance versus the conventional horizontal reinforcement under earthquake loads is analysed. The results of various models are produced using finite element geotechnical engineering software, PLAXIS. Through comparison between the vertical-horizontal reinforcement model and the horizontal reinforcement model, the response features of these two systems are evaluated.

1. Fundamental Behaviour of Vertical Reinforcement Soil deformation develops tensile or compressive stresses in the reinforcement. Soil shearing resistance builds from frictional contact among soil particles subject to the effective compressive stress. The magnitude of stresses depends on the reinforcement inclination in the direction of tensile or compressive stresses in the soil. The mobilised reinforced force, ultimately limited by the available bond, acts to alter the force equilibrium in the soil mass. Shear deformation in the soil causes tensile forces to be mobilised in the horizontal reinforcement, and provides two additional components (tangential and normal) of resistance in the slope. The tangential component of the reinforcement force directly resists the disturbing shear force in the soil, while the normal component of the force mobilises the additional frictional shearing resistance. Figure 1 shows the concept of applying vertical reinforcement and its additional components illustrated by Shrestha and Khabbaz [5]; Tvr cos θ resisting the disturbing shear force, and Tvr Sinθ , normal component of the force, providing extra frictional shearing resistance Tvr Sinθ tan θ . Beside this, vertical reinforcements can confine soil in different units alongside layers by horizontal reinforcement, and produce intact effect in the soil mass.

Figure 1. Effect of reinforcement on equilibrium allowing for horizontal and vertical reinforcement [5]

2. Numerical Modeling To investigate the performance of the reinforced soil wall by introducing vertical reinforcement comparing with conventional reinforcements, a 2D numerical analysis was carried out using the finite element software, PLAXIS. The height of retaining wall was assumed to be 10m with inclined facing of 1 in 20 having reinforcement of 7m in

239

B. Shrestha et al. / Performance Comparison of Vertical-Horizontal

length. The construction analysis of the wall was conducted layer-by-layer following the sequence as in conventional construction, which consists of thirteen layers of conventional horizontal reinforcement underlain by medium dense soil. It can be noted that this study is a preliminary attempt; hence a simple but reliable calculation method is chosen. For the sake of simplicitly, the water table level was assumed to be far below the wall foundation. Mohr-Coulomb failure criteria were assigned to all soils for the static and seismic analysis. For the seismic analysis of the model, the earthquake load of 1995 Kobe earthquake was used, which had a local magnitude of 7.2 in Richter scale and peak acceleration of 0.833g, where g is the earth's gravitational acceleration. The material properties, which are typical parameters for the selected soils, are summarised in Table 1. Two cases of numerical analysis were conducted; one is with conventional reinforcement and the other case is associated with vertical-horizontal reinforcement, as shown in Figures 2 (a) and 2 (b), respectively.. The backfill is the material to be compacted between geogrid layers; the fill material is employed to fill up the gap between the reinforced soil wall and the natural ground; and the facing soil is used to represent the relatively less compacted soil close to the facing elements. Parameters of the vertical reinforcement and horizontal geogrids are given in Table 2. Concrete facing elements were represented by dishrag wall (plates) and the vertical reinforcement was represented by the use of node to node anchor with elastic material behaviour. Parameter Material model Type of material Soil unit weight Horizontal Young's modulus Poisson's ratio Cohesion Friction angle Dilatancy angle Interface reduction

Name Model Type γunsat Kx Еref ν Cref Ф Ψ Rinter

Table 1: Sand and interface properties Backfill Ground Fill Mohr-Coulomb Drained 19 18 17 0.15 0.75 1 60000 50000 40000 0.3 0.2 0.3 0.5 0.5 5 38 33 30 4 3 2 0.85 0.75 0.67

Soil facing

17 0.5 30000 0.3 5 32 4 0.7

Unit kN/m3 m/day kN/m2 kN/m2 0 0 -

Table 2: Properties of the Vertical Reinforcement and Geogrids Elements Vertical reinforcement

Parameter Normal stiffness Spacing out of plane Strength

Geogrids

Normal stiffness Tensile strength

Symbol EA Ls Fmax, compression Fmax, tension EA Np

Value 1500 1 0 2.5 5000 100

Unit kN m kN kN kN/m kN/m

240

B. Shrestha et al. / Performance Comparison of Vertical-Horizontal

(a)

Conventional Horizontal Reinforcement.

(b) Vertical-Horizontal reinforcement Figure 2. Numerical analysis models

3. Analysis of Results The results of the finite element analysis are examined in both cases of reinforced walls with and without inclusion of the vertical reinforcement under seismic loading. The horizontal deformations at the face of wall in all layers of reinforcements were calculated, compared and presented in Figure 3. The deflection of retaining wall with inclusion of vertical reinforcement is drastically less than that of the conventional reinforcement wall. The maximum lateral displacement of vertical and conventional reinforced walls are 2.6 mm in top of the wall and 42 mm in one-third height of the wall, respectively. This result clearly indicates the potential benefits of vertical elements over conventional soil reinforcement for reducing horizontal deformations.

B. Shrestha et al. / Performance Comparison of Vertical-Horizontal

241

Figure 3. Lateral deflection of reinforced soil walls

To investigate the reason of improvement after the application of vertical elements, the axial stress in horizontal reinforcement were observed and found almost similar values in both cases. Madhavi Latha et al. [6] observed similar results with different type of reinforcing materials and concluded that this is due to the very low strain levels in the reinforcement layers. Axial forces on the vertical reinforcements were varied from 1x10 -4 to 3x10-3 kN/m, but the the higher values are concentrated towards opposite sides of the toe. This might be due to a rocking effect of the reinforced wall. Those outputs demonstrate that the action of vertical elements can hold the horizontal reinforcement layers in place during seismic loading application. Likewise, the examination of vertical shear stress between each layer revealed a promising outcome as the values are quite lower in case of inclusion of vertical reinforcement. The values of equivalent vertical forces from shear stress diagram between each layer are presented in Figure 4. Kinking of some points could be due to deformation of wall during seismic force. The results can be inferred that the connection of horizontal layers can increase integrity of the system due to confinement and the wall acts as a block during seismic load events.

Figure 4. Comparison of equivalent vertical force between horizontal layers

242

B. Shrestha et al. / Performance Comparison of Vertical-Horizontal

4. Conclusions Retaining walls of reinforced soil proved to be an appropriate solution for preventing wall failure, caused by earthquakes. Seismic responses of reinforced soil retaining walls with and without vertical reinforcement were analysed, by developing proper numerical models in PLAXIS software. Connection of each consequent layer resulted in remarkable reduction of the face panels’ displacement. It was observed that the ultimate tensile strength of horizontal reinforcement had an insignificant effect on the use of vertical reinforcement, whereas the vertical reinforcement induced integrity of the wall, reducing the lateral deflections. The findings of this study indicate that the proposed inclusion of vertical components to reinforced soil walls provides hefty stability compared with the conventional reinforced systems under earthquake loading.

References [1] D.H. Gray, H. Ohashi, Mechanics of fiber reinforcement in sand. Journal of Geotechnical Engineering, ASCE 109 (3) (1983), 335–353. [2] R.M. Arenicz, R.N. Choudhury, Laboratory investigation of earth walls simultaneously reinforced by strips and random reinforcement, Geotechnical Testing Journal 11 (4) (1988), 241–247. [3] M. X. Zhang, A. A. Javadi, and X. Min, Triaxial tests of sand reinforced with 3D inclusions, Geotextiles and Geomembranes, 24 (2006) 201-209. [4] M. X. Zhang, H. Zhou, A. A. Javadi, and Z. W. Wang, Experimental and theoretical investigation of strength of soil reinforced with multi-layer horizontal-vertical orthogonal elements, Geotextiles and Geomembranes 26 (2008), 1-13. [5] B. Shrestha, H. Khabbaz, Improving Reinforced Soil Performance Incorporating Vertical Reinforcement, GeoShanghai 2010 International Conference, Ground Improvement and Geotechniques, ASCE Geotechnical Special Publication 207 (2010), 249-254. [6] G. Madhavi Latha, and A. Murali Krishna, Seismic response of reinforced soil retaining wall models: Influence of backfill relative density, Geotextiles and Geomembranes 26 (4) (2008), 335-349.

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 243 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-243

The behaviour under excavation of the Luanda’s sandy formation: case studies NOBRE, Duarteª; PINA, Joãob; XAVIER, Baldomirob; ªTeixeira Duarte S.A. Delegação de Luanda b Teixeira Duarte S.A. Abstract. The aim of this paper is to describe the geotechnical solutions that are being applied in the sandy formation of the city of Luanda. For that purpose the author will describe the case study of the excavation, the retaining wall and the foundation solution of two buildings over twenty storeys high and with several underground floors. Besides the constructive solutions, such as diaphragm walls as the retaining wall system, ground anchors, struts, jet-grouting slab and top-down slabs as the stabilization system, raft and combined pile raft foundation as the foundation system and injections as ground treatment, special emphasis will be given to the geological description and the ground behaviour during excavation. Keywords. Retaining wall, diaphragm wall, ground anchors, jet-grouting slab, raft foundation, combined pile raft foundation, ground treatment, injections and monitoring.

Introduction As a consequence of the evident reduction of work in Portugal, promoters, project designers and contractors, as well as others, have started looking at the external market for alternatives. The stabilization of the political situation in Angola, allied to its wealth and financial capacity, as well as to the close relationship that Portugal keeps with this country, makes this market very attractive to Portuguese investment. As a consequence, the work in Luanda has increased, with strong incidence in the construction of tall buildings, frequently with more than twenty storeys above ground and several underground floors. However, the non-existence of previous deep excavations in Luanda, coupled with a lack of knowledge of local geotechnical conditions, such as the geotechnical behaviour during excavation, brings a great deal of uncertainly to the design of structural solutions and its implementation. ELYSÉE TOWER

GES TOWER

Figure 1. Downtown Luanda with the location of the Elysée and GES Towers.

244

D. Nobre et al. / The Behaviour Under Excavation of the Luanda’s Sandy Formation

The aim of this paper is to describe and characterize the behaviour of the sandy formations of Luanda during excavation works and to present examples of geotechnical solutions compatible with the local ground conditions. For that purpose, two case studies of the excavation, retaining wall and foundation solutions of two buildings in downtown Luanda will be presented (Figure 1), one in the lower city area, near Luanda’s Bay, with twenty-two storeys and three underground floors, with surface groundwater (Elysée Tower), and another in the higher city area, near Kinaxixe, with twenty-five storeys, six underground levels and a partially embedded ground floor (GES Tower, the tallest building in Luanda, inaugurated in September 2009).

1. Geology The geology of Luanda’s City substratum can be grouped into three distinct geological ages: one Recent Complex from Quaternary, one Pleistocene Complex also from Quaternary and one Miocene Complex from Tertiary [1]. The Recent Complex is generally composed of inorganic expansive clay of high plasticity (Cazenga’s Formation), with variable thicknesses of up to 4.0m. However, in the lower area of the city, the Recent Complex appears as a form of very loose to loose granular medium sands (Littoral Line Formation), sometimes muddy, with variable thicknesses. The Pleistocene Complex, usually named ‘The Muceque Formation’, is composed of loose to medium sands which reveals a collapsible behaviour when saturated and under load. This Complex occupies most of the surface of Luanda and has a variable thickness of up to 17m. The Miocene Complex, usually named ‘The Luanda Formation’, is the most heterogeneous formation and is characterized by clays of low to high plasticity, sometimes with expandable behaviour, silts and compact to very compact sand. In the upper layers an occurrence of limestone rock has also been identified, sometimes with high concentration of shells. The Luanda Formation occurs at the surface or, commonly, under Recent and Pleistocene Complexes and extends to depths of more than 50m. Most of the tallest buildings foundations (shallow and deep) as well as the majority of deep excavations are carried through in this formation. Groundwater mainly occurs at the lowest areas of the city, near the littoral line, and is synchronized with tide’s level. 0

20

40

ELYSÉE TOWER

60

0.00 2.50 5.00

C1A

7.50 10.00

DEEP (m)

0 LEGEND

C1A

LOOSE SAND

C1B

MEDIUM SAND WITH COBBLES AND BOULDES

MIOCENE - LUANDA FORMATION

22.50

C2A

MEDIUM TO DENSE SAND (NSPT˜ 30 A 40 BLOWS)

C2B

DENSE TO VERY DENSE SAND (NSPT˜ 40 A 60 BLOWS)

25.00 30.00

2.50 5.00 7.50 10.00

20.00

C2B

20

40

GES TOWER

60

PLEISTOCENE - MUCEQUE FORMATION C2B

C2E

22.50 25.00 C2F

30.00 32.50

35.00

35.00

37.50

37.50

40.00

40.00

LOOSE TO MEDIUM CLAYED SAND

MIOCENE - LUANDA FORMATION

20.00

32.50

C1

C2C

17.50

27.50

LEGEND

C1 C2A

12.50 15.00

GEOLOGICAL PROFILE

± 100.00

0.00

RECENT - LITTORAL LINE FORMATION

C2A

17.50

27.50

NUMBER OF BLOWS (SPT)

C1B

12.50 15.00

GEOLOGICAL PROFILE

± 0.00

DEEP (m)

NUMBER OF BLOWS (SPT)

C2A

LIMESTONE

C2B

MEDIUM SAND (NSPT˜ 30 BLOWS)

C2C

MEDIUM TO DENSE SAND (NSPT˜ 40 BLOWS)

C2E

DENSE SAND (NSPT˜ 50 BLOWS)

C2F

DENSE TO VERY DENSE SAND (NSPT˜ 60 BLOWS)

Figure 2. Geotechnical profile at the Elysée and GES Towers.

Although being situated in separate zones of the city, the two case studies described at the article have essentially been developed on the sands of the Luanda Formation (Figure 2). However, in the GES Tower, the excavation took place under dry conditions while in the Elysée Tower it was carried out in the presence of high groundwater level. In the GES Tower the presence of a limestone layer was in display

D. Nobre et al. / The Behaviour Under Excavation of the Luanda’s Sandy Formation

245

at the top of the Luanda Formation. The geotechnical site investigation was carried out through the execution of standard penetration tests (SPT) in eight boreholes, five in the Elysée Tower and three in the GES Tower, each one 40m deep.

2. Elysée Tower case study The Elysée Tower is property of “Sociedade de Participação Angolana” and is located at the lower area of downtown Luanda, less than 300m from the coastline, in the confluence of “Rua Rainha Jinga” and “Rua Pereira Forjaz”, near “Largo Julius Futchik”. The tower, with twenty-two stories and three underground floors, 30m long by 25m wide. In order to achieve the maximum depth excavation of approximately 11m, a retaining wall solution of a 17m deep and 0.50m thick reinforced concrete diaphragm wall was designed (Figure 3a). As a result of the high permeability of the sandy formations and in order to avoid erosion situations during ground anchors drilling, essentially induced by the groundwater gradient, a stabilization solution without the execution of ground anchors under the groundwater level was designed. Horizontal equilibrium was guaranteed by two stabilization levels, one active and one passive. The active stabilization level, positioned approximately at 2m depth, was assured by 46 temporary ground anchors of 1.5m spacing and with four 0,6” stands, 15m free length, 9m bound length and a lock off load of 600kN. The passive stabilization level, positioned approximately at 11.5m deep, was assured by a 1m thick jet-grouted slab comprising more than one thousand jet grouting type 1 columns with 1100mm diameter and a uniaxial compressive strength of 1.5MPa. This slab, carried out before the beginning of the excavation, allowed the adoption of wide vertical spans (of around 9.00m between the anchor heads and jet-grouting slab) thus avoiding the use of ground anchors under the groundwater level. In one of the retaining wall façades, the presence of a nearby five-story building with a shallow foundation forced the execution of two additional stabilization levels, one at the first and other at the second underground floors, each one assured by a topdown slab of 25m long, 5m wide and 0.30m thick. The support for the top-down slabs was guaranteed by vertical 12m HEB220 steel columns, sealed in the ground under the jet-grouting slab.

Figure 3. a) Excavation and retaining wall solution. b) Foundation solution.

In order to guarantee global stabilization and avoid the failure of the jet-grouting slab, which had just been designed for diaphragm wall horizontal stabilization, a pumping program composed of nine deep well-points, each 22m in length, was executed before the excavation (Figure 3a). The grooved length of the deep well points was positioned under the jet-grouting slab, between 12 and 18m deep. For the

246

D. Nobre et al. / The Behaviour Under Excavation of the Luanda’s Sandy Formation

measurement of the drawdown six 18m long piezometers were installed under the jetgrouting slab. At the end of the excavation a total pumping discharge of 600m 3/h was measured. For the foundation solution (Figure 3b), due to the site’s geotechnical condition, namely as a result of the groundwater level and as a consequence of the loads transmitted by the superstructure columns (8.5MN service loads), a combined pile raft foundation with a reinforced concrete slab was designed, of variable thicknesses varying between 1.5 and 2.0m and founded on 68 reinforced concrete bored piles of 1000mm diameter and 13m length. With the aim of evaluating the safety conditions during the excavation and in order to validate the design considerations, a monitoring plan was implemented, composed of twenty five topographic marks, positioned on the diaphragm wall and nearby buildings, and four 17m long inclinometers, installed inside the diaphragm wall. From the inclinometer measurements it was possible to observe that the measured displacements were quite different from the estimated ones (Figure 4). PROFILE 1

PROFILE 2

DISPLACEMENT (m) -0.005

0.005

0.015

DISPLACEMENT (m) 0.025

0.035

-0.005

0

0

1

1 GROUND

2 3

2

0.015

0.025

0.035

GROUND ANCHOR

3

TOP-DOWN SLAB

4

4

5

5

6

6

TOP-DOWN SLAB

7

7 DEEP (m)

DEEP(m)

0.005

8 9

8 9

10

10

11

11

12

JET-GROUTING SLAB

12 JET-GROUTING SLAB

13

13

14

14

15

15

16

16

17

17 09-10-2005 14:58

14-10-2005 9:44

27-10-2005 16:51

09-10-2005 14:58

14-10-2005 9:44

27-10-2005 16:51

08-11-2005 10:03

23-11-2005 9:46

ESTIMATED

08-11-2005 10:03

23-11-2005 9:46

ESTIMATED

Figure 4. Inclinometer measurements in two representative profiles.

In profile 1, located in the retaining wall front with top-down slabs, the measured values were lower than the estimated design values. This behaviour was associated to a higher stiffness of the top-down slabs, probably induced by the existing corner struts, as well as to the overvaluation of the permanent and live loads transferred by the nearby building. After finishing the excavation, a displacement increment of just 5mm was measured. In profile 2, located at the single anchored retaining wall front (without top-down slabs), the ground anchor head displacement was similar to the estimated value but, on the span, a maximum displacement of around 30mm was measured, approximately 50% higher than the estimated value. This response might have resulted of having taken a considerable higher stiffness for the jet-grouting slab in the design (instead of 10GN/m2 it should have been 2GN/m2). It is possible to predict that if the jet-grouting slab displacement had been approximately null, the displacements would have been identical.

3. GES Tower case study The GES Tower is property of “ESCOM – Espírito Santo Imobiliária, SARL” and is located in the upper area of downtown Luanda, near the Kinaxixe neighbourhood and between “Rua Marechal Tito” and “Rua Conselheiro Ornelas”. The tower, with twentyfive storeys, six underground levels and a partially embedded ground floor, has a 2500m2 quadrangular area of 50m wide.

D. Nobre et al. / The Behaviour Under Excavation of the Luanda’s Sandy Formation

247

In order to achieve the maximum depth excavation of approximately 25m, a retaining wall solution of a 0.50m thick reinforced concrete diaphragm wall with 25 to 28m length was designed (Figures 5a and 5b). The horizontal equilibrium of the diaphragm wall was guaranteed by the execution of five stabilization levels with 328 temporary ground anchors of 1.5 to 3.0m spacing and with four 0,6” stands, 6 to 18m free length, 6 to 9m bound length and a lock off load of 600kN. At the corners, the stabilization was assured by five levels of 2 and 5m long HEB200 struts. As a consequence of the ground surface heterogeneity and, mainly, of the limitation of local equipments for diaphragm walls, which could only drill up to a maximum depth of 30m, the working platform had to be lowered. For that purpose a reinforced shotcreted wall of 0.25m thick and 3.5m height was carried out through a “Berlinese Type” construction process. The horizontal load stabilization of this wall was assured by 39 temporary ground anchors of 3 to 4m spacing and with four 0,6” stands, 6m free length, 6m bound length and a lock off load of 300kN.

Figure 5. a) Excavation, retaining wall and foundation solution. b) View of diaphragm wall.

To evaluate the safety conditions during the excavation and in order to validate the design considerations, a monitoring plan composed of topographic marks positioned on the diaphragm wall was implemented. During excavation a maximum displacement of almost 30mm was measured at the top of the diaphragm wall. Considering the depth of excavation, this value was considered perfectly admissible, despite being a little higher than the estimated one. Due to the site’s geotechnical condition and as a consequence of the base tension transmitted by the superstructure columns (20MN for service loads), a reinforced concrete raft foundation was designed with variable thicknesses ranging between 2.0 and 2.5m at the most loaded area and 0,6 to 1,4m at the remaining area (Figure 5a). For the design, an allowable bearing capacity of 600kN/m2 was considered at the foundation level. Before finishing the excavation works and in order to confirm the geotechnical parameters assumed in the foundation design, namely to verify soil integrity, a complementary ground investigation composed of four additional boreholes, also with the execution of standard penetration tests, was carried out (Figure 6a). The SPT tests revealed, at the first 6.00m depth under the foundation level, an average of 25 blows instead of the 60 blows that were initially identified. This occurrence, which was not an SPT blow correction issue, was associated to an unusual decompression of the sandy formations induced by the temporary vertical stress release during the excavation process. In order to assess the real foundation conditions and to estimate the decompression consequences, a plate load test was carried out using a circular plate of 1.50m diameter up to a load of 900kN/m2 and 600kN/m2 at the first and second cycle, respectively (Figures 6b and 6c). Maximum settlements of 45mm were measured at the first cycle and 20mm at the second, which were correlated to a ground deformation modulus of

248

D. Nobre et al. / The Behaviour Under Excavation of the Luanda’s Sandy Formation

around 15000kN/m2 in charge and 25000kN/m2 in recharge. As a consequence of ground properties variation and in order to avoid settlements not foreseen in the design, a solution was developed that could restore the initial geotechnical properties. For that purpose, a ground treatment solution based on the injection of cement grout under the raft with a borehole mesh of 1.5m wide and with 6m length was designed. 0

20

40

2.50 5.00 7.50

60

TEST PROCEDURE PLATE PRESSURE (kN/m2)

NUMBER OF BLOWS (SPT) 0.00

90

900

60

750

60

600

60

60

450

60

60

300

60

60

150

60 60

0 TIME (min)

10.00 12.50

PLATE PRESSURE(kN/m2) 0

30.00 32.50 35.00 37.50 40.00

450

600

750

900

10 15

22.50 27.50

300

5

20.00 25.00

150

0

17.50

20 SETTLEMENT (mm)

DEEP (m)

15.00

25 30 35 40 45 50 55 60 65 70

Figure 6. a) Complementary investigation tests results. b) Reaction structure used in plate load test. c) Load test procedure and test results.

To evaluate foundation behaviour during superstructure construction, as well as during the injection works, a monitoring plan supported on the measurement of topographic marks positioned on the structure columns was developed. During the construction phase, which took almost one year, a maximum settlement of 5mm was measured. These very low displacements corroborate the efficiency of the ground treatment.

4. Conclusions The described case studies are two of the first deep geotechnical works carried out in the Luanda city ground. With these two examples it has been possible to illustrate most of the structural and construction solutions that are currently being applied. As an excavation and retaining wall solution, the diaphragm wall methodology is being widely used and, in most cases, this has proved to be an excellent option. In the presence of high groundwater, the retaining wall stabilization solutions with jetgrouting slabs combined with ground anchors at the top and/or top-down slabs, have shown to be efficient, specially in the displacement control. The drawdown solutions through deep well points have also revealed to be efficient, despite their excessive water pumping volumes. At the foundations, and depending on the depth of the Luanda Formation, shallow foundations such as rafts, deep foundations such as piles, and mixed foundation solutions such as combined pile raft foundations, have demonstrated to be viable solutions for the tallest buildings. However, the sand decompression identified not only in the GES Tower but also in other excavations of over 15m deep, is a phenomenon that should not be overlooked in future design.

References [1] Silva, H & Teixeira, G. (1973). “Geotechnical Map of the Luanda Region – 1st Approximation”, Laboratório de Engenharia de Angola, nº 183.

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 249 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-249

Theoretical Evaluation of the Influence of Cohesion on Lateral Support Design Jacobus BREYL, Gavin WARDLE and Peter DAY Jones & Wagener Consulting Engineers, Rivonia, South Africa Abstract. The amount of support required for vertical or near-vertical excavations in partially cemented or moist fine grained soils is significantly influenced by the “apparent” cohesion of the soil. This is a parameter that is often difficult to determine and its value is influenced by a number of factors. A theoretical evaluation was carried out to quantify the influence of cohesion on the overall factor of safety for various face heights. The dangers of relying too heavily on cohesion are highlighted and backed up with simple probability analyses which show that slopes designed to a given factor of safety have a higher probability of failure when the stability of the slope is derived largely from cohesion. Practical recommendations are provided to assist in lateral support designs. Keywords. Cohesion, lateral support, factor of safety, probability of failure.

Introduction A typical South African soil profile will often include partially saturated, partially cemented, residual or moist fine-grained soils. All of these soil types exhibit some degree of cohesion. This is readily apparent from the ability of these soils to stand vertically in excavation sidewalls, often to considerable heights. There are, however, numerous instances where such excavations have failed without warning, often with tragic consequences. The value of cohesion in these soils depends on a number of factors such as degree of saturation, jointing, length of exposure, etc causing it to vary over time or at different locations along the excavation face. There is also the added problem that the cohesion measured in the laboratory may not be a true reflection of the strength in the field due to changes in moisture content, sample disturbance or the rate of shearing used in the test. The design of lateral support is influenced significantly by the assumed cohesion in the soil and it is obvious that higher cohesion values will lead to higher factors of safety. However, in view of its variability, the use of cohesion in the analysis could produce a non-conservative design. A case in point is the excavation of Bank City in central Johannesburg where the presence of slickensided joint planes had a significant effect on the shear strength of the retained cohesive soil [1]. The main objective of this paper is to point out the danger of relying too heavily on cohesion by means of simple analyses in which the effect of cohesion is determined.

250

J. Breyl et al. / Theoretical Evaluation of the Inﬂuence of Cohesion on Lateral Support Design

1. Scenarios Analysed and Methods of Analysis 1.1. Anchor Force Required to Obtain a Specified FOS for Various Values of Cohesion Figure 1 shows the simple wedge failure method that was used for determining the anchor force required [2] and values of the assumed parameters. ϭϬŬWĂƐƵƌĐŚĂƌŐĞ

ϭϬŽ d ;ĂŶĐŚŽƌĨŽƌĐĞͿ ,

;ĨĂĐĞŚĞŝŐŚƚͿ

&ĂŝůƵƌĞƐƵƌĨĂĐĞ β ;ǀĂƌŝĂďůĞͿ

^ŽŝůWƌŽƉĞƌƚŝĞƐ͗ ϮϬŬEͬŵϯ ĞŶƐŝƚǇ;γ Ϳ &ƌŝĐƚŝŽŶĂŶŐůĞ;φ ͛Ϳ ϯϬŽ EŽǁĂƚĞƌƚĂďůĞ

Figure 1. Wedge failure analysis with assumed parameters

Factors of safety of 1,0, 1,25, 1,5 and 2,0 were chosen and the analysis carried out for cohesion values ranging from 0 kPa to 20 kPa in increments of 2 kPa. For each value of cohesion, the anchor force (T) required to achieve the desired factor of safety was determined. The process was repeated for face heights (H) of 3 m, 5 m, 7,5 m and 10 m. Figures 2 and 3 show the anchor force required to achieve the desired factor of safety for wall heights of 3 m and 7,5 m. ϭϬϬ

,сϯŵ

d;ŬEͬŵͿ

ϴϬ ϲϬ ϰϬ ϮϬ Ϭ Ϭ

ϱ

ϭϬ ĐΖ;ŬWĂͿ

Figure 2. Anchor forces for 3 m high face

ϭϱ

ϮϬ

J. Breyl et al. / Theoretical Evaluation of the Inﬂuence of Cohesion on Lateral Support Design

ϱϬϬ

251

,сϳ͕ϱŵ

d;ŬEͬŵͿ

ϰϬϬ ϯϬϬ ϮϬϬ ϭϬϬ Ϭ Ϭ

ϱ

ϭϬ

ϭϱ

ϮϬ

ĐΖ;ŬWĂͿ Figure 3. Anchor forces for 7,5 m high face

1.2. Relationship between Factor of Safety and Probability of Failure A simple probabilistic analysis was carried out on the 7,5 m high wall. The bivariate point estimate method [3] was used to determine the probability of failure of a wall designed for a factor of safety of 1,25. In this analysis, the coefficient of variation of φ’ was assumed to be 7% and that of c’ 40% based on representative values given by Harr [3]. Two correlation coefficients were considered, 0 and -0,3. By assuming FOS to be normally distributed, the probability of a factor of safety below 1,0 was determined. The resulting probability of failure for various cohesion values for a 7,5 m high wall designed for a factor of safety of 1,25 is shown in Figure 4. ϭϴ

WƌŽďĂďŝůŝƚǇŽĨ&ĂŝůƵƌĞ;йͿ

ϭϲ

ŽƌƌĞůĂƚŝŽŶĐŽĞĨĨŝĐŝĞŶƚсϬ

ϭϰ

ŽƌƌĞůĂƚŝŽŶĐŽĞĨĨŝĐŝĞŶƚсͲ Ϭ͘ϯ

ϭϮ ϭϬ ϴ ϲ ϰ

&K^сϭ͕Ϯϱ

Ϯ Ϭ Ϭ

ϱ

ϭϬ

ϭϱ

ŽŚĞƐŝŽŶ;ŬWĂͿ

Figure 4. Probability of failure for 7,5 m high face designed for a FOS of 1,25

ϮϬ

252

J. Breyl et al. / Theoretical Evaluation of the Inﬂuence of Cohesion on Lateral Support Design

2. Discussion of Results 2.1. Reduction in Required Force with an Increase in Cohesion Table 1 summarises the percentage reduction in the anchor force achieved by introducing a cohesion of 20 kPa into the calculation. It can be seen that the effect of cohesion is much more dramatic in low walls than in high walls. For a 3 m high wall, the inclusion of a 20 kPa cohesion reduces the anchor force required to achieve a factor of safety of 1,5 by 97% of that required for zero cohesion. In the case of a 10 m face the reduction is only 36%. This points to the fact that low faces are more sensitive to changes in cohesion than high faces. Table 1. Reduction in anchor force achieved by including a cohesion of 20 kPa in calculation. Face height FOS = 1 3.0 m 100%* 5.0 m 100%* 7.5 m 82% 10.0 m 63% * face is self supporting – no anchor required

FOS = 1,25 100%* 84% 59% 46%

FOS = 1,5 97% 65% 46% 36%

FOS = 2 67% 45% 32% 25%

2.2. Increase in Required Anchor Force if Assumed Cohesion Cannot be Obtained When using a high value of cohesion in a design, the designer needs to be sure that the value used is a reasonable estimate of that likely to be present in the retained soil. This can be seen in Figure 3 where, if the designer assumes a cohesion of 15 kPa, an anchor force of 227 kN/m is required to achieve a FOS of 1,5 for a 7,5 m high face. However, if there is a section of the face where the cohesion is not 15 kPa but only 5 kPa due to (say) adverse jointing, the FOS drops to below 1,25. In order to achieve a FOS of 1,5 over this section of the face, the force would have to be increased by 37% to 313 kN. For the 3 m high face, the effect is even greater. If a cohesion value of 15 kPa is used, then a force of only 17 kN is required to obtain a FOS of 1,5. If, over a section of the face, the actual cohesion is only 5 kPa, the wall would fail (FOS<1,0). To maintain the desired FOS, a force of 39 kN is required – an increase of 130%. This illustrates the point made earlier that lower faces are more sensitive to a change in cohesion than higher faces. 2.3. Relationship between the Factor of Safety and Probability of Failure Figure 4 shows that the probability of failure of a 7,5 m high wall increases significantly for a cohesion greater than 5 kPa despite the wall being designed for a factor of safety of 1,25 for all values of cohesion. An indication of acceptable probabilities of failure can be obtained from Kirsten [4] who suggests a probability of failure for medium term, semi-permanent slopes where public access is discouraged of between 1,5% and 5%. This is the category into which most temporary lateral support designs fall. If a limit of 5% is applied to Figure 4 it can be seen that the probability of failure becomes unacceptable for cohesion values higher than 12 kPa despite the use of a FOS of 1,25. For a cohesion of 20 kPa, a FOS of approximately 1,4 would be needed to achieve the desired probability of failure. This demonstrates the limitations of the use of a global FOS in slope design.

J. Breyl et al. / Theoretical Evaluation of the Inﬂuence of Cohesion on Lateral Support Design

253

3. Code Requirements for Cohesion Values Used in Design 3.1. South African Lateral Support Codes The dangers associated with the use of cohesion in the design of lateral support have been recognised for as long as deep basements have been constructed in South Africa. One of the earliest guidelines was issued by the Johannesburg City Engineer’s Department in 1962 or 1963. This undated, typewritten document with hand drawn sketches stated “In general, the value of c used in the design should be taken as zero, unless it can be shown that the material is uniform and not fissured. In any event, the value of c should not exceed one-quarter of the cohesive strength as determined by laboratory tests. Attention is drawn to the effect of service trenches alongside excavations which will destroy the cohesive strength of the material”. The current code [2] echoes this theme and states that “the cohesion in terms of effective stress (c’) should be taken as zero unless it can be established that the material is intact and not fissured. A non-zero value of c’ should be supported by an appropriate testing programme taking due account of the jointing, fissuring or slickensiding of the soil mass or rock mass”. The code goes further by requiring that the minimum lateral earth pressure assumed in the design should not be lower than that resulting for a friction angle of 40 o and a cohesion of zero. 3.2. Limit States Design Codes In response to the deficiencies of the global factor of safety design approach highlighted above, a number of modern codes including the Eurocodes and the recently published South African “Loading Code” [5] use the Limit States design approach. Many versions of the limit states design approach are based on the application of both partial load and resistance factors to carefully chosen characteristic values to obtain the design values used in the calculation. The design is verified by checking that the design resistance exceeds the design effect of actions (loads). The required “safety margin” is achieved by appropriate choice of characteristic values and the application of partial load and material factors. The load and resistance factors are (in theory) chosen to ensure a uniform level of reliability for all types of structures and for the full load spectrum. This is in contrast to the global factor of safety approach which determines the ratio of the resistance to the effect of actions, both determined using unfactored input parameters. In SANS10160-5:2010 (Basis of geotechnical design and actions) [5], the characteristic value of a geotechnical parameter (e.g. soil strength) is taken as a “cautious estimate of the value affecting the occurrence of the limit state”. If statistical methods are used, it is selected such that the likelihood of a worse value occurring is less than 5%. The value chosen should be representative of the soil mass (including joints) and not just of the samples tested. The partial material factor applied to undrained shear strength (cu) is 1,4 and that applied to the friction angle (tanφ’) is 1,25 reflecting the difference in the coefficient of variation of these two parameters. It is interesting to note that the partial material factor applied to cohesion in terms of effective stress (c’) is also 1,25. This is because, under “true” effective stress conditions, the value of c’ is close to zero even for clayey soils. What is being measured in partially saturated or cemented soils is an “apparent” cohesion that should be treated more like an undrained cohesion than effective stress cohesion.

254

J. Breyl et al. / Theoretical Evaluation of the Inﬂuence of Cohesion on Lateral Support Design

4. Practical Recommendations In the light of the above, the following practical recommendations are made when considering the use of cohesion in a lateral support design: • Ensure that the samples tested are representative of the soil mass and not just of the intact material. Where this is impractical, an assessment of the effect of soil structure (jointing, etc) on the shear strength should be made. • Ensure that the tests are carried out at the appropriate strain rate and at a moisture content that represents the worst conditions likely to occur in the field. In general, the chosen strain rate should be sufficient to ensure full dissipation of excess pore water pressures and, if there is any possibility of the material becoming saturated in the field, the samples should be tested in the saturated condition. • Consider the effect that drying out of the exposed soil and dissipation of negative pore water pressures could have on “apparent” cohesion derived from partial saturation of the soil. • Always check the design for the φ’ = 40o, c’ = 0 kPa condition as recommended by the lateral support code. • Take adequate cognisance of the effect of service trenches including their effect of cohesion and the possibility that these could encourage the formation of tension cracks that could fill with water. • Carry out sensitivity analyses to assess the effect of changes to input parameters, cohesion in particular. • When using limit states design methods, ensure that the characteristic values chosen are a conservative estimate of the strength of the soil mass through which the failure surface will pass. Unless the tests are carried out on saturated samples under full effective stress conditions (i.e. full dissipation of excess pore water pressures on the failure plane), use the higher partial material factor applicable to undrained cohesion in the analysis. • Inspect the excavation face during construction for any signs of adverse jointing, increased moisture content or other factors that could affect the strength of the material. • Monitor the performance of the excavation throughout the construction process and investigate any unpredicted or adverse behaviour. Acknowledgement: The authors acknowledge with thanks the assistance of Mzwakhe Dlamini with the calculations.

References [1]

[2] [3] [4] [5]

Day, P.W., Wardle, G.R. and Krone, B, Design, Construction and Performance of Deep Basement Excavations in South Africa and Zimbabwe, Proc. 11 ARC SMFE, Cairo, 11-15 December 1995, Vol 2, p592-600. SAICE, Lateral support in surface excavations – Code of practice, Geotechnical Division, S.A. Institution of Civil Engineers, 1989. M.E. Harr, Reliability based design in civil engineering, McGraw-Hill, New York, 1987. Kirsten H.A.D., Significance of the probability of failure in slope engineering, The Civil Engineer in South Africa, January 1983. SANS10160:2010, Basis of structural design and actions for buildings and industrial structures, Parts 1 – 8, SABS, Pretoria, 2010.

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 255 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-255

Internally Instrumented Soil Nail Pull Out Tests a

Jacobus BREYLa,1 and Gavin WARDLE a Jones and Wagener Consulting Civil Engineers, Rivonia, South Africa Abstract. In order to evaluate the ultimate bond stress used in modelling soil nail behaviour for lateral support, soil nail pull out tests are conducted on the site. In the analysis of the test results it is assumed that the free length given to the test nail is completely free and that the load distribution along the grout body is uniform at failure. Three instrumented soil nail pull out tests were conducted to evaluate the validity of these assumptions. The current in-house practice for obtaining a free length proved to be effective. More testing is required to evaluate the typical load distribution at failure. Suggested next steps for further research in this regards are given. Keywords. Soil nail pull out testing, bond stress

Introduction The pull out resistance of a soil nail is governed by the bond stress on the grout-soil interface, the bond between the bar and the grout and the strength of the bar itself. The last two parameters can be verified off site, but the first parameter needs to be confirmed on site. Current practice in soil nail design in South Africa is for the designer to assume a bond stress for the grout-soil interface based on soil parameters [1]. The bond stress is then verified on site by means of soil nail pull out tests. The designer usually specifies a required number of nails to be tested. There is, however, no accepted standard in South Africa for soil nail pull out testing. This leads to a variety of testing methods being used with different outcomes. For example, during a recent design audit of a lateral support wall, the authors reviewed a range of soil nail pull out tests conducted in different soils. The test nails were all approximately 8 m long. Upon analysis it was noticed that the majority of the test nails failed at the same load – that of the bar strength. The only information that could be gleaned from this was that, the “bond” value exceeded the capacity of the bar. No further information was known about the actual bond value for the different soils. Current practice at Jones & Wagener (J&W) for soil nail testing is to keep the length of the test nails short to give the grout-soil bond the opportunity to fail before the bar reaches its ultimate capacity. The test nail is also provided with a free length (de-bonded length) over the first meter behind the face. If the grout column of the test nail extends up to the face (as the case for a working nail) then the grout column will push against the back of the gunite face during the test and increase the measured “bond” of the nail. The free length is obtained by lining the drilled hole with a uPVC 1

Corresponding Author: Geotechnical Engineer, Jones & Wagener Consulting Engineers, P.O. Box 1434, Rivonia, 2128, South Africa; Email: [email protected]

256

J. Breyl and G. Wardle / Internally Instrumented Soil Nail Pull Out Tests

pipe greased on the outside. Several load cycles are then applied during which the maximum load is incrementally increased. The displacement of the nail head is measured throughout. The applied load is plotted against the measured displacement and the failure point identified by a change in slope of this line. This usually occurs when the displacement is approximately 10% of the grout body diameter. The failure load (kN) is divided by the bond length (m) to give a pull out resistance in kN/m and divided by the circumference of the grout body to give the bond stress in kN/m 2 or kPa Several assumptions are made in this procedure. The free length is assumed to be completely de-bonded from the soil and the load distribution at failure is assumed to be uniform along the bonded length of the nail. A testing programme was set up to investigate the validity of these assumptions. The objectives of the tests were: • To examine whether the current method of de-bonding is effective in obtaining a free length; • To examine the load distribution along the bond length as the load is increased, • To evaluate the load distribution along the bond length at failure.

1. Preparation and Installation of Instrumented Soil Nails The soil nails were instrumented by placing 3 load cells on each nail in the form of strain gauged Wheatstone bridges. The nails were 4 m long and the bridges were installed at 1 m intervals. Each bridge consisted of two strain gauges welded longitudinally to opposite sides of the bar. At each position, a 1-2 mm recess was machined to provide a level surface for each strain gauge. The longitudinal orientation meant that the gauges would elongate when the bar was tensioned. Each bridge was completed with two 350 Ω precision resistors located close to the logger. The resistance of the cables connecting the precision resistors to the rest of the circuit brought a slight imbalance to each bridge, but there was enough capacity left to measure the applied loads. After testing the circuit, the bridges were sealed with silicon and a tar patch to protect them against the water from the grout. The purpose of the load cells is to measure the load at different points along the steel bar when the nail is pulled. The output from the Wheatstone bridges are in microvolt and a relationship between microvolt and applied load (kN) had to be found for each bridge. This was done by determining the relationship between applied load and microvolts obtained for each bridge by applying known tension loads to the full length of soil nail bar and measuring the outputs in the laboratory. Three instrumented test nails were installed at the bottom of a 16m deep basement excavation in Parktown, Johannesburg. A 102 mm diameter hole was drilled, flushed with air and filled with grout. The instrumented nails were inserted with care into the homing grout to ensure the wiring did not get damaged. Nails 1 and 2 were installed in the Parktown shales whilst Nail 3 was installed in a diabase dyke intrusion. The nails were installed after the guniting and lateral support installation in that area had been completed.

J. Breyl and G. Wardle / Internally Instrumented Soil Nail Pull Out Tests

257

2. Testing of the Instrumented Nails Once the grout had cured sufficiently, the pull out tests were conducted. The equipment used included a 20 ton (200 kN) hollow cylinder jack, a hand pump, a pressure dial gauge, displacement dial gauges, an electronic pressure transducer and a data logger to log the change in microvolt as well as the change in oil pressure. A diagram of the test nail set up can be seen in Figure 1. After setting up the equipment a small load was applied to check if all the instruments were taking readings. Three load cycles were then applied in which the load was incrementally increased to 50%, 75% and 100% of the expected failure load. If no failure had been reached at this point, a fourth cycle was applied during which the load was increased up to failure. Nails 1 and 2 could not be failed with the 20 ton jack, but Nail 3 (which was in the diabase) did reach failure. A jack with a larger capacity was then used to pull Nails 1 and 2. Nail 2 was pulled to the limit of the bar’s capacity (230 kN) but still failure could not be reached. It appears as if the strain gauges on Nail 1 got damaged during the application of the larger load and no useful readings could be obtained.

Figure 1. Testing set up of the instrumented soil nail

3. Data Analysis and Results The applied load was divided by the assumed bond length of 3 m to obtain a pull out resistance in kN/m. The load for each bridge was obtained by multiplying the microvolt readings for each Wheatstone bridge with the calibration factor. Figures 2 and 3 show the pull out resistance versus displacement for Nails 1 and 3. (Note: the displacement plotted is the actual measured nail head deflection and does not take into account elongation of the bar.) It can clearly be seen that Nail 1 did not fail whilst Nail 3 failed during the second load cycle.

258

J. Breyl and G. Wardle / Internally Instrumented Soil Nail Pull Out Tests ϳϬ

WƵůůŝŶŐ&ŽƌĐĞ;ŬEͬŵͿ

ϲϬ

ǇĐůĞϰ

ϱϬ ϰϬ

ǇĐůĞϯ

ϯϬ

`` ````

ǇĐůĞϮ

ϮϬ

ǇĐůĞϭ

ϭϬ

ϵ

ϴ

ϳ

ϲ

ϱ

ϰ

ϯ

Ϯ

ϭ

Ϭ

Ϭ

ŝƐƉůĂĐĞŵĞŶƚ;ŵŵͿ

Figure 2. Pulling force and displacement of Nail 1

WƵůůŝŶŐ&ŽƌĐĞ;ŬEͬŵͿ

ϲϬ ϱϬ

ǇĐůĞ ϰ

ϰϬ

ǇĐůĞϯ

&ĂŝůƵƌĞ

ϯϬ

``

ǇĐůĞϮ ϮϬ ϭϬ

ǇĐůĞϭ ϰϬ

ϯϱ

ϯϬ

Ϯϱ

ϮϬ

ϭϱ

ϭϬ

ϱ

Ϭ

Ϭ

ŝƐƉůĂĐĞŵĞŶƚ;ŵŵͿ

Figure 3. Pulling force and displacement of Nail 3

The effectiveness of the de-bonding method employed by J&W can be seen if the applied load is compared with the load recorded by Bridge 3 located at the start of the free length. Figure 4 compares these two values for Nail 1. The average bond stress was calculated for each section of the bonded length as the load difference between the two bridges on either side of the section divided by the perimeter area of the grout body in that section. The premise is that the difference in load must have been shed to the soil via the grout in that section. For example, the load difference between Bridge 3 and 2 must have been shed to the soil in Section 3 (see Figure 1). The bond stress at the end of the nail is zero and thus the load shed in Section 1 is equal to the load at Bridge 1.

J. Breyl and G. Wardle / Internally Instrumented Soil Nail Pull Out Tests

259

Figures 5 and 6 below show the development of the average bond stress per section for each load cycle applied to Nails 1 and 3 compared to the bond stress calculated for the 3 m soil nail shown in italics. ϮϬϬ

>ŽĂĚ;ŬEͿ

ϭϲϬ

ϭϮϬ

ƉƉůŝĞĚůŽĂĚ;ŬEͿ

ϴϬ

ϰϬ

ƌŝĚŐĞϯ >ŽĂĚ ;ŬEͿ

Ϭ ϭϰ͗Ϯϰ͗ϬϬ

ϭϰ͗ϯϴ͗Ϯϰ

ϭϰ͗ϱϮ͗ϰϴ

ϭϱ͗Ϭϳ͗ϭϮ

ϭϱ͗Ϯϭ͗ϯϲ

dŝŵĞ

Figure 4. Free length evaluation of Nail 1 ϮϱϬ

ǇĐůĞϰ;ϭϵϭ ŬWĂͿ

ŽŶĚ ƐƚƌĞƐƐ;ŬWĂͿ

ϮϬϬ

ǇĐůĞϯ;ϭϬϵ ŬWĂͿ

ϭϱϬ

ǇĐůĞϮ;ϴϮŬWĂͿ ϭϬϬ

ǇĐůĞϭ;ϱϱŬWĂͿ ϱϬ

Ϭ Ϭ

;^ĞĐƚŝŽŶϯͿ

ϭ

;^ĞĐƚŝŽŶϮͿ

Ϯ

ŽŶĚůĞŶŐƚŚEĂŝůϭ;ŵͿ

Figure 5. Average stress per bond length section for Nail 1

;^ĞĐƚŝŽŶϭͿ

ϯ

260

J. Breyl and G. Wardle / Internally Instrumented Soil Nail Pull Out Tests ϮϬϬ

ǇĐůĞϰ;ϭϱϲ ŬWĂͿ

ϭϴϬ ϭϲϬ

ǇĐůĞϯ;ϭϭϭ ŬWĂͿ

ŽŶĚ ƐƚƌĞƐƐ;ŬWĂͿ

ϭϰϬ ϭϮϬ

ǇĐůĞϮ;ϴϮŬWĂͿ

ϭϬϬ ϴϬ ϲϬ ϰϬ

ǇĐůĞϭ;ϱϭŬWĂͿ

ϮϬ Ϭ Ϭ

;^ĞĐƚŝŽŶϯͿ

ϭ

;^ĞĐƚŝŽŶϮͿ

Ϯ

;^ĞĐƚŝŽŶϭͿ

ϯ

ŽŶĚůĞŶŐƚŚEĂŝůϯ;ŵͿ

Figure 6. Average stress per bond length section for Nail 3

4. Discussion of Results Figure 4 shows a gap between the applied load and the load measured by Bridge 3 for loads lower than 150 kN. For loads higher than 150 kN the gap gets significantly smaller. This was also the case with Nails 2 and 3. A possible explanation for this is that the applied load is based on oil pressure readings taken at the pump. At low loads, losses due to jack friction and de-bonding of the free length play a greater role than at high loads. This effect can also be seen in the difference between the readings on either side of cycle 4 where the applied load is lower than the load at Bridge 3 due to hysteresis of the system due to friction. The error that could result due to the load applied to the soil nail below 150 kN in the case of J&W’s setup is that the bond stress calculated from the test result can easily be overestimated, especially for nails that fail at a low load. A solution would be to always measure the load at the beginning of the free length, but this would be expensive. Another option would be to factor the applied load below 150 kN, when calculating the bond stress from the applied jack force. Also the calculation of the average bond stress per section is based on the assumption that the drilled boreholes for the soil nails are uniform in diameter and that no over-break has occurred. Figures 5 and 6 shows a general increase in the middle section of the nail, with the first metre taking slightly more load than the third metre. When borne in mind that Nail 3 failed during the second cycle and that Nail 1 did not fail, it is evident that the current data in hand is insufficient to make substantial conclusions about the load distribution. If the displacement is logged electronically the mobilisation of shear stress on the grout/soil interface can be investigated in more detail and the development of the load distribution examined more closely.

J. Breyl and G. Wardle / Internally Instrumented Soil Nail Pull Out Tests

261

Nail 3 is the only nail which clearly failed under ultimate load during in the testing as shown in Figure 2. This occurred during the second load cycle. At failure, the bulk of the resisting load was being carried by the middle section of the nail. Further testing is required to establish if there is a typical load distribution pattern at failure. It is interesting to note that the applied load (and calculated bond stress) kept on increasing even after the nail had failed with a displacement of more than 35 mm measured at the nail head. This increase in load may have been caused by resistance of the cone of soil between the grout body and the back of the gunite face against which the jack presses during the test. This effect cannot be relied upon for actual wall behaviour, because the gunite face moves with the nail head once ultimate load for the soil nail has been reached.

5. Conclusions and Suggested Next Steps for Further Research The current method adopted by J&W of de-bonding the first metre of the nail appears to be effective. However at low applied loads (less than 150 kN), jack and de-bonding friction could result in an over estimation of the bond stress calculated. The calculated bond stress based on the bond length and the applied load is an average as the middle portion of the test nail exhibited a higher bond stress than the first and last sections. More tests are required to establish the amount that friction in the system affects the results and the typical shape of the bond stresses along the length of the test nail, whether it is consistent or varying depending on the length of the bonded section of test nail etc. Suggested next steps for further research are: • • • • •

Measuring the discrepancy between the applied loads compared to the load at the start of the bonded section in order to determine a relationship with which the applied load can be factored during routine soil nail tests; Looking at the displacement versus bond stress per section of the grout body. Electronic displacement logging will assist in this; Tests with more Wheatstone bridges per bond length to see if the shape of the load distribution curve can be established more accurately; Evaluating the influence of the bar elongation on the total displacement of the nail head, and Exposing soil nails after the tests to examine the grout body to look for cracking and possible over break portions that could have influenced the readings.

References [1]

Heymann, G., Rohde, A. W., Schwartz, K. and Friedlaender, E. Soil nail pull out resistance in residual soils. Proc. Int. Symposium on Earth Reinforcement Practice, Fukuoka, Japan, November 1992, pp.487-492.

The authors acknowledge with thanks the assistance of ESOR Africa, Mzwakhe Dlamini, Khethile Mbatha and Professor SW Jacobsz in the collection of data for this paper.

262 Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-262

Reinforced Soil Retaining Wall Systems Reach New Heights in the Middle East Peter G. WILLS1, Chaido DOULALA-RIGBY2 Tensar International Ltd, Blackburn, UK Abstract. Reinforced Soil Retaining Wall (RSRW) Systems were first introduced into major civil engineering projects in the United Arab Emirates (UAE) around 2000. The project boosting this trend was the Dibba Idhn Tawaian Road, Dubai. With a total wall face area exceeding 40,000 m2, it established RSRW Systems in the UAE proving them well suited to construction through difficult mountainous terrain and under extreme climatic conditions. Since the completion of this project in 2003 more than 40 major projects have been successfully completed in the UAE region for major clients, with the total area of wall face completed to date exceeding 500,000 m2. The most impressive one, which is nearing completion now, is the The Dubai Fujairah Freeway Project for The Ministry of Public Works, which is the focus of this paper. The paper will discuss and illustrate the ERW system components as well as the design and construction techniques adopted by both designer and contractor for The Dubai Fujairah Freeway Project and highlight why SRW’s are well suited to the logistical problems faced under such difficult conditions.

Keywords. Reinforced Soil Retaining Wall Systems (RSRW), geogrid, mechanical polymeric connector, United Arab Emirates

Introduction Since its introduction over ten years ago into the UAE, RSRW systems have become accepted for use in major highway retaining walls. Their economy and durability as well as ease and speed of construction have made this system an attractive choice for many project infrastructure developers in both the public and private sectors. With the ever increasing demands in heights of retaining structures in the region, RSRW systems have become a popular choice for Contractors for retained heights in excess of 10m without having to consider a much more expensive and complex solution such as bored pile retaining structures or a viaduct. As an additional benefit, Contractors have often favored this system for its versatility and the limited space that is needed for its construction. 1

Peter G. Wills: International Business Manager, Tensar International Ltd, Cunningham Court, Blackburn, UK, BB1 2QX; Email: [email protected] 2

Chaido Doulala-Rigby: Chief Civil Engineer, Tensar International Ltd, Cunningham Court, Blackburn, UK, BB1 2QX; Email: [email protected]

P.G. Wills and C. Doulala-Rigby / RSRW Systems Reach New Heights in the Middle East

263

The pioneer project of this kind in the UAE was Dibba Idhn Tawaian Road, UAE. The project involved the construction of a 22 m wide highway that abutted the existing mountainous terrain on one side and was supported on RSRWs on the other side. The construction was a cut and fill operation in order to achieve the required alignment of the planned route. The route cut through remote, rugged and elevated terrain with exposure to wind and high temperatures. Numerous gullies and steep sided valleys formed by heavy rainfall events cut across the landscape. Construction of 22 no. retaining walls with a total face area of 45000 m2 was required to achieve the necessary grades. The walls, reaching heights up to 19 m, incorporated large culverts to cope with massive storm water discharges from major storm water valleys or wadis. The project was successfully completed in 2003 (figure 1).

Figure 1. Dibba Road: RSRWs of up to 19 m high

1. The Dubai Fujairah Freeway Project The success of Dibba Road led the way for the new technology in the region. Fujairah Freeway, which is located in the same geographical region as Dibba followed. Fujairah is the fifth largest Emirate of the UAE situated in the Arabian Gulf. Already a busy port, Fujairah is featured to become a strategic trading place for both Abu Dhabi’s oil companies and Dubai’s financial and industrial sectors. Set in the Arabian Gulf, the relatively undeveloped area is also expected to follow in Dubai’s footsteps and emerge as a popular tourist destination. Despite the desert-covered UAE, Fujairah is the only Emirate that is almost totally mountainous. Current indigenous routes in the region are limited to basic tracks. The desolate, rugged terrain makes construction in these parts challenging, a contributing factor to the undeveloped nature of the area. In order to facilitate the development of the region, the Ministry of Public Works decided to construct a 4-lane motorway (10 lanes in total, including 2 hard shoulders), which will link Fujairah with Dubai. When completed, this freeway will be 80 km long and is expected to half the travel time between Dubai and Fujairah. The project is split into 3 packages. Package 1 has already been completed and involved mainly cut slopes and unreinforced, low gradient embankments. Package 2, which is the one presented in this paper, involves the construction of 45km freeway through very rough and rugged terrain. The value of the works for Package 2 is estimated at US$300 million. Package 3 had not yet commenced at the time this paper was written.

264

P.G. Wills and C. Doulala-Rigby / RSRW Systems Reach New Heights in the Middle East

1.1 Local Topography The UAE is situated on the east and west coasts of the northern part of the Arabian Peninsula. Fujairah and Dibba are both situated on the eastern side of the peninsula whereas Abu Dhabi and Ras Al Khaima are situated on the western side. While the western side is dominated by wide low-lying coastal plains, the eastern side forms part of a very narrow coastal plain that comprises a mixture of coastal beach and sabkha deposits and alluvial outwash sediments from the nearby range of the mountains of Oman. Available geological literature (Ref. 5) indicates the Emirate of Fujairah to form part of the Oman mountain range composed of basic igneous rocks, predominantly Gabbro, with associated wadis and outwash deposits. The Arabian Gulf is an area of extensive carbonate sedimentation with the nature and distribution of the sediments being governed by the recent geological history and structural setting of the Gulf, the orientation of the coastline, the prevailing winds and the torrential, once in a year, rainfall events. The affect of extreme climatic conditions are visible on the exposed mountain face where the rock is often exfoliated and highly fractured. 1.2 Route Alignment Proposal The 10-lane road was aligned to navigate through the mountainous landscape. Intense rainfall which occurs typically once a year has formed deep gullies and valleys within the terrain. The new road had to bridge over the existing valleys and cut through existing severe gradients of the mountains. The original conforming design was the construction of major viaducts across the valleys. However, the construction of viaducts would have been incredibly costly and challenging due to the lack of access, water resources and difficulties of facilitating concrete curing in the extreme temperatures and the arid conditions. Following the Dibba Road success in similar topographic settings, the Contractor opted for RSRW Systems. 1.3 Fujairah Freeway RSRW System Design In early 2006, Geosynthetics specialist Tensar, who manufacture geogrids, was commissioned with the design and supply of 29 individual reinforced soil retaining walls for the project. Twenty five of them form single tier walls with maximum heights up to 22 m and four are major tiered structures made of two or three tiers with maximum heights up to 60 m. The total reinforced soil wall area when completed is estimated to be in excess of 100,000 m2. In Package 2, the freeway passes through the rugged Hajar Mountains involving rock-cutting with depths of up to 100 m. The level differences are nearly as high at the locations of the valleys, approaching up to 60 metres. At the base of these valleys a number of wadis have been formed over the years carrying enormous volumes of fast flowing water during storm events. The presence of these localised deep valleys meant that the reinforced soil retaining walls had to bridge the valleys whilst building over existing wadis. In order to maintain the expected storm water flow in the wadis during the infrequent but severe rainstorms, large culverts had to be incorporated at the base of the RSRW systems. All culverts were designed and constructed by the Contractor and Tensar was called upon to design the walls above and around them.

P.G. Wills and C. Doulala-Rigby / RSRW Systems Reach New Heights in the Middle East

265

Figure 2. Large culverts constructed at the base of the RSRWs

In summary there are three types of RSRW systems designed for the project: • Single tiered walls with 86° face angle up to 22 m high • 2-tiered walls comprising two single-tier walls with a horizontal setback ranging from 5 to 15 m, forming an integral structure with cumulative height up to 40 m; • 3-tiered walls comprising three single-tier walls with two horizontal setbacks ranging from 5 to 20 m, forming an integral structure with cumulative height up to 60 m.

1.3.1 Design components Tensar’s RSRW systems comprise four major components, namely the concrete modular face block, the HDPE uniaxial geogrid, the polymeric mechanical block connector and the reinforced fill material. Tensar’s facing units are produced from an automated factory process using a semi-dry concrete mix. These units are produced locally in the UAE and their minimum crushing strength at 28 days is 30 MPa. Tensar’s uniaxially orientated HDPE geogrids are produced in the UK and transported to the UAE. Testing is done in accordance with internationally recognised standards such as ISO and ASTM. The geogrids carry the European CE registration and the independent British Board of Agrément (Ref. 1). Tensar’s mechanical polymeric connectors are also made of HDPE to provide a high level of load transfer at the grid-block connection at all levels whilst allowing the transfer of horizontal shear loads between adjacent blocks. The shape and feature of these connectors is designed specifically for the ‘System’, is durable in all conditions and provides high efficiency connection strength. The reinforced fill material specified in the project is well graded granular fill (site won gabbro) with maximum particle size of 37.5 mm. Multiple samples of this material were tested in shear boxes in accordance with the British Standards guidance on soil identification and description (Ref. 2). The internal angle of friction was found to be in excess of φ’ = 45°. A conservative peak value of φpk’ = 42° was adopted for design.

266

P.G. Wills and C. Doulala-Rigby / RSRW Systems Reach New Heights in the Middle East

Constant volume (cv) values were also derived in order to satisfy the design methods’ requirements, as discussed later in the paper. Similar fill material but of slightly lesser quality was used as backfill behind the reinforced soil zone with a design value of internal angle of friction (peak) of φpk’ = 40°. The foundation soils were taken as competent, as advised by the Contractor. Although the majority of the walls were founded on rock, there were areas where the walls were founded across existing wadis where the in-situ soils comprised loose alluvial deposits at the surface. Under the close supervision and as part of the Contractor’s responsibility, any such deposits were removed and replaced by engineered rock fill providing a competent foundation for the walls. In summary, the soil parameters that were used for design are tabulated in the table below: Table 1. Soil properties used for design c’ Density ʕpk’ / cʕ v’ Soil Type (°) (kPa) (kN/m3) Reinforced fill (crushed 42 / 38 0 22.4 Gabbro) Reinforced fill (crushed 40 / 36 0 22.4 Gabbro) Foundation soils (in-situ Gabbro or 40 / 36 0 22.4 engineered rock fill) 1.3.2 Design Method The single tiered walls were designed using the Bautechnik (Deutsches Institut fur Bautechnik) Method (Ref. 3), which is a 2-part wedge German design method for geosynthetic reinforcing materials utilising constant volume (cv) soil shear strengths. This method checks internal stability of the reinforced soils block, i.e. checks pull out failure of geogrids as well as failure against geogrid rupture, which are the two main required internal stability checks in a reinforced soil retaining structure. The method also checks sliding along the base of the whole reinforced soil block and bearing capacity utilising the Meyerhof pressure distribution, in accordance to DIN4017. Global (overall) stability of the walls was also checked with Bishop’s Simplified method of slices in a limit equilibrium analysis. Bishops method utilised peak values of soil shear strength. The 2-tiered and 3-tiered walls were analysed as integral structures taking into account the effect of the upper tiers onto the lower tiers and vice versa (Ref. 4). The walls were first modelled individually with any upper tier modelled as surcharge and finally the whole multi-tiered structured was modelled in a slope software and analysed with Bishop’s simplified method of slices checking for internal, compound and global failure slip circles. A 20 kPa temporary surcharge was modelled at the crest of all walls in order to model the future freeway traffic.

P.G. Wills and C. Doulala-Rigby / RSRW Systems Reach New Heights in the Middle East

267

1.4 Construction The reinforced fill material was all site won and sourced from the adjacent cut slope excavations. The cut slopes were blasted from the mountain face and the detached rock boulders were collected and transferred to a stock pile collection area. There the boulders were broken into smaller pieces, sorted and eventually were transported to a nearby crushing plant, where they were crushed and sorted into the required gradation with a maximum particle size of 37.5 mm. To construct the walls, the site won fill was compacted and reinforced with uniaxial geogrids. Characterised by long, slim apertures, the high density polyethylene material is stretched in one direction during manufacture to produce a geogrid with significantly higher strength in the direction of roll than in the cross direction. The geogrids are secured to the modular block facing by the polymer mechanical connector. The 200 mm wide, 200 mm deep, 400 mm long modular concrete blocks are laid dry without using mortar, removing the need for any water-based products to be used in the process. Fill is placed and compacted to 200 mm thickness. The geogrid is laid inbetween the compacted fill layers at a vertical spacing typically varying from 200 mm to 600 mm.

Figure 3. Reinforced soil wall under construction

After compaction, each layer of fill is tested on site to ensure that 95% or better compaction is achieved. All the fill material used for the construction of the walls was obtained from the sections in cut. An intensive blasting operation was necessary to remove over 14M.m³ of material. As well as minimising cost of material, the reuse of site-won Gabbro also substantially cut carbon emissions by reducing transport to and from site. Between 100m² to 150m2 of wall is being completed per day. Upon completion of each wall, appropriate scour protection is provided at base level to protect the foundations of the walls from the effects of severe rainstorms.

268

P.G. Wills and C. Doulala-Rigby / RSRW Systems Reach New Heights in the Middle East

Figure 4. Construction of a 3-tier RSRW up to 60 m high

2. Conclusions The Arabian Gulf region has embraced the RSRW system and adopted it to construct a number of spectacular structures. Its economy and durability as well as ease and speed of construction have made this system an attractive alternative construction choice for projects situated in remote, rugged and elevated terrain with exposure to wind and high temperatures. For the Fujairah Freeway project, the construction of RSRW system provided an alternative solution to tunnelling and viaduct construction offering a robust, cost effective, sustainable, environmental friendly and a low maintenance engineering solution. RSRWs comprising integrated geogrid modular block wall systems have developed in to a popular choice for major civil engineering structures and are now used in some of the most prestigious projects in the Middle East.

References [1] [2] [3] [4] [5]

BBA Certificate No. 99/R109, 1999: Tensar RE and RE500 Geogrids for Reinforced Soil Wall and Bridge Abutments Systems, British Board of Agrément, Watford, UK British Standards Institution. 2002. Geotechnical investigation and testing - Identification and classification of soil - Part 1: Identification and description (ISO 14688-1:2002). The Deutches Institut fur Bautechnik, Approval Certificate Number Z 20.1-102 for reinforced soil structures, Berlin, Germany (1990 ) Geoguide 6, 2002: Guide to reinforced fill structure and slope design, Geotechnical Engineering Office, Civil Engineering Department, The Government of the Hong Kong Special Administrative Region UAE Ministry of Energy, Petroleum and Minerals Sector, Minerals Department. 2006. 1:50,000 Scale Geological Sheet 50-4, Al Fujairah. Solid and Drift Geology. (Produced by BGS under Contract to UAE Ministry of Energy, Petroleum and Minerals Sector, Minerals Department).

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 269 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-269

Deep Excavations in Luanda City Centre Alexandre Pinto and Xavier Pita JetSJ Geotecnia Lda., Lisbon, Portugal

Abstract. The aim of this paper is to present some case studies of deep excavations in Luanda down town, where the geological and geotechnical conditions are very complex, mainly due to surficial location of the ground water table, installed on granular and high permeability soils. The adopted solutions allowed the control of the deformations both at the earth retaining structures and at the neighborhood structures and infrastructures, as well as the control of the ground water inflow to the excavation pit, allowing the safe fulfilling of both the works schedule and budget. Keywords. Deep excavations, diaphragm walls, jet grouting.

Introduction Luanda down town has a very complex geological and hydrogeological scenario, mainly due to surficial location of the ground water table, installed on granular and high permeability sandy soils, resulting from the fill regularization of the Luanda bay banks. The recent construction of new buildings, as for example the Baía, the Kilamba and the Kianda buildings (Figure 1), with deep basements and sensitive neighbouring conditions, has demanded the need for the execution of deep excavations using adequate technologies. Taking into account the described scenario, in this paper are presented integrated solutions of excavation and foundations, which have allowed the control of the deformations both at the earth retaining structures and neighborhood structures and infrastructures, as well as the control of the ground water inflow to the excavation pit, allowing the fulfilling of both the works schedule and budget in safe conditions, as already happened in other works [1]. As main example it is pointed out the case of the Baía Building, which could be extrapolated to the other two ones. The main constraints of this building are pointed out, particularly the need to build 5 underground floors, the geological and hydrogeological conditions (submersed granular sandy soils), as well as the neighbouring conditions. Also emphasized are the main design criteria of the adopted solutions, mainly the diaphragm retaining walls, braced during the excavation works by an horizontal jet grouting sealing slab, located below the excavation final level, and by two levels of reinforced concrete slabs and steel trusses, acting as stiff diaphragms at the level of floors -2 and -4. Finally, the design models, as well as the main results of the instrumentation and observation during the excavation works are also presented and analysed.

270

A. Pinto and X. Pita / Deep Excavations in Luanda City Centre

Figure 1. Site location of the described buildings.

1. Baía Building The excavations works for the Baía building has finished in 2010, comprising an area of about 1000m2, with a rectangular shape of about 37x27m2 and 12m of overall depth for the construction of 5 underground floors and 25 upper floors. As main design criteria it was considered the need to avoid the executions of ground anchors below the ground water table, as well as the control of the ground water inflow to the excavation pit. According to this criteria it was adopted a solution of peripheral diaphragm walls, braced during the excavation works by an horizontal jet grouting sealing slab, located below the excavation final level, and by two levels of reinforced concrete slabs and steel trusses, acting as stiff diaphragms at the level of floors -2 and -4 (Figure 2). The jet grouting sealing slab allowed also the limitation of the ground water inflow through the excavation base. As solution for the foundations, barrettes, caped for a reinforced concrete raft, were adopted. During the excavation works the barrettes allowed the nailing of the jet grouting sealing slab against uplift, resisting to tension loads and transferring them to the soil by lateral resistance.

Figure 2. Adopted solution for the excavation works at the Baía building: perspective and cross section.

A. Pinto and X. Pita / Deep Excavations in Luanda City Centre

271

1.1 Main Restraints As main restraints, it is possible to emphasize the following points: •

Geological and geotechnical conditions: high permeability sandy soils with the ground water table located about 1.5m below the ground surface, and oscillating due to the influence of the tides at the Luanda bay (Figure 3).

Figure 3. Main geological and geotechnical conditions.

•

Neighborhood conditions: the Sonangol Clinic building, less than 1m from the excavation pit border, with 2 upper floors and 1 basement, resting over spread foundations, as well as the Ambiente roundabout and Direita de Luanda street, both with important road traffic (Figure 4).

Figure 4. Execution of diaphragm walls very close to the Sonangol Clinic building

1.2 Main Earth Retaining and Foundations Solutions Taking into account the main restraints, the following main solutions were adopted, on an integrated way, for the earth retaining and foundations (Figure 5): •

Earth retaining solution: peripheral diaphragm walls with 0.60m thickness, braced during the excavation works by an horizontal jet grouting sealing slab with 3m thickness and 1.50m diameter columns, located below the excavation final level, and by two levels of reinforced concrete slabs (integrated on the

272

A. Pinto and X. Pita / Deep Excavations in Luanda City Centre

basement final structure) and steel trusses (where the final basement slabs were not horizontal) acting as stiff horizontal diaphragms, at the level of floors -2 and -4 (figure 2). Those diaphragms were externally supported by the peripheral walls and internally by steel profiles embedded inside the internal foundation barrettes. At the Sonangol Clinic side and above ground water table level, one row of ground anchors was performed in order to control the wall deformability. Behind the joints of the diaphragm wall panels one jet grouting column was performed in order to increase the resistance against the water inflow. The panels were embedded 8m bellow the final excavation level. •

Foundations solution: reinforced concrete raft with minimum thickness of 0.5m, cast over the jet grouting sealing slab and capping barrettes, embedded 5m on the compact sands. During the excavation works the barrettes allowed also the nailing of the jet grouting sealing slab against uplift, accommodating tension loads, transferred to the soil by lateral resistance.

Figure 5. Plan of the earth retaining and foundations solutions

1.3 Design The design of both earth retaining and foundations solutions, was performed using 2 FEM programs (Plaxis V8 and SAP2000V14) on an interactively way (Figure 6).

Figure 6. Main loads and calculation models

A. Pinto and X. Pita / Deep Excavations in Luanda City Centre

273

1.4 Monitoring and Survey A wide monitoring and survey plan was implemented in order to manage the safety of the excavation works, through the previous assessment of both the diaphragm walls and the neighbor building behavior. The following devices were installed: •

18 topographic targets (TT) placed at the diaphragm walls and at the neighbor building façade.

•

1 load cell (LC) on a ground anchor.

•

1 inclinometer (IN) inside the diaphragm at the neighbor building side.

The obtained results confirmed the majority of the design assumptions. The maximum observed displacements were: 18mm on vertical and upper direction at the end of the excavation works and 10mm of vertical and inward direction at the diaphragm walls (figure 7). The maximum displacements at the Sonangol Clinic building were not bigger than 7mm and no cracking or any other damage phenomena were observed.

Figure 7. Main monitoring results at Sonangol Clinic building side

2. Kilamba Building The excavations works for the Kilamba building is being continued during 2011, comprising an area of about 1800m2 with a rectangular shape of about 47x38m2 and 10.5m of overall depth, for the construction of 3 underground floors. The geological and geotechnical conditions were very similar to the Baía building ones. The main design criteria were similar to the Baía building. However, foundations solution, bored piles (1.0 diameter), using Kelly bar, were used, instead of barrettes. Before the beginning of the excavation works, pile full scale load tests, with Osterberg cells, are being performed in order to confirm the lateral resistance.

274

A. Pinto and X. Pita / Deep Excavations in Luanda City Centre

3. Kianda Buildings Complex The excavations works for the Kianda buildings complex is being continued also during 2011, comprising an area of about 100,500m2, with a trapezoidal shape of about 80x130m2 and 9m of average depth for the construction of 2 underground floors. The geological and geotechnical conditions were very similar to previous ones (Figure 8). The main design criteria were also similar to the Baía building. However, as foundations solution, bored piles (1.0 and 0.8m diameter), using Kelly bar, as well as tubular steel micropiles, were used, instead of barrettes. Micropiles were adopted in the areas where the permanent loads were small (outside the towers areas). Due to the site overall area, as well as the excavation depth, the bracing of the diaphragm walls (0.60m thickness) is being performed using one row of ground anchors, at the top of the diaphragm walls, and a jet grouting sealing slab (2.0m diameter columns with topographic and GPS implantation), at the final excavation level.

Figure 8. Perspective of the building complex and execution of the diaphragm walls

4. Main Conclusions As main conclusion it should be pointed out the good performance of the adopted solutions, mainly due to the “box effect” induced by the diaphragm walls and the jet grouting sealing slab, built before the beginning of the excavation works, allowing the efficient control of both the deformations and the ground water inflow. Also important for the deformations control are the stiff bracing systems, including the structural elements, the execution of ground anchors above ground water table, as well as the jet grouting sealing slab at final excavation level. Due to its predictability the adopted solutions have allowed also the fulfilling, in safe conditions, of both the works schedule and budget.

References [1] A. Pinto et al., Ground Improvement solutions at Sana Vasco da Gama Royal Hotel, 17th International Conference on Soil Mechanics and Geotechnical Engineering – Egypt (2009), p 2180–2183.

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 275 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-275

Geotechnical Innovation in Shaft Sinking in the Zambian Copper Belt G.C. Howell SRK Consulting Johannesburg, South Africa Abstract. Substantial capacity enhancement of mines in the Copper Belt of Zambia has required that geotechnical expertise and innovative solutions be applied while sinking new shafts from surface through sub soil (saprolitic) and incompetent rock to a considerable depth. One shaft is now in operation and boasts the tallest steel headgear in the south hemisphere. The surface geology of the area exposes typical sub-tropical soils where the upper 25 to 80m consists of very weathered saprolitic soils overlying more competent granitic or conglomerate rocks. At the surface, the colluvial materials exhibit large collapse potential. Shaft sinking in these conditions is problematic, geotechnically challenging and is complicated but the high cost and time needed to ship equipment into Zambia. To complicate matters further, groundwater is common and rainfall is generally high which creates unique and fundamental geotechnical conditions. This paper describes the geotechnical conditions on site and the unique and innovative methods used to develop the shaft infrastructure for a project that encapsulates all the major elements of geotechnical engineering while providing a practical solution. Keywords. Geotechnical Engineering, Shafts, Soils, Lateral Support

1. Introduction The solution to a problem can occur to the designer at any time provided you are thinking along the right lines. The solution to this particular problem occurred to me while I was sitting in the hot Zambian sun under the only scraggly tree on the mine. The design of this shaft had been done assuming that 5 m long pile sockets could be constructed in the soft rock at a depth of 25 m below surface. I was on site to witness the installation of the piles, but the auger rig on site was battling to even make an impression on the very soft rock in the sequence above and I knew that we were heading for delays if this situation persisted. Bringing a more powerful machine to site from South Africa would mean a delay of more than 3months and we were already on the critical path. What options were there? The background to this project was the following: the shaft depth is 1 500 m; the total working load on each leg of the headgear is 10 000 kN (1 000 tonnes) making is the tallest steel head gear structure in Southern Africa at 89 m; the shaft shape is oval with dimensions 9.4 by 7 m (semi-circular ends with 2.4 m straight sides) which presented its own problems; and the site consisted of deeply weathered saprolitic soils overlain by 5 to 6 m of transported materials exhibiting large collapse potential.

276

G.C. Howell / Geotechnical Innovation in Shaft Sinking in the Zambian Copper Belt

The solution was to employ an innovative, yet untried but unique method to ensure stability of the shaft structure from surface to 25m depth. This paper describes the context and the construction procedure used to ensure a structurally competent upper shaft infrastructure for one of the largest copper mines in Zambia.

2. Philosophy of Shaft Sinking Shafts are permanent structures that provide access to underground workings for people, services (air, water and power) and materials while providing a vital extraction function for ore, waste air and water. Any disruption during development or operation will have a significant impact on safety and profitability. As a geotechnical and structural element, therefore, shafts constitute the most stringent of design requirements, particularly in cases where suspect conditions are apparent. The philosophy of shaft design juxtaposes the geotechnical hazards, on the one hand, with the geotechnical contribution (or intervention) on the other, as shown conceptually in Figure 1. The identification of the geotechnical hazards are determined by a competent subsurface investigation using boreholes, test pits and laboratory testing under the supervision of a competent Engineering Geologist supported by a Geotechnical Engineer. The purpose of the investigation is to identify the geotechnical zones to be intersected by the shaft, but moreover, to identify the constitutive, groundwater and jointing characteristics of the materials. For example, of major important for the surface works are bearing capacity, settlement, swelling and collapse potential aspects for the upper, often completely weathered or transported, materials, while deeper in the shaft, the occurrence of adverse jointing and water bearing strata are important. Common nomenclature for both soil and rock identification and characterisation should be used (MCCSSO and Q, RMR classifications). Geotechnical interventions are the appropriate engineering response to the hazard. Settlement and collapse potential may be dealt with using piles, excavation and engineered fill or soil improvement techniques (of which pressure grouting, stone piles and dynamic compaction are options), while slope stability issues may require lateral support in the form of secant, contiguous or closely spaced piles depending on the inherent strength of the insitu materials with or without soil-nails, anchors and/or rock bolts. Under appropriate circumstances, combinations of techniques may be considered. Squeezing (from clays) or inward pressures from unconsolidated sediments are most effectively countered by the ring compression nature of the shaft wall, but structural problems arise when the shaft is subjected to unsymmetrical loading. Similarly, non-circular shafts (such as straight-sided ovals) can increase the circumferential bending moments (and therefore the reinforcement requirements) by up to 7 times. Elliptical or circular profiles are preferable from a geotechnical and structure perspective but may not be convenient for conveyance design - an issue that designers should be cognizant of. Water pressure must be considered in all instances since stresses developed can be significant.

G.C. Howell / Geotechnical Innovation in Shaft Sinking in the Zambian Copper Belt

277

Figure 1. Philosophy of Shaft Design

Interventions within the rock-quality materials depend on variability, joint spacing and material quality. Adverse wedge formation requires support in the form of rockbolts, mesh and/or shotcrete. Since permanent concrete linings are expensive and time consuming to install, clients are loath to countenance their use, but safety, risk and structural longevity issues should be carefully considered in potentially unlined shafts. Particular attention must be paid to friable and material capable of weathering insitu as wedge formation may result in time (even during construction) which could put men and machinery at risk. Even in massive rock mass conditions, onion skin development may occur due to weathering or stress effects. This design philosophy has become the standard used rigorously for the shafts and declines designed by SRK Consulting. This paper further describes the outcome to a particular set of geotechnical hazards encountered for a project in Zambia.

3. Geotechnical Context The shaft is situated in northern Zambia where the engineering geology is typical of tropical and semi-tropical areas consisting of transported and deeply weather saprolitic/lateritic materials. The depth of weathering and the consistency of the materials vary dramatically independently of the surface profile and therefore a thorough geotechnical investigation using boreholes, large diameter auger holes, DPSH probing and surface test pits were used to accurately characterize the profile. An example of an auger-hole log is shown in the Appendix, where the detail descriptions are evident.

278

G.C. Howell / Geotechnical Innovation in Shaft Sinking in the Zambian Copper Belt

The results of the investigation showed a geological profile consisting of: •

Surface to -5 m: Fill, hillwash and transported materials generally of loose to medium dense consistency reporting as fine to medium sand. With depth the material increases in clay and silt content. The most important aspect of this material is that it is pin-hole voided and susceptible to up to 10% collapse. Evidence of collapse potential was clearly visible in the existing building on site and further collapse settlement was experienced during construction due to uncontrolled surface water run-off.

•

-5 m to -7.5 m: Rounded and sub-rounded gravels, pebbles and cobbles in a matrix of medium dense silty fine to medium sand. This pebble marker stratum varies in thickness across the site (in some places it is absent) and clearly illustrates the interface between the erosion surface of the parent rock and the transported materials above. The pebble marker, in this case, also represents the depth extent of the collapsible materials.

•

-7.5 m to -22 m: Residual quartzite, siltstone and sandstone highly weathered to fine and medium sand of medium dense to dense consistency with pockets of loose to medium dense fine sand. This material represents the bulk of the shaft pre-sink in soil materials.

•

-22 m to -25 m: Residual sandstone of very soft rock strength. This material represents the transition of the weathering profile to rock quality materials. In this case the transition occurred at this depth, but in other part of northern Zambia, this stratum can be at 80 m depth or more.

•

-25 m to -70 m: Residual sandstone improving in strength with depth from soft rock to medium hard rock. The bedding is very apparent dipping shallowly to the west and intersected by random sub-vertical joints, making the barrel susceptible to the formation of small to large blocks/wedges.

From these descriptions, it is clear that the geotechnical context covers a number of potential hazards including collapse potential, stability, bearing capacity, settlement and construction safety issues. A design was therefore required to address all these aspects.

4. Geotechnical Engineering Context The geotechnical hazards alluded to in the previous section required an initial design which consists of the following elements: •

The shaft consists of two reinforced concrete collars. The bank level at +0.5 m supports the headgear and all ancillary equipment, while the sub-bank level at -5.5 m allows access to the hoist. The two collars form the pile caps and are connected vertically by retaining walls.

G.C. Howell / Geotechnical Innovation in Shaft Sinking in the Zambian Copper Belt

279

Figure 2. Typical Layout at Shaft Position

•

Lateral support “shaft” piles were needed to create a stable perimeter for the shaft to address the stability hazard. This consisted of 28 no 750 mm diameter auger cast piles arranged at close spacing in the form of an ellipse as shown in Figure 2. The clear spacing between piles was of the order of 450 mm since arching between piles was possible due to the slightly cohesive nature of the material below 6 m which allowed the material to be excavated and shotcreted without undue sloughing. The depth of the shaft piles was designed to

280

G.C. Howell / Geotechnical Innovation in Shaft Sinking in the Zambian Copper Belt

intersect the soft rock between -22 and -25 m. The accuracy of the pile installation was expected to be less than 1% of the pile length (about 250 mm), but control of the auger process ensured that the maximum deviation was less than 100mm in practice. •

The headgear piles are located at the 4 corners of the collar in groups of 6 x 750 mm diameter. These pile groups form the main support for the structural steel headgear which is reputed to be the tallest steel structure of its type in the world (or at least the southern hemisphere). These piles were designed to be socketed 5 m into the hard rock at and below -25 m where the bearing capacity and settlement profile was adequate to resist the dead and live loads of 10 000 kN per leg.

•

Lateral support of the excavation of the sub-bank level was provide but 4 sets of 4 piles along each side which also doubled as additional support of the bank level collar.

•

Once the collars had been cast, excavation of the shaft was to proceed in 3 m lifts, followed by casting of the permanent lining so that the lateral support piles were exposed for span less than 6m as a “fixed ended beam” or 3 m as a “cantilever”.

The double ring support system was therefore designed to address numerous geotechnical hazards while ensuring safety during construction and stability for the operational mine for the next 30 years. The above description is the design based on the operational requirements and the geotechnical parameters for the site and, in a sense, is standard practice. But piling equipment in Zambia at the time was just not capable of installing the piles according to the design specifications. That’s why I was on site in the hot Zambian sun. The auger rigs had literally ground to a halt on the soft rock and advance was very slow, even non-existent. To bring more powerful rigs to Zambia would mean a delay of many months and the shaft was on the critical path of the project. I had to do something and the solution occurred to me on site that day. I discussed the concept with the contractor’s site agent and developed the alternative innovative design.

5. Geotechnical Engineering Solution The solution to the problem was to build a spread-footing at depth. The vast majority of structures are founded on spread footings near surface, but this one would have to be founded at a depth of 25 m. The question was how this was to be done? The lateral support ‘shaft’ piles had already been installed (or were being installed) when the decision was made, but there was still a lot of work to do before the 25 m level was reached. As a result, the collar, consisting of the bank and sub-bank levels, was constructed according to the original design, supported at surface on the shaft and headgear piles. During this operation, the stage and stage winder was commissioned so that shaft excavation, followed immediately by the reinforced concrete permanent shaft barrel, could be commenced. The permanent lining was kept within 3m of the bottom of the

G.C. Howell / Geotechnical Innovation in Shaft Sinking in the Zambian Copper Belt

281

excavation to ensure that no instability of the piles could occur, thereby ensuring an incremental “encastre” condition at all times. This method progressed to the -23 m level where the ends of the piles were starting to be exposed in the shaft excavation. With the permanent lining at the -22 m level, the shaft was excavated to -25 m and the piles exposed. From a geotechnical perspective the ‘contact’ between the soil above and the soft rock was clearly visible. Advantageously, the soil horizons exhibited significant cohesion at this depth and therefore the possibility of instability in the overlying material was negated. Similarly, no running water was encountered and the material was at worst described as ‘slightly moist’. The design now called for the octagonal shape to be subdivided into 8 segments called ‘adits’ which were excavated two-at-a-time on opposite sides. These adits exposed the ends of the piles in each ‘head wall’. The pile concrete was moiled back to expose the pile reinforcement which was then integrated with the lower ring beam reinforcement, shutters placed and concrete cast. Dowel bars were installed to connect adjacent adits. Progress continued on an alternate adit basis until the full lower ring beam was complete. The result of the operation is shown in part section in Figure 3 where the maximum bearing pressure is of the order of 200 kPa, which is well within the capacity of the soft rock at founding level.

Figure 3. 3D sectional view of completed Shaft

The completed design essentially represents a conventional spread-footing at depth, supporting reinforced concrete columns (the piles) with the collar atop. Subsequently, construction of the rest of the shaft barrel, headgear, winders and underground works continued apace. This shaft is now operational and the construction method has been hailed as a supreme success.

282

G.C. Howell / Geotechnical Innovation in Shaft Sinking in the Zambian Copper Belt

6. Conclusion Faced with a practical problems related to the inability of the available equipment to successfully complete the postulated design, an alternative and innovative design method was found to complete this shaft project on time and within budget. The circumstances under which this design was completed are unique in our experience. We often develop our design premise using sophisticated ‘1 st world’ technology, as we did in this case. But we must realize that the appropriate means is not always available to achieve the specifications, as we found out in this instance. It is however a credit to the client (for realizing the importance), the contractor (for being receptive) and to ourselves (for the technical ability) to conceptualize and construct an alternative foundation system which has proven to be successful and innovative in the field of geotechnical engineering. The solution is also unique - we can find no references of similar techniques being used anywhere else in the world. We therefore consider ourselves justified in claiming accolades for our innovative approach.

Acknowledgements We would like to acknowledge the support of KCM Zambia for their buy-in to the method used and Grinaker-LTA Mining (in particular Richard Simpson and Brett Pollington) for their competent construction and encouragement. Finally, my thanks to SRK’s Angus Bracken, Derek Warwick and Melt Bester whose engineering geological expertise and conceptualization skills allowed the engineering solution to become a reality.

G.C. Howell / Geotechnical Innovation in Shaft Sinking in the Zambian Copper Belt

Appendix: Typical Borehole log (part)

283

284 Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-284

The use of reinforced soil to construct steep sided slopes in order to create a safer highway - Ruhengeri to Gisenyi road, Rwanda Peter ASSINDERa,1, Heribert SCHIPPERS b and Giuseppe Ballestrac a Huesker Synthetic GmbH, Gescher, Germany b Strabag International GmbH, Cologne, Germany c Afriace Limited, Ventimiglia, Italy

Abstract. Parts of the existing Ruhengeri to Gisenyi road in the North Western part of Rwanda winds along steep mountainous hillsides which stand at ~8,000 feet. The large number of trucks travelling to the Democratic Republic of Congo has resulted in the edge of the road falling away down the hillside. The project formed part of the African Development Bank and European Commission collaboration for Rwanda’s transport improvement program, with this particular part of the project financed by the European Commission. Part of the large rehabilitation project to reconstruct and improve the general alignment and safety of the road, included a 350m section where a reinforced soil solution was adopted. The 70° steep reinforced soil slope raised to a maximum height of 8m. The use of reinforced soil allowed the road to be widened and the use of locally won fill from a local quarry to be used, thus negating the use of concrete structures or other expensive material solutions. Additionally, a vegetated slope face allowed the structure to blend in with the existing landscape. This paper looks at the design methods used and also provides a photo chronological sequence to clearly demonstrate this type of construction. The paper also highlights some of the techniques that can result in high quality construction of reinforced soil structures. This form of construction also provided a sustainable solution with respect to carbon footprint and the paper includes information on recent research undertaken on comparison of construction techniques with respect to carbon emissions and sustainable construction. Keywords. Road construction, geosynthetics, reinforced soil, sustainability

Introduction The Ruhengeri to Gisenyi road in North Western Rwanda is the main transit route for goods into the Democratic Republic of the Congo. At the section around project reference +15km, the road is cut into the mountainside at ~8,000 feet (Figure 1). It is understood that the original road was constructed approximately 30-40 years ago. 1

Corresponding Author: Peter Assinder, Huesker Synthetic GmbH, Fabrikstrasse 13-15, D-48712, Gescher, Germany: E-Mail: [email protected]

P. Assinder et al. / The Use of Reinforced Soil to Construct Steep Sided Slopes

285

An increase in traffic volume of heavy goods vehicles and general deterioration of the existing road has resulted in the edges of the pavement construction falling away from the main carriageway. Additionally, parts of the existing carriageway had insufficient highway width and a lack of suitable safety barrier. Funding was obtained from the European Commission to improve the quality of the road with respect to its long-term stability and safety. At this section the carriageway required widening by approximately 1m with an additional 2m ‘hard shoulder’ and safety barrier. The widening requirements resulted in the pavement extending into the existing steep hillside (Figure 2). Therefore, a green faced steep reinforced soil structure was deemed the most appropriate, economical and sustainable solution to extend the existing carriageway, whilst maintaining the optimal volumes of construction materials required. The Ruhengeri to Gisenyi road re-construction and upgrade commenced in July 2007 and was completed in March 2010, with the reinforced soil slope section constructed in 2009.

Figure 1. Road before rehabilitation

Figure 2. Start of tree clearance

1. Design theory The proposed structure comprised a 70 degree steep reinforced soil slope. The design was undertaken in Europe to German design guidelines, assuming the partial safety concept. The calculations were carried out in accordance with the appropriate DIN Standard [1] (Figure 3), EBGEO guidelines [2] and EC7 [3]. Load Case 1 (normal situation with traffic) was assumed for the static calculations with a 20 kPa traffic loading and a 5 kPa pedestrian loading. The project consultant undertook the following stability checks • • • • • • •

Global Stability (GZ 1C) – Krey/Bishop method Sliding Stability (GZ 1B) Bearing Capacity (GZ 1B) Reinforcement strength (GZ 1B and GZ 2) Internal Stability (GZ 1B and GZ 1C) Validation of geogrid anchor length (front and rear) (GZ 1B and GZ 1C) Settlement of structure (GZ 2)

286

P. Assinder et al. / The Use of Reinforced Soil to Construct Steep Sided Slopes

Figure 3. Limit States to satisfy according to DIN 1054

The ground conditions were to a degree assumed using standard parameters (Table 1). The resultant geometry of the reinforced soil slope (Figure 4) which remained stable following the check of the required failure mechanisms comprised layers of polyester geogrid with an ultimate tensile strength of 55 kN/m and anchor lengths varying from 3.5m to 6m. For lower sections of slope the amount of geogrid and required anchor lengths reduced accordingly.

Table 1. Summary of soil parameters Name

Reinforced Fill Rock Existing road base Natural Soil Sub Base

Weight density γk/k’ (kN/m3) 19/10 24/14 19/9 18/8 18/8

Angle of Friction Φk (°) 28 36 34 34 34

Cohesion ck (kN/m2) 5 100 5 10 5

*Dpr ≥ 100%

Figure 4. Cross section of Reinforced Soil Slope

Modulus of Stiffness Es,k (MN/m2) 30….50* ≥100 40 50 60

P. Assinder et al. / The Use of Reinforced Soil to Construct Steep Sided Slopes

287

2. Construction practice The construction of steep reinforced soil slopes forms part of a general earthworks procedure. The first construction stage comprised excavation of the existing hillside (Figure 5) to the required level. A nonwoven geotextile was installed along the cut face to provide separation and filtration. A drain was placed at the heel of the excavation to control groundwater ingress. Localised persistent rock outcrops at proposed formation required removal.

Figure 5. Excavation for structure

The process of constructing the reinforced soil block and steep sided slope face is a multi-phased repetitive process (Figure 6) which results in a ‘wraparound’ type construction. The steep slope face was formed using a climbing temporary shutter (Step 1 of Figure 6) which was set at the desired slope face angle and held in place with metal arms. The required strength and length of geogrid from the geotechnical design was then placed with sufficient length placed over the front of the shutter to provide face anchorage as the lift is completed (Step 2). A secondary geotextile was placed within the front face to hold the topsoil wedge in place (i.e. to stop the finer grained organic material falling through the apertures of the geogrid) (Step 3 and Figure 7). The specified Fill from the design was then placed in the correct layer thickness and compacted to an end specification (proctor density of 100%) (Steps 4 and 5). When the layer was near completion, the geogrid face anchor length (and front face geotextile) was folded back into the construction fill and secured with the final layer of fill (Steps 6-8). The first layer was then complete (Figure 8) and the temporary shutter lifted and placed on top to start the next layer (Step 9) with the process continued to completion of the structure (Figure 9). The climbing shutter form of construction is an effective and economical approach to constructing steep reinforced soil slopes. However, care and attention is required with respect to compaction and anchorage at the front face. The topsoil wedge within the front face of the wrapped lift is there to provide an organic medium to allow vegetation to develop, thus creating a green faced slope. There can be a quandary between providing too much compaction at the front face (which can result in over compacted topsoil which is no longer providing a satisfactory growing medium), or not enough compaction at the front face (which can result in the face ‘slumping’ as the temporary shutter is lifted away).

288

P. Assinder et al. / The Use of Reinforced Soil to Construct Steep Sided Slopes

Experience has shown that a 300mm wedge of topsoil should be sufficient to sustain vegetation growth and this topsoil wedge can be ‘heeled in’. Innovative solutions have been used in Europe to isolate the topsoil wedge from the main reinforced fill body, such as using a cardboard box within the wrapped front face, which is then filled with topsoil, and thereby kept completely separate from the main fill.

Figure 7. Reinforced soil lift

Figure 8. Reinforced soil lift layer completed

Figure 6. Construction sequence for reinforced soil slope

Figure 9. Reinforced soil structure near completion

3. Resource and Sustainability The use of reinforced soil in place of more traditional structures (e.g. gravity mass concrete retaining walls) can result in a more sustainable solution [5]. The concept of carbon foot-printing in slope engineering is a developing theory [6] in Europe, America and Japan.

P. Assinder et al. / The Use of Reinforced Soil to Construct Steep Sided Slopes

289

Sustainability is now a requirement of planning policy and legislation in many countries. Reinforced soil slopes offer great opportunity for a reduction in carbon footprint/embodied energy values, as they can readily enable the re-use of site won, near site materials or the incorporation of recycled or modified fills. They can also allow the more rapid construction of perhaps a less conventional, but no less valid, alternative to a more tried and tested solution such as a reinforced concrete retaining wall or steel sheet piles. Carbon footprint calculators for different construction processes are now becoming more refined, resulting in a more detailed calculation. Therefore, allowable comparisons are starting to be made between different construction solutions with respect to the amount of carbon used. The project solution did not use any concrete or steel (both components have high level carbon output in comparison to geosynthetics and soil). The fill was sourced very close to the project site (thereby reducing transportation, which is a high end user of carbon).

4. Conclusions Geosynthetic reinforced soil structures are a mature construction solution, with supporting state governed guidelines and accepted design processes. With quality manufactured geosynthetics, which carry appropriate external accreditation, high levels of construction can be achieved. As sustainable solutions are now becoming more and more relevant, these types of geosynthetic slope engineering structures are at the forefront of low carbon emission solutions, whilst also remaining economically advantageous.

Figure 10. Completed road section

References [1] DIN 1054:2005-01 Baugrund; Sicherheitsnachweise im Erd- und Grundbau [2] EBGEO 1997: Empfehlungen für Bewehrungen aus Geokunststoffen - Deutsche Gesellschaft für Geotechnik e.V. (DGGT) [3] DIN EN 1997-1, Eurocode 7: Geotechnical design - Part 1: General rules; EN 1997-1:2004 [4] Brau, G., 2008, EBGEO – German Recommendation for Reinforcement with Geosynthetics, Eurogeo 4, Edinburgh [5] WRAP 2010, Sustainable Geosystems in civil engineering applications, Report MRF116 [6] O’Riordan, N., Phear, A., Nicholson, D., Hughes, L. 2011 “Examining the carbon footprint and reducing the environmental impact of slope engineering options”. Ground Engineering 28-30

This page intentionally left blank

Section 6 Materials Testing

This page intentionally left blank

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 293 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-293

Characterization of Shear Strength of Abandoned Dumpsite Soils, Orita-Aperin, Nigeria Kolawole Juwonlo OSINUBI a, 1 and Afeez Adefemi BELLO b Department of Civil Engineering, Ahmadu Bello University, Zaria, Nigeria b Department of Civil Engineering, Osun State University, Osogbo, Nigeria

a

Abstract. The results of the laboratory investigation carried out on abandoned dumpsite soil in order to define the ranges of the water content and dry unit weight at which compacted test specimens would have adequate shear strength are presented. Test specimens were compacted with British Standard light energy over a range of water content namely -2, 0, 2 and 4% of optimum moisture contents. Specimens with dry unit weight of 16.6kN/m3 prepared in the 13.9 - 18.1% moulding water content range recorded shear strength values equal to or greater than 200 kN/m2 which is the minimum acceptable for materials to be used as hydraulic barriers in containment structures. Keywords: Abandoned dumpsite soil, British Standard light, Compaction, Containment structure, Dry unit weight, Optimum moisture content, Shear strength.

Introduction Hydraulic barriers used for waste containment structures in landfills design play a vital role in impeding fluid flow and attenuating inorganic contaminants. For a compacted natural soil to be used as hydraulic barrier it must possess a hydraulic conductivity value less than or equal to 1 x 10-9 m/s, volumetric shrinkage upon drying (maximum of 4%) and shear strength (minimum of 200 kPa). The structural integrity of these hydraulic barriers must be ensured by constructing facilities that have adequate shear strength. Edil et al. (1992) [1] stated that the material should have a minimum unconfined compressive strength of 200 kN/m2 and be durable to withstand destructive forces of alternating wet/dry and freeze/thaw cycles. This strength is the lowest value for very stiff soils based on consistency classification (Peck et al., 1974) [2]. An abandoned dumpsite soil lies beneath a landfill during its inactive (post closure) period. It is expected that the natural soil structure, physical composition and chemical properties should have been affected and changed since leachate would have probably infiltrated into the soil. It is also expected that the soil sample will behave the same way natural processed soil and geosynthetic clay liners behave when used as containment material. However, this research work presents the results of the effect of compaction water content and dry unit weight on the shear strength of the sample soils to determine its potential for use as hydraulic barrier material in waste containment structures.

294

K.J. Osinubi and A.A. Bello / Characterization of Shear Strength of Abandoned Dumpsite Soils

1. Background Shear strength test is one of the common methods of measuring undrained strength of undisturbed samples. Lambe and Whiteman (1979) [3] stated that shear strength test is the best general purpose test but it underestimates strength because disturbance decreases effective stress. In this paper, unconfined compressive strength was used to depict the structural integrity of the compacted abandoned dumpsite soil propsed for use in containment facilities. The unconfined compressive strength (UCS) value can be estimated using the following equations: δ = R x CR x 1000 kN/m2 A

(1)

and A = 100Ao 100 – ε %

(2)

then δ = RCR (100 – ε %) x 1000 kN/m2 100 Ao

(3)

where Ao , the initial area of cross-section (mm2), ε, % = x/Lo, ε, % = strain percent, Lo = initial length of specimen, D = diameter of specimen (mm), CR N/division = mean calibration of load ring, R divisions = load ring reading at strain, ε; A (mm 2) = area of cross-section at strain, R x CR, Newtons = load on specimen at strain, ε ; σ = compressive stress at strain. The ring calibration CR can be assumed to be constant. 2. Materials and Methods 2.1. Materials The soil samples used in this study are naturally occurring yellowish brown material obtained from a borrow pit at Orita - Aperin abandoned dumpsite, Ibadan (Latitude 7o30’ and Longitude 4o56’), Oyo state, Nigeria using the method of disturbed sampling. The soil samples were obtained at a depth of 1.80 m and designated as AB1, AB2 and AB3. 2.2. Methods • •

Laboratory tests were carried out to determine the index properties of the sample specimen in accordance with British Standard Institution [4]. The specimens were prepared by mixing the relevant quantity of dry soil samples previously crushed to pass through BS No.4 sieve (4.76 mm aperture) as outlined in BSI 1377 [4] and ASTM [5]. The specimens were

K.J. Osinubi and A.A. Bello / Characterization of Shear Strength of Abandoned Dumpsite Soils

•

295

prepared using moulding water content in the range 6.5 – 22.5%. The compaction method used is the British Standard light (BSL) that is easily achieved in the field and similar to that described by Daniel and Benson (1990) [6]. Unconfined compression test was carried out on soil specimens previously mixed with tap water and compacted at moulding water contents in the range 6.5 – 22.5% using BSL energy. Compacted specimens were sealed in plastic bags and allowed to stand for at least 24 hours before trimming and testing. At least three specimens (38 mm diameter by 76 mm high) per moulding water were used in the unconfined compression tests.

3. Results And Discussion 3.1. Index properties The index properties of the soil samples are summarized in Table 1. The particle size distribution curves are shown in Figure 1. The soils are classified as A-7-6 according to the Association of American States Highway and Transportation Officials Classification System (AASHTO) [7] and as lean clay with sand (CL), according to the Unified Soil Classification System (USCS) [4]. The clay mineralogy of the soil samples determined using x-ray diffraction (XRD) is kaolinite. Micrographs of the soils reported by Osinubi and Bello (2010) [8] indicate that kaolinite flakes and smectite tubes are present in their fabrics. Table 1. Index properties of abandoned dumpsite soils Property

Soil Samples

Natural moisture content, %

AB1 4.1

AB2 4.3

AB3 4.2

Specific gravity

2.61

2.61

2.64

Liquid limit, %

36

43

40

Plastic limit, %

23

27

24

Plasticity index, %

13

16

16

Linear shrinkage, %

6.25

6.4

8.59

% Passing BS No. 40 sieve

74.15

71.6

75.95

% Passing BS No. 200 sieve

56.15

57.9

58.99

% < 2 µm

18.2

18.9

19.92

Maximum dry density, Mg/m3

18.05

17.8

17.17

Optimum moisture content, %

14.4

14.5

14.2

A-7-6 (4)

A-7-6(6)

A-7-6(7)

USCS classification

CL

CL

CL

Activity

0.7

0.75

0.75

AASHTO classification

Derived Parameters Grading modulus

0.61

0.67

0.6

Plasticity product

889.7

947.2

982.4

1121.4

1236

1304.8

Plasticity modulus

296

K.J. Osinubi and A.A. Bello / Characterization of Shear Strength of Abandoned Dumpsite Soils

120

percentage passing (%)

100

80 AB1 60

AB2 AB3

40

20

0 0.001

0.01

0.1

1

10

particle size (mm)

Figure 1. Particle-size distribution curves for Orita-Aperin abandoned dumpsite soils

3.2. Chemical composition The chemical composition of the abandoned dumpsite soils is summarized in Table 2. The concentrations of Fe2O3 are in the range 5.5 – 6.4% for soil samples in agreement with findings reported by Nwaiwu (2004) [9]. The organic carbon content of the soil samples is low thus indicating low loss on ignition. Table 2. Chemical composition of soil samples Oxide

Soil Samples AB1

Fe2O3 (%)

6.4

CaO (%)

7.1

MnO3 (%)

0.2

K2O (%)

1.1

Cr2O3 (%)

0.2

Al2O3 (%)

1.6

AB2 6

AB3 5.5

4

18.1

0.23

0.73

0.6

1.9

0.14

0.11

2.4

3.2

3.4

1.1 0.11

SiO2 (%)

3.8

Organic Carbon

0.13

0.18

6.3

6.7

7

0.25

0.35

0.29

pH EC µmhos/cm

3.3. Effect of compaction moulding water content The variation of unconfined compressive strength (UCS) with moulding water content is shown in Figure 2. UCS generally increased to peak values at 16.2% moulding water

K.J. Osinubi and A.A. Bello / Characterization of Shear Strength of Abandoned Dumpsite Soils

297

Unconfined compressive strength (KN/m2)

450 400 350 AB1 300 AB2 250 AB3 200 150 100 50 0 6

8

10

12

14

16

18

20

22

24

Moulding water content (%)

Figure. 2. Variation of unconfined compressive strength of soils with moulding water content

content and thereafter decreased. However, the UCS values of specimens prepared within 13.9 - 18.1% moulding water content range satisfied the minimum 200 kN/m2 requirement for a material to be used in waste containment applications. Such an increase in strength could probably be due to the formation of very weak bonds between the soil particles and the available water molecules (Osinubi et al., 2007) [10]. 3.4. Effect of dry unit weight The variation of UCS with dry unit weight is shown in Figsure 3. Generally, UCS

Unconfined compressive strength (KN/m2)

450 400 350 AB1

300

AB2

250

AB3 200 150 100 50 0 15

15.5

16

16.5

17

17.5

18

18.5

Dry unit weight (KN/m3)

Figure. 3. Variation of unconfined compressive strength of soils with dry unit weight

298

K.J. Osinubi and A.A. Bello / Characterization of Shear Strength of Abandoned Dumpsite Soils

increased to a peak and decreased non-linearly with dry unit weight. UCS values greater than 200kN/m2 were recorded at dry unit weight greater than 16.6 kN/m3. The findings are within the range of values reported by [9, 10]. 4. Conclusion The abandoned dumpsite soils obtained from Orita-Aperin, Ibadan, Nigeria were classified as A-7-6 or lean clay with sand (CL) according to AASHTO and USCS. The soils were compacted using British Standard light energy to determine the effects of compaction water content and dry unit weight on their shear strength when used in waste containment application. Unconfined compressive strength (UCS) values of the soils generally increased to peak values at moulding water content of 16.2%, and thereafter decreased to very low values as water content increased. UCS values recorded in the 13.9 - 18.1% moulding water content range are equal to or greater than 200kN/m 2 minimum acceptable value required for containment structure. Generally, UCS values increased non-linearly to peak values and thereafter decreased with increase in dry unit weight. The study established that UCS values of abandoned dumpsite soils greater than 200 kN/m2 can be achieved when prepared at moulding water content in the range 13.9 – 18.1% and compacted to a dry unit weight greater than 16.6 kN/m3 using British Standard light energy. Consequently, the soils can be used as hydraulic barriers in waste containment applications. References [1] T. B. Edil, K. Sandstrom, and P.M. Berthouex, Interaction of inorganic leachate with compacted pozzolanic fly ash, Journal of Geotechnical Engineering, ASCE, 118 (1992), 1410-1430. [2] R.B. Peck, W.E. Hanson and T.H. Thornburn , Foundation Engineering, John Wiley and Sons, Inc., New York, 1974. [3] T.W. Lambe and R.V. Whitman, Soil Mechanics, SI Version, John Wiley and Sons, New York, 1979. [4] BS1, Methods of testing soils for civil engineering purposes, British Standards Institution, BS 1377 1990, London. [5] ASTM, Natural Building Stones, Soil and Rock, Annual Book of American Society for Testing and Materials (ASTM) Standard 4 (8), Philadelphia. [6] D.E. Daniel and C.H. Benson, Water content-density criteria for compacted soil liners, Journal of Geotechnical Engineering, ASCE, 116 (1990), 1811-1830. [7] AASHTO, Standard Specifications for Transportation Materials and Methods of Sampling and Testing, American Association for State Highway and Transportation officials, 14th Ed. Washington, D.C, 1986. [8] K.J. Osinubi and A.A . Bello, Attenuative capacity of compacted dumpsite soils, Book of Abstracts 9th Nigerian Materials Congress ‘NIMACON 2010’ Enugu, Nigeria, 23 – 26 November, 47 – 48, 2010. [9] C. M. O. Nwaiwu, Evaluation of Compacted lateritic soils as hydraulicbarriers in Municipal solid waste cointainment systems. Unpublished Ph.D, Thesis. Dept. of Civil Engrg., Ahmadu Bello University, Zaria., 2004. [10] K.J. Osinubi, T. S. Ijimdiya and K. Kasai, Evaluation of strength of reconstituted laterite for use as liner and covers, Proceedings of Bi-Monthly Meetings/Workshops, Materials Society of Nigeria, Zaria Chapter, 1 – 8, 2007.

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 299 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-299

The Use of the Crumb Test as a Preliminary Indicator of Dispersive Soils Amrita MAHARAJ CSIR Built Environment, Pretoria, South Africa Abstract. Dispersive soils are prevalent in many areas around the world and the presence of these soils has always posed a serious problem on potential construction sites. The use of dispersive soils in hydraulic and other engineering structures such as roadway embankments can also lead to serious failures if the problem is not properly identified and addressed appropriately. Although the causes and consequences of dispersion are well understood, one of the main problems is the inability to positively identify such soils and thereby to reduce the potential for failure of many engineering structures. Many identification methods have been proposed but none has been completely successful. The primary test methods that are currently used for the identification of dispersive soils are the Pinhole Test; the SCS Double Hydrometer test; the crumb test and various chemical analyses of the soils with the crumb test being the most basic and unsophisticated test to perform. No single test and even the use of a combination of methods are reliable and it is possible that the reason lies in the actual testing procedures. A study involving the collection of various samples and execution of a single standard dispersive laboratory test, namely the crumb test, has identified some shortcomings. This paper discusses some of the various problems identified in the crumb test method and suggests some solutions to overcome them. Keywords: Dispersive soils, dispersion, failure, identification, crumb test, shortcomings

Introduction Dispersive soils are those soils, which when immersed in relatively pure and still water will deflocculate causing the clay particles to go into suspension. These soils are prevalent in many areas around the world and their presence has always posed a serious problem on potential construction sites. The use of dispersive soils in hydraulic and other engineering structures such as roadway embankments can also lead to serious failures if the problem is not accurately identified and appropriately compensated for. Although the causes and consequences of dispersion are well understood, one of the main problems is the inability of existing test methods to positively identify such soils and thereby assist in reducing the potential for failure of many engineering structures. Many identification methods have been proposed but none has been completely successful. The primary test methods that are currently used for the identification of dispersive soils are the Pinhole Test; the Soil Conservation Service (SCS) Double Hydrometer test; the crumb test and various chemical analyses of the soils with the crumb test being the most basic and unsophisticated test to perform. No single test and even the use of a combination of methods are reliable and it is possible that the reason lies in the actual testing procedures.

300

A. Maharaj / The Use of the Crumb Test as a Preliminary Indicator of Dispersive Soils

A study involving the collection of four materials of differing dispersivity and execution of a single standard dispersive laboratory test, namely the crumb test, has identified some inconsistencies in the test method. The effect of different variables on the results of the crumb test was, therefore, investigated and modification of the test procedure to improve test reliability proposed.

1. Background The crumb test, as an indicator for dispersive soils, is the simplest and easiest of the physical tests and was first described by Emerson in 1967 [1]. Emerson [1] found the interaction of clay-sized particles in water to be a major determining factor in the stability of a soil in an agricultural context. Based on this deduction, simple physical tests were devised to qualitatively divide soils into eight different classes. Remoulded soil crumbs were also used in one of the tests to simulate the effect of cultivation on the soil. Samples from a variety of soils were tested and their chemical properties determined for comparative purposes. Figure 1 illustrates the flow chart developed by Emerson for the classification of soils.

Immerse dry aggregates in water

No slaking

Slaking

Complete Dispersion (Class 1)

Some Dispersion (Class 2)

Swelling (Class 7)

No Dispersion

No Swelling (Cla ss 8)

Remould at water content equivalent to field capacity, immerse in water

Dispersion (Class 3)

No Dispersion

Carbonate or gypsum present (Class 4)

Ca rbonate or gypsum a bsent

Make up 1:5 aggregate-water suspension

Dispersion (Class 5)

Flocculation (Class 6)

Figure 1: Flow chart for the classification of soil crumbs (Adapted from Emerson, 1967)

An evaluation of the existing literature has indicated that many researchers appear to misquote Emerson’s work and use his findings incorrectly. There have been many cases in which the method has been misinterpreted with regard to variables such as moisture content and dispersing medium [2],[3],[4],[5]. An ASTM standard is also available for the crumb test (ASTM D6572-00) [6]. The standard, however, takes other

A. Maharaj / The Use of the Crumb Test as a Preliminary Indicator of Dispersive Soils

301

variables such as temperature, which has no effect on the dispersivity of a soil, into account. The standard also calls for remoulding of the sample into a specific size, which again has no effect on the dispersivity. The method mostly followed currently, which can be carried out in the field or a laboratory, involves placing a crumb of soil in a beaker of solution and observing the reaction as the crumb begins to hydrate. The test is primarily used as a visual assessment of the behaviour of the soil as it indicates the tendency of the particles to deflocculate in solution. After a certain time, usually 5-10 minutes, the soil crumb and the solution in the beaker are observed and the soil is classified according to the quantity of colloids in suspension [4],[5]. The soil can be at its in situ moisture content, air dried, oven dried or remoulded before being immersed in the beaker. The solution in which the crumb is immersed is commonly distilled water but a dilute sodium hydroxide solution is also known to be used instead of water. Four grades of dispersivity can be noted ranging from no reaction to strong reaction (Table 1).

Table 1: A table depicting the description of grades for a crumb test (Walker, 1997). Grade

Reaction

Description

1

No reaction

Crumbs may slake, but no sign of cloudiness caused by colloids in suspension

2

Slight reaction

Bare hint of cloudiness in water at surface of crumb.

3

Moderate reaction

Easily recognisable cloud of colloids in suspension, usually spreading out in thin streaks on bottom of beaker.

4

Strong reaction

Colloid cloud covers nearly the whole bottom of the beaker, usually as a thick skin.

A literature search has found that one of the main problems associated with this test is the inconsistency of results. It was also found that no standard protocol regarding variables like immersion solution and condition of crumb is employed. Tests are carried out using dilute NaOH or distilled water and samples are either air dried, oven dried, remoulded or in situ. All of these variables can have significant effects on the outcome of the test and thus the classification of the dispersivity of the soil. Figure 2 illustrates the difference in results obtained when the crumb test was carried out on the same material but with different variables. The first test was carried out on a remoulded crumb in dilute 0.001N NaOH solution (a), the second was carried out on an air dried crumb in distilled water (b) and the third on a remoulded crumb in distilled water (c). One of the consistent observations, however, that has come up many times is the time taken to “run” the test. It is most commonly stated that observations on the dispersivity (or suspension cloud) should be taken 5 to 10 minutes after the crumb is immersed in water. It should, however, be noted that if a soil is dispersive, the colloidal suspension will not settle and will still be present after a few hours. Figure 3 gives an idea of what the colloidal suspension of a dispersive soil should look like after more than an hour.

302

A. Maharaj / The Use of the Crumb Test as a Preliminary Indicator of Dispersive Soils

Figure 2: Comparison of results for same material based on different variables

Figure 3: Colloidal suspension of a highly dispersive soil- Grade: 4- strong reaction

2. Methodology For this study, the crumb test was carried out on four samples, two of unknown classes, one highly dispersive (based on previous work and field performance) and one nondispersive sample (also based on previous work). The test was carried out on each sample using different variables. These variables include condition of the crumb or moisture content and immersion medium. Six tests were carried out on each sample. Two solutions were used, namely distilled water and dilute 0.001N NaOH. For each solution, three conditions of crumbs were tested. The condition of the materials included air dried, oven dried and remoulded crumbs. For the remoulded specimens, air dried samples were crushed and distilled water added until the soil was at a consistency to mould into approximately spherical “crumbs” about 3cm in diameter.

A. Maharaj / The Use of the Crumb Test as a Preliminary Indicator of Dispersive Soils

303

The tests were run for approximately 16hrs with observations made at 10 minutes, 2 hours and 16 hours to evaluate changes in the colloidal suspension over time. In the case of a truly dispersive soil, the colloidal suspension should not settle over time.

3. Results The results of the testing are summarised in Table 2. Table 2: Results obtained for crumb test under different variables. Sample Solution

Crumb condition

Air dried

Distilled Water

Oven dried

Remoulded

Air dried

0.001N NaOH

Oven dried

Remoulded

Time

ZT114

DD

KNP

ND

10 min

4

2

1

2

2 hrs

4

2

1

1

>16 hrs

4

1

1

1

10 min

4

3

1

1

2 hrs

4

2

1

1

>16 hrs

4

2

1

1

10 min

4

3

3

2

2 hrs

4

3

3

1

>16 hrs

4

2

2

1

10 min

3

2

2

1

2 hrs

4

1

1

1

>16 hrs

4

1

1

1

10 min

1

2

2

1

2 hrs

3

1

1

1

>16 hrs

3

1

1

1

10 min

4

4

2

2

2 hrs

4

4

1

1

>16 hrs

4

4

1

1

Results show that after 10 minutes most samples observed would be classified as being dispersive to some degree (Grade 2-4). Settlement of particles began after approximately 30 minutes and the maximum settlement was attained after 2 hours. Figures 4 and 5 graphically illustrate the variation in results obtained at the 10 minute and 16 hour readings.

304

A. Maharaj / The Use of the Crumb Test as a Preliminary Indicator of Dispersive Soils

Figure 4: Variation in classification at 10 minutes and 16 hours in distilled water.

Figure 5: Variation in classification at 10 minutes and 16 hours in dilute NaOH solution.

Figure 6 illustrates the difference in colloidal suspension observed at 10min, 2 hours and 16 hours for a non-dispersive sample. At 10 minutes, the sample can be classified as dispersive, with a classification grade of 3-4. The suspension, however, settles after some time and the 2 hour reading classifies the sample as being nondispersive. This illustrates that the results of a crumb test after 10 minutes are not reliable as some of the fine particles, which are not necessarily dispersive, can still be in suspension. As discussed before, the colloidal suspension of a dispersive soil should not settle over time. It is likely that the inconsistency of results associated with the crumb test is primarily due the time of observation and it is recommended that a minimum waiting period of 1 hour be practiced.

A. Maharaj / The Use of the Crumb Test as a Preliminary Indicator of Dispersive Soils

10 min

2 hrs

305

16 hrs

Figure 6: Illustration of the difference in colloidal suspension observed at 10 minutes, 2 hours and 16 hours.

There are slight differences between the results of samples immersed in distilled water and dilute NaOH. The significant difference occurred between the remoulded crumbs and those that were dried. Remoulding the samples appeared to have the effect of enhancing the dispersive behaviour of the soil. Sample DD demonstrated dispersive behaviour when remoulded and slight or non-dispersive behaviour when dried. The known dispersive sample, ZT114, gave a less dispersive reaction when oven dried and immersed in NaOH solution. Remoulded sample KNP classified as dispersive when immersed in distilled water, as opposed to being non-dispersive in the other tests.

4. Discussion Results obtained from the 6 crumb tests on each of the samples illustrates the variations that can occur due to the lack of a standard protocol for testing dispersivity of soils and leading to differences in classifications. A variety of factors affects the dispersive behaviour of soils resulting in contradictory results, which are likely to pose a problem when faced with the task of treating the soil for construction purposes. Changing the solution in which the crumb is immersed has a significant effect on the results. It is likely that the different solutions have different effects on various soils and carrying out the test using both solutions should provide more useful results. Oven dried samples demonstrated the most inconsistent results. This is due to the fact that the physiochemical properties of the soil pore-water and adsorbed water may be changed when exposed to high temperatures. Remoulded samples showed relatively consistent results since remoulding the samples appears to enhance the dispersive behaviour in the soil. Remoulding the samples also simulates the action of the working and compaction processes on the soil in the field, and is likely to give more realistic results. All samples showed some variation in results due to the different variables. The only sample, however, that was not affected by the different variables was the totally non-dispersive sample (ND).

306

A. Maharaj / The Use of the Crumb Test as a Preliminary Indicator of Dispersive Soils

5. Conclusions Investigations into the crumb test method most commonly used for the identification of dispersive soils have highlighted some of the differences that can be obtained on a single soil, as a function of variations in test procedures. This is due to the numerous variables in the test procedures resulting in different interpretations of the test methods, and consequently misleading results. Research and experience show a number of ambiguities in the test procedures which can be interpreted differently by different laboratories. The major problems observed with the crumb test method have been discussed in this paper and suggestions to overcome them proposed. In light of the results of this study, the current procedures for the identification of dispersive soils by the use of the crumb test should be reviewed and the need for a detailed, simple and repeatable test protocol acknowledged. In order to reduce the variation/inconsistencies in results, it is essential that the test method is reviewed and the optimum procedure developed. The procedure should be simple and have as few ambiguities as possible so that no misinterpretations can occur. Work to improve and standardise the test protocol is currently being extended to include more materials.

References [1] W.W. Emerson, A classification of soil aggregates based on their coherence in water, Australian Journal of Soil Research, 2 (1967), 211-217. [2] R.T. Heinzen and K. Arulanandan, Factors influencing dispersive clays and methods of identification, In Proceedings Symposium on Dispersive clays, related piping and erosion in geotechnical projects, ed. J L Sherard and R S Decker, ASTM Special Publication 623 (1977). 94-109. [3] F.G. Bell and R.R. Maud, Dispersive soils: A review from a South African perspective, Quarterly Journal of Engineering Geology & Hydrogeology 27 (1994), 195-210. [4] D.J.H. Walker, Dispersive soils in KwaZulu-Natal, MSc Thesis, University of Natal, Durban, 1997. [5] F.G. Bell and D.J.H. Walker, A further examination of the nature of dispersive soils in Natal South Africa, Quarterly Journal of Engineering Geology and Hydrogeology 33 (2000), 187-199. [6] ASTM, Standard test method for determining dispersive characteristics of clayey soils by the crumb test, ASTM D 6572-00, ASTM International, Pennsylvania, 2000.

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 307 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-307

Some Engineering Properties of Fine and Coarse Grained Soil Before and After Dynamic Compaction Brian HARRISONa and Eben BLOM b Inroads Consulting cc, Johannesburg, South Africa b Esor Africa (Pty) Ltd., Johannesburg, South Africa

a

Abstract. Over the last 50 years Dynamic Compaction (DC) has been used extensively as a soil improvement technique to densify a range of soil types and fills over considerable depth. The efficacy of the technique is frequently determined by means of in-situ penetrometer and large diameter plate load tests. This paper presents the results of a range of in-situ and laboratory tests for two sites underlain by fine and coarse grained soils, both before and after treatment with DC. Keywords. Dynamic Compaction, Soil Treatment, Soil Properties

Introduction Since Louis Menard first introduced Dynamic Compaction (DC) in the mid 70’s [1], it has been increasingly used worldwide as a means of soil improvement. Briefly, the process entails repeatedly dropping a heavy weight, or pounder, on to the ground from a pre-determined height, on a grid pattern, to densify the underlying soil or fill. The repeated application of this energy enables densification to be achieved to a considerable depth, which is manifested as craters at the ground surface. By densifying the soil in this manner its bearing capacity is improved and post-construction settlements reduced. The technique has been used to densify a wide range of soil types, including silts and clays and a variety of fill deposits, as well as landfill. The efficacy of the technique is most often determined by means of in-situ measurements, mainly penetrometer testing and large diameter plate load tests. This paper presents the results of some laboratory tests and a range of in-situ tests carried out within two sites to determine the effectiveness of DC in improving the stiffness of the compressible soil horizons underlying them. One of the sites was underlain by loose gravels and the other by soft to firm clayey silt.

1. Dynamic Compaction Process Two test sites underlain by compressible soils, and located near to each other, were investigated to establish the degree to which the stiffness of these soils improved after being compacted utilising the DC technique. On both of the sites the DC process was

308

B. Harrison and E. Blom / Some Engineering Properties of Fine and Coarse Grained Soil

undertaken using a 60 ton crawler crane dropping a 13 ton 1,1 m square pounder for the primary and secondary prints. A 12,5 ton 2,4 m sided octagonal pounder was used for the “ironing” phase. The drop height of both pounders was of the order of 18 m. The compaction process entailed concentrating energy at the primary print positions with the square pounder, after which secondary prints were set out and compacted as for the primary prints. After filling the craters formed by the pounders with the surrounding soil, the entire area was provided with an “ironing” phase comprising one blow of the octagonal pounder with a predetermined overlap. The penetration of the pounder into the ground surface forms a crater, and on each site the depth of the crater was recorded after each blow and this is illustrated in Figure 1. The information provides an indication as to the optimum number of blows required to achieve maximum densification for a given pounder mass, shape, drop height and grid spacing. On the two sites between ten and thirteen blows were applied and it was found that after five blows little significant additional penetration of the pounder was achieved. Based on this the number of blows applied for both the primary and secondary compaction phases on both sites were limited to seven.

Figure 1. Typical pounder penetration vs pounder blows on Site 1 and Site 2

2. Soil Conditions on Site 1 2.1. Soil Profile This area is blanketed by moist, soft to firm, clayey silt, of transported aeolian origin to depths of 3 to 4 m. It is in turn underlain by any combination of loose to medium dense nodular ferricrete, dense ferricrete and ferricrete boulders, typically between 1 and 2 m thick. The clayey silt layer again underlies the latter and extends to the bottom of the test pits at 6 m below surface. 2.2. Laboratory Tests Typical indicator properties of the soil are presented in Table 1. According to the Unified Soil Classification [2] (USC) system the soil is an inorganic clay of low to

B. Harrison and E. Blom / Some Engineering Properties of Fine and Coarse Grained Soil

309

medium plasticity, whilst the AASHTO classification [3] groups the soil as a plastic clay. Table 1. Typical mean results of indicator properties of aeolian soil on Site 1 Liquid Limit

Plasticity Index

Grading Modulus

36

14

0,45

Percent Particle Size MIT Classification Gravel Sand Silt Clay 3 33 42 22

Classification Unified CL

AASHTO A-6

Its maximum dry density at modified AASHTO compactive effort measured 1890 kg/m3 at an optimum moisture content of 17 %. Saturated drained triaxial tests were carried out on undisturbed block samples recovered from the test pits excavated in the area, both before and after DC. One cell only was tested for each sample under a confining pressure approximately equivalent to the overburden pressure. The object of the test was primarily to estimate the elastic modulus of the soil. Due to space constraints only the mean results of the 4 tests carried out are presented in Table 2 below. Table 2. Mean results of triaxial tests on aeolian soil at Site 1

Test

Dry density kg/m3

Before After DC DC Drained triaxial 1320 1620 * Secant modulus at 1 % strain.

Degree of saturation % Before After DC DC 73 77

Degree of compaction % Before After DC DC 70 86

Modulus MPa* Before DC 4

After DC 11

The test results show that the average dry density increased by about 22 % after compaction, which in terms of the modified AASHTO effort is an improvement from 70 % to 86 % of the maximum dry density, which is still low. On average the elastic moduli improved almost three-fold from a mean of 4 to 11 MPa. 2.3. In-situ Tests In addition to the laboratory tests, in-situ hand held shear vane and cross-hole plate load tests were carried out at a range of depths in the test pits, and both Dynamic Probe Super Heavy (DPSH) and vertical plate load tests were conducted from surface. The mean modulus of three plate tests, summarized in Table 3 below, also show a threefold increase in the modulus. Table 3. Summary of plate load tests at Site 1 Average modulus from 300 mm dia. Average cross-hole plate tests depth below (MPa) surface (m) Before DC After DC 2,5 3 9

Modulus from 1000 mm dia. vertical plate test (MPa) After DC 14

310

B. Harrison and E. Blom / Some Engineering Properties of Fine and Coarse Grained Soil

The shear vane tests are summarized in Figure 2 below, and show that the average “undrained” shear strength over the depth measured improved from 122 to 136 kPa, an increase of a little over 11 %, which is nominal.

Figure 2. In-situ “undrained” shear strength before and after DC at Site 1.

DPSH tests were done within the test section prior to and after DC in an attempt to determine the depth to which the soil properties had improved, however, the presence of ferruginous gravels, cobbles and boulders resulted in highly variable penetration rates. One of the better comparative plots is illustrated in Figure 3.

Figure 3. In-situ DPSH tests before and after DC at Site 1.

Before DC penetration rates averaged 4 blows per 300 mm up to a depth of about 3,5 m were recorded in the aeolian layer. After DC this improved to an average of 11 blows per 300 mm.

B. Harrison and E. Blom / Some Engineering Properties of Fine and Coarse Grained Soil

311

3. Soil Conditions on Site 2 3.1. Soil Profile This site is blanketed by 2 m thick layer of loose to medium dense nodular ferricrete in a slightly moist, clayey silt matrix below which residual shale occurs as tightly packed, angular, medium and coarse shale gravels also in a slightly moist, clayey silt matrix, and with an overall consistency of loose. It extends to depths of the order of 5 to 6 m below surface where highly weathered shale is encountered as a very stiff, silt. 3.2. Laboratory Tests No undisturbed block samples could be taken of the gravel horizons and thus grading analysis and Atterberg limits only were carried out on a sample of the shale gravel, the results of which are presented in Table 4. Table 4. Typical results of indicator tests on shale gravels at Site 2 Liquid Limit

Plasticity Index

Grading Modulus

37

14

1,93

Percent Particle Size MIT Classification Gravel Sand Silt Clay 58 15 16 10

Classification Unified GC

AASHTO A-2-6

According to the USC and AASHTO systems the soil is described as clayey gravels and granular materials with plastic clay respectively. 3.3. In-situ Tests DPSH penetrometer tests were undertaken before and after DC in an attempt to determine the depth to which densification occurred. The test results are presented in Figure 4 and illustrate that before DC penetration rates up to a depth of 5 m averaged 17 blows per 300 mm and after compaction improved to an average of 40 blows per 300 mm.

Figure 4. In-situ DPSH tests before and after DC at Site 2

312

B. Harrison and E. Blom / Some Engineering Properties of Fine and Coarse Grained Soil

Cross-hole and vertical plate jacking tests were carried out in the test pits and at surface. A summary of the average moduli determined from the tests are presented in Table 5 and illustrates that an almost three-fold improvement in the soil stiffness is achieved through DC. The higher modulus determined from the vertical test could be due to the influence of ferricrete cobbles or boulders in the profile beneath the test position. Table 5. Summary of plate load tests on shale gravels at Site 2 Average modulus from 300 mm Modulus from 1000 mm dia. Average dia. cross-hole plate tests vertical plate test depth below (MPa) (MPa) surface (m) Before DC After DC After DC 2,5 7 18 34

4. Conclusions Based on laboratory and in-situ testing undertaken both before and after DC the clayey silt aeolian horizon underlying Site 1 was considered unsuitable for treatment utilising the DC technique. Although its stiffness improved up to threefold it was not sufficient to limit settlements of structures founded within it to tolerable levels. Better improvement may have been achieved had the soil been drier. Even though partially saturated at 73 % saturation, the pore water pressures generated during the compaction process were sufficient to hamper densification. More complete dissipation would possibly have been achieved by reducing the rate at which the pounder was dropped, or by phasing the compaction to allow the pore water pressures to dissipate sufficiently. From the results of DPSH and plate load testing on site 2, it appears that the gravels compact well and to a sufficient degree and depth to render them adequate for supporting the structural loads on this site. This is despite the fine and plastic nature of the matrix, which is similar to that of the aeolian horizon, but drier. It appears that even if partially saturated, fine grained soils respond best to DC when they are relatively dry and at worst slightly moist.

References [1] L. Menhard, and Y. Broise Theoretical and Practical Aspects of Dynamic Consolidation, Geotechnique Vol. 25 (1976), 3-18. [2] Unified System Classification of Soils for Engineering Purposes, ASTM Designation D-2487 (1967). [3] AASHTO – Classification of Soils and Soil-Aggregate Mixtures, AASHTO Designation M-145 (1970).

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 313 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-313

Using electro-osmosis technique in the improvement of a Ugandan clay soil Denis KALUMBA a, Brenda UMUTONI b, Robinah KULABAKO b and Stephanie GLENDINNING c a University of Cape Town, Cape Town, South Africa b Makerere University, Kampala, Uganda c Newcastle University, Newcastle upon Tyne, United Kingdom

Abstract. Although electroosmotic ground improvement on fine grained soils was pioneered way back in the 1940s, and the technique has been tried in various places around the world, no single investigation of the potential of this innovative method in dewatering and stabilising soft soils has ever been attempted in Uganda. Therefore, the effect of electro-osmosis treatment on a typical Ugandan clay was undertaken. The study focused on evaluating the performance of this process on the shear strength and moisture content of the soil. The experimental study was carried out in a bench scale electroosmotic cell with dimensions: 201 mm by 102.5 mm by 90 mm. The results demonstrated that the electro-osmosis process could significantly enhance reduction of the soil water content resulting in increased soil shear strength. The quality of electroosmotic improvement was enhanced by longer processing time and increased potential gradient. However, the improvement was limited by the drying of the anode area. Keywords. Soil stabilisation, ground improvement, dewatering, electro-osmosis , electrokinetics, shear strength

Introduction The continuous rising population in the East African country of Uganda has resulted in increased construction activity to meet the housing, motorway and industrial development demands (UBOS, 2009). As a consequence, it has become increasingly difficult to find sites with suitable soil properties. In fact, majority of the new projects in cities like Kampala have encountered difficult ground conditions - typically low lying swampy areas comprising soft water-logged clays. Moreover, for construction to proceed in such problematic soils, ground improvement has inevitably been necessary to strengthen the foundation soils so that they are able to support imposed loads. The traditional practice in this part of the world has for long relied upon mechanical means such as compaction and preloading with a surcharge; which of course, are technically inefficient to stabilize and consolidate soft clays in-situ. This is because the time required to drain such material by these conventional methods is excessive. Although never used in Uganda before, fine-grained soils such as clays can effectively be improved by applying an external electric current across the specimen causing several effects, among which is pore fluid movement from one end [anode side] to the other [cathode side] (Kalumba et al. 2009). This process is known as

314 D. Kalumba et al. / Using Electro-Osmosis Technique in the Improvement of a Ugandan Clay Soil

electro-osmosis (Acar, 1992). Electro-osmosis has the advantage of treating fine grained soils without significant excavation or soil disturbance, which is a big benefit over the other methods such as preloading. Besides, this process works best in soils with low hydraulic conductivities like clays (Mitchell, 1993; Stephanie et al., 2006; Kalumba 2006). The procedure involves application of a low direct current (DC) across a moist mass resulting in improved soils engineering properties. The most important of these being a reduction in water content, accelerated consolidation, as well as an increase in shear strength. This study, therefore, was conducted to investigate the potential of electroosmotic stabilisation and consolidation of a typical Ugandan clay thereby presenting an opportunity for its possible application in future soil treatment projects in Uganda.

1. Test Materials and Methods 1.1. Soil Bulk samples of clay were obtained from Busega swamp located along the Northern Motorway by-pass found in the northern suburb of Kampala city. This recently completed motorway passes through extended portions of swamps draining into Lake Victoria. Before construction could proceed, the road footprint had to be preloaded for a whole year to obtain the desired strength and settlement. The redeemed soil was classified as well graded clay with plastic and liquid limits of 17% and 37% respectively. It was grayish brown in colour and had a specific gravity of 1.63, while the natural moisture content was 28.6%. The specimen pH was determined as 3.9. 1.2. Electrodes Electrodes used for both anode and cathode were made from 19 mm wide steel stirrups. The stirrups were cut to dimension and welded together to create an electrode with 3 sleeves as shown in Figures 1(a) and (b). Steel was used for the stirrups due to local availability, while electrode configuration was based on material and electric field optimisation. 1.3. Laboratory Scale Model. In the absence of any relevant standard methods or apparatus, a transparent plastic ‘box’ model was chosen for the study (Figure 1(c)). This rectangular open-to-thesurface configuration was deemed to be representative of the field conditions. The adopted dimensions of the box were 207 mm by 125 mm by 96 mm for the length, width and height respectively. Nine holes, each 8 mm in diameter, were drilled on the cathode end to allow for the effluent passage. 1.4. Test procedure. To ensure the worst case scenario, the moisture content of all test specimens was increased to 60±2% (gravimetric) by thoroughly mixing the required amount of water to the collected samples. For each experiment, the prepared soil was spooned into the

D. Kalumba et al. / Using Electro-Osmosis Technique in the Improvement of a Ugandan Clay Soil 315

test box to a height of 90 mm, tamping it carefully each time to remove any air trapped. Two electrodes were then inserted directly into the soil - one on either side (200 mm apart) - by gently pushing them vertically until they reached the bottom of the box. The electrode on the perforated side of the cell was connected to the negative terminal of the DC power supply (cathode) so that the electroosmotic fluid could easily drain from the cell into the measuring cylinder (Figure 1(d)). Voltage gradients of 0.5, 1.0 and 1.4 V/cm and process durations of 3.5, 7 and 28 days were used in the study. In each test case, the current on the power supply was set to a maximum so as to achieve a current density in the range of 1–10 A/m2. Constant voltage conditions were used in all tests to keep constant net rates of electrolysis reactions at all times (Acar et al. 1992). During treatment, the effluent volume and the current were monitored. At stoppage, the temporal moisture content (MC) along the test cell and the direct shear strength of specimens sampled from the mid-cell were determined. The direct shear tests were carried out in accordance with BS 1377-7:1990 using a 60 mm x 60 mm shear box apparatus with the horizontal displacement rate set as 1.51 mm/min. The relationship between the soil’s moisture content, shear strength and the electric current passing through the soil sample were determined.

Figure 1. (a) and (b) Electrode used; (c) Model Plastic Box; (d) Schematic of Experimental Set up

2. Results Results show that the highest current was measured at the beginning of each experiment (Figure 2). Higher currents were observed for higher voltage gradients and longer experiment duration. The current finally stabilised at 1.5 mA for most of the tests.

316 D. Kalumba et al. / Using Electro-Osmosis Technique in the Improvement of a Ugandan Clay Soil

Figure 2. Variation of current with experiment duration (after Umutoni, 2010)

Figure 3. Variation of moisture content with distance along test cell (after Umutoni, 2010)

D. Kalumba et al. / Using Electro-Osmosis Technique in the Improvement of a Ugandan Clay Soil 317

Figure 5. Variation of friction angle with experiment duration (after Umutoni, 2010)

Figure 6 Variation of cohesion with process duration (after Umutoni, 2010)

Figure 4. Variation of cumulative effluent volume with duration (after Umutoni, 2010)

318 D. Kalumba et al. / Using Electro-Osmosis Technique in the Improvement of a Ugandan Clay Soil

Comparison of the weekly changes in electrical current for all the experimental setups (Figure 2) shows that by the end of the first week, the current had reduced significantly by nearly 90% of the original value. This demonstrates that the largest amount of electrical energy was consumed during the initial stages of the treatment. Beyond the second week, however, there was no further decrease in current. It is evident from Figure 3 that the MC at the different positions of the cells reduced with time. The figures show that the MC reduced from the initial value of 60% at all positions of the cells. The MC was also lower at different positions of the cell for longer experiment durations and higher voltage gradients used. MC was highest closest to the cathode and decreased with distance from the cathode It is clear from Figure 4 that the volume of effluent collected per unit treatment time was greatest at the start of the experiment for all the tests when the measured current was observed to be highest. A continual gradual decrease in effluent volume was observed with time until no more effluent was coming off after the first week. There was increased volume of effluent extracted for experiments with higher voltage gradients. In addition, the longer the treatment time, the greater the cumulative volume of effluent that was collected. There was an 11.9% increase in total cumulative volume of effluent discharged for an increase in voltage gradient from 0.5V/cm to 1.0V/cm and a 19.3% increase for an increase in voltage gradient from 1.0V/cm to 1.4V/cm for the four week experiments. Figures 5 and 6 indicate enhancement in soil shear strength (measured in terms of soil friction angle, φ, and cohesion, c,) with higher voltage gradients and longer treatment times. For instance, for the 1.4V/cm experiment, the specimen friction angle and cohesion increased by 1) 34.7% and 57.8% respectively when the process duration was increased from 3.5 days to one week and, 2) 12.1% and 48.9% respectively when the treatment time was increased from 1 week to 4 weeks. Considering the four weeks period, there was a 17.2% improvement in the friction angle and 68.1% increase in cohesion when the voltage gradient was doubled from 0.5V/cm to 1.0V/cm. Additionally, a 4.3% and 11.1% increase in the friction angle and in cohesion respectively were achieved when the voltage gradient of the cell was increased from 1.0V/cm to 1.4V/cm. During electro-osmosis treatment, the current across the soil sample was highest at the beginning and decreased gradually resulting in the progressive decrease of dewatering flow rate (Figure 2). The current was high initially because of the high ionic concentration. The ionic concentration decreased at later stages as a result of physiochemical changes and the migration of the cations and anions towards their respective electrodes and this migration reduces the current (Kalumba, 2006). The continuous loss of pore water, at the cathode end, as treatment progressed gave rise to drying of the sample as shown in Figure 3. At higher voltage gradients and treatment times, larger volumes of effluent were expelled leading to lower overall MC of the cell. The action of expelling fluids resulted in reduction of the void space within the soil (porosity) which led to densification/consolidation thereby enhancing the soil’s shear strength. Densification initially occurred rapidly at high MC but the rate was lower at later stages of testing when MC and current were lower. Additionally, densification reduced the material permeability, which, constrained the flow paths, and thus inhibited the electroosmotic flow. That is why no more effluent was collected after the soil specimens had reasonably densified. The volume of effluent was higher at the beginning of the experiment and deceased gradually with time until no more effluent was being given

D. Kalumba et al. / Using Electro-Osmosis Technique in the Improvement of a Ugandan Clay Soil 319

off (refer to figure 4). Generally, effluent discharge ceased after 72 hours of process treatment.

3. Conclusions A series of tests were performed to assess the effectiveness of the electro-osmosis treatment on Ugandan clay. Based on the results of the experiments, the following conclusions can be made: • Electro-osmosis can be used to dewater Ugandan clay soils too. This was achieved by application of a suitable electric potential to induced water flow – which confirmed the feasibility of using the technology for this material stabilization. • Significant reduction in the moisture content of clay soil samples was achieved for the investigated voltage gradients of 0.5, 1.0 and 1.4 V/cm applied up to 4 weeks. • For all treated samples, the MC was highest closest to the cathode and decreased with distance from the cathode. • More water was expelled initially and at higher voltage gradients for extended testing times. Consequently, the MC of clays after electroosmotic treatment was lower for larger voltage gradients and higher process periods. • The progressive decrease in the MC corresponded to increased soil shear strength. The attained strengths were higher for higher voltage gradients and times. • Consolidation initially occurred rapidly but the rate was lower at later stages of treatment when the MC and current were lower. .

REFERENCES [1] Y.B. Acar, Electrokinetic soil processing (A review of the state of the art), Grouting/Soil Improvement and Geosynthetics (1992), 1420-1433. [2] Y.B. Acar, H. Li, and R..J. Gale, Phenol removal from kaolinite by electrokinetics, ASCE Journal of Geotechnical Engineering (1992), 1837. [3] S. Glendinning, J. Lamont-Black, and A. Fourie, Dewatering of tailings using electrokinetic geosynthetics - EKG, Proc., 5th ICEG, Environmental Geotechnics, Cardiff, Wales, U.K., Vol II (2006), 878–885. [4] D. Kalumba, Remediation of heavy metal contaminated fine grained soils using electrokinetic geosynthetics, Ph.D. thesis, University of Newcastle upon Tyne, U.K, 2006. [5] D. Kalumba, S. Glendinning, C.D.F. Rogers, D.I. Boardman, and M.A.. Tyrer, Dewatering of tunnelling slurry waste using electrokinetic geosynthetics, ASCE Journal of Environmental Engineering, Vol 135, Issue 11 (2009), 1227-1236. [6] J.K. Mitchell, Fundamentals of Soil Behaviour, 2nd Edition, New York, John Wiley, 1993, [7] B. Umutoni, Investigation into using the electro-osmosis process to improve the shear strength of clay soils in Uganda- case study: Busega swamp, BSc Thesis, Makerere University, 2010. [8] UBOS, Uganda Bureau of Statistics – 2009 Statistical Abstract, http://www.ubos.org/onlinefiles/uploads/ ubos/pdf%20documents/2009Statistical_%20Abstract.pdf, 2009.

320 Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-320

Derived Empirical Relations and Models of Vital Geotechnical Engineering Parameters Based on Geophysical and Mechanical Methods of Testing a

John MUKABIa,1 Kensetsu Kaihatsu Consultants, Nairobi, Kenya

Abstract. ǡ ǡ Ǥ ǡ ǡ ǡ ǡ ǤǡǦ Ǧ Ǥ ǡ Ǥ Ǥ ǡ ǡ ǡǦǡǤ Keywords. ǡǡǡǡǡǡ

Introduction The recent tendency by most geotechnical engineers is to adopt field techniques for the measurement of stiffness due to large error factors that are associated with laboratory testing within the linear elastic range, which various researchers have determined to be very small, averaging for clayey geomaterials [1]. These errors are mainly as a result of sampling disturbance, bedding errors and system compliance. Menzies [2] states that many field tests, however, are unsuitable for measuring soil parameters in the direction of foundation load application whereas plate loading tests simulating the size and loading of real foundations are normally very expensive. This necessitates the employment of geophysical methods which measure the ground property in terms of maximum stiffness as distinct from a ground parameter such as 1

John MUKABI: E-mail: [email protected]

J. Mukabi / Derived Empirical Relations and Models of Vital Geotechnical Engineering Parameters 321

strength which is dependent on the method of measurement. Consequently, establishing a correlation of appreciable precision between the elastic/shear modulus and shear strength for application in modeling ground behavior over a wide range of strain, design and construction control is considered essential. In this study, Geotechnical Investigation (GI) was undertaken at four bridge sites for purposes of determining design parameters for bridge foundations in Juba, the capital of South Sudan. The GI was part of the Juba Urban Transport Infrastructure (JUTI) Study, which was grant aid funded by The Government of Japan, through its implementing Agency, Japan International Cooperation Agency (JICA).

1. Testing and Geotechnical Investigation The Standard Penetration Test (SPT), Dynamic Cone Penetration Test (DCPT), Geophysical Survey and Soil Classification were carried out at 1m intervals. The geoelectromagnetic probing was carried out with two coils; a transmitter and a receiver, which was placed within the borehole vicinity on land as shown in Figure 1.

Figure 1. Sampling, testing and measurement at the bridge sites in Juba, South Sudan

The system is based on the transient electromagnetic method (TEM) sounding technology which enables the conduction of near surface and subsurface soundings to a depth of up to 300m, depending on the geological formation and the frequency applied, whereby lower frequencies sense deeper into the ground. The measured data presented in Figure 3 was determined using a similar method of geophysical survey. Disturbed samples were extruded at 1m intervals or at points where the soil type and/or characteristics varied. This was for purposes of measuring basic physical and mechanical properties such as specific gravity, bulk and dry densities, Atterberg Limits, grading, moisture content and compaction characteristics. The data applied for modeling was determined from sophisticated laboratory tests and field seismic survey.

2. Typical Test Results ʹǦͲͳǤ ȋȌǤ

322 J. Mukabi / Derived Empirical Relations and Models of Vital Geotechnical Engineering Parameters

ĞƉƚŚsƐZĞƐŝƐƚŝǀŝƚǇĨŽƌŽƌĞŚŽůĞϬϭ

WĞŶĞƚƌĂƚŝŽŶĞƉƚŚǀƐ^Wd EͲsĂůƵĞĨŽƌ,ͲϬϭ ^WdEͲsĂůƵĞ

ZĞƐŝƐƚŝǀŝƚǇ ;KŚŵͲŵͿ

Ϭ

ϮϬϬ

ϰϬϬ

ϲϬϬ

ϴϬϬ

Ϭ

ϭϬϬϬ

Ϭ͘ϬϬ

ϮϬ

ϯϬ

ϰϬ

ϱϬ

ϲϬ

Ϯ͘Ϭ

Ϯ͘ϬϬ

ϰ͘Ϭ

ϰ͘ϬϬ

ϲ͘Ϭ

WĞŶĞƚƌĂƚŝŽŶĞƉƚŚ;DͿ

ϲ͘ϬϬ

ĞƉƚŚ;ŵͿ

ϭϬ

Ϭ͘Ϭ

ϴ͘ϬϬ ϭϬ͘ϬϬ ϭϮ͘ϬϬ ϭϰ͘ϬϬ

ϴ͘Ϭ

EхϱϬƚŚĞƌĞĨŽƌĞEK ŵĞĂƐƵƌĞŵĞŶƚƐƚĂŬĞŶ

ϭϬ͘Ϭ ϭϮ͘Ϭ ϭϰ͘Ϭ

ϭϲ͘ϬϬ

ϭϲ͘Ϭ

ϭϴ͘ϬϬ

ϭϴ͘Ϭ ϮϬ͘Ϭ

ϮϬ͘ϬϬ

Figure 2. Comparison of geophysical survey and Standard Penetration Test (SPT) data

3. Derived Basic Empirical Relations The mathematical and regression analysis carried out with consideration to various environmental factors yielded the following equations applicable for ρ >25Ωm. N SPT = A N ln (ρ ) − B N (1)

CBR = A BC ln( ρ ) − B BC (%) qu =

[A

ln( ρ ) − B q

q

(2)

] (MPa)

(3) E

max

= A E ln( ρ ) − B

E

(MPa)

(4)

On the other hand, the relations between the N-value and Unconfined Compressive Strength (UCS) as well as initial stiffness expressed as Emax and UCS are presented in Eqs. (5) and (6) respectively. q u = { [N

E

max

[

SPT

+ B

N

]×

A q / A N } − B q (MPa)

]

= { q u + B q A E / A q } − B E (MPa)

(5) (6)

where, NSPT is the number of blows from a Standard Penetration Test (SPT), CBR is the California Bearing Ratio (a measure of bearing capacity), qu is the Unconfined Compressive Strength (UCS), Emax is the elastic (Youngs’) modulus, ρ is the geoelectromagnetic resistivity, and AN=19.6, BN=48.4, ABC=40, BBC=98, Aq=0.96, Bq=2.37, AE=1024 and BE=280 are material and ground related constants.

4. Application of the Empirical Relations for Design and Modeling Figure 3 depicts the geotechnical engineering parameters derived over a geo-formation thickness of up to 56metres. These parameters were quite useful in determining the appropriate type and mode of piling as well as simulating the interactive behavior.

J. Mukabi / Derived Empirical Relations and Models of Vital Geotechnical Engineering Parameters 323

Figure 3. Computed design parameters over relevant depth of geological formation Note 1: The UCS, qu, the elasctic modulus, E0 and and shear modulus, G0 are measured in MPa.

5. Introducing the GECPRO Empirical Model Rigorous testing can be economically, technically, and time-wise prohibitive and yet, with the increasing necessity for the development of mega structures against the backdrop of the prevalence of large scale destructive natural and manmade disasters, it is practically compelling, on the part of the geotechnical engineer, to determine relatively precise and reliable Value Engineering (VE) based cost-effective design parameters taking into serious account prevalent seismicity, environmental factors and structural sustainability. An appreciably versatile geo-mathematical model (GECPROM) is proposed. GECPROM is designed to probe and estimate changes in vital geo-properties for clayey geomaterials and ground for a wide range of strain (very small strains to large pre-failure strains). The significant advantage of this model is that various geotechnical changes and geo-structural behavior can be modeled from a single sophisticated experimental test, whilst simultaneously catering for the effects of drainage conditions, loading rate, and consolidation stress-strain-time history even in the small strain region. Consequently, as one of the most integral modules of the GECPROM, a method of mathematically determining the Elastic Limit Strain (ELS) within which range the elastic stiffness and shear modulus can be measured or derived more precisely, is developed [Eqs. (7) ~ (11)]. This is extended to determining the subsequent sub-yield strain limits. The importance and application of this method is demonstrated in Fig. 5.

324 J. Mukabi / Derived Empirical Relations and Models of Vital Geotechnical Engineering Parameters

Figure 4. Graphical representation of stress~strain parameters applied for YSL [Mukabi, 1995a]

Considering the intersection properties of the parameters in Figure 4, a square relation is developed as expressed in Eq. (7). a [(ε a )i − (ε a )o ] − a [(ε a )i − (ε a )o ] . Δ q i − Δ q o = E max 2

(7)

The tangent moduli α at point α and β at point β are thence expressed as; α = ( ε )ε α = − ʹ [(ε

)α

− (ε

) ]

[

a E β = (dq d ε a )ε β = E max − 2 a (ε a )β − (ε a )o

•

(8)

]

(9)

(ε ) : Initial Yield Strain (ELS)

(ε a )Y

I

(ε a )β

=

a ⎡ E max − Eα ⎤ − ⎢ ⎥ × ( 2 − E E β )⎥⎦ α ⎣⎢

[(ε

a

)β

(ε a )α ] .

⎛ ⎡ α − β ⎜ (Δ − Δ ) + ⎢ ⎜ ⎢⎣ ʹ (ε )β − (ε ⎜ ⎜ × [(ε ) − (ε ) ]ʹ = ⎝ ʹ [(ε ) − (ε ) ]

{

(10)

−

)α

⎤ ⎥ ⎥⎦

}

⎞ ⎟ ⎟ ⎟ ⎟ ⎠ × ͳͲͲ

(11) where,

(ε )α •

(ε )

=

β

α

× (ε

)Δ , β

. = β ε = Ͳ ǤͲͳ Ψ , α = α

: Secondary and Tertiary Yield Strain

J. Mukabi / Derived Empirical Relations and Models of Vital Geotechnical Engineering Parameters 325

The secondary and the tertiary yield strains, which quantitatively define the intermediate strain boundary limit and the pre-failure large scale plastic strain boundary limit respectively, are computed from the equations proposed by Mukabi [3].

Figure 5. (a) Yield surfaces traced from computed YS and, (b) Zoning of sub-yield surfaces based on computed Yield Strains for Long Term (LT) and Short Term (ST) Consolidated specimens

The basic equation developed from CSSR functions [4] for the GECPROM shear modulus module is expressed as: (12) where,

is the initial shear modulus at a variable stress point

arbitrary or designated consolidation stress ratio traced to

,

is the is the initial shear

modulus determined at in-situ overburden pressure, =0.95 and =0.35 are material constants, the values of which are applicable for most natural stiff and hard clayey geomaterials, while =1.16 and =0.4 for stress states in the 1st quadrant and =-1 for stress states in the 4th quadrant accordingly [3]. In developing the model functions it is important to consider the relativistic rates of change of ͳȓǯǡȔ ǡ α α ǡ ǣ (13) (14) ǯ ǣ (15) Based on the Analytical Function Model (AFM) [3], the following initial subyield zone and shear strength GECPROM functions for simulating various geo-changes,

326 J. Mukabi / Derived Empirical Relations and Models of Vital Geotechnical Engineering Parameters

are derived. I. Stress States ( ) BL The basic generalized equation defining the impact of stress states is expressed as:

(16) ǣ

α

ȋȌ

(17) ʹǤǤ ǡ

(18) ǤǤ ǡ

(19)

, which is indicated in the lower BL of the second component, which defines the interface of the transition and transposition from the end of Ǥ ȋͳͶȌ ǯ Ǣ (20)

II. Consolidation Stress-strain history (

)

(21)

J. Mukabi / Derived Empirical Relations and Models of Vital Geotechnical Engineering Parameters 327

where

=1.9×10-3 is a strain level dependent constant,

is the secondary

consolidation factor, is the overconsolidation factor and is the initial yield strain determined under normally consolidated conditions at a standard time period designated after the end of primary consolidation [5]. III. Drainage conditions (

)

(22) where threshold of

are the effective axial and radial stresses respectively determined at the is the stress ratio during consolidation, =-1,

=+1 ( : drained and : undrained) and is the Poisson’s ratio. For perfectly drained conditions 0.2 and 0.5 for perfectly undrained state. IV. Cyclic prestraining ( ) (23) V. Strain rate (24) where, the subscripts SR denote Strain Rate, ASR: Applied Strain Rate during testing or arbitrary designation and RSR: Reference Strain Rate. VI. Deriving shear strength from elastic stiffness and introducing some modeled results Developing an appreciably reliable correlation between shear strength at failure and the elastic (Initial) stiffness is important in determining a more precise parameter (qmax) closely related to a ground property (E0). The GECPROM module for this relation is: (25)

Figures 6 and 7 depict the experimental and modeled characteristic curves for small stress ~ strain behavior and the decay of stiffness plotted as a function of strain level respectively. Figure 7(b) includes curves that predict ageing effects on the characteristics of the decay curves. The results conform to the tendency of experimental data reported by various researchers [1] [6]. In these figures all the curves modeled by the GECPRO show a very good agreement in comparison to the experimental trends. It can also be inferred that the curves for the reconstituted specimens deviate significantly from those of the intact ones in all cases.

328 J. Mukabi / Derived Empirical Relations and Models of Vital Geotechnical Engineering Parameters

Figure 6 Comparison of modeled and experimentally measured relations for intact and reconstituted Overconsolidated (OC) OAP clay from Osaka, Japan, for (a) small stress ~ strain characteristics ( and, (b) very small stress ~ strain characteristics (

Figure 7 Comparison of modeled and experimentally determined strain level dependency of stiffness decay curves for Overconsolidated (OC) OAP clay from Osaka, Japan, for (a) Intact and reconstituted specimens and, (b) Prediction curves for ageing effects for intact and slightly disturbed specimens

It can be further derived that the elastic modulus measured from field seismic survey, plotted in Figure 7(b) confirms the reliability of the laboratory experimental data. ǡ Ǥ ȋʹ͸Ȍ ȋʹ͹ȌǤȋͳ͸Ȍ̱ȋʹͶȌǤ ȋʹ͸Ȍ

ǡ

ǡ

ȋʹ͹Ȍ αͲǤͲͲͳΨ αͲǤͲͳΨǤ

J. Mukabi / Derived Empirical Relations and Models of Vital Geotechnical Engineering Parameters 329

6. Conclusions Vital geotechnical engineering parameters were determined from experimental testing based on geophysical surveys and mechanical methods of testing. The main conclusions that can be drawn from this study include the following. 1.

2.

3. 4.

The proposed empirical relations for bearing capacity, strength and elastic modulus (initial stiffness) may be useful in correlating design parameters for versatility, enhanced precision and actual ground simulation. As demonstrated in this paper, the method of quantitatively determining the zonal sub-yield strain magnitude proposed in this study is not only important in modeling but can also be useful in the prediction of the magnitude of ground movement under loading and construction control. The proposed GECPROM is versatile and appreciably effective in probing, simulating, modeling and predicting geotechnical changes in ground. Remolding (reconstitution) completely destroys the intrinsic structure of structured clays. Theories and concepts based on reconstituted clay behavior are therefore inadequate in modeling the behavior of natural clays.

References [1] Mukabi J.N, Deformation Characteristics at Small Strains of Clays in Triaxial Tests, PhD Thesis, Institute of Industrial Science, University of Tokyo, Tokyo, Japan, 1995. [2] Menzies, B.K., Near-surface site characterisation by ground stiffness profiling using surface wave geophysics," Instrumentation in Geotechnical Engineering. H.C.Verma Commemorative Volume, Oxford & IBH Publishing Co. Pvt. Ltd., New Delhi, Calcultta. pp 43-71, 2001. [3] Mukabi J.N., Characterization and Modeling of Various Aspects of Pre-failure Deformation of Clayey Geomaterials, to be published. [4] Mukabi J.N., Kotheki S. (2010a) - Mathematical Derivative of the Modified Critical State Theory and its Application in Soil Mechanics. Procs. 2nd International Conf. on Applied Physics & Mathematics, 2010 IACSIT, Kuala Lumpur, Malaysia. [5] Mukabi J.N, Tatsuoka F. (1999c) - Effects of stress path and ageing in reconsolidation on deformation characteristics of stiff natural clays. Proc. 2nd I.S Pre-failure characteristic of geomaterials, Torino, Italy. [6] Tatsuoka F., Jardine R.J., Presti D.L., Benedetto H. D. Kodaka T. ( 2000) – Characterising the pre-failure deformation of geomaterials. XIV ICSMFE in Hamburg, Theme Lecture.

330 Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-330

Quantitative Analysis to Verify the Theory of Soil Particle Agglomeration and Its Influence on Strength and Deformation Resistance of Geomaterials Sirmoi WEKESAa,1, John MUKABI b, Vincent SIDAIc, Sylvester KOTHEKId, Joram OKADOe, Julius OGALLOf, George AMOYOg and Leonard NGIGIh a,b,d,e,g&h Kensetsu Kaihatsu Consultants, Nairobi, Kenya c Bamburi Cement-The Lafarge Group, Nairobi, Kenya f Department of Civil Engineering, University of Nairobi, Nairobi, Kenya Abstract.

Soil particle agglomeration is one of the most important characteristics required in geomaterials for their application in the construction of geo-structures. Fundamentally, when soil particles are subjected to secondary (creep) or long-term consolidation within confined boundary conditions, they undergo mechanical changes of their intrinsic properties as a result of ageing. In the case of clays this may alter their minerals as in weathering, changes in concentration of ions in pore water, and/or replacement with ions of different valance, precipitation, cementation, mineral leaching, internal erosion and time related secondary consolidation associated with creep (∂ε‘a/∂t=0) and thixotropy basically defined as a gain in strength at constant water content. In the soil mechanics perspective of agglomeration mechanisms and matrix of soil particles, it is considered that the effects of these processes will have limiting boundary values with time tending to a residual state as the material densification is enhanced. On the other hand, Lime and/or cement stabilization of soils is principally based on ion exchange capacity, pozzolanic reaction, carbonation and suction/swelling effects through heat of hydration. Nevertheless, hardly any research has been comprehensively carried out regarding the quantitative analysis of such agglomeration mechanisms. By applying data from innovative experimental testing and research, this Study undertakes quantitative analysis to verify the theory of soil particle agglomeration by comparing the strength and deformation resistance characteristics for geomaterials with cementing agents of varying intrinsic properties, particle size, texture, shape, and enhanced mechanical stabilization based on optimum batching and/or through soil particle interaction with reinforcing systems such as geogrids.

Keywords. Soil, particle, agglomeration, quantitative, stabilization, ageing

Introduction Under this relatively new topic, research has been initiated to geo-scientifically characterize soil agglomeration mechanisms and their contribution to the development or retardation in strength, bearing capacity and deformation resistance quantitatively 1

John MUKABI: E-mail: [email protected].

S. Wekesa et al. / Quantitative Analysis to Verify the Theory of Soil Particle Agglomeration

331

defined in terms of elastic limit strain and stiffness. In this Study therefore, the fundamental theory of particle agglomeration is verified through quantitative analysis by adopting concepts that are mainly related to cementation, ageing and stabilzation. The theory of agglomeration fundamentally postulates that the soil particle, in the initial state of reference, is extremely small, globally confined in a finite space, but locally suspended independently in an infinitely small space. The physio-chemical changes of the particles are kinematically dictated by the centre of mass and gravity, moment of inertia, energy system and Quantum Gravitational (GQ) forces within a relativistic framework. The initial state of the soil particles can be defined by the Theory of Suspended Particles [1]. However, [2] noted that basically, at any given time, the theory is only applicable when agglomeration occurs in the absence of external forces and is only dependent mainly on time related cementation and localized physiochemical changes that may ultimately culminate in solidification tending to rocky like crust formation. Furthermore, the theory considers that within finite boundaries, there is a cyclic and sequential recurrence of agglomeration and subsequent dispersion resulting in further agglomeration that forms larger particles, reduces the voids and fluid action leading to the increase in effective stresses, strength and deformation resistance of the composite and/or confined geomaterials.

1. Main Considerations for the Testing Regime and Analysis The testing regime was designed such that particle agglomeration could be adequately monitored, evaluated and quantitatively analyzed. The analytical concepts are basically formulated to define the mechanism through the Initial State → Agglomeration → Dispersion → Cluster Formation, whilst considering micro-scale effects, kinematic changes, probabilistic functions and phase transformations. Retrospective prediction and correction analysis are undertaken within the framework of numerical analysis. For the sake of simplified analysis, it is assumed that within any given time and space, the particles will maintain a uniaxial state of motion towards agglomeration and that transformation within the system will not occur at that particular point.

2. Test Results and Analysis Figure 1 depicts the relation between the Consolidation Stress Ratio (CSR) function, δCSR and the degree of soil particle agglomeration defined in terms of the magnitude of cementation. It can be noted that the degree and rate of particle agglomeration is dependent on the type of geomaterial and magnitude of cementation. This may implies, by inference that the rate and mode of agglomeration varies with the intrinsic properties and degree of Mechanical Stabilization (MS) of the geomaterial. This fact can also be derived from Figure 2. The CSR function, δCSR is defined in Eq. (1) as proposed by [3].

δ CSR =

φ ' − Bβ

′ φCalculated

; Aβ 㧩0.279, B β =14.6 and Aβ = 0.31δ CSR + 14.95 (1)

332

S. Wekesa et al. / Quantitative Analysis to Verify the Theory of Soil Particle Agglomeration

ŽŶƐŽůŝĚĂƚŝŽŶ^Z&ƵŶĐƚŝŽŶ͕ɷ ^Z

ϭϮϬ ϭϬϬ ϴϬ ϲϬ

,ŝŐŚůǇtĞĂƚŚĞƌĞĚ ^ŽĨƚ^Žŝů

ϰϬ

sŽůĐĂŶŝĐ ZĞĚƐŚ

ϮϬ Ϭ

Ϭ Ϯ ϰ ϲ ϴ ĞŐƌĞĞŽĨ ŐŐůŽŵĞƌĂƚŝŽŶĂƐĂ&ƵŶĐƚŝŽŶŽĨĞŵĞŶƚĂƚŝŽŶ;йͿ

Figure 1. Dependency of degree of soil particle agglomeration on type of geomaterial defined in terms of consolidation stress ratio function, δCSR

It can also be inferred from these Figure 1 that as soil particle agglomeration progresses, the shear properties are enhanced notwithstanding the type of geomaterial. On the other hand, Figure 2 shows the effect of mechanical stabilization on the deformation resistance defined in terms of elastic stiffness. As can be noted, the elastic modulus increases as the Optimum Batching Ratio (OBR) tends towards an optimum value. These results are consistent with the Cyclic Prestraining (CP ) models introduced by [4], which are presented in Figures 3(a) and (b).

ůĂƐƚŝĐ^ƚŝĨĨŶĞƐƐ͕ŵĂǆ;Ě͘EͬĐŵϮͿ

WĂƌƚŝĐůĞŐŐůŽŵĞƌĂƚŝŽŶĨĨĞĐƚYƵĂŶƚŝĨŝĞĚŝŶdĞƌŵƐŽĨ DĞĐŚĂŶŝĐĂů^ƚĂďŝůŝǌĂƚŝŽŶĂŶĚůĂƐƚŝĐDŽĚƵůƵƐ

DŽĚĞͬĞŐƌĞĞŽĨDĞĐŚĂŶŝĐĂů^ƚĂďŝůŝǌĂƚŝŽŶ Figure 2. Particle agglomeration effect quantified in terms of mechanical stabilization and Emax

Basically, the models consider that CP may enhance the inherent microstructure linear elastic and recoverable properties, as well as the deformation resistance and shear strength; or otherwise damage or even completely destroy the soil structure (microstructure, fabric) of well structured and cemented geomaterials. In other words, application of low amplitude, and a low number of loading cycles can initially enhance the geotechnical engineering properties of clayey geomaterials, whereas further

S. Wekesa et al. / Quantitative Analysis to Verify the Theory of Soil Particle Agglomeration

333

escalation of CP past a certain threshold results in the progressive damage of the agglomerated particles eventually de-structuring and destroying the intrinsic structure.

Figure 3. Cyclic Pre-straining Model for; (a) Initial Young’s/ Shear modulus and, (b) Initial Yield Surface

ŐŐůŽŵĞƌĂƚŝŽŶh^;DWĂͿ

The quantitatively analyzed effects of soil particle agglomeration on the Unconfined Compressive Strength (UCS) for varying modes of stabilization are depicted in Figure 4. The results were adopted in the detailed design of the runway pavement, aprons, taxiway and parking for the Isiolo Airport located in the North Eastern Province of Kenya [5].

ϰ͘ϱϬ ϰ͘ϬϬ ϯ͘ϱϬ ϯ͘ϬϬ Ϯ͘ϱϬ Ϯ͘ϬϬ ϭ͘ϱϬ ϭ͘ϬϬ Ϭ͘ϱϬ Ϭ͘ϬϬ

EŶŚĂŶĐĞĚ ^ƚƌĞŶŐƚŚ ĚƵĞ ƚŽ WĂƌƚŝĐůĞ ŐŐůŽŵĞƌĂƚŝŽŶĨŽƌsĂƌǇŝŶŐDŽĚĞƐ ϳĂǇƐƵƌĞͬ^ŽĂŬ ϯĂǇƐƵƌĞͬ^ŽĂŬ ϮϴĂǇƐƵƌĞͬ^ŽĂŬ

EĞĂƚKZDKWDKWDн'ĞŽŐƌŝĚ DŽĚĞŽĨ^ƚĂďŝůŝǌĂƚŝŽŶ

Figure 4. Enhanced strength as a result of particle agglomeration for varying modes of stabilization Notes: OBRM: Optimum Batching Ratio Method; OPMC: Optimum Mechanical and Chemical Stabilization

It can be noted from this figure that; a) the UCS is significantly enhanced as the soil particles agglomerate progressively with time particularly for the cemented geomaterial, b) soil particle agglomeration is dependent on the mode of stabilization, c) the effects of geogrid reinforcement are more noticeable with increased agglomeration possibly due to the influence of confinement in increasing the degree of interlocking of particles culminating in reduction of voids ratio and, d) the degree of soil particle agglomeration is higher under OPMC stabilization in comparison to geogrid reinforcement thus confirming the theory that, in comparison to other influencing factors, cementation processes have greater impact on agglomeration mechanisms .

334

S. Wekesa et al. / Quantitative Analysis to Verify the Theory of Soil Particle Agglomeration

Figure 5 shows the coupled effects of degree of cementation and soaking on the agglomeration of particles with respect to strength for different types of OPMC stabilized geomaterials, whilst the effects of ageing as defined by [6], on the same, are depicted in Figures 6 (a) and (b) for varying pavement structural configurations. The effects of soaking are seen to be virtually insignificant on soil particle agglomeration.

Figure 5. Effect of cementation and soaking on the soil particle agglomeration of varying geomaterials quantified in terms of shear strength [Mukabi and Kotheki, 2010d]

Figure 6. Influence of particle agglomeration on strength and deformation resistance as a result of time dependent cementation and consolidation for; a) UCS and, b) Elastic modulus Notes: Pavement structure constructed in swampy areas with expansive Black Cotton Soil for; 1) Type II-1: without OPMC stabilization and without subgrade improvement, 2) Type II-2: with OPMC Level 3 stabilization and OBRM stabilization for Base/Subbase and partial subgrade improvement and, 3) with OPMC Level 5 and OBRM stabilization for Base/Subbase and also with well improved subgrade.

The existence of soil particle agglomeration as a result of cementation and ageing and its’ ability to enhance engineering properties of geomaterials can be quantitatively verified from Figures 5 and 6 respectively. However, it can be noted that these effects are more significant on the elastic modulus than the peak shear strength. This phenomenon can be attributed to the fact that the range of the Initial Yield Surface defined as the region of linear elastic and recoverable behavior increases significantly with ageing hence yielding at much higher stress levels resulting in higher ratio of elastic modulus when compared with shear strength as reported by [4] and confirmed in

S. Wekesa et al. / Quantitative Analysis to Verify the Theory of Soil Particle Agglomeration

335

this Study. As can be further noted from Figure 6, soil particle agglomeration occurs at a much higher rate in the initial period of ageing eventually tending to a residual state with increased progression of time. This phenomenon verifies the theory of soil particle agglomeration introduced by [2] from the perspective that, with time, the soil particles form larger clusters through solidification, tending to exhibit minimal particle movement as the clustered mass tends towards an inert state.

^ƚƌĞŶŐƚŚĚĞǀĞůŽƉŵĞŶƚĂƐĂZĞƐƵůƚŽĨsĂƌǇŝŶŐ

h^;DƉĂͿ

Ϯ͘ϬϬ ĞŵĞŶƚĂƚŝŽƵƐ^ŽŝůWĂƌƚŝĐůĞŐŐůŽŵĞƌĂƚŝŽŶ ϭ͘ϱϬ ϭ͘ϬϬ

WŽǌǌŽ WŽǁW

Ϭ͘ϱϬ

WŽǁD

Ϭ͘ϬϬ Ϭ

ϭ

Ϯ

ϯ

ϰ

ϱ

ϲ

ĞŵĞŶƚĂƚŝŽƵƐ^ŽŝůWĂƌƚŝĐůĞŐŐůŽŵĞƌĂƚŝŽŶ;йͿ

ϳ

Figure 7. Influence of cementation properties on the rate of gain in strength due to particle agglomeration Notes: Legend Abbreviation of cement type: 1) Pozzo: Pozzolanic; 2) PowP: Power Plus; 3) Power Max

The degree and influence of cementation properties on the rate of gain in strength due to soil particle agglomeration is shown in Figure 7. The variation in the mode of agglomeration due to qualitative and quantitative changes in cementation properties can clearly be inferred. 3. Conclusion The Theory of Soil Particle Agglomeration has been further verified and its’ contribution to the enhancement of strength, and deformation resistance confirmed by undertaking quantitative analysis in this Study. The results have also shown that soil particle agglomeration is dependent on the intrinsic properties of geomaterials, degree and influence of cementation properties, consolidation characteristics, period of ageing, and mode of stabilization. References [1] Mukabi J.N, Application of the Mechanics of Tropical soils in Geotechnical Engineering: Some Recent Advances and Introduction of Futuristic Concepts on Geoscientific, Global Energy Conservation Systems GECS, Proceedings. of IEK International Conference (2008), IEK, Nairobi, Kenya. [2] Mukabi J.N, Kotheki S., Proposed theory of soil particle agglomeration and profound reciprocal effects on the behavior of composite Geomaterials, Proceedings of the 2010 IEEE International Geoscience & Remote Sensing Symposium (2010), Hawaii, USA.

336

S. Wekesa et al. / Quantitative Analysis to Verify the Theory of Soil Particle Agglomeration

[3] Mukabi, J.N., F. Tatsuoka, Influence of Reconsolidation Stress History and Strain Rate on The Behaviour of Kaolin Over a Wide Range of Strain, Proceedings 12th ARC Geotechnics for Developing Africa pp. 365 (1999), Durban, South Africa. [4] Mukabi J.N., Characterization and Modeling of Various Aspects of Pre-failure Deformation Clayey Geomaterials, to be published. [5] Kensetu Kaihatsu Limited, Isiolo Airport Pavement Design, Engineering Report No. ISAT 0211/O2 (2011), submitted to Kenya Airports Authority, Government of Kenya, Nairobi, Kenya. [6] Leroueil S., Vaughan P.R, “The general and congruent effects of structure in natural soils and weak rocks,” Geotech., 40, No.3, pp.467-488 (1990).

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 337 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-337

Characterizing bulk modulus of finegrained subgrade soils under large capacity construction equipment Joseph ANOCHIE-BOATENG1 CSIR Built Environment, Pretoria, South Africa

Abstract. This paper focuses on characterizing the volumetric stiffness behavior of fine-grained subgrade soil at three different moisture states using a newly proposed hydrostatic compression test procedure. The deformation properties obtained from a laboratory testing program were used to determine bulk modulus at varying hydrostatic stress states, and moisture states chosen at optimum moisture content, 3% below and 3% above the optimum. The test results are analyzed, and used to develop regression correlation models for the soil sample tested. These models can be used for evaluating the impact of moisture on bulk modulus of fine-grained soils with similar characteristics for their sustainable use in foundation applications under off-road construction and compaction equipment. Keywords. Bulk modulus, hydrostatic stress, subgrade soils, construction equipment,

Introduction The routine operations of large capacity off-road construction equipment on finegrained cohesive soils have become a concern to the construction and equipment manufacturing sectors. A major problem is the mobility (trafficability) of large haul trucks and shovels during field operations on these soils. Cohesive fine-grained and cohesionless granular soils constitute the foundation of highway and airport pavements as well as railroad track. These materials would exhibit different load bearing capacities at different stress and moisture states under construction equipment. To understand behaviour of these foundation materials under large capacity construction and compaction equipment it is important to properly address the true volumetric deformation characteristics under all-around uniform normal stress conditions. Bulk modulus is an important material property that describes the resistance to volume change when an element of soil is subjected to all-around hydrostatic loading [1]. In this paper, bulk modulus is determined in the laboratory for a selected finegrained cohesive soil using a newly developed hydrostatic compression test procedure [2]. The test procedure considers field loading characteristics of off-road construction haul trucks and shovels to determine bulk modulus at varying hydrostatic stress states, and moisture states chosen at the optimum moisture content, 3% below and 3% above 1

Corresponding Author. Senior Researcher, CSIR Built Environment, Transport Infrastructure Engineering, Bldg 2C, P O Box 395, Pretoria, 0001, South Africa; E-mail: [email protected]

338

J. Anochie-Boateng / Characterizing Bulk Modulus of Fine-Grained Subgrade Soils

the optimum. The bulk modulus together with shear modulus will be used to obtain the elastic modulus and Poisson’s ratio of the soil sample.

1. Laboratory Testing Program 1.1. Properties of Soil Sample The fine-grained cohesive soil investigated in this paper was obtained from Caterpillar Inc. field demonstration test sections in Illinois, and was shipped to the University of Illinois Advanced Transportation Research and Engineering Laboratory (ATREL) for testing. The sample was a clayey soil, a “CL” according to the United States unified soil classification system or an “A-6” according to the American Association of State Highway and Transportation Officials (AASHTO) classification, with a liquid limit (LL) of 27.2, a plasticity index (PI) of 13.1, and composition of 0.3% gravel, 29.5% sand, 40.9% silt, and 29.3% clay. Accordingly, the soil sample was designated herein as SA-6. From the standard Proctor [3] test procedure, the maximum dry density obtained was 18.4 kN/m3 at an optimum water content (wopt) of 14.3 %. The dry densities at the moisture states of 3% below and 3% above the optimum were 18.0 kN/m3, and 17.6 kN/m3, respectively 1.2. Laboratory Testing Procedure The loading characteristics of off-road large capacity construction and compaction equipment dictate field loading stress states and therefore directly influence the volumetric deformation and stiffness behavior of soils in the field. For instance, Joseph [4] noted from field studies that a Caterpillar 797B off-road haul truck could produce vertical stresses of about 800 kPa with confining stresses ranging between 250 and 300 kPa. He also observed that the P&H 4100 type BOSS shovels generated a static vertical loading of up to 220 kPa, and could induce a ground confinement of about 70 kPa [4]. It is expected that the soil would undergo anisotropic loading conditions. However, under the laboratory triaxial testing conditions (σ2 = σ3), thus, an isotropic conditions was used in this study to simulated the volumetric hardening of the SA-6 soil. An innovative advanced triaxial testing device, the University of Illinois FastCell (UI-FastCell) integrated with an Universal Testing Machine (UTM) loading device at ATREL could be used to achieve the field loading conditions. The UI-FastCell offers unique capabilities in laboratory material characterization including measurement of on-sample vertical and radial displacements, and a bladder type horizontal confinement chamber with a built-in membrane which can be inflated to apply hydrostatic stresses to simulate high field loading conditions on granular and bituminous materials in the laboratory [5]. Figure 1 shows the UI-FastCell test setup.

J. Anochie-Boateng / Characterizing Bulk Modulus of Fine-Grained Subgrade Soils

339

Figure 1. UI-FastCell test setup.

1.3. Hydrostatic Test Procedure The UI-FastCell was used for applying hydrostatic stresses on the fine-grained soil specimens. The hydrostatic compression test was conducted on 150 mm diameter by 150 mm high pneumatic vibratory compacted specimens. During testing, compacted soil specimens were subjected to a sequence of different applied hydrostatic (isotropic) compression stresses of 20.7, 41.4, 69, and 138 kPa under drained conditions, with volumetric change measurements. Specimens were loaded from zero stress conditions to these individual hydrostatic stresses, unloaded to zero, and then, reloaded to the next stress state until the maximum hydrostatic stress of 276 kPa was reached (i.e., 0 ψ 20.7 kPa ψ 0 ψ 41.4 kPa ψ 0 ψ 69 kPa ψ 0 ψ138 kPa ψ 0). A pulsed wave shape with 60-second loading and 60-second unloading was applied on the test specimens at each stress state. The loading rate was maintained in such a way that no pore pressure was induced. The axial static loading was controlled by the vertical load cell, and the radial loading was measured by a pressure transducer. To achieve isotropic condition, the UTM software was adjusted to ensure that equal radial and vertical loads were applied to the sample. Both axial and radial deformations were measured by two symmetrical linear variable displacement transducers (LVDTs) for each load cycle and the corresponding axial and radial strains (ε 1 and ε3) are computed for the test specimens. Two replicate tests were performed for the soil sample at three moisture states of 11.3%, 14.3% and 17.3%. Overall, 12 tests were conducted on the soil sample at the three moisture conditions.

2. Analyses of Test Results The applied hydrostatic stresses and measured volumetric strains obtained from hydrostatic compression tests are used to calculate bulk modulus. A plot of the applied

340

J. Anochie-Boateng / Characterizing Bulk Modulus of Fine-Grained Subgrade Soils

isotropic compression stress against volumetric strain gives a nonlinear curve for soils [6-8]. Vesic and Clough [7] suggested that the soil’s elastic properties could conveniently be obtained from the nonlinear curve by straight line approximations that linearly relate increments of both the isotropic stress and volumetric strains. In this study, the straight line approximation concept was used for analyzing the test results of the samples. The bulk moduli K of the soil sample was calculated from the ratio of the incremental hydrostatic stress Δσ to the incremental volumetric strain Δεv. Eq. (1) was used to define the bulk modulus of the soil sample tested: Δσ1 + Δσ 2 + Δσ3 Δσ = Δεv Δεv

K=

(1)

where the volumetric strain εv is computed from the axial strain ε1 and radial strain ε3 as εv = ε1+ 2ε3; for triaxial compression tests, hydrostatic stress is given by σ = σ1 = σ2 = σ3. A total of about 270 stress-strain data sets for each test were analyzed for the bulk modulus of the soil sample at one moisture state. Each data set represents an average value from the two replicate specimens. Figure 2 shows a plot of the applied hydrostatic stress against the total volumetric strain for SA-6 soil sample at the three moisture states. It can be demonstrated from figure 2 that the behaviour of the SA-6 soil could be linear (i.e., constant K) up to a hydrostatic stress of about 80 kPa, and therefore, complying with Eq. (1). However, it can be seen from the figure that above certain threshold hydrostatic stress states, the bulk modulus–hydrostatic stress relationship presented in Eq. (1) for soils may not necessarily be applicable under certain conditions. The straight line approximation was used to obtain the incremental hydrostatic stresses and corresponding volumetric strains. The bulk modulus was then computed at each hydrostatic loading stress using Eq. (1). 150 w = 11.3%

Hydrostatic Stress, kPa

120

wopt = 14.3% w = 17.3%

90

60

30

0 0

2

4

6

Volumetric Strain, % Figure 2. Variation of stress with strain.

8

10

341

J. Anochie-Boateng / Characterizing Bulk Modulus of Fine-Grained Subgrade Soils

Table 1 lists test results of the SA-6 soil at the three moisture states. As expected, the soil sample at dry of optimum gave the highest bulk modulus values whereas the lowest bulk modulus values were obtained at wet of optimum. The average bulk modulus value increases by 1.1 MPa from optimum (dry density of 18.3 kN/m3) to dry of optimum (dry density of 18.0 kN/m3), and decreases by an average of 1.2 MPa from optimum to wet of optimum (dry density of 17.6 kN/m3). Thus, a change in water content of 3% below the optimum resulted in about 38% increase in the bulk modulus of the soil sample, whereas a change in water content of 3% above the optimum resulted in about 42% decrease in the modulus values. The high lubrication of soil particles at wet of optimum water content weakens the soil sample. Therefore, the modulus of the sample becomes low at wet of optimum when compared to dry of optimum, or the soil becomes less sensitive at dry of optimum. Figure 3 shows the correlations between bulk modulus as linear functions of hydrostatic stress for the soil sample at the three moisture states. The significantly high coefficients of correlation values indicate that the straight line incremental approximation concept (Eq. 1) used for the analyses performed well for the SA-6 sample at all three moisture states. Table 1. Test results for SA-6 soil at three moisture states. w =11.3%

Δσ (kPa)

wopt = 14.3%

w = 17.3%

Δεv (%)

K (MPa)

Δεv (%)

K (MPa)

Δεv (%)

K (MPa)

20.7 – 41.4 (20.7)

0.68

3.15

0.88

2.43

1.35

1.58

41.4 – 69.0 (27.6)

0.62

4.45

0.82

3.36

1.65

1.67

69.0 – 138.0 (69.0)

1.00

6.90

1.60

4.31

2.60

2.65

Bulk Modulus K (MPa)

8.0

w = 11.3%

7.0

wopt = 14.3%

6.0

w = 17.3%

y = 0.0717x + 2.0128 R² = 0.9483

y = 0.0335x + 2.0534 R² = 0.8486

5.0 4.0 3.0 2.0

y = 0.0229x + 1.0697 R² = 0.9979

1.0 0.0 0

30 Hydrostatic Stress s (kPa)

60

Figure 3. Correlations between bulk modulus and hydrostatic stress at three moisture states.

90

342

J. Anochie-Boateng / Characterizing Bulk Modulus of Fine-Grained Subgrade Soils

3. Summary and Conclusions Hydrostatic triaxial compression tests were performed on a fine-grained cohesive soil sample in the laboratory using a newly developed hydrostatic loading test procedure. The laboratory tests were performed to determine bulk modulus at three moisture states of 11.3%, 14.3% and 17.3%, representing dry of optimum, optimum and wet of optimum, at dry densities of 18.0 kN/m3, 18.3 kN/m3 and 17.7 kN/m3, respectively. The test procedure applies low to high hydrostatic stress levels on the specimens to simulate the laboratory loading behavior of fine-grained soils under construction and compaction equipment. Moisture content affected the bulk modulus properties of the soil sample as it was evident that at the high moisture state, the sample exhibited low bulk modulus when compared to low moisture state, at which the sample had high bulk modulus. The test results provide a database of bulk modulus properties for the soil at the three moisture states. Based on the test results, bulk modulus correlations in the form of linear functions of the applied hydrostatic stress were established for the soil sample at the different moisture states. The anticipated use of the regression correlation equations would provide essential guidelines for predicting volumetric deformation behavior of the finegrained subgrade soil in the field. In addition, the bulk modulus data obtained through this study will be useful for engineers and construction equipment manufacturers to estimate volumetric loading characteristics and stiffness behaviour under construction haul trucks and shovels in the field for the soil tested, and other fine-grained cohesive soils with similar characteristics.

Acknowledgements The author would like to acknowledge Professor Erol Tutumluer of University of Illinois at Urbana-Champaign, and Dr. Liqun Chi of Caterpillar, Inc. of Peoria, Illinois for their support in this study.

References [1] S.L. Kramer, Geotechnical Earthquake Engineering, Prentice Hall, New Jersey, 1996. [2] J.K. Anochie-Boateng. Advanced testing and characterization of transportation soils and bituminous sands, PhD Thesis, University of Illinois at Urbana-Champaign, Urbana, 2007. [3] AASHTO T 99-01. Standard Method of Test for Moisture-Density Relations of Soils Using a 2.5-kg (5.5lb) Rammer and a 305-mm (12-in.) Drop, American Association of State and Highway Transportation Officials, Washington DC, 2004. [4] T.G. Joseph, Physical, static and inferred dynamic loaded properties of oil sand. Final Progress Report, Phases I, II, & III, submitted to Caterpillar, Inc., Peoria, 2005. [5] E. Tutumluer, and U. Seyhan, Laboratory determination of anisotropic aggregate resilient moduli using a new innovative test device, Journal of Transportation Research Board 1687, (1999). 13 – 21. [6] K. Terzaghi, and R. B. Peck, Soil Mechanics in Engineering Practice, 2nd ed. Wiley, New York, 1967. [7] A.B. Vesic, and G.W. Clough, Behavior of granular material under high stresses. Journal of the Soil Mechanics and Foundation Division 94 (1968.): 661 – 668. [8] B.S. Quabin, V.N. Kaliakin, and J.P. Martin, Variable bulk modulus constitutive model for sand, Journal of Geotechnical and Geoenvironmental Engineering 2 (2003), 158 – 162.

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 343 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-343

Aspects Géologiques et Géotechniques Associés au Projet et à la Construction d’un Tronçon de l’Autoroute de Dakar (Sénégal) Rui FREITAS a, Virgílio REBELOa, Luís FERREIRAb et André CABRALb a COBA, SA, Lisbonne, Portugal, b MSF, SA, Lisbonne, Portugal

Résumé. Dans le tronçon autoroutier, objet de l’étude, qui s’étend sur environ 4 km dans la ville de Dakar (Sénégal), le contexte géologique et géotechnique est caractérisé par des terrains sableux et nappe phréatique près de la surface. Le scénario géologique et géotechnique existant, le milieu urbain environnant, ainsi que le drainage naturel déficient ont soulevé la question des matériaux de construction à utiliser pour les remblais. Les principaux aspects liés aux études des différentes solutions techniques envisagées, et soumises à l’APIX (autorité sénégalaise en charge des grands projets et Maître de l’Ouvrage), ainsi que les aspects liés aux travaux d’exécution, à la charge de MSF, et de l’Assistance Technique correspondante, à la charge de COBA, sont présentés dans cette communication. Mots clés. sables, marécageuse, nappes phréatiques, mélange de sols

Introduction La réalisation d’une autoroute s’impose actuellement dans la région de Dakar (Sénégal), compte tenu de l’encombrement de la circulation, presque en permanence et quasiment paralysé aux heures de pointe. Cette autoroute qui, dans le tronçon Patte d’Oie – Pikine, s’étend sur environ 4 km traverse une zone d’occupation mixte à usage résidentiel, commercial et industriel et est contigüe à la RN1, principale voie d’accès au centre de Dakar. Étant donné que la totalité des matériaux nécessaires à la construction proviendront d’emprunt, cet aspect a été très contraignant pour l’ouvrage, en raison de la difficulté d’obtention de matériaux à de courtes distances dans la région.

1. Contexte Géologique de l’Ouvrage La zone de l’autoroute s’insère dans un contexte géologique de sables quaternaires. Des dépressions humides sont visibles au fond desquelles se forment généralement des mares et des lacs liés aux fluctuations de la nappe phréatique (zones plus foncées dans la Figure 1). L’ouvrage se situe dans la zone des dépressions inter-dunaires (“Niayes”), où affleurent des sables humifères, avec une nappe phréatique proche de la surface. Ainsi, les terrains d’influence de l’ouvrage correspondent à des sables fins à limoneux, mal gradués et d’origine dunaire, parfois argileux.

344

R. Freitas et al. / Projet et Construction d’un Tronçon de l’Autoroute de Dakar (Sénégal)

Figure 1. Localisation et encadrement du tronçon Patte d’Oie – Pikine de l’autoroute de Dakar.

2. Problématique Associée à l’Ouvrage Le projet ayant fait l’objet d’appel d’offres (APD) prévoyait la construction de remblais de sables de dune avec géométrie des talus à 2/3 (v/h), revêtus avec 0,20 m de perré maçonné, étant nécessaires environ 1.200.000 m3 de matériaux. La zone d’emprunt la plus proche (sables de dune de N’Diaye Lo, Tableau 1) est située à 25 km au Nord du lieu d’implantation de l’ouvrage, sachant qu’il s’agit de la seule exploitation autorisée dans la région. Tableau 1. Caractéristiques des sables de dune de N’Diaye Lo / Rufisque de la catégorie D1 (LCPC) Essai Fines (Ø < 0,08 mm) 2 mm > Ø > 0,08 mm ES (Equivalent de sable) WOPM (Proctor modifié) γd max (Proctor modifié) CBR à 95% WOPM

Gamme de valeurs 1,5 à 7% 93 à 98,5% 62 à 81% 7 à 7,2% 17,1 à 17,2 kN/m3 15%

Leurs principales caractéristiques sont l’absence de cohésion et leur perméabilité élevée. Leur granulométrie mal graduée et de petit calibre rend ces sols très érodables et d’une traficabilité difficile. Les essais de cisaillement direct réalisés en phase d’APD sur des sables remaniés prélevés le long du tracé, de même nature que les sables de la zone d’emprunt, ont conduit aux paramètres de résistance suivants: c = 0 kPa; Ø = 27 a 32o, qui constitueront leur limite inférieure. Cependant, pour les sables mal gradués, compactés à 95% de la densité sèche maximum (essai Proctor), il est prévisible que l’angle de frottement interne soit de l’ordre de 37 o ± 1o (Bureau of Reclamation, 1973). S’agissant de matériaux de nature purement de frottement, le facteur de sécurité (FS) des talus de remblais projetés à 2/3 (v/h), (correspondant au talus β incliné à 33,7o), sera obtenu par FS = tg Ø / tg β = tg 36º / tg 33,7o = 1,09 < 1,5 (facteur de sécurité minimum acceptable). A partir du calcul automatique (SLOPE/W) on a obtenu FS~1,12.

R. Freitas et al. / Projet et Construction d’un Tronçon de l’Autoroute de Dakar (Sénégal)

345

Vu les caractéristiques des matériaux proposés dans l’APD pour la construction des remblais, a été considéré que les conditions adéquates de stabilité globale des remblais n’étaient pas assurées, en particulier pour les plus hauts. Ainsi, l’exécution de ces remblais n’était, du point de vue technique, pas possible sans avoir recours à des solutions complémentaires pour garantir le confinement latéral du remblai.

3. Solutions Alternatives Envisagées Le tronçon autoroutier se développe toujours en remblai, avec des hauteurs de 1 et 14 m (phase APD) et environ 1.200.000 m3 de matériaux, qui devraient être obtenu par emprunt, seront nécessaires. Après l’optimisation des caractéristiques du tracé, le volume des matériaux pour les remblais a ainsi été substantiellement réduit, à environ 575.000 m3. Néanmoins, en ce qui concerne la structure du corps des remblais, diverses solutions types ont été envisagées et considérées alternatives à la solution de l’APD. 3.1. Remblai Homogène (Solution type de l’APD) Pour considérer un profil en remblai homogène avec des sables de dune, comme il était prévu dans l’APD, la solution pour le problème de la stabilité des talus passerait par l’adoption d’une pente inférieure (ex. 1v/2h, correspondant à β = 26,6 o), conduisant à un facteur de sécurité de FS = tg Ø / tg β = tg 36º / tg 26,6o = 1,45. Le FS~1,5 se confirme pour la pente minimale recommandée de 1/2 (v/h) à travers la méthode de Bishop simplifiée, en utilisant le programme de calcul automatique SLOPE/W. 3.2. Remblai Zoné Comme alternative au remblai homogène, une solution possible serait l’adoption d’un remblai zoné, en utilisant les sables de dune disponibles pour la zone intérieure et des matériaux plus résistantes pour les épaules. Ces matériaux pourraient être de 3 types : • • •

Grave latéritique – matériau à granulométrie bien graduée, pourcentage de fines ce qui lui apporte des propriétés de cohésion non négligeables; Enrochement (“tout-venant”) – matériau grossier, très drainant, ce qui apporte un angle de frottement élevé lorsqu’il est dument compacté; Sable de dune mélangé avec matériaux plus fins – devient un matériau à granulométrie mieux graduée et avec pourcentage plus élevé de fines, ce qui lui apporte une composante de cohésion significative pour la stabilité des talus.

On admet qu’une solution en remblai zoné pourrait permettre de maintenir la pente de projet de 2/3 (v/h), aspect qui a été validé avec des calculs de stabilité. Le principal inconvénient de cette solution est que les zones d’emprunt des différents matériaux sont plus éloignées que celles des sables de dune. En ce qui concerne l’origine de l’enrochement, la distance de transport est supérieure à toutes les autres alternatives, étant inférieure dans le cas des matériaux pour produite le mélange avec le sable de dune.

346

R. Freitas et al. / Projet et Construction d’un Tronçon de l’Autoroute de Dakar (Sénégal)

3.3. Remblai Renforcé avec Géosynthétiques La troisième alternative serait l’adoption de talus renforcés avec du géotextile ou géogrille, permettant même l’adoption de pentes un peu plus accentuée que la solution actuellement prévue. Néanmoins, cette solution devrait être analysée, non seulement en termes de coût mais également en termes d’intégration paysagère.

4. Étude des Matériaux Analysés Compte tenu des solutions alternatives présentées, une étude de caractérisation a été menée pour les matériaux indiqués dans chaque solution, à savoir : les sables de dune et sables argileux, latérites et enrochement. 4.1. Sables de dune et sables argileux Pour les sables de dune, on a eu recours à la zone au Nord de Dakar, dans une vaste zone du cordon dunaire, avec 4 zones d’emprunt actuellement en exploitation. Les sables argileux de l’exploitation de Dougar Peulh sont les matériaux de nature plus fine ayant un plus grand potentiel pour les mélanges de sols. Les caractéristiques moyennes des terrains sableux, de chaque exploitation, sont indiquées dans le Tableau 2. Tableau 2. Caractéristiques des sables de dune et sables argileux (Dougar Peulh)

Zone d’emprunt Deni Biram Ndao Keur Ndiaye Lo N’Diakhirat Tivaouane Dougar Peulh

Analyse Granulométrique <2,0 <0,5 <0,08 mm mm mm 100 100 5,0 100 100 3,0 100 100 1,5 100 100 0,7 98,5 95,5 55,0

Limites Atterberg LL IP (%) (%) NP NP NP NP NP NP NP NP 27,5 14

E.S. (%) 62 63 78 92 -

Proctor Modifié WOPM γd max (%) (kN/m3) 9,0 16,9 9,0 16,8 7,2 16,9 15,8 16,7 9,4 20,7

CBR CBR95 (%) 12,5 12,5 13,5 12 4

Exp. (%) 0 0 0 0 1,65

4.2. Latérites Les latérites utilisées dans l’étude des matériaux provenaient des exploitations de Sindya et Mont Rolland, à environ 50 à 60 km du tracé. Les caractéristiques de compactage des latérites de chaque exploitation sont indiquées dans le Tableau 3. Les latérites correspondent à des sols sablo-graveleux très limoneux, classifiés dans la catégorie B6 (LCPC) avant l’essai CBR et dans la catégorie B5 (LCPC) après l’essai CBR, sachant que le pourcentage de fines a pratiquement doublé et que les caractéristiques de plasticité ont légèrement diminué. 4.3. Matériaux rocheux Le matériau rocheux à utiliser pour le zonage des remblais était situé à environ 120 km du tracé (carrière de Diack).

347

R. Freitas et al. / Projet et Construction d’un Tronçon de l’Autoroute de Dakar (Sénégal)

Tableau 3. Caractéristiques des latérites – Essais de compactage et CBR

Zone

Proctor Modifié γd max WOPM (%) (kN/m3)

Mont Rolland

9,3

20,3

Sindya

9,9

19,2

N (coups) 10 25 55 10 25 55

γd (kN/m3) 18,31 19,24 20,00 17,25 18,17 19,18

CR (%) 90,2 94,8 98,5 89,8 94,6 99,9

CBR W (%) 15 13,6 11,8 16,6 15,3 14,5

Exp. (%) 0,98

0,77

CBR (%) 20 58 118 15 44 108

CBR95 (%) 60

50

5. Solution Adoptée La solution la plus avantageuse, du point de vue technique, serait la réalisation de remblais au profil zoné, corps de remblai construits avec des sables de dunes et les épaules avec matériau latéritique. Cette solution aurait l’avantage d’assurer un degré de fiabilité de stabilité du remblai plus élevé. Néanmoins, le volume significatif de latérites à transporter, sur une longue distance, engendrerait un impact significatif sur le trafic lourd de la RN1 et aurait des répercussions au niveau du délai des travaux. On a alors sélectionné, en tenant compte des aspects techniques et économiques, la solution de construction des épaules avec mélange de sols. Les avantages de cette solution sont la plus grande proximité des zones d’emprunts de ces sols, leur coût relatif, les impacts socio-environnementaux moins contraignants (elle franchit des zones moins occupées et avec moins de circulation) et facilite la livraison et l’accomplissement des délais prévus. Les principaux inconvénients sont le besoin d’effectuer une étude spécifique pour l’optimisation du mélange et le besoin d’assurer l’homogénéité du mélange et le contrôle du compactage. En raison du caractère érodable des matériaux de mélange de sols, qui constituent la face des talus, un système de protection superficielle en perré maçonné a été considéré.

6. Études de la Composition des Mélanges Après l’analyse des caractéristiques des sables de dune pour les remblais, une étude de formulation de mélanges a été débutée, dans le but d’améliorer la cohésion et la portance des matériaux, en ayant été menée sur différents mélanges de sables de dunes (différentes exploitations) avec des sables argileux de Dougar Peulh (DP) avec des pourcentages de sables de dune entre 70 à 80 %. Les valeurs obtenues dans les essais d’identification et caractérisation des échantillons sont répertoriés dans le Tableau 4. Ces résultats ont permis de constater qu’il y a une amélioration substantielle de la portance des sols, en passant de CBR de 12-13% (sables de dune) à 18-28% (mélanges). Les meilleurs résultats ont été obtenus dans le mélange des sables de Tivaouane (Tiv.) avec les sables argileux de Dougar Peulh (DP). Pour l’analyse de la stabilité des talus, des essais de cisaillement ont été effectués sur tous les mélanges (Tableau 5). Le mélange de sols ayant obtenu le meilleur résultat, au niveau des caractéristiques de compactage et de résistance, a été le mélange constitué par 70% de sable de dune de Tivaouane et 30% de sable argileux de Dougar Peulh. Ce mélange a, ainsi, été

348

R. Freitas et al. / Projet et Construction d’un Tronçon de l’Autoroute de Dakar (Sénégal)

sélectionné pour le zonage des remblais et a permis la définition de la largeur des épaules des remblais en fonction de ses caractéristiques géotechniques. Tableau 4. Caractéristiques des mélanges de sols

Keur Ndiaye Lo (KNL)

Tivaouane (Tiv.)

Deni Biram Ndao (DBN)

Zone

Mélange 80 % DBN + 20 % DP 75 % DBN + 25 % DP 70 % DBN + 30 % DP 80 % Tiv. + 20 % DP 75 % Tiv. + 25 % DP 70 % Tiv. + 30 % DP 80 % KNL + 20 % DP 75 % KNL + 25 % DP 70 % KNL + 30 % DP

Analyse Granulométrique <2,0 <0,5 <0,08 mm mm mm 99

98,5

10

Limites Atterberg LL IP (%) (%) -

-

E.S (%)

Proctor Modifié WOPM γd max (%) (kN/m3)

21

-

18,6

CBR Exp (%)

CBR95 (%)

-

-

0,120,15 0,290,35

16,519,5

99

98,5

14

16-17

10-11

-

8,510,2

99,5

98,5

18,5

17-18

11-12

-

9,4-9,8

19,0

98,5

97

10,3

-

-

24

9,2

18,0

0

19

98,5

96,5

14,5

-

-

22

10,1

18,9

0

27

98

96

14,5

-

-

20

10,9

19,4

0

28

99

98,5

10,0

-

-

19

-

-

-

-

99

98

16

15

9

-

-

-

-

99

98

17

1516,3

8,8-9

9,610,2

18,7

0,120,18

1415,5

17-18

Tableau 5. Caractéristiques de résistance des mélanges avec sables de dune de Tivaouane

Mélange

Classe LCPC

70% Tiv. + 30% DP 75% Tiv. + 25% DP 80% Tiv. + 20% DP

A1 A1 A1

Proctor Modifié WOPM γd max (%) (kN/m3) 10,9 19,4 10,1 18,9 9,2 18,0

Cisaillement Cohésion Angle de (kPa) frottement (º) 22,0 30 14,9 31 10,5 32

7. Remarques Finales Dans une zone densément occupée, où prédominent les sables de dune dans les alentours de l’ouvrage, une solution en remblai zoné, impliquant l’utilisation dans la bande latérale des matériaux issus du mélange des sables de dune avec des sables argileux, peut être une solution appropriée du point de vue technique, économique et environnemental, comme on a tenté de le démontrer.

Références [1] COBA. Autoroute à péage Dakar - Diamniadio. Tronçon Malick Sy-Pikine. Lot 2 - Patte d’Oie-Pikine. Note Technique n. º 7 - Analyse de la Stabilité des Remblais et Rapport 2.3 - Etude Géologique et Géotechnique. Avril-Juin, 2007. Lisbonne, Portugal. [2] MSI LAB. Autoroute à péage Dakar - Diamniadio. Tronçon Malick Sy-Pikine. Lot 2 - Patte d’Oie Pikine. Caractérisation d’Emprunts de Sable. Avril, 2007. Dakar, Sénégal.

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 349 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-349

Characterization of Granular and Bitumen Stabilised Materials using Triaxial Testing Kim JENKINS a, and William MULUSA a Stellenbosch University, South Africa

a

Abstract. In pavement engineering, there is a need for improved methods of characterisation of materials. This would contribute to improved mix design, pavement design and quality control of the materials. In particular, granular and bitumen stabilised materials (BSMs) need for more reliable testing procedures than CBR, UCS and ITS tests for the characterisation and QA/QC (with bitumen and emulsion binders, has long been recognised by the roads industry. Granular materials and BSMs incorporating either foamed bitumen or emulsion can be viewed in the same light because of similarities in their behaviour i.e. stress dependency and shear resistance. Triaxial testing for the evaluation of shear parameters is widely recognised as a reliable method of measuring the critical performance properties of granular and bitumen stabilised materials. However, the triaxial test in its current state as a research test has little chance of extensive use by practitioners and commercial laboratories, because of complexity, cost and time issues. Major adaptations to the research triaxial test are necessary, therefore, if such a useful test can have a chance of being accepted by road practitioners. A study was carried out to investigate possibilities of developing a simple, economical, reliable and robust test for characterizing granular and bitumen stabilized materials, with a link to performance. This is achieved through innovative design and manufacture of a prototype triaxial cell capable of accommodating 150 mm diameter by 300 mm deep specimens. Ideas were developed based on some existing test apparatus such as the Texas Triaxial. The cell that has been developed is simpler than the research (geotechnical) triaxial cell and the operational protocols have been streamlined, thereby reducing the time and steps required in assembling specimens and testing them. A simple triaxial cell has been designed, comprising of a steel casing with a latex tube inside, making a less sophisticated test configuration, which eliminates the need for the use of O-rings, membrane and tie rods. This paper provides details of the cell that has been design and monotonic triaxial tests that have been conducted using the simple triaxial cell. Results from the simple triaxial have been correlated with parallel test results from the research (geotechnical) triaxial apparatus. Explanations are provided as to the theory for processing of results to provide permanent deformation calculations as a result of repeated loading applied to the material. Keywords. Road materials, granular, bitumen stabilized materials, triaxial test

Introduction In excess of 90% of materials used in road pavements in southern Africa are granular. In addition to this, bitumen stabilized materials (BSMs), which exhibit partly granular characteristics, are being used increasingly to enhance the performance of granular materials in the base and sub-base layers of roads, by creating a flexible material with improved shear strength, stiffness and resistance to moisture damage. The challenge is to find reliable methods of characterization of granular materials and BSMs, so that these materials can be judiciously selected for application in roads. The appropriateness of such materials needs to be evaluated in three ways, namely engineering properties (strength), load spreading (stiffness) and resistance to permanent deformation (rut resistance). All of these evaluations can be carried out in a

350

K. Jenkins and W. Mulusa / Characterization of Granular and BSMs Using Triaxial Testing

triaxial test set-up. In the case of granular materials and BSMs, the engineering property of primary interest is the shear strength. There are several laboratory tests in geotechnical engineering in existence for the determination of the shear strength parameters. These include direct shear test, triaxial shear test, simple shear test, using different drainage conditions (drained or undrained), rate of loading, range of confining pressures, and stress history. In pavement engineering however, these tests are uncommon. Their use is limited to research. CBR is the commonly used test in pavement engineering for evaluating the strength and bearing capacity of road materials. This test however is an empirical-phenomenological test method whose results cannot be linked directly to shear parameters. From different types of tests used to determine the shear strength parameters, triaxial test in principle (with or without adaptations effectively simulates the stressdeformation behaviour of road materials. This is supported by various stressdeformation tests reported by Rodriguez et al., [1].

1. Development of a Triaxial Test for Road Materials Significant numbers of research publications have shown the benefits of using triaxial tests for pavement materials, for example [2] and [3]. It is not the scope of this paper to deal with the application of triaxial testing, but rather to delve into the way in which the test is carried out. First some details of service conditions need to be borne in mind. Some of the primary differences between road materials in layer-works and those evaluated in geotechnical tests are: • material processing, where road materials disturbed and compacted and geotechnical materials are generally undisturbed and consolidated, • particle size and grading, where road materials are can have up to 50mm maximum particle size in a matrix of controlled grading whilst geotechnical materials generally, but not always, finer and of variable grading, • moisture conditions, where the in-service moisture conditions of road materials are generally below optimum moisture content (OMC) in unsaturated conditions whilst geotechnical evaluations are carried out at a range of moisture conditions, • loading conditions, are generally dynamic with variable speed and intensity and geotechnical loads can be static or dynamic. Given the similarities and differences in the materials and conditions of the materials in pavement and geotechnical evaluations, the triaxial testing needs to be adapted accordingly. Some of the factors for consideration are outlined below. 1.1. Triaxial Test Types for Road Materials Given the loading conditions and performance evaluations required for pavement layers, three types of pavement material evaluation are required: •

Shear strength evaluation using monotonic loading in quick undrained (unsaturated) conditions, to obtain the shear parameters Cohesion (C) and Friction Angle (φ)

K. Jenkins and W. Mulusa / Characterization of Granular and BSMs Using Triaxial Testing

• •

351

Resilient Modulus (Mr) evaluation using short duration dynamic loading at different stress ratios in sinusoidal wave form, with or without a rest period, on unsaturated specimens, Permanent deformation evaluation using long duration dynamic loading at different stress ratios in sinusoidal wave form, with or without a rest period, on unsaturated specimens to evaluate the rate of permanent vertical strain (εv)

1.2. Specimen Size In order to obtain repeatable, reproducible triaxial test results on granular pavement materials and BSMs the ratio of specimen diameter to maximum particle size needs to be a minimum of 7.5:1, refer [4]. Triaxial tests on soils and clays in geotechnical field uses specimen diameters in the range of 35 mm to 100 mm, which more than satisfies this criterion. A monotonic triaxial test on pavement materials, which is the focus of this study, requires a specimen diameter of 150 mm in order to obtain d specimen/dmax-particle greater than 7.5:1 for a dmax-particle up to 20 mm. The height of a specimen needs to be at least 1.5 but preferably 2 times the diameter of the specimen in order to eliminate the friction effects on the platens influencing the results. A specimen of 150 mm diameter by 300 mm high would then satisfy the requirements for the triaxial testing of most granular pavement materials. 1.3. Need for an efficient “simple” triaxial test set-up In a bid to develop a simple triaxial test relevant to the local road construction industry, the authors conducted a survey aimed at investigating facilities, testing capacity and resources that are currently available with civil engineering laboratories in South Africa. The findings from the survey of Mulusa [4], had provided guidance with regard to the nature and sophistication of any new tests to be developed. The survey paved way to the conceptualization process that saw several ideas around the general approach of the simple triaxial cell development considered and given reality checks. The main concepts that were considered include the Tube, Bottle and Sandwich, and Encapsulated Tube Concepts. Amongst the concepts considered, and after it was established through trials that the special tube needed could be made locally at the University of Stellenbosch laboratories the Tube Concept proved practical and is the basis on which the design of the Simple Triaxial Cell, refer [4]. After analysing the specimen assembly procedures of the Texas triaxial test procedure by TxDOT [5] and the monotonic triaxial test procedure obtained in the Technical Memorandum [6], it was concluded that two main factors contribute to the complexity of the geotechnical triaxial cell namely the time it takes to assemble the specimen accurately in the cell resulting from paying attention to many details such as placing membrane with its O-rings on the specimen and on platen disks. This takes time especially with the care to be taken not to damage the edges of the specimen and that the specimen must be centrally positioned on the base plate and the centre of the top cap must be aligned with the centre of the specimen. Secondly the inherent design of the cell which makes it water and/or air tight at relatively high pressures. Therefore, the general approach of the simple triaxial cell development was aimed at finding simple solutions to these factors.

352

K. Jenkins and W. Mulusa / Characterization of Granular and BSMs Using Triaxial Testing

The purpose of the simple triaxial cell design was then to overcome the drawbacks of standard triaxial testing cell through considerable simplification by means of a new structure and procedure of assembly of specimen into the cell. This was aimed at specifically reducing time and steps required in the procedure. 1.4. Design and Modelling of Simple Triaxial Cell The basic design of the simple triaxial cell is to use a steel casing comprising a latex tube which is then introduced around the specimen sitting on a base plate. This approach eliminates the use of membrane, O-rings on the specimen and tie rods, as shown in Figure 1. The overall dimensions of the cell are 244 mm diameter by 372 mm height. The cell comprises basically of the base, hollow cylindrical steel casing, latex tube and top disk. The casing is introduced, with the tube in it, onto the base and held into position by simple mechanical clamps. Regulated air pressure is applied through pressure inlet valve. This apparatus can only be used for monotonic tests as LVDTs cannot be mounted on the specimen due to hindrance by the tube.

Latex tube

Top

Specimen 150mm Ø x 300mm height

Galvanized steel casing

Grooved ring handle

Pressure inlet Base plate

Figure 1. Design Models of Simple Triaxial

2. Validation of Simple Triaxial Apparatus Following the design and manufacture of an electro-galvanised simple triaxial test (STT) apparatus, the need arose to verify the STT result output relative to the conventional research triaxial test (RTT) that uses a membrane and O-rings to seal the specimen during confinement. Validation was only investigated for monotonic triaxial testing.

K. Jenkins and W. Mulusa / Characterization of Granular and BSMs Using Triaxial Testing

353

Reclaimed asphalt pavement (RAP), Hornfels with maximum aggregate size of 19mm and graded within limits set for BSMs in TG2 [7], was used in this study. Hornfels (RAP) was collected from N7 rehabilitation project in the Western Cape, South Africa. Selected materials were stabilized with bitumen emulsion (ANiB SS-60). The residual binder content for both Hornfels was 2 %. Stabilised materials were tested with both 0 % and 1 % active filler (i.e. cement). The test matrix involved two mixes producing a total of 16 specimens for both STT and RTT tests. Table 1 shows the matrix of the tested mixes and aggregate type used. The binder used in this study was bitumen emulsion type B which is a stable grade anionic emulsion (60 % residual binder and 40 % emulsion water). The bitumen emulsion content of 3.3 % (i.e. 2 % residual binder) was used for the treatment of the Hornfels RAP blends. Table 1: Matrix of tested mixes and aggregate type Item Hornfels (RAP) + 2% Residual Binder No Specimens Confining Press σ3 (kPa) x No specimens

Simple Triaxial Test (STT) Emulsion + Emulsion + 0% Cement 1% Cement 3 5 50 x 1 50 x 1 100 x 1 100 x 3 200 x 1 200 x 1

Research Triaxial Test (RTT) Emulsion + Emulsion + 0% Cement 1% Cement 3 5 50 x 1 50 x 1 100 x 1 100 x 3 200 x 1 200 x 1

800 700 600 500 400 300 200 100 0

RTT STT

0.0 1.0 2.0 3.0 4.0 5.0 Strain [%]

STT versus RTT at 50 kPa

Applied Stress [kPa]

Applied Stress [kPa]

Parallel monotonic failure testing between the RTT and STT was carried out on specimens listed above, all of comparable density and moisture. Parallel testing with the Research Triaxial Test was conducted according to the triaxial testing protocol that was developed at Stellenbosch University [6]. Exhaustive results are included in the research of Mulusa [4] and only selective graphs are shown here in Figure 2. The synthesis is based on the comparison of triaxial results obtained using STT and RTT methods on comparable specimens from the same mix tested at same confining pressure. The synthesis also includes determination and presentation of the shear parameters (cohesion and angle of friction) using results from both STT and RTT methods on same mixes.

1400 1200 1000 800 600 400 200 0

RTT STT

0.0 1.0 2.0 3.0 4.0 5.0 Strain [%]

STT vs RTT at 200 kPa

Figure 2. Stress – strain plots of STT and RTT at 50 and 200 kPa ʍ3 on Hornfels (RAP) + 3.3 % Emulsion + 0 % Cement mix

354

K. Jenkins and W. Mulusa / Characterization of Granular and BSMs Using Triaxial Testing

Linear regression analysis of results above was performed to obtain MohrCoulomb failure envelope lines in order to approximate the mechanical properties of cohesion and internal angle of friction as determined by STT and RTT results on the same mix; see summary of mechanical properties in Table 2 below. Table 2. Summary of Shear Parameters from STT and RTT results

Test STT RTT

Specimen No. E+0C_1 E+0C_2 E+0C_3 E+0C_4 E+0C_6 E+0C_5

σ3 [kPa] 50 100 200 50 100 200

σa,f [kPa] 645 937 1390 696 853 1211

σ1,f [kPa] 649 941 1394 748 955 1413

Cohesion [kPa]

Internal Friction Angle [o]

Correlation Coefficient [R2]

95

41.4

0.996

123

39.3

0.999

3. Conclusions Going through situation analysis, conceptualization, design, manufacture, assembly, testing and analysis of test results of the Simple Triaxial Test prototype, it can be concluded in accordance with the main objective of the study that the Simple Triaxial test has been developed. Its simplicity stems from the following features of the STT cell. The STT yields acceptable results when compared to the research triaxial setup, thus enabling more reliable and efficient pavement material evaluation for the roads industry.

References [1] Rico Rodriguez A., del Castillo H., Sowers G. F. Soil Mechanics in Highway Engineering. Trans Tech Publications. ISBN 0-87849-072-8. 1988. [2] Ebels LJ, 2008. Characterisation of Material Properties and Behaviour of Cold Bituminous Mixtures for Road Pavements. PhD Dissertation, University of Stellenbosch, South Africa. [3] Jenkins, K.J. and Ebels, L.J, 2007. Determination of Shear Parameters, Resilient Modulus and Permanent Deformation Behaviour of Unbound and Bound Granular Materials Using Tri-Axial Testing on 150mm Ø x 300mm High Specimens. Technical Memorandum. Stellenbosch, South Africa, 2007. [4] Mulusa, W.K, Development of Simple Triaxial Test for Characterising Bituminous Stabilised Materials, MSc.Eng Thesis Stellenbosch University, South Africa, 2009. [5] Texas Department of Transport, 2002. Triaxial Compression for disturbed soils and base materials, TxDOT Designation: Tex-117-E, August 2002. [6] Jenkins, K.J., Ebels, L.J., Mathaniya, E.T., Kelfkens, R.W.C., Moloto, P.K., Mulusa, W.K., 2008. Updating Bituminous Stabilised Materials Guidelines: Final Draft Mix Design Report, Phase II. Technical Memorandum. Stellenbosch, South Africa, 2008. [7] Asphalt Academy, TG2, Second Edition, Bitumen Stabilised Materials. A Guideline for the Design and Construction of Bitumen Emulsion and Foamed Bitumen Stabilised Materials. ISBN 978-0-7988-55822, Asphalt Academy, Pretoria, South Africa. 2009

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 355 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-355

The Effect of Iron Oxide on the Strength of Soil/Concrete Interface F. OKONTA and A. DERRICK University of Johannesburg, Johannesburg, South Africa

Abstact, The effect of free iron oxide on the direct shear strength parameters of a weathered quartzite was investigated in a conventional shear box device with smooth and rough concrete bases. The result indicate that increase in the free iron oxide content result in increase in soil cohesion, and decrease in friction angle when sheared without soil concrete interface. The introduction of rough concrete surface result in significant scatter or weak trend in the iron oxide – cohesion/friction angle curves. However decrease in strength with increase in roughness was indicated for the mix with the highest concentration of iron oxide which may be due to the shift of the failure plane from the interface to the soil. Keywords. Interface, Shear zone, Iron oxide.

1. Introduction The ultimate shearing resistance between soils and construction materials is important because it determines the stability of friction piles, retaining walls, anchor rods, earth reinforcement, submarine pipelines, offshore gravity structures and geomembranes. The stability is dependent on the roughness of the interface material, the properties of the soil i.e. grain size distribution, shape of the particles, rate of shear displacement and magnitude of the normal stress. The peak shear resistance of clay soils shearing against solid surfaces is dependent on the stress history, method of sample preparation, water content, clay content, rate of shear and roughness of the surface [1]. For compacted clay soil, [2] noted that the interface shear strength determined by the direct shear test was different from the shear strength obtained from a simple shear test for a rough concrete surface. Experimental results by [3] showed that both the internal friction angle of sand and the interface friction angle increase as the shearing rate increases. The maximum shear stress also increases as shearing rate increases for smooth and rough surfaces. [1] defined two types of surface roughness: macro-roughness due to undulations on the surface and micro-roughness due to the interaction of the particles with the interface. The shearing resistance of soil at an interface is affected by the surface roughness, thus surface roughness and particle diameter should both be evaluated. [4] observed that the soil particle diameters should be considered when evaluating surface roughness. Much earlier [5] performed a series of tests to determine

356

F. Okonta and A. Derrick / The Effect of Iron Oxide on the Strength of Soil/Concrete Interface

the interface friction angle of soil and concrete and concluded that the skin friction of c – φ′ soils is a function of grain size distribution, moisture content, normal load, type of construction material and roughness of the construction material surface. In situ profiles of residual soils exhibit marked lateral and vertical variation in physical and mechanical properties [1]. Fersiallitic soil is one of the three distinct categories of naturally cemented residual soils, it is formed in subtropical or Mediterranean climates (precipitation is 500 – 1000mm per annum, average temperatures of 13 - 20°C, with a hot dry season [6]. Elements freed by primary weathering are retained in the soil profile. Free iron is often more than 60% of the total iron content. True cohesion results from the bonding between soil particles, due to cementation, electrostatic and electromagnetic attractions, or primary valence bonding (adhesion). Cementation is bonding due to cementing agents (calcium carbonate or iron oxide, or artificially by Portland cement). For design of structures the effect of free iron in the strength of soil - contruction material interfaces is vital.

2. Soil Samples and Test Method. In the investigation performed, samples were taken from an excavation in western Johannesburg, at least 1.5m below the ground surface. The Formation is part of the West Rand Group which forms part of the Witwatersrand Supergroup. The Witwatersrand Supergroup is made up of a thick sequence of shales, quartzite and conglomerates with two intercalated lava flows. The performance of the quartzite of Hospital Hill Subgroup is known to be affected by the degree of weathering. X-ray diffractometer test revealed that quartz occurred at a relatively higher percentage in the soil followed by hematite, muscovite, garronite and chloritoid respectively. The red color of the soil is attributed to the presence of hematite (Fe2 O3). The direct shearbox apparatus was used to evaluate the shear performance of the naturally cemented soil/concrete interface. Concrete samples were made in moulds so as to fit into the lower half of the shearbox snugly (60mm × 60mm) following [5]. 2-3mm and 5-6mm aggregates were pressure sprayed on the concrete paste to two different surface roughnesses. The blocks were soaked in water prior to testing so that they would not absorb water from the soil when tested. The cement, sand and stone was mixed in the ratio 1:3:3 respectively. Four different soil mixes with 0%, 4%, 8% and 12% iron oxide were prepared. The soil was saturated with warm water at approximately 60°C and mixed for 10 mins and then placed in an oven and allowed to slowly dry at a temperature of approximately 40°C. Samples were saturated and sheared after consolidation to standard pressures of 50kPa, 100kPa, 200kPa and 400kPa at slow strain rate of 0.2mm/min.

F. Okonta and A. Derrick / The Effect of Iron Oxide on the Strength of Soil/Concrete Interface

357

Figure 1. Lower half of shear box containing concrete specimen

3. Test Results and Discussions Table 1. The physical properties of the soil. PARTICLE SIZE CLAY = 37% SILT = 8% FINE SAND = 40% COARSE SAND = 12% GRAVEL = 3%

SPECIFIC GRAVITY 0% Fe2O3 = 2.84 4% Fe2O3 = 2.91 8% Fe2O3 = 3.0 12% Fe2O3 = 3.0

COMPACTION PROCTOR MDD = 1950 kg/m3 OMC = 12.2 % ASSTHO MDD = 2203 kg/m3 OMC = 11.5 %

ATTERBER GE LIMITS LL = 25% PI = 11% LS = 3%

The physical properties shown in Table 1, revealed that the residual quartzite is a low plasticity soil of uniform grading. The mix sample shows an increase in specific gravity with increasing free iron oxide content. A total of 64 shear box tests were performed. The results detailed in Figure 2, Figure 3 and Table 2 are based on peak shear stress or shear stress at 20% strain. Figure 2 and Table 2 indicate that when the samples are sheared without concrete interface, the angle of friction decreases with increase in iron oxide content. With the introduction of rough surface the trend becomes erratic although a weak trend of decrease in friction angle with increasing iron oxide, as the failure plane moves from the interface to the high iron oxide soil, is indicated. Figure 3, however, revealed a general trend of increase in cohesion with increasing iron oxide and roughness.

358

F. Okonta and A. Derrick / The Effect of Iron Oxide on the Strength of Soil/Concrete Interface

Figure 2. Angle of Friction versus % Free Iron oxide

Figure 3. Soil Cohesion versus % Free Iron oxide

F. Okonta and A. Derrick / The Effect of Iron Oxide on the Strength of Soil/Concrete Interface

359

Table 2. Summary of Friction Angle and Cohesion INTERFACE

PERCENT IRON OXIDE

NO INTERFACE

0% 4% 8% 12% 4% 8% 12% 4% 8% 12%

3mm AGGREGATE SIZE INTERFACE 6mm AGGREGATE SIZE INTERFACE

FRICTION ANGLE 37 34 34 33 37 44 38 43 42 39

COHESION (kPa) 9 22 25 36 12 34 44 23 58 49

4. Conclusions • •

• •

•

Most samples tended to contract as the test progressed, but some samples dilated at a low normal load (50kPa). The internal friction angle of the soil was marginally reduced by the coating of iron oxide. This is an inherent property of the soil used and is expected to remain relatively constant for all the tests performed. The results show that the cohesion within the soil tends to increase as the concentration of iron oxide within the soil increases. According to [1], the shear strength of the soil would tend to decrease as the roughness of the concrete increases. This trend was not immediately apparent, however a weak trend indicated that as the roughness of the concrete surface increased, the soil failure plane shift further away from the rough concrete surface interlocking and into the high iron soil with reduced strength. The presence of smooth concrete surface had very limited effect on the strength of the soil.

References [1[ Lemos, L.J.L., Vaughan, P.R., (2000), Clay-interface shear resistance, Géotechnique, Volume 50, Issue 1, February 2000, page 55 – 64, Database: ICE Virtual Library. [2] Shakir, R.R., Zhu, Jungao, (2008), Behaviour of compacted clay-concrete interface, In Frontiers of Architecture and Civil Engineering in China, Higher Education Press, Volume 3 Number 1 March 2009, Database: Springer-Verlag GmbH. [3] Al-Mhaidib, A.I., 2005, Shearing Rate Effect on Interfacial Friction Between Sand and Steel, Proceedings of the Fifteenth International Offshore and Polar Engineering Conference South Korea, June 19 – 24 2005. [4] Kishida, H., Uesugi, M., (1987), Tests of the interface between sand and steel in the simple shear apparatus, Géotechnique, Volume 37, Number 1, page 45 – 52, Database: ICE Virtual Library. [5]Potyondy, J.G., (1961), Skin friction between various soils and construction materials, Géotechnique, Volume 11, Issue 4, December 1961, page 339 – 353, Database: ICE Virtual Library. [6]Fookes, P.G., (1997), Tropical Residual Soils: A Geological Society Engineering Group Working Party Revised Report, The Geological Society, The Alden Press, Oxford.

360 Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-360

Moisture retention characteristics of some mine tailings a

S. K.Y GAWUa and J. YENDAW b Kwame Nkrumah University of Science and Technology, Kumasi, Ghana b University of Mines and Technology, Tarkwa, Ghana

Abstract. Tailings, like natural clayey soils, are prone to unsaturated behavioural changes as a result of changes in water content and suction. The equilibrium relationship between the moisture content and suction for soil, also known as soil water characteristic curve (SWCC), is commonly obtained in the laboratory using pressure plate equipment. This paper presents results of the laboratory determination of SWCC of four mine tailings and applies the van Genuchten curve fitting equation to estimate their air-entry values. High air-entry values were obtained for two tailings types and low values of the other two. The curves also show that the residual water contents of these tailings correspond to a value of suction of approximately 1000 kPa. Keywords. Soil water characteristic curve, pressure plate equipment, air-entry value, suction

Introduction Many geotechnical problems associated with mine tailings deposition involve unsaturated flow conditions and hence are related to the unsaturated behaviour of soils. Thickened tailings, like natural clayey soils are prone to unsaturated behavioural changes due to water loss by drainage, a de-saturation process especially when deposited in arid and semi-arid regions. Although an increase in the effective stress of the tailings is a potential advantage through this de-saturation process, drainage produced by matric suction may become environmentally unacceptable due to the risk of groundwater pollution and crack formation that allows significant influx of oxygen. This may lead to the production of acid mine drainage, the largest single environmental problem facing the mining industry today [1]. For wet tailings having the potential to generate acid mine drainage, significant ingress of oxygen is undesirable. A soil-water characteristic curve (SWCC) depicts an equilibrium relationship between the moisture content and suction (the difference between pore water pressure and air pressure) of the soil and is commonly obtained in the laboratory using pressure plate equipment in which the volume change of the soil skeleton is assumed to be zero or the change in volume of the soil is a measure of the change in water content. Typically, six to eight data points are required to define the essential features of the soil-water characteristic curve, namely the air-entry value (defined as the matric suction at which drainage of the pores begins), the residual water content (where a large suction change is required to remove additional water from the soil) and the slope of

361

S.K.Y. Gawu and J. Yendaw / Moisture Retention Characteristics of Some Mine Tailings

the curve (indicating the water storage potential of the soil over a specific range of matric suctions). This paper presents results of the laboratory determination of soil-water characteristic curves (SWCC) of four mine tailings and applies the van Genuchten curve fitting equation to estimate the essential parameters.

1. Materials The materials used for the investigation were gold tailings, two mineral sand tailings dubbed MS1 and MS2 and zinc tailings. The basic geotechnical index characteristics and mineral composition of the materials are presented in Table 1 and their particlesize distribution curves are shown in Figure 1. Table 1. The basic physical parameters and mineral composition of the tailings tested Tailings / Properties Specific gravity Liquid limit (%) Plasticity index (%) Linear shrinkage Shrinkage limit (%) Particle-size distribution Clay (<2 μm; %) Sand (%) Fines (<75 μm; %) Fines (<20 μm; %)

Gold

Mineral composition

MS 1

MS 2

Zinc

2.74 29 3 2 24

2.76 66 37 12 28

2.84 69 34 17 32

2.77 37 1 2 33

13 21 79 55

56 13 87 80

68 14 86 79

11 26 74 37

Quartz, kaolinite, pyrophyllite

Quartz, kaolinite, muscovite

Quartz, kaolinite, haematite

Quartz, Muscovite, anhydrite, microcline

2. Soil-water measurement In this investigation, a pressure plate apparatus consisting of six cells (Figure 2), designed and constructed at the University of the Witwatersrand, Johannesburg, was used to acquire data for soil-water characteristic curves. In the pressure plate, up to 150 g of de-aired saturated tailings were placed on a previously saturated high air-entry ceramic base to a height of about 10 to 15 mm and sealed in an air-tight chamber. Graduated cylinders were then placed directly below the apparatus to measure the volume of drained water. To prevent evaporation of the drained water, a layer of coloured kerosene was poured onto the surface of the water in the cylinder. The chamber was then pressurized to obtain a desired value of suction.

S.K.Y. Gawu and J. Yendaw / Moisture Retention Characteristics of Some Mine Tailings

% passing

362

100 90 80 70 60 50 40 30 20 10 0

Gold MS 1 MS 2 Zinc

0.001

0.01

0.1

1

10

Grainsize (mm)

Figure 1. Particle-size distribution of the tested tailings

Figure 2. Pressure plate apparatus Air pressures were controlled with precision Norgren regulators and air pressure measurements were made using a Psi-Tronix model 2000 precision digital pressure gauge. Each specimen was left to equilibrate at an applied air pressure before the next increment in pressure was made. Equilibrium condition was assumed under the applied suction when no more water drained from the tailings. The time required for the measurement of one data point was approximately seven days. After equilibrating at the final pressure, the wet mass was determined. The tailings were then oven-dried and the gravimetric water content at each pressure increment was determined by back calculation.

S.K.Y. Gawu and J. Yendaw / Moisture Retention Characteristics of Some Mine Tailings

363

3. Results The changes in gravimetric water content versus suction relationship for all the tailings tested are combined in Figure 3. The curves have two linear portions, almost flat gradients at low suctions followed by steeper gradients at high suctions or low water contents. The position of the curves can be seen generally as a function of the percentage fines content; the higher the fines content (Table 1), the more shift of the curve to the right. A similar observation was made when soils of different origins were tested with the same apparatus [2]. The suction level of the tailings undergoing drying is therefore a function of percentage fines content. The desired parameters are not easily discernible for these tailings as presented. To estimate these parameters, the van Genuchten curve fitting equation was used.

Matric suction (kPa)

1000 800 600 400 200 0 0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

Gravimetric water content (w/w) Mineral sand1

Mineral sand 2

Zinc

Gold

Figure 3. Soil-water retention curves for the tailings. 3.1. Estimation of essential parameters Several empirical models have been proposed to improve the description of soil-water retention relationships by providing estimates of air-entry values as well as the rate of water extraction from soils once the air entry value has been exceeded. A smooth threeparameter function with the flexibility of fitting a wide range of soils is the equation of van Genuchten [3], which can be written as follows:

[ (

ww = wr + (ws − wr ) 1 + αψ n

)]

−m

(1)

where ww is the gravimetric water content at any soil suction, wr is the residual gravimetric water content, ws is the saturated gravimetric water content, ψ is the soil suction; α, n and m are material parameters. The parameter α is related essentially to the air-entry value of the soil, n is a curve fitting parameter related to de-saturation rate after the air entry and m (m = 1- 1/n, is often used) is an empirical constant affecting the shape of the retention curve. The estimation method used in this investigation ensues from fitting the experimental data to Eq. (1), using the RETC code [4]. Data for each tailings type were used as input to the RETC (RETention Curve) computer program to calculate the fitting parameters. Initial water contents, back calculated from the oven-dry samples,

364

S.K.Y. Gawu and J. Yendaw / Moisture Retention Characteristics of Some Mine Tailings

were used as ws whilst wr was assumed to be the water content after the test was stopped. While adjusting the model parameters to the experimental data, the RETC code calculates α, the inverse of which is taken as the air-entry value [5]. Figure 4 compares experimental and van Genuchten fitted retention curves for the zinc tailings. Similar curves were obtained for other three tailings types. Values of the fitted parameters obtained with the RETC computer program are listed in Table 2. There is a very good agreement between the fitted values and experimental data as indicated by coefficient of correlation, r2. Table 2. Fitted hydraulic parameters from the van Genuchten [3] equation for the retention curves α

n

r2

(w)

(1/w)

(-)

(-)

0.0527

0.5156

0.0151

1.4872

0.993

66

Gold

0.0102

0.8431

0.0563

1.7278

0.996

18

MS1

0.4142

2.6485

0.0360

4.9389

0.999

28

MS2

0.3310

1.3360

0.2683

1.5806

0.985

4

Tailings type

wr

ws

(w)

Zinc

Air-entry value (kPa)

Water content (w)

0.6

0.5

0.4

0.3 van Genuchten

experimental points

0.2 1

10

Suction (kPa)

100

1000

Figure 4. Experimental and van Genuchten fitted retention curve for zinc tailings

4. Discussion Equations presented by Deschamps et. al [6] show that the production of acid mine drainage in wet tailings can be controlled by limiting high ingress of oxygen. In near saturated conditions, the influx is greatly reduced implying that when tailings dry out there should be limiting water content, here assumed to be the air-entry value, when ingress of oxygen may be significant with its associated crack development. An enhanced environmental protection that can be provided by the thickened tailings disposal technique therefore requires knowledge of air-entry values for proper monitoring and controlling of the hydrology of de-saturating tailings. The calculated air-entry value for the zinc tailings (Table 2) indicates that significant drainage by gravity and influx of air will not occur up to a suction of about 6.6 m of water (66 kPa).

S.K.Y. Gawu and J. Yendaw / Moisture Retention Characteristics of Some Mine Tailings

365

The flocculated gold, MS1 and MS2 will also remain saturated for values of matric suction up 18 kPa, 28 kPa and 4 kPa respectively. Hence, if these tailings were allowed to drain under gravity only, the phreatic surface would have to be 6.6 m, 1.8 m, 2.8 m and 0.4 m respectively beneath surface before air enters the tailings. MS2 contains hematite, an oxide of iron that commonly occurs in large masses but frequently with radial or concentric structure. The occurrence of hematite may have contributed to an initial open structure responsible for the low air-entry value. It was also observed, when testing this sample, that cracks developed at very low suction pressures. This situation may also be responsible for the low air-entry value. The values of n also indicate that the rate of de-saturation after air-entry values are exceeded is highest for MS1, followed by flocculated gold, then MS2 with the zinc tailings showing a relatively slow de-saturation rate. In addition, the soil-water characteristic curves show that the residual water contents correspond to a value of suction of approximately 1000 kPa. This conclusion is similar to that reached by Barbour et al. [1], when the hydraulic properties of thickened tailings from an active zinc-copper mine was evaluated.

5. Conclusion The van Genuchten curve fitting equation was applied in the estimation of the air-entry values as well as the water storage potential of four mine tailings. Higher air-entry values were obtained for the zinc and MS1 tailings whilst comparatively lower values were obtained for the gold and MS2 tailings. Air-entry values are particularly significant for acid generating tailings in that oxygen influx into the tailings occurs at suctions exceeding air-entry values. Below these values, acid mine drainage is insignificant. The curves also show that the residual water contents of these tailings correspond to a value of suction of approximately 103 kPa. The estimated residual water content for most soils however is of the order of 106 kPa [7].

References [1] S.L. Barbour, G.W .Wilson, St. L. Arnaud, R.J. Salvas, D. Bordin, Aspects of environmental protection provided by thickened tailings disposal. Innovative Mine Design for the 21st Century, Bawden, W.F., Arhibald, J.F (eds). Proceedings of the International Congress on Mine Design, Ontario, Canada (1993), 725 –736. [2] B. A. Harrison and G. E.Blight, The determination of soil-water characteristic curves from indicator tests, In: Wardle, G. R., Blight, G. E. and Fourie, A. B. (eds.) Geotechnics for Developing Africa, Balkema, Rotterdam (1999), 325 – 329. [3] M. Th. van Genuchten, A close-form equation for predicting the hydraulic conductivity of unsaturated soils, Soil Science Society of America Journal 44 (1980), 892 – 898. [4] M. Th.van Genuchten, F. J. Leij, and S. R Yates, The RETC Code for quantifying the hydraulic functions of unsaturated soils, Report no. EPA/600/2 – 91/065, U.S EPA, Office of Research and Development, Ada Oklahoma, 1991. [5] W. J. Rawls, T. J.Gish, and D. L. Brakensied, Estimating soil water retention from soil physical properties and characteristics, Advances in Soil Science, 6 (1991), 213 – 234. [6] T. Deschamps, M. Benzaazoua, B. Bussière, M. Aubertin and T. Bedem, Microstructural and geochemical evolution of paste tailings in surface disposal conditions, Minerals Engineering 21 (2008), 341-353. [7] S. K. Vanapalli, D.G. Fredlund, and D. E. Pufahl, The influence of soil structure and stress history on the soil-water characteristics of a compacted till, Geotechnique, 49 (1999), 143 – 159.

366 Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-366

Prediction of Over-Consolidated-Ratio for African Soil a

Diganta SARMA a,1 and Moumy DSARMA a Transportation and Geotechnical Consultant, Gaborone, Botswana Abstract. Consolidation properties of soil are essential for prediction of the deformation characteristics, determination of which involves considerable time, cost, and rigorous testing process. Further, natural state of saturation is not simulated and thus for partially saturated soils the standard odometer test gives misleading results of the evaluated consolidation parameters particularly the OverConsolidation-Ratio (OCR), which is an important parameter that influences many soil properties of practical significance. In this paper an experimental investigation for a simple yet reliable method has been presented for prediction of OCR of partially saturated soil from the simple index properties. Correlation of simple index properties with the consolidation parameters has been evaluated through modified odometer tests and simulating natural state of saturation and interpreted for African Soil. Further, advantages of this simplified process, which is confined to shallow depth, may essentially be useful for the pavement engineers for rapid and reliable results saving considerable time and cost in detailed investigation. Keywords. Consolidation, deformation characteristics, partially saturated, OverConsolidation-Ratio, simple index properties of soil

Introduction Consolidation properties of soils indicate an insight on the compressibility behaviour of soils with associated expulsion of water. However, determination of such properties involves considerable time, cost and rigorous testing process. Further, natural state of partial saturation and soil-moisture is not simulated in the standard consolidation procedures. The sampling technique is also not specific for the Oedometer tests and sampling disturbance influences the results considerably. As such, modified methodologies of Odometer test for field simulation as well as simple correlations of the consolidation parameters with fundamental properties are always preferred by practising engineers. Over-Consolidation-Ratio (OCR) as a consolidation parameter is important in soil mechanics as it influences many soil parameters of practical significance as discussed under Section 1. OCR in the simple sense is related to the past stress history of soil and is defined as the ratio of the preconsolidation pressure to the present overburden pressure. Most of the soils in nature exist at overconsolidated (OC) state due to the experience of higher pressure in the past. Further, the process of desiccation, weathering, leaching, repeated rise and fall in water table and various other processes also impart overconsolidation to soil. Therefore, emphasis has been given for determination of OCR using simple parameters and modified methodologies. It is 1

Corresponding Author: P. O. Box 81897, Gaborone, Botswana, Email: [email protected]

D. Sarma and M. Dsarma / Prediction of Over-Consolidated-Ratio for African Soil

367

reported that void ratio at liquid limit (eL) correlates with some of soil properties [1, 2]. Accordingly, the possibility of development of a correlation of OCR with void ratio at liquid limit (eL) and in-situ void ratio (e) has been explored. The e and e L are measured considering the respective moisture contents.

1. Importance of Over-Consolidation-Ratio (OCR) The value of OCR influences significantly some of the fundamental properties of soil. The difference between the shear strength of undrained and drained conditions depends upon the value of OCR. For high OCR the difference may be very large. OCR controls the position of Mohr’s circle in unconfined compression tests which passes through the origin for unconsolidated undrained test but shifted away from the ordinate with increase in the value of OCR, simulating laboratory condition different from the field and apparently results in lesser compressive strength value in the field. Higher the OCR value, larger is the difference between the laboratory unconfined compressive strength and field compressive strength. Further, for OCR > 4 to 8, Mohr’s failure envelope becomes curved. Higher the OCR, greater is the extent of curvature and Φ value changes at every point, however, becomes straight after crossing the threshold of preconsolidation pressure. The degree of overconsolidation, expressed in terms of OCR, is one of the reliable indicators for assessing the presence of fissures due to the platy arrangement of particles and in consequently the basis of design whether to adopt peak strength or residual strength. Value of OCR distinguishes normally consolidated clay for application of mathematical relationship [3] for evaluation of Compression Index (Cc) from Liquid Limit (LL). Besides above, the other notable influences are as follows. The relationship for uncemented OC soil [1] involving the in-situ void ratio (e), void ratio at Liquid Limit (eL), overburden pressure (σO′ in kPa) and pre-consolidation pressure (σ′C in kPa) ensuring the relation between consistency indices and OCR is as follows. e = 1.122 − 0.188 log σ C '−0.0463 log σ O ' eL

(1)

Shearing process of highly OC clays may result in fall of pore pressure and may even be negative due to associated increase in volume influencing pore pressure parameter ‘A’ suggested by Skempton [4]. His suggested ‘Af’ (value of ‘A’ at failure) values varies from +0.5 to +1 for NC clays, 0 to +0.5 for light OC clays and –0.5 to 0 for heavily OC clays. Overconsolidation increases coefficient of earth pressure at rest (K0), approximate value of which was suggested by Meyerhof [5] as follows. K O = (1 − Sinφ ′)(OCR )

(2)

However, besides OCR, coefficient of earth pressure at rest (K 0) is also influenced by PI of the soil as shown by Brooker et al [6] from experimental data. It was found that within the PI values from 10 to 30 the average K0 varies from 0.5 to 28 with variation of OCR from 1 to 32.

368

D. Sarma and M. Dsarma / Prediction of Over-Consolidated-Ratio for African Soil

Overconsolidation tends to increase the Young’s Modulus (E) of soils [7] such that the E/σc (where, σc is the consolidation stress in isotropic system) tends to increase with increase in OCR till limiting value of 6 beyond which the trend is unclear. Such variation is more pronounced at higher factor of safety (Refer Figure 1).

800

FS =

(σ 1 − σ 3 ) f (σ 1 − σ 3 )

= 3.0

600

E/σ'c 400 200 0 1

2

5

10

20

50

Overconsolidation Ratio Undisturbed Kawasaki

Remoulded V.B.C.

Remoulded B.B.C.

800

FS =

(σ 1 − σ 3 ) f (σ 1 − σ 3 )

Undisturbed Amuay

= 1.5

600

E/σ'c 400 200 0

1

2

5

10

20

Overconsolidation Ratio

Figure 1. Effect of overconsolidation on Young’s modulus [7] for two different factor of safeties

OCR also influences settlement to larger extents. The consolidation settlement (Sc) is calculated using the following equation: S C = CC

⎛ σ '+ Δσ ' ⎞ HO ⎟⎟ log⎜⎜ O 1 + eO ⎝ σO' ⎠

(3)

where, CC is the compression index, HO the thickness of compressible layer, eO the in-situ void ratio, σO′ the present effective overburden pressure, and Δσ′ the average effective pressure increment on the layer. This equation is valid only for NC soil. For OC soil, CC must be replaced by recompression index Cr. Thus for use of proper index, it must be ascertained whether soil is overconsolidated (refer Figure 2a and 2b).

369

D. Sarma and M. Dsarma / Prediction of Over-Consolidated-Ratio for African Soil

2. Objectives and relevance of the present study It is apparent from the foregoing discussions that OCR is an important parameter for understanding of the behaviour of soil and knowledge of the parameter provides an insight into its practical implications. Generally OCR is determined from the results of the oedometer test which is a time consuming and involves tedious process. The objective of the present study is to explore the possibility of correlating OCR with common soil parameters which can be obtained easily. σ'o+Δσ' σ'o

σ'o

σ'c Δe

Void ratio

Δe1 Δe2

Cr 1

Virgin Cc compression

Logarithm of effective stress (a)

eo e

σ'o+Δσ'

Void ratio

Recompression 1

σ'c

Logarithm of effective stress (b)

Figure 2. Effect of preconsolidation pressure on settlement for (a) σ’O + Δσ’ <σ’C, (b) σ’O + Δσ’ >σ’C

Such correlations can save considerable time and money, particularly in handling bulk samples needed for pavement construction and rehabilitation projects of highways and airports. Further, in case of urgent defence need when tractability on natural ground is required for operation of combat vehicles and equipments, information on the state of overconsolidation of the surface or near surface soils may constitute a very important parameter. Therefore, this experimental study is confined to shallow depth of soil strata, which is of utmost interest for the pavement engineers.

3. Methodology This study is directed towards correlating OCR with in-situ void ratio (e) and void ratio at liquid limit (eL) because of the proven applicability of the ratio (e/e L) to various soil characteristics [1, 2, and 8]. Also it has been established that void ratio varies with overburden pressure (refer Figure 3) in a similar manner for different soils having different liquid limits [9]. Therefore, liquid limit has been taken as a common denominator because of its uniqueness in relation to soil properties. From classical terminology, liquid limit is a measure of type and amount of clay fraction present in a soil. The common clay minerals (Kaolinite, Montmorillonite and Illite) give rise to negatively charged surfaces due to replacement of central cation of the unit cell (sheet structure or basic stacking pattern) by another cation of low valency, which causes adsorption of polar water molecules as well as cations (from the salt present in the solution) to the particle surfaces resulting in formation of diffused double

370

D. Sarma and M. Dsarma / Prediction of Over-Consolidated-Ratio for African Soil

layer. Hence, greater the amount of clay fraction, greater will be affinity towards water resulting higher liquid limit. When two clay particles come close enough, their diffused double layers interact. It causes higher cation concentration in between two particles (near the faces of the platy particles) resulting an osmotic potential which is also influenced by the properties of clay minerals as well as the pore fluid. The relationship of half space distance ‘d’ with logarithm of net repulsive force is unique irrespective of the clay minerals for a particular physico-chemical environment [10]. This was supported experimentally by Nagaraj et al [11] for a usual environment encountered in the field as follows: d = 97.67 − 29.343 log(R − A) = a − b log(R − A)

(4)

where, d is the half space distance, and R-A the net repulsive force which can be identified as the effective stress of classical soil mechanics [3]. 4.5 Wl, =120 P.I.=80 Highly colloidal clays

4.0 3.5

Wl, =80 P.I.=50 Colloidal clays

Void Ratio

3.0 2.5

Wl, =50 P.I.=25 Clays

2.0

Wl, =30 P.I.=12 Silty clays

1.5 1.0 0.5

Silts

0.0 0.1

1.0

10

102

103

104

105

Effective Overburden Stress

2

Figure 3. Approximate relation between void ratio and overburden stress for clay sediments, as a function of the Atterberg’s liquid limit (LL) and Plasticity index (PI), After Lambe and Whitman [9]

Several researchers [12 and 13] presented the liquid limit (LL) of fine-grained soils corresponds to a unique equilibrium consolidation/ suction pressure of about 6 kPa with a shearing resistance of about 1.7 to 2 kPa. Extending the logic of d versus log (RA) to the physico-chemical environment at LL, it is likely that the interparticle separation distance at liquid limit, dL, is same for all soils since stress conditions are of same order. Assuming parallel plate configuration, eL = G.γ W .S .d L

(5)

where, eL is the void ratio at liquid limit, G the specific gravity, the unit weight of water, S the specific surface, and dL the interparticle separation distance at liquid limit. Identifying ‘d’ as the void ratio, equation (5) can be expressed as: eL = a ′ − b′ log σ L

(6)

D. Sarma and M. Dsarma / Prediction of Over-Consolidated-Ratio for African Soil

371

where, a′ and b′ are constants like in equation (4), and σ L the effective stress at liquid limit. Due to difference in specific surfaces of different soils, e L can be different for the same order of σL and dL. Hence liquid limit can be regarded as a state at which the separation distance between particles or their aggregated units are under force-fieldequilibrium and eL can be a normalisation parameter at macro level to generalise the behaviour of different fine grained soils. Thus, all the water held at liquid limit of soils can be considered as interacting water directly under the influence of interparticle forces which is also dependent upon pore size distribution. Attractive force is predominant only within a distance of 20 Ao and practically no force exerted beyond a distance of about to 300 Ao. It has also been experimentally established that the pore size distribution curves for different soils at their liquid limits are of the same type. Test data on permeability indicate that at liquid limit water contents, the permeability coefficient, k, is of the same order for all soils. Considering the state of soil in volume basis, the weight of solid particles is inversely proportional to the liquid limit water contents for unit volume of soils, i.e. the weight of the soil particles in unit volume will be such as to provide same order of surface area and hence the same order of physico-chemical potential for all soils. Thus, the resulting microstructure, depending upon the physico – chemical potential in unit volume, can be of the same pattern. These unique conditions of same consolidation / suction pressure, constant shear strength, and same order of permeability at liquid limit, can be represented as a datum state in relation to which all other state and stress conditions can be normalised. In particular, the compression equation of normally consolidated uncemented saturated soils, upon normalisation, would result in the form [14]: e = a − b log p eL

(7)

where, ‘e’ is the in-situ void ratio, and effective stress ‘p’ equals to ‘σ – u’. Above explanation and formulation can be used not only for pure clays but also for natural soils containing coarser particles as the clay particles form a coating around the coarser particles preventing a direct contact between them or the coarse particles float in a matrix of clay particles. It is proved experimentally that coarser particles reduce the physico-chemical potential of the soil proportionately without altering the basic mode of stress release. Hence eL should correspond to the modified liquid limit of the soil as a whole taking into account for the reduction in physico-chemical potential. F ⎞ ⎛ WLmod ified = WL⎜1 − ⎟ ⎝ 100 ⎠

(8)

where, WL modified is the liquid limit of the soil as a whole, WL the liquid limit for soil fraction finer than 425 micron, and ‘F’ the fraction of soil coarser than 425 micron expressed as a number. Double layer theory can be applied to soils if the modified liquid limit value is appreciable, at least to the extent of 30-35%. From the above discussion, it is obvious that the determination of void ratio (e L) at liquid limit and correlating it to the in-situ void ratio (e) and other important parameter like OCR is of utmost importance because of the unique characteristic of soils at liquid limit so that the cumbersome procedure of determining the OCR by Oedometer test can be avoided.

372

D. Sarma and M. Dsarma / Prediction of Over-Consolidated-Ratio for African Soil

In this experimental study, OCR has been evaluated following Casagrande’s method. To simulate field compressibility at the laboratory, the soil samples were consolidated at field moisture contents as per suggestion put forward by Sowers [15]. Field cone penetration and unconfined compression test were carried out to obtain the undrained strength of soils. These results provide supplementary information on comparative extents of overconsolidation of different soils tested.

4. Investigation site The site selected for investigation comprised of scattered hills, eroded plains, filled-up and low-laying areas. The climate of such region is seasonally varying from heavy rainfall to occasional dry with little or no rainfall. Samples were collected from 0.30 m to 1.1 m below ground level.

5. Findings and discussion The samples tested were all fine-grained soils. The variation of liquid limit and plastic limit with very high OCR values implied a wide range of properties for these soils. The difficulty experienced in sampling from greater depth indicated generation of high matrix suction through the partially saturated narrow pores resulting large shear strength [16 and 10]. It is inferred that where the proportion of air content in the soil is less than 5%, air is held in position under high pressure by surface tension, since it is not easy for expulsion or compression of such small quantity air. Further in such situation, the relative humidity of the pore air also remains higher. If the degree of saturation is low and the air space is continuous, there may be considerable migration of water vapour blocking the pore spaces of soil mass supplementing additional strength under stress. This might be one of the reasons besides others for causing a major contradiction, where it has been observed that higher the OCR values, higher were the in-situ void ratios. The further explanation is that within the sampling depths the soil repeatedly returns to desiccated state due to alternating wet and dry cycles and subsequently concentration of various salts (viz., Ca, Mg, Al and Fe) continues for increasing desiccation bonds with natural cementing compounds. These desiccation bonds are responsible for crumbling the soil having skeletal formations where the particle-to-particle contacts at corners offer stronger shear resistance. This skeletal formation associated with the small shrinkage and hair cracks, present in the highly overconsolidated soils, can cause high in-situ void ratio. Due to such desiccation bonds the rebound curves become flat in their undisturbed state. In other words, small expansion on unloading, is an indication that swelling is due to elastic rebound and straightening of bent particles, indicating that the soil is of non-swelling type and thus its volume change behaviour is governed by the shearing resistance at particle contacts [16]. For laboratory tested undisturbed soil, the presence of desiccation bonds between soil particles augments the intrinsic effective stress present between them and thus increases the shearing resistance at particle contact points. The larger shearing resistance present at a particular state enables the undisturbed soil to support the external load at higher void ratio. Upon loading, the volume change occurs by shearing displacements or sliding between particles.

D. Sarma and M. Dsarma / Prediction of Over-Consolidated-Ratio for African Soil

373

The location exhibiting high OCR with high undrained cohesion appears to be cemented as evident from e - log p curve perhaps generated due to repeated wetting and drying progressively gaining strength and stiffness [17]. The similarities shown by the stress-strain behaviour of desiccated soils to the reported behaviour of soils subjected to the similar environment depicts that perhaps desiccation bonds are induced by repeated wetting and drying for which OCR have been observed. For the range of data considered in this study, the e/eL versus OCR relationship has been found to have the following equation: e = 0.808 log (OCR ) − 0.380 eL

(9)

This suggests a straight line with coefficient of correlation 0.941. For partially saturated uncemented soil Nagaraj et al [1] postulated the following equation: ⎛ e ⎜⎜ ⎝ eL

⎞ 13 ⎟⎟ S r = 1.19 − 0.208 log σ C '−0.047 log σ O ' ⎠

(10)

where, Sr is the in-situ degree of saturation, σC′ the preconsolidation pressure, and σO′ the effective overburden pressure. Rearranging the parameters in equation (10) a family of curves can be derived for OCR versus (e/eL)Sr1/3 for a particular overburden pressure σO′. While plotting the experimental points of the present study in the family of curves as per Nagaraj et al., a partial conformity is observed. Conformity diverges apparently more with increase in OCR values perhaps due to increasing cementation bonds. It is also inferred that for partially saturated soils, the stress fields of air, water and solids act independently, measurement of which are difficult. The effective stress in such condition becomes a function of total stress, difference of pore air and water pressure, surface tension and a parameter dependent mainly on degree of saturation [18]. Inadequate measurement of any parameter leads to an impact on the correct evaluation of OCR. Compressibility characteristics of soils are also affected by the sampling disturbances. The difference becomes prominent in e - log p curve for a specimen in case of piston sampling and block sampling. Therefore in this experimental study block sampling was done for reducing the disturbance due to sampling thus simulating as close as possible to the actual structure of the soil. The testing procedure also influences the results due to the friction developed between the sample and the Oedometer ring. Errors in the effective stress exceeding 6% have been reported [18]. The discrepancies observed between the laboratory tests and in-situ observations are also for enforcing assumptions that soil particles and pore fluid are incompressible is close to the truth for any saturated soil. The practice of one dimensional consolidation in laboratory also does not corroborate the form of consolidation in the field under a load of limited extent. In case of low hydraulic gradient as the case in Oedometer test, Darcy’s law might also be deviated. It is also inferred that the testing procedures influence the results considerably. Testing a specimen in Oedometer at partially saturated condition is different from testing at saturated stage as most of the desiccation bonds collapse during saturation. Therefore emphasis was given in this experimental work to carry out the consolidation test at natural moisture content to simulate the partially saturated state of the soils.

374

D. Sarma and M. Dsarma / Prediction of Over-Consolidated-Ratio for African Soil

The shapes of the e - log p curves indicate heavily overconsolidated cemented soils in conformity with similar results of Nagaraj et al [19] except their postulation that cementation develops only in marine environment. Attempt has been made to draw a comparison with Skempton’s chart for c/p′ vs. Ip, for the extent of overconsolidation. The plotting confirms that the soils are highly overconsolidated.

6. Prediction for African soil Fine grained soils within the depth of foundation of structure are common in Africa, so as the wider variation of LL, PL with high OCR considered in the present study. The apparent higher OCR due to various reasons stated above, namely, high matrix suction, partially saturated narrow pores, surface tension due to less air content, relative humidity of pore air, degree of saturation, continuous air space, migration of air vapour, desiccated state, salt concentration, bond with natural cementing compounds, skeleton formation associated with small shrinkage, high in-situ void ratio, etc, although common in Africa, it is important that the practising engineers render due consideration of the appropriate situation before use of these correlations for field application. With such careful consideration the correlations presented in this paper for prediction of OCR are useful for African soil too. The sensitivity of the above factors is being reconfirmed for some African soils with completed investigation and being published elsewhere, with further continuation for extended investigation.

7. Conclusion A correlation has been developed involving overconsolidation ratio (OCR) of partially saturated soils, their in-situ void ratio (e), and void ratio at liquid limit (e L). This approach has been adopted because of the versatile nature of the parameter e/eL which has been reported to be related to various soil properties including consolidation. It is also found that most of the soils in nature exist in partially saturated state. Characterisation of compressibility of soils involves sophisticated and time consuming laboratory tests and complicated field works. Therefore, it is often emphasised on evolving simple methods to predict such behaviour of soil from basic index properties. Block sampling has been adopted to reduce the disturbance and the effect is observed in the e - log p curves. Emphasis has been given for modifying conventional testing procedure to simulate the natural state of partial saturation. The effect is observed from the prominent peaks of the e - log p curves. The soils tested are found to be highly overconsolidated and cemented as depicted by the patterns of the e - log p curves, which are caused by the repeated wetting and drying cycles followed by desiccation. This implies that cementation is not necessarily developed only in marine environment. The natural moisture contents of the soils are found to be at their respective plastic limits and fulfil Skempton’s criteria. The range of data considered in this study fulfils the empirical correlation within the scope of study. Such correlation can save considerable time and money particularly in handling bulk samples needed for pavement construction and rehabilitation projects of highways and airports and can be applicable for African soil.

D. Sarma and M. Dsarma / Prediction of Over-Consolidated-Ratio for African Soil

375

Acknowledgement The experimental investigations referred to in this paper were part of postgraduate research work of Author b, which were conducted under the supervision of Prof. P. K. Bora, PhD (Birmingham), Head of Civil Engineering (Retired), Assam Engineering College, India. His encouragement is gratefully acknowledged. The essence of the research in context to the African condition is the contribution of both the Authors.

References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19]

Nagaraj, T. S. and Srinivasa, Murthy, B. R. Prediction of Compressibility of Overconsolidated Uncemented Soils, Journal. Geotechnical Engineering. ASCE 112:4 pp 484-488, 1986. Nagaraj, T. S., Srinivasa, Murthy, B. R., and Vatsala, A. Prediction of Soil Behaviour, Part I – Development of Generalized Approach, Ind. Geotech Jl. 20:4, 1990. Terzaghi K., and Peck, R. B. Soil Mechanics in Engineering Practice, 2nd Edition. John Wiley and Sons, New York (USA), 1948. Skempton, A. W. The Pore Pressure Coefficients A and B. Geotechnique 4, pp. 143-147, 1954. Meyerhof, G.G. Bearing Capacity and settlement of Pile foundation. Journal of Geotech Engineering Division, ASCE 102:3, pp. 137-227, 1976. Brooker E. W, Earth Pressure at rest related to stress history. Canadian Geotechnical J.2, Feb. pp 1-15, 1965. Ladd C. C. Stress-strain Modulus of Clay from Undrained Triaxial Tests. Proceeding. ASCE 90:3 Sept, 1964. Nagaraj, T. S., Srinivasa, Murthy, B. R., and Vatsala, A. Prediction of Soil Behaviour, Part III – Cemented Saturated Soils. Ind. Geotech Jl. 21:2, 1991. Lambe, T. W. and Whitman, R. V. Soil Mechanics, John Wiley and Sons, Inc. New York (USA) pp 320, 1969. Sridharan, A. Altschaeftl, A. G. and Diamond, S. Pore Size Distribution studies, Journal of ASCE, Soil Mechanics Division 97: 5: 771 – 787, 1971. Nagaraj, T. S. and Srinivasa, Murthy, B. R. A Critical Reappraisal of Compression Index Equations, Geotechnique. London 36:1, 1986. Russel E. R. and Mickle, J. L. Liquid Limit values of Soil Moisture Tension, ASCE Journal of Soil Mechanics and Foundation Engineering. Division, 96: 967-987, 1970. Wroth, C. P. and Wood, D. M. The Correlation of Index Properties with some Basic Engineering Properties of Soils, Canadian Geotechnical Journal. Ottawa, May 15:2:137 – 145, 1978. Nagaraj, T. S., Srinivasa, Murthy, B. R., and Vatsala, A. Prediction of Soil Behaviour, Part II – Saturated Uncemented Soils. Ind. Geotech Jl. 21:1, 1990. Sowers G. F. Introductory Soil Mechanics and Foundations, Geotech. Engg. Macmillan Publishing Co., Inc. New York, 1979. Sridharan A. Some Studies on the Strength of Partly Saturated Clays, PhD Thesis, Purdue University, Lafayette, Indiana, 1968. Alam, M. M. and Sridharan, A. Effect of Wetting and Drying on Shear Strength. Journal of the Geotechnical Engineering Division, Proceeding. ASCE, 107: G T4: 421-438, 1981. Scott, C. R. An Introduction to Soil Mechanics and Foundations. 3rd Edition. Applied Science Publishers Ltd., London, 1969. Nagaraj, T. S., Srinivasa, Murthy, B. R., and Vatsala, A. Prediction of Soil Behaviour, Part IV – Partially Saturated Soils. Ind. Geotech Jl. 21:3, 1991.

376 Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-376

The Strength of Compacted Sand in a Modified Shear Box Apparatus

a

F. OKONTAa and D. SCHREINERb University of Johannesburg, Johannesburg, South Africa b University of Kwazulu Natal, Durban, South Africa

ABSTRACT. Samples of compacted Berea sands were tested in a modified shear box apparatus that was modeled after the Cambridge type simple shear device to impose two dimensional simple shear strain on soil sample. Identical samples were also tested in the conventional direct shear box apparatus. The result indicates no significant difference in drained strength mobilized by the test samples in the two devices for the range of applied normal stress. Samples tested in the modified device indicated larger vertical strain at large normal stress range.This may be due to the different mode of deformation imposed on the sample in the two devices, for unlike in the mod device, the conventional shear box subject a limited band of the test sample to shear deformation. Keywords. Berea Sand, Simple Shear Deformation, Shear Modulus

1. Introduction Berea Red Sands are predominantly quartzitic sands that are weakly bonded by clays and and ferromagnesian precipitates of leaching. Berea type formations are weathered coastal plain sands that underlie most of the near coastal, coastal and port cities of Southern Africa from Durban through parts of Mozambique to the port city of Mombasa [1]. Soil structure interaction modeling and geotechnical design of Piles and Retaining Walls are important components of the design of Harbors, Ports and Coastline structures. While the conventional shear box apparatus can be used to determine shear strength, it does not impose shear strain on soil samples and thus the test data cannot be used to determine shear modulus. The conventional shear box was modified to impose simple shear deformation on test samples, as devices that impose simple shear deformation on samples are preferred to triaxial devices for the study of soil structure interaction problems [2, 3, 4]. The major objectives are to develop a device that subjects a typical test sample to shear conditions in which the entire thickness becomes a shear band, and allow for easy measurement of shear induced soil volume changes that are useful in soil structure interface studies. Two main types of devices that impose simple shear deformation on test samples are currently in use. The Norwagian Geotechnical Institute simple shear apparatus, made of

F. Okonta and D. Schreiner / The Strength of Compacted Sand in a Modiﬁed Shear Box Apparatus 377

steel wire - reinforced membrane, was designed to investigate the shear strength of quick clays and soft soils. The device cannot be used for the testing of stiff clays and coarse sands because of significant horizontal strain and bulging of the membrane [5]. The stiff plates of the Cambridge Simple shear apparatus ensures that changes in sample volume can be directly correlated to changes in sample thickness. In addition, the Cambridge apparatus imposes more uniform strain than the Norwegian apparatus mainly because of the better-defined side boundary conditions [6, 7]. The rigid end boundaries in a conventional shear box apparatus remain fixed and impose an overall restriction of zero lateral strain in the soil along the central plane during the testing phase. The same restriction had been observed independently in the Cambridge simple shear apparatus [8, 9, 10]. It is thus necessary to compare results of conventional shear box tests with that of a modified shear box apparatus that was modeled after the Cambridge simples shear device, in order to ascertain whether the results of tests in the modified shear box device can be used for the determination of routine study of effective stress parameters and soil structure interaction modeling.

2. The Modified Shear Box Apparatus The device runs on ball bearing tracks of the conventional shear box frame. The ends of the soil sample are confined by steel end flaps, which were designed to allow for rotation of up to 60 degrees on either side of the central position. A problem common to many Cambridge type simple shear devices is that the axis of rotation of the end plates is not exactly at the intersection of the faces of the end plates and the base [8, 3]. In the new apparatus, modeled after the device by Ansel and Brown (1978), the above problem was overcome by using a bearing centered on the intersection of the faces of the end plates and the base, the bearing being housed in circular sidewall sockets. The development of different loading models and calibration of components of the device shown in Figure 1 are detailed in Okonta (2005). The model showed in Figure 1 subjects soil samples of sizes 100mm X 100mm X 30mm to simple shear deformation through the reaction to the right hand end plate via a horizontal rod. The resistance per unit horizontal area of the sample is a measure of the average shear stress on the sample due to the imposed rotation of the end plates. The rotation of the end plates is expressed in degrees equivalent to the ratio of the horizontal displacement of the top of the sample to the initial sample height. A conventional shear box apparatus that can test 100mm X 100mm samples was also used in this investigation.

3. Physical Properties and Sample Preparation Method The soil samples investigated were taken from a depth of 2.0m – 2.5m alongside M13, the Pinetown – Pietermaritzburg Highway. The result of wet sieve and hydrometer test show that the soil is well graded clayey sand consisting of 18% Clay, 5% Silt, 73% Fine Sand (0.075mm - 0.425mm) and 4% coarse sand (0.425 mm – 4.57mm). Series of tests to determine the physical properties reveal that the specific gravity is 2.70, and the

378 F. Okonta and D. Schreiner / The Strength of Compacted Sand in a Modiﬁed Shear Box Apparatus

proctor maximum dry density and optimum moisture content are 1788kg/m3 and 10.3% respectively. Measured masses of soil samples that were air dried at controlled Laboratory temperature of 20˚C were mixed are 20% moisture contents and statically compacted into the conventional and modified shear boxes in 30mm thick layers with dry densities of 1445 kg/m3 – 1455 kg/m3. The samples were subsequently dried to an average insitu moisture content of 8% and sheared at a slow strain rate of 0.5mm /min. A second set of samples were compacted to dry densities of 1645 kg/m3 – 1655 kg/m3 and similarly tested. All the samples were tested at normal stresses of 400kPa, 100kPa, 200kPa and 400kPa.

Figure 1, Modified Shear Box Apparatus

F. Okonta and D. Schreiner / The Strength of Compacted Sand in a Modiﬁed Shear Box Apparatus 379

4. Results and Discussions The stress strain curves shown in Figure 2 and Figure 3, for samples compacted to densities of 1445 kg/m3 – 1455 kg/m3 and 1645 kg/m3 – 1655 kg/m3, represents the general trend indicated by the result of a total of 18 tests conducted in the modified shear box device and the conventional shear box device respectively. The stress strain response is similar at the early stages of imposed shear deformation but separated as deformation increases. The minimal difference in mobilized stress at large imposed deformation indicated in Figure2 and Figure 3 is due to difference in uniformity of internal stress and strain within the samples in the two devices as well as different modes of deformation and evolution of failure planes [12]. In the direct shear box a horizontal plane of failure is imposed on the sample at the split of the box. The stress and strains curves shown in Figure 2 and Figure 3 are based on average measurements as no measurement over the middle third (i.e. core measurements) were conducted. While the stress strain curves continue to increase with increase in imposed strain in the two apparatus and indicated marginal difference in mobilized shear stress, the vertical strain shear deformation curves revealed significant difference in shear induced vertical strain. While samples tested in the two apparatus indicated shear induced volume compression and reduction in vertical strain, the result shown in Figure 2 and Figure 3, show that the samples tested in the modified shear box apparatus exhibited more compressive vertical strain. The relatively small compressive strain exhibited by the samples tested in the conventional shear box shown in Figure 3, may be due to the narrow band of particles close to the imposed horizontal plane of shear that were involved on the mobilization of shear stress. In the modified shear box device the entire sample thickness constitutes the active shear band [6]. A weak trend of decrease in stiffness with increasing applied normal stress was indicated by the modified shear box device and needed to be further investigated as the modified device should better reflect the stiffness of the entire sample.

380 F. Okonta and D. Schreiner / The Strength of Compacted Sand in a Modiﬁed Shear Box Apparatus

Figure 2. Shear stress and vertical strain curves of medium dense Berea Sand in Modified Shear Box Device

Figure 3. Shear stress and vertical strain curves of medium dense Berea Sand samples tested in the Conventional Shear Box Device

F. Okonta and D. Schreiner / The Strength of Compacted Sand in a Modiﬁed Shear Box Apparatus 381

References [1] Brink, A.B.A. (1984), Engineering Geology of Southern Africa. Vol. 4. Building Publications. Silverton. Pretoria. [2] Randolph. M.F., and Wroth, C.P.(1981), Application of the failure state in undrained simple shear to the Shaft capacity of driven piles. Geotechnique, Vol. 31, No1. pp 143-157. [3] Ansell, P, and Brown, S.F. (1978), A cyclic simple shear apparatus for granular materials. Geotechnical Testing Journal, Vol1, No2, p 82-92. [4] Blight, G. E. (1997), Properties of Residual Soils, A. A. Balkema, Rotterdam, pp236. [5] Vucetic, M., and Lacasse, S. (1982). Specimen size effect in simple shear tests. Proceedings of ASCE, 108, (GT12), pp1567-1585. [6] Budhu, M. (1984). On comparing Simple shear and Triaxial test results. Journal of Geotechnical Engineering.Vol 110, No 12 [7] Airey, D.W,. Budhu, M. and Wood, D.M. (1988), Some aspect of the behavior of soils in simple shear. Proceedings of the 5th Australian and New Zealand. Geomechanics Conference, Sydney, pp18-39. [8] Roscoe, K.H. (1953), An apparatus for the application of simple shear to soil samples. Proceeding of the 3rd International Conference and Foundation Engineering. Zurich, Vol.1, pp186-191. [9] Dyer MR (1985) Photoelastic Investigation of reinforced sands Proceedings of Int Conf on Reinforced Soils and Rocks pp 437 - 443 [10] Jewel, R. A. (1989), Direct shear behavior of sands. Geotechnique 39, No. 2, 309- 322. [11] Okonta F N (2005) Capacity of Vertically loaded piles in low density sands. PhD thesis, University of Natal [12] Maccarini. M, Laboratory studies of a weakly bonded artificial soil. PhD thesis, University of London.

382 Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-382

Experimental Study on use of Mechanically Stabilized Residual Soils for Pavement Layers in Magoe, Mozambique Raphael NDIMBO

TÉCNICA Engenheiros Consultores, Maputo, Mozambique Abstract. This study is an investigation of the effects of stabilizing 2 types of residual soils (sample A and B) with natural river sand (sample C) in Tete Province, Mozambique. Various combinations of soils and sand (sample A+C, B+C and A+B+C) are assessed for the purpose of improving their strength to conform as pavement materials. Laboratory classification tests (grading and Atterberg Limits) were conducted on each of the 3 natural materials and strength tests (CBR and compaction) were performed on each of the 2 soil samples (sample A and B) and on the mixtures of soils and sand (A+C, B+C and A+B+C) in different proportions of the blended materials. The results of strength tests indicated that the addition of sand to sample A reduced the maximum dry density (MDD) and the optimum moisture content (OMC) and the CBR values (for 1-day soaked) decreased. For sample B, addition of sand caused a decrease in OMC and an increase of the MDD and CBR (1-day soaked). The optimum proportions of the blended material were obtained with 20% sand (C), 20% of soil sample B and 60% of soil sample A which gave a CBR of 98% after 4 days of soaking. A trial section of base course with materials dumped in the ratio of 1 truck of soil sample B, 1 truck of sand sample C and 3 trucks of soil sample A gave a 4-day soaked CBR of 79%. The study indicates that sand may be mixed with several soils from different borrow pits to improve the properties for road construction materials. Laboratory and field tests are presented to compare the properties of the mechanically stabilized trial section and the optimum design parameters. Keywords. Stabilization, treated sample, blending, compaction

Introduction Use of local materials should always be explored in road construction works in order to reduce costs associated with long haulage distances. However, naturally occurring soils seldom meet the required geotechnical properties for roadworks. The SATCC specifications [1] require a gravel base material to have a soaked CBR>80% at 98% mod. AASHTO density, plasticity index PI<6% and minimum grading modulus of 2. Modification of the properties of available local materials is sometimes needed in order to produce suitable road construction materials [2]. Soil stabilization has been widely used to improve the handling and engineering characteristics of soils for pavement layers. It is the process by which natural soils are treated by addition of certain materials to it by mechanical, chemical or electrical means [3]. The aim of this study is to understand the behaviour of the local materials and design an appropriate mix to achieve satisfactory performance when constructed in the pavement layer. This report presents the results of an investigation aimed at evaluating

383

R. Ndimbo / Experimental Study on Use of Mechanically Stabilized Residual Soils

the influence of sand on the strength of the mechanically stabilized mixtures. Sand was added in various proportions to 2 borrow pit soils in order to determine the effect on the strength (CBR) of the modified soils.

1. Properties of Materials The residual soil samples used in this study were collected from 2 existing borrow pits along the Estima – Magoe road area in Mozambique. Both samples were excavated from test pits at a depth of between 0.5m and 1.5m. A sample designated “A” was calcrete in the form of weathered hardpan and was dug at a borrow pit Km 101 RHS. The second sample “B” was excavated from a borrow pit located at Km 96+920 RHS and the soil was nodular calcrete. The sand sample, “C”, used in the experimental tests was river sand type collected from a dry seasonal stream at km 98+000. Sampling and testing was done between January and February 2010. All the samples used in the study were air dried and then prepared for testing according to the required standard test methods of AASHTO – 2000. 1.1 Identification and Classification of Materials The results of tests for identification and classification of natural materials before modification of samples are summarized in Table 1. Table 1. Classification test results of untreated samples Material properties Gradation, % Passing 37.5mm 19mm 9.5mm 4.75mm 2.0mm 0.425mm 0.075mm Grading Modulus, GM Atterberg Limits: LL PI LS Soil Classification: AASHTO USCS Material description

Sample A

Sample B

95.6 78.9 64.0 48.7 31.0 19.2 12.7 2.4 37.5 11.4 6.7 A-2-6(0) GC

100.0 83.6 70.6 57.3 42.0 29.4 20.2 2.1 33.7 13.7 8.0 A-2-6(0) GC

Clayey gravel

Clayey gravel

Sample C

100.0 98.8 85.3 15.0 2.7 NP

A-1-b SP Poorly graded sand

Soil sample A is classified as GC with 51% gravel, 36% sand and 13% fines while sample B is clayey gravel with 43% gravel, 37% sand and 20% fines. Sample C has a coefficient of uniformity 5.7 and a coefficient of curvature 1.5 and with 96% sand can be classified as poorly graded sand. The tested soils have percentage passing 0.075mm of less than 35% which indicates granular materials. The geotechnical properties of these materials may be rated as fair to poor for pavement layers but can be modified for better performance. It

384

R. Ndimbo / Experimental Study on Use of Mechanically Stabilized Residual Soils

can be noted from Table 1 that the sand may reduce fines content and Atterberg limit values of blended materials when added to the 2 soil samples in proper proportions. The gradation curves of all the materials used in the laboratory blending are presented in Figure 1.

Figure 1. Gradation curves of materials used in the blending tests

1.2 Compaction and CBR of Materials The compaction effort used in this study for Proctor and CBR tests was 4.5 kg rammer (modified effort). The test procedures for CBR were carried out on materials passing 19mm sieve in accordance with AASHTO designation T 193 and the related compaction tests according to AASHTO T180. The results of compaction and CBR tests for soil samples with and without addition of sand are presented in Table 2. The preliminary mixtures of sand with soil samples “A” and “B” separately gave unsatisfactory results of CBR. However, from the laboratory results it was noted that a 20 - 25% sand content gave comparatively good gains in strength. It can be observed that in all cases of blending, addition of sand reduced the OMC of the resultant mixture. The CBRs of unsoaked samples were high but strength of soaked samples was relatively low due possibly to water changing the clay micro structure and affecting the stress and pressure in the soil.

385

R. Ndimbo / Experimental Study on Use of Mechanically Stabilized Residual Soils Table 2. Summary of laboratory test results of compaction and CBR

Sand Content (%)

Sample Designation

0 16.7 20 25 33.3 50 0 16.7 20 25 33.3 50

A Borrow Pit Km 101 RHS

B Borrow Pit Km 96+920RHS

Modified Proctor

CBR at 100% MDD (%)

OMC (%)

MDD (kg/m3)

Unsoaked

10.6 9.1 8.9 8.0 7.3 6.5 9.2

2011 1931 1938 1964 1994 1936 2014

102

8.8

2064

1 Day soaked

4 Days soaked 52

71

57 14 40 36 17 10 20

74

29

12

2. Experimental Results and Discussion 2.1 Test Procedure It was further decided to investigate the strength properties of a mixture of both soils with sand in proportions of 10%, 16.7% and 20%. These proportions were decided upon due to encouraging results obtained from simulation of gradation with the 2 soil samples. Practical aspects of dumping ratios of truck loads with various materials at site also influenced the choice of sand proportions. CBR and modified Proctor tests were performed on mixtures which weighed exactly 6.0 kg in total out of the proportions indicated in Table 3. The OMC and MDD obtained from each compaction test of the mixtures were used to mould CBR specimens at 100% MDD. The CBR penetration tests were performed for unsoaked samples and after soaking in water for 24 hours and after 4 days. 2.2 Test Results of Laboratory Blending of Samples Table 3. Test Results of strength properties of combined soils treated with sand Sample Mix Proportions (%)

Modified Proctor

Mix No.

A B/Pit 101+000

1 2 3 4

80 50 60 40

CBR at 100% MDD (%)

B B/Pit 96+920

C River Sand

OMC (%)

MDD (kg/m3)

Unsoaked

1 Day soaked

4 Days soaked

10 33.3 20 40

10 16.7 20 20

12.0 11.1 11.4 10.4

1940 2024 1992 2061

93 105 110 117

71 83 86 63

80 41 98 60

386

R. Ndimbo / Experimental Study on Use of Mechanically Stabilized Residual Soils

Table 3 presents strength test results from the laboratory blending of both soil samples combined in different proportions with various percentages of river sand. From Table 3, it can be noticed that generally the CBR values of the combined modified soils are higher than those of the individually treated soils (from Table 2). The highest CBR value of 98% was reached after 4 days of soaking a combined sample with 20% sand added to 20% of soil “B” from borrow pit at km 96+920 RHS and 60% of soil “A” from borrow pit at km 101 RHS. From the results of strength tests in Tables 2 and 3, it can be noticed that the OMC decreases with an increase in sand content. This is due to the decrease in the total surface area of particles caused by addition of coarse particles of sand. Increased amounts of sample B caused an increase in MDD and unsoaked CBR possibly due to its denser particles. Generally the grading of the blended materials shifted from course graded to fine graded with the increase of sample B (see Figure 2). The PI values of 11.4% and 13.7% (from Table 1) are both above the maximum required value of 6.0% for natural gravel base course of class G4. Generally, the Atterberg Limits results of both soil samples as shown in Table 1 are above the maximum values recommended for base course material of natural gravel by TRH 14 (1985) – Guidelines for Road Construction Materials. However, the mechanical stabilization by addition of sand should decrease the PI and other consistency limits to acceptable values. The linear shrinkage values of the borrow pit samples are below 8% which indicate that the soils are inexpansive and hence have low swelling potential. 2.3 Test Results of Field Blended Base Course The construction of the 150mm base course layer on the trial section of 100m involved initially dumping of materials on the prepared subbase layer. The sequence of dumping was in the form of sample A, B, A, C, A for every 20m section of the road. The materials were then mixed, watered and processed by motor graders. Field compaction was performed by vibratory rollers. A summary of test results of mechanically stabilized soil samples with river sand from the experimental section are given in Table 4. Construction, sampling and testing of the trial section was conducted between April and May 2010. The materials used in the trial section were excavated from the borrow pits at a depth between 1.5 – 2.5m while the materials used in the laboratory mixtures were dug by hand tools at a depth of 0.25 – 0.75m. The test results of materials from the two excavations should not be compared directly because the materials are not homogeneous. However, the results indicate the trend of changes caused by blending different materials. Table 4. Test results of field stabilized base course LL

PI

LS

41.8

8.9

5.3

37.5 mm 90.0

4.75 mm 49.9

0.075 mm 13.9

GM 2.3

MDD kg/m3 1982

OMC % 9.1

Soaked CBR @ 100% MDD 79

Particle size distribution chart of the sample taken from the road after compaction is plotted in Figure 2.

R. Ndimbo / Experimental Study on Use of Mechanically Stabilized Residual Soils

387

Figure 2. Field and simulated gradation curves of blended materials

According to the AASHTO classification system, the blended material from the road is A-2-5. It is silty gravel with 36% sand and 50.1% gravel and 13.9% fines. The shape of the grading curve, in Figure 2, is centrally placed within the grading envelope, which indicates good compactability. The plasticity index has decreased due to addition of sand up to 8.9% in comparison to preliminary component materials. The CBR of the blended material is lower than predicted from preliminary laboratory tests. However it is higher than the CBR of untreated materials. The decrease in CBR may have been caused by poor proportioning of the constituent materials of the blended sample. The reached CBR of 79% and PI of 8.9% classifies the blended material as G5 as per SATCC specification.

Conclusions The laboratory test results in this study show that sand may be used in mechanical stabilization of residual soils for improvement of strength properties. The results of mechanical stabilization are influenced by the process of choosing the optimum stabilized proportions of various ingredients of the blended materials. Addition of sand decreases the plasticity index of the stabilized material. The simulation of gradation curves for blended materials can assist in choosing the proper materials for optimum proportions of ingredients of mechanically stabilized mix. The laboratory tests carried out on the mixtures of 60% of soil from borrow pit at km 101RHS, 20% of soil from borrow pit at km 96+920RHS with 20% of river sand revealed marked increase in strength to meet the project specified requirements for base course material of low-volume roads. This study has revealed that tropical residual soils can be mixed in proper proportions with river sand in order to comply with the strength requirements for pavement materials.

388

R. Ndimbo / Experimental Study on Use of Mechanically Stabilized Residual Soils

This type of mechanical stabilization may achieve high CBR values and low PI values if the different component materials are properly dumped on the road. With proper awareness starting from the Designer to the Contractor, it may be a cheaper construction method than chemical stabilization.

References [1] [2] [3]

SATCC 1998 - Standard specifications for Road and Bridge Works. B. K. Sahu – Improvement in CBR of Various Soils in Botswana by Fly Ash. O. Omotosho, O. J. Eze-Uzomaka – Optimal Stabilization of deltaic laterite

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 389 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-389

Effects of Compaction on Engineering Properties of Residual Soils of Tete Mozambique Carlos QUADROS and Raphael NDIMBO TÉCNICA Engenheiros Consultores, Maputo, Mozambique

Abstract.

Several samples of residual base course material were studied to investigate the effects of compaction on their physical and mechanical properties. Test results of compaction and CBR obtained for borrow pit material are compared with repeatedly used soil samples from the road. From analysis of changes of strength of materials before, during and after construction, and also site observations, it is recommended to establish not only the minimum requirements for compaction density, but also the maximum density. Some guidelines are discussed for control of the level of compaction at the site.

Keywords. Granular base, Compaction, California Bearing Ratio, Plasticity, Field Density

Introduction In most tropical areas, residual soils are used as a road construction material and they form subgrade, subbase and base layers of roads which carry low to medium traffic. Tropical residual soils are formed by the in situ decomposition (chemical weathering) and/or the disintegration (physical weathering) of the underlying parent rock under the intense conditions of tropical climate. The weathering process of tropical soils is ongoing and the materials exist at various stages of transformation from slightly weathered near the fresh bedrock to completely weathered near the top soil. The weathered particles are sometimes mechanically weak and liable to breakdown under handling, compaction or loading. Nevertheless, the breakdown of coarser particles during field rolling may be considerably less than under laboratory test conditions. Strength of materials after field compaction tends to be higher than is obtained from laboratory test results. The geotechnical properties of residual soils are quite varied and only in rare cases do the natural materials satisfy the minimum specification requirements for granular base course [4]. However, there are many examples, especially in Southern Africa, where marginal quality materials, which do not conform to normal specifications for bases, have performed successfully [2]. The aim of this paper is to illustrate the effect of the level of compaction on some engineering characteristics of residual soils based on laboratory and field test results.

390

1.

C. Quadros and R. Ndimbo / Effects of Compaction on Engineering Properties of Residual Soils

Test Materials and Methods

1.1 Material Quality Requirements The SATCC specifications [1] requires a gravel base material to have a soaked CBR>80% at 98% mod. AASHTO density, plasticity index PI<6% and minimum grading modulus of 2. Such a material is classified by the South African TRH 14 Guidelines for the Road Construction Materials [3] as G4 material. A study conducted by TÉCNICA of the results of the borrow pit testing along the project road revealed availability of suitable granular base material from km 0 to km 42. The road section from km 42 to km 134, within which falls the study area, indicated mostly subbase quality material (soaked CBR of 30 – 50%). With reference to some research programs on highway engineering materials in the SADC region, it was proposed for the low-volume road project a 150mm of granular base with a soaked CBR of 45% compacted to 100% mod. AASHTO density. 1.2 Sampling and Preparation of Materials The materials used in this study were residual soils excavated in March 2010 from an existing borrow pit at a depth of between 1.5 and 2.5m below ground level. Sampling for the test materials was performed from 2 sections of the road after spreading the material by motor grader and also from the same 2 areas after field compaction by vibratory rollers. The samples for laboratory compaction tests were collected in November 2010 from the same borrow pit at a depth of 2.5m, near the bedrock. All the soil samples were air dried and then prepared for testing according to standard methods of preparing disturbed soil samples (AASHTO T87). 1.3 Test Methods Grading analyses were conducted by wet sieving according to the AASHTO standard test method T146. Liquid Limits and Plastic Limits were determined on materials passing the 0.425mm sieve in accordance with AASHTO T89 and T90 respectively. Linear shrinkage tests were performed as outlined in BS 1377-4: 1990, Classification tests [5]. CBR tests were performed on materials passing 19mm sieve according to AASHTO T193 with related moisture contents and maximum dry densities in accordance with AASHTO T180 (Method D). Standard and modified compaction tests were done in accordance with standard procedures outlined in AASHTO T99 and T180 respectively. Compaction tests, for materials sampled from the road, were carried out according to AASHTO T180 (modified Proctor) for soils with particles susceptible to crushing. It required separate compaction for each sample of soil with different moisture contents. Compaction tests for samples collected from the borrow pit were conducted according to AASHTO T90 (standard Proctor) and T180 for both separate compaction using different soil batches and repeated compaction, using the same soil batch with different moisture contents.

C. Quadros and R. Ndimbo / Effects of Compaction on Engineering Properties of Residual Soils

2.

391

Results and Discussions

2.1 Soil classification tests The results of classification tests performed on soil samples collected from the road are summarized in Table 1. Table 1: Soil classification test results Grading, % Passing Sample location

LL %

PI %

LS %

km

Max Size mm

4.75 mm

µm

75 GM

Soil Classif. A U A S S H C S T O

Remarks

Before field compaction (bfc 1) After field compaction (afc 2)

79+50080+000

29.0

8.0

4.7

63

43.0

16.3

2.30

A-2-4

GC

79+50080+000

30.7

8.4

5.3

37.5

57.2

18.0

2.01

A-2-4

GC

78+50079+000

31.8

9.6

5.3

53

48.3

19.5

2.22

A-2-4

GC

Before field compaction (bfc 3)

78+50079+000

33.9

10.1

6.0

37.5

65.2

21.8

1.92

A-2-4

SC

After field compaction (afc4)

According to the Unified Soil Classification System (USCS), all the 4 samples shown in Table 1 lie above the A-line in the plasticity chart indicating clayey materials. The Atterberg Limit values of both samples taken after field compaction (afc 2 and afc 4) are higher compared to corresponding values of samples collected before field compaction (bfc 1 and bfc 3). The increase in Atterberg Limit values is a result of a decrease in particle size, caused by disintegration of coarser particles increasing the soil plasticity. However, the Liquid Limits of the samples are all less than 35% which indicates low plasticity of the materials. The gravel fraction of samples after field compaction decreased compared to before road compaction from 57.0% to 42.8% for one sampling area and from 51.7% to 34.8% for the second section. The percentage of sand increased after field compaction from 26.7% to 39.2% and from 28.8% to 43.4%. The fines content (passing 0.075mm sieve) increased after compaction from 16.3% to 18.0% and from 19.5 to 21.8%. From the grading test results plotted in Figure 1, it can be observed that field operations caused shifts in the gradation curve shape and position from coarse gradings near the bottom limit of envelope to fine gradings near the top limit. These changes are also reflected in the decrease of grading modulus values of samples collected from the road after field compaction.

392

C. Quadros and R. Ndimbo / Effects of Compaction on Engineering Properties of Residual Soils

Figure 1: Gradation curves of gravel base materials

2.2 CBR and Modified Compaction Tests Table 2 presents the results of strength tests on the samples taken before and after compaction of the road. Table 2: Summary of strength tests Sampling Location Km 79+50080+000 79+50080+000 78+50079+000 78+50079+000

MDD (kg/m3 )

OMC (%)

4-day soaked CBR @ 100% MDD (%)

Remarks

2018

9.1

72

Before field compaction (bcf 1)

1997

9.6

65

After field compaction (afc 2)

2006

8.7

60

Before field compaction (bfc 3)

1973

9.8

52

After field compaction (afc 4)

From Table 2, it can be noted that an increase of OMC for a sample before and after field compaction causes a decrease of MDD. This may be because of increased clay content, since it is reflected in the increase of PI as indicated in Table 1. The CBR values, as shown in Table 2, of samples after field compaction are relatively lower compared to corresponding samples before compaction. This is possibly because the breakdown of soil particles, as reflected in Figure 1, reduced the compactability of the samples which caused a decrease in density and CBR strength.

C. Quadros and R. Ndimbo / Effects of Compaction on Engineering Properties of Residual Soils

393

2.3 Standard and Modified Compaction Tests A summary of laboratory compaction tests performed on gravel base materials using different compactive efforts are shown in Table 3. Borrow pit materials were sampled from a stockpile and base layer materials were sampled after field compaction of the materials from the same borrow pit at km 83+300 RHS. Table 3: Summary of laboratory compaction tests

Sampling Location

Compactive Effort

Type of compaction

MDD (kg/m3)

OMC (%)

1

B/Pit

2.5kg rammer

Repeated

2042

15.8

2

B/Pit

2.5kg rammer

Separate

2029

14.0

3

B/Pit

4.5kg rammer

Repeated

2212

10.3

4

B/Pit

4.5kg rammer

Separate

2128

10.2

5

Base Layer

2.5kg rammer

Repeated

2098

13.1

6

Base Layer

2.5kg rammer

Separate

2043

12.5

7

Base Layer

4.5kg rammer

Repeated

2183

11.1

8

Base Layer

4.5kg rammer

Separate

2108

10.4

Remarks Borrow Pit Km 83+300RHS Borrow Pit Km 83+300RHS Borrow Pit Km 83+300RHS Borrow Pit Km 83+300RHS Sampled after field compaction Sampled after field compaction Sampled after field compaction Sampled after field compaction

It can be observed from Table 3 that as the compactive efforts increased from the standard effort, using 2.5kg rammer, to the modified effort, using 4.5kg rammer, the values of MDD increased while the corresponding values of OMC decreased. Comparison of compaction results obtained for separately and repeatedly compacted soil samples shows that repeatedly used soils have higher MDDs. High values of OMC for repeatedly used samples may be caused by the breakdown of coarser particles. This in turn increases the specific surface area of the sample and increases the water required to lubricate the particles. 2.4 Field DensityTests Conventional compaction control is by dry density from field density determinations expressed as a percentage of the maximum dry density obtained in the laboratory using a modified (4.5 kg rammer) compactive effort. The SATCC specification requires for gravel base material to have minimum field compaction levels of 98% or 100% of the modified AASHTO maximum dry density. The results of the in situ tests for the base materials, conducted by using sand replacement method, are given in Table 4. The average density, for the road section from km 79+500 to 80+000, is 100.3% as shown in Table 4. The value is lower than the required minimum average density of 100.6% for the 6 samples, according to SATCC specifications. The average density, for the section from km 78+500 to 79+000 is 103.2% and is higher than the required minimum of 100.7 for 7 samples. The minimum single test values are in both cases

394

C. Quadros and R. Ndimbo / Effects of Compaction on Engineering Properties of Residual Soils

higher than the minimum specified values. However, the in-situ CBR tests by DCP, also presented in Table 4, show a higher CBR value of 105% for the section from km 79+500 to 80+000. Table 4: Field density test results of base layer

Location CH. 79+500 R 79+575 L 79+650 C 79+715 R 79+900 L 79+975 C 78+500 C 78+575 R 78+650 L 78+725 C 78+800 R 78+875 L 78+950 C

Dry density (kg/m3 )

Moisture Content (%)

Max. Dry Density (kg/m3 )

1994

9.4

2018

9.1

98.8

2022

9.2

2018

9.1

100.2

2046

8.8

2018

9.1

101.4

2019

9.0

2018

9.1

100.0

2030

9.1

2018

9.1

100.6

2038

8.9

2018

9.1

101.0

2074

8.9

2006

8.7

103.4

2070

8.6

2006

8.7

103.2

2065

9.0

2006

8.7

102.9

2083

8.8

2006

8.7

103.8

2068

9.2

2006

8.7

103.0

2062

9.5

2006

8.7

102.8

2069

8.7

2006

8.7

103.1

OMC (%)

Relative Compaction (%)

Average Density (%)

Min. Single value, (%)

DCP CBR (%)

100.3

98.8

105

103.2

102.8

92

From Table 4, one can observe that for the section of the road from km 79+500 80+000 the average field moisture content is lower than the OMC and field dry densities are higher than the MDD. This is appropriate for a material indicating no excessive break down of coarse particles. For the other section (km 78+500 – 79+000) the field moisture content is higher than the OMC and the dry densities are high. This may have been caused by either excessive handling and disturbance of soil particles during mixing and processing the base layer or excessive compaction during field rolling. Though in both sections the percentage of fine materials passing 0.075mm sieve increased after field compaction, it seems that the breakdown of particles in the section from km 78+500 to 80+000 was more detrimental to the gradation and strength properties. The classification of the material changed from clayey gravel before field compaction to clayey sand after compaction, as shown in Table 2. Visual inspection of excavated field density holes (km 78+500 – 79+000) showed a thin layer around the surface and a densely compacted surface below. The thin layer

C. Quadros and R. Ndimbo / Effects of Compaction on Engineering Properties of Residual Soils

395

may have been caused by overcompaction and a subsequent breakdown of coarser particles.

Conclusions Test results indicate that Atterberg Limit values increase with decreasing particle size produced during field operations. Atterberg Limit results are used to identify the type of soil and they may also be used for correlations with other engineering properties such as strength and permeability. An increase of compactive effort from standard compaction to modified compaction caused the increase of MDD values and the decrease of OMC values. This may be true particularly in cases where the breakdown of particles is insignificant. However, additional compaction energy above the maximum density would be damaging to particles as breakdown may occur to cause a decrease in density. The compaction data shows that repeatedly used soil samples give higher values of MDD and OMC than separately compacted samples. This seems to be the case mostly for soil particles less susceptible to crushing.

Recommendations In order to minimize excessive compaction, it should be encouraged to construct trial sections and establish the appropriate number of passes of rollers which guarantee acceptable densities. The SATCC specification [1] sets only the minimum acceptance limits for the average densities and single test values. This creates a high degree of variability which should be restricted by upper limits.

References [1] [2] [3] [4] [5]

SATCC 1998 - Standard specifications for Road and Bridge Works. A. Lionjanga, TRL, UK 1987 -Development of specifications forCalcretes in Botswana. TRH 14 – Guidelines for Road Construction Materials, S.A. B. Clegg (1983)– Design Compatible Control of Base Course Construction, Australian Road Research pp112-122. BS 1377-4:1990 Methods of Testing Soils for Civil Engineering Purposes, BSI, London.

396 Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-396

Selection of pavement foundation geomaterials for the construction of a new runway Joseph ANOCHIE-BOATENG1 CSIR Built Environment, Pretoria, South Africa

Abstract. The selection of foundation geomaterials including base/subbase unbound granular aggregate is critical for the design of flexible airfield pavements in the United States. These materials must meet the Federal Aviation Administration (FAA) selection criteria for airport pavement construction. A comprehensive testing study was conducted to select granular materials for the construction of a proposed runway North Carolina, United States. The objective was to evaluate granular materials and subgrade soils within the Greensboro area, to establish their pertinent engineering properties including strength and deformation parameters. This paper focused on three granular materials selected for the runway construction. The test results including grading, CBR, shear strength and resilient modulus indicated that all the three samples met the FAA selection criteria as base/subbase materials for the construction of the runway.

Keywords. Resilient modulus, Shear strength, CBR, runway, granular materials

Introduction Road and airfield pavement materials found at selected sites and within the locality are often evaluated and considered for use as construction materials. The in-place materials may be removed and replaced with a higher quality material, or they may be modified in some manner to provide qualities that meet construction specifications. A comprehensive laboratory testing study was recently conducted at the University of Illinois at Urbana-Champaign to support an expansion project of an international airport in North Carolina. The project involved the construction of a hub facility and a new runway capable of accommodating large aircrafts. As part of the study, three granular materials were selected for evaluation. Generally, granular base/subbase materials and subgrade soils i.e., geomaterials, constitute the pavement foundation materials. The granular base layers serve a variety of purposes including reducing the stress applied to the subgrade layer and providing drainage for the pavement. This paper presents the engineering properties of three granular materials for a runway. The properties include, CBR, shear strength and resilient modulus. These properties are used by United States Federal Aviation Administration (FAA) to characterize granular materials for airport pavement thickness design [1, 2]. 1

Corresponding Author. Senior Researcher, CSIR Built Environment, Transport Infrastructure Engineering, Bldg 2C, P O Box 395, Pretoria, 0001, South Africa; E-mail: [email protected]

397

J. Anochie-Boateng / Selection of Pavement Foundation Geomaterials

1. Granular Materials and Properties Granular materials (crushed aggregates) from three quarries in North Carolina were selected for this study. Representative samples were obtained from Jamestown, North and Pomona quarries in the Greensboro area. The samples were tested to determine the CBR, compaction density and optimum water content and gradation. Additional tests for shear strength and resilient modulus were performed on the sample from the Pomona quarry. A detailed description of sample preparation, testing equipment and test procedures used for this study are provided by [3]. All test specimens were prepared in accordance with ASTM and AASHTO testing procedures that have been recommended by the FAA for airport pavement designs.

2. Laboratory Testing Program 2.1. Sieve Analysis Test Sieve analysis tests were conducted to determine grading of the crushed aggregates at the three stone quarries. The tests were performed in accordance with America Society of Testing Materials (ASTM) test method [4]. Grading of granular materials is an indicator of aggregate performance and it is one of the criteria used by the FAA for selecting crushed aggregate as a base material for an airport pavement. The FAA specification requires the maximum fines content of 8% for base materials [5]. Figure 1 compares the grading results of the three samples studied. The FAA minimum and maximum requirement of base course materials is compared in the figure. The fines content were approximately 5.5% for the Jamestown sample, 8.0% for the Pomona sample, and 8.7% for the North sample. It can be seen that the North sample did not meet the FAA criteria of 8% fines although it barely failed to meet this criteria. All the three samples have nearly the same grading characteristics. The materials may therefore, have similar strength and engineering properties. 100 Jamestown Quarry

90

North Quarry

Percent Passing by Mass

80

Pomona Quarry

70

FAA Min Limit

60

FAA Max Limit

50 40 30 20 10 0 1000

100

10 1 Grain Size (mm)

0.1

0.01

0.001

Figure 1. Grading of granular materials from Jamestown, North and Pomona quarries.

398

J. Anochie-Boateng / Selection of Pavement Foundation Geomaterials

2.2. Moisture-Density Test A laboratory moisture-density test was performed on the three samples using [6]. This test was performed to determine the water content needed to achieve the maximum dry unit weight of the granular materials. These values would be used to control field compaction during the construction of the runway. Compactions of aggregate materials generally increase density, shear strength, and stiffness, and decreases permeability. Thom and Brown [7] reported that resistance to rutting in granular materials under repetitive loading improves when the density is increased. It is well know that unbound granular materials resilient modulus decreases as the moisture content increases. Figure 2 shows the compaction properties results of the three samples studied. The optimum water content of the North, Jamestown and Pomona samples were 5.5%, 6.2% and 6.5%, respectively and the corresponding maximum dry densities were 23.6 kN/m3, 23.2 kN/m3, and 23.1kN/m3. The results indicate that the compaction properties of all three samples are close, although the same from the North quarry has slightly better compaction properties than the samples from Jamestown and Pomona quarries. 24.5 North 24

Jamestown Pomona

Dry Density (kN/m3)

23.5

23

22.5

22

21.5

21 0

2

4 6 8 Moisture Content (%) Figure 2. Moisture-density relationship of Jamestown, North and Pomona granular materials.

10

2.3. California Bearing Ratio (CBR) Test California Bearing Ratio (CBR) value is a strength parameter used by the FAA for airport pavements design. The FAA AC 150/5320-6D specifies that the minimum CBR value of 80 is required for a crushed stone to be used as a base material for airport

399

J. Anochie-Boateng / Selection of Pavement Foundation Geomaterials

pavements. The CBR test was conducted on the granular materials samples using the ASTM D 1883 [8]. Soaked CBR tests were performed on the three quarry materials. Figure 3 shows the CBR test results of the three samples. The results indicate that all the three granular materials had CBR values greater than 100. Accordingly, all the three granular materials met the FAA specifications. A comparison of the CBR values and the maximum densities of all the three quarries were made. The North sample with the highest maximum dry density had the highest CBR value of 230 when compared with Jamestown sample (CBR = 200) and North sample (CBR = 180). It follows that compaction characteristics had an effect on the strength of the materials. 30000

25000

Bearing (kPa)

20000

15000

10000 Pomona North 5000

Jamestown crushed stone

0 0

2

4

6

8

10

12

Penetration (mm)

Figure 3. Comparisons of load-penetration curves of the three granular materials with standard crushed-rock.

3. Stiffness and Shear Strength Tests Resilient modulus and shear strength tests were performed on the Pomona sample to evaluate the stiffness and strength properties. Only Pomona sample was enough to prepare samples for these tests. A 150 mm diameter by 300 mm high cylindrical specimens of the granular base materials were prepared for conducting both the shear and resilient modulus tests. The specimens were prepared at the optimum water and maximum dry density, and compacted using a pneumatic vibratory compactor. 3.1. Resilient Modulus Testing on Pomona Sample Repeated load triaxial tests were conducted on the Pomona sample to determine its resilient modulus properties following the standard test procedure [9]. The data recorded in this test were bulk stress θ, resilient modulus MR and the deviator stress σd

400

J. Anochie-Boateng / Selection of Pavement Foundation Geomaterials

at 15 stress states. The test specimen was compacted at the optimum water content of 6.5%, and the maximum dry density of 23.1 kN/m3. Figure 4 shows the resilient modulus results for the Pomona sample. It can be seen that resilient modulus increased with bulk stress, which is typical of granular materials. The phenomenological K-θ model presented in Equation 1 was used to estimate the modulus value of the sample. The resilient modulus test data were analyzed used to develop the parameters (n, k) for the model.

M R = kθ n

(1)

where MR = resilient modulus; k and n are material constants obtained from regression analysis. The MR model was expressed in logarithmic relationships to transform the power functions into linear expressions having two separate terms. A generalized resilient modulus model obtained for the Pomona is represented by Equation 2. M R = 64 .255 × θ 0.2202

(2)

Resilient Modulus, MPa

500

y = 64.255x0.2202 R² = 0.7718

100 50

Bulk Stress, kPa

500

Figure 4. Resilient modulus-bulk stress relation for Pomona granular material.

3.2. Shear Strength Results for Pomona Sample The shear strength tests were performed at confining pressures of 35 kPa, 69 kPa and 103 kPa to determine the friction angle (φ) and cohesion (c) used to define the MohrCoulomb failure envelope. Using the University of Illinois in-house shear test

J. Anochie-Boateng / Selection of Pavement Foundation Geomaterials

401

procedure, the deviator stress was applied axially at a constant displacement rate of 38mm/second (strain rate of 12½ percent per second). The linear relation between the shear stress and the normal stress was used to determine the friction angle and cohesion of the granular material. The relation is expressed as: τ = c + σ tan φ

(3)

where τ is the shear stress and σ is the normal stress. Figure 5 shows the test results represented by Mohr circles at failure for the three confining stress states of the Pomona sample. High c value is associated with high resistance of the granular material to shearing stresses, and high φ value implies ability of the Pomona sample to develop strength and resist rutting under aircraft loading on the runway. Shear strength parameters were obtained by drawing a straight line that is tangent to the circles. The results obtained (φ = 58 deg, c = 96.6 kPa) can be used to determine the maximum shear strength of the granular material. These strength parameters are put in the relation, τ = c + σ tan φ to model the granular material. 1400

τmax = 1.6σn + 96.6

1200

35 kPa 69 kPa 103 kPa

Shear Stress τ n (kPa )

1000 800 600 400 200 0 0

500

1000 Normal Stress σ n (kPa )

1500

2000

Figure 5. Shear strength properties of Pomona granular material.

4.

Summary and Conclusion

This paper presented laboratory evaluation results of three granular materials selected for consideration as base/subbase material for the construction of a new runway in the United States. Specimen preparations and testing procedures conform to

402

J. Anochie-Boateng / Selection of Pavement Foundation Geomaterials

ASTM and AASHTO standard procedures. All tests were conducted to meet Federal Aviation Administration (FAA) specifications and requirements of granular materials used as base/subbase for the design and construction of airport pavements. Based on the FAA specifications, it was evident that all the three granular materials evaluated meet the FAA selection criteria of base/subbase course materials for the airport runway and associated taxiways and aprons. The study has shown that granular materials from the selected quarries could be used for future design and construction of pavements in North Carolina.

Acknowledgements/Disclaimer This paper is prepared from a study conducted in the FAA Center of Excellence for Airport Technology at the University of Illinois at Urbana-Champaign. The study was funded by the Federal Aviation Administration under Piedmont Triad International Airport in Greensboro, North Carolina and Trigon Engineering Consultants Inc. in Greensboro, North Carolina. The contents of this paper reflect the views of the author who is responsible for the facts and accuracy of the data presented within. The contents do not necessarily reflect the official views and policies of the FAA.

References [1] FAA Advisory Circular 150/5320-6E. Airport pavement design and evaluation. U.S. Department of Transport, Federal Aviation Administration, Washington DC, 2009. [2] FAARFIELD. FAA Rigid and Flexible Iterative Elastic Layered design. U.S. Department of Transport, Federal Aviation Administration, Washington DC, 2009. [3] Anochie-Boateng. J.K. Evaluation of granular base materials and subgrade soils for the design of a new runway, taxiways, and aprons for Piedmont Triad International Airport (PTIA) in Greensboro, North Carolina. MS Thesis, North Carolina A&T State University, Greensboro, 2002. [4] ASTM C136 Standard Test Method for Sieve Analysis of Fine and Coarse Aggregates, West Conshohocken, PA, 2006. [5] FAA Advisory Circular AC 150/5370-10B. Standards for Specifying Construction of Airports. Federal Aviation Administration, Washington DC, 2005. [6] ASTM D1557 Standard Test Methods for Laboratory Compaction Characteristics of Soil Using Modified Effort (56,000 ft-lbf/ft3). West Conshohocken, PA, 2000. [7] Thom, N.H. and Brown, S.F., 1988. Elastic properties of granular materials from repeated load laboratory testing. 12th International Conference of the International Society of Soil Mechanics and Foundation Engineering, Rio de Janeiro [8] AASHTO T 307. Standard Method of Test for Determining the Resilient Modulus of Soils and Aggregate Materials. American Association of State and Highway Transportation Officials, Washington DC, 1999. [9] ASTM D1883. Standard Test Method for CBR (California Bearing Ratio) of Laboratory-Compacted Soils. American Association of State and Highway Transportation Officials, Washington DC, 1999.

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 403 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-403

Suggested Improvements in Site Investigation and Numerical Characterization Procedures for House Foundation Design John Terry PIDGEON, Rachael GOVENDER The Centre for Excellence in Foundation Engineering, CSIR, Pretoria, South Africa

Abstract. Current geotechnical investigation practice in South Africa should at least be carried out in accordance with GFSH-2 which dictates the frequency of test pits and soil samples per hectare. Very many geotechnical reports have been studied and the laboratory test results both plotted on ternary charts and contoured using a ‘minimum curvature’ technique. This investigation has revealed a number of shortcomings in current practice, and these are as follows: • In spite of the existence of GFSH-2, the number of soil samples is rarely adequate to determine foundation design input parameters accurately. • There are often discrepancies between the description of the soil in the profile and that found by plotting the results of the grading analysis on the US Corps ternary chart. • The spatial variability of the Plasticity Index (PI) and 0.425 fraction at the study site when compared with the visual assessment revealed that the trend of describing a distinct soil layer in a profile as being the same as a corresponding layer in another test pit based purely on a visual analysis, rarely gives an accurate prediction of this layers soil properties. • A method has been proposed for determining the effective PI of the whole sample for use in heave prediction equations which incorporate this parameter. Keywords. Heave prediction, soil characterization, settlement potential, linear shrinkage, swell potential, EEMM.

Introduction Geotechnical investigations in South Africa have for a number of years been required by the NHBRC to comply with GFSH-2 [2]. This should be superseded in the near future by SANS 634 which is currently under preparation, and subsequently by Eurocode 7[3]. In spite of the introduction of these guidelines and codes there are still serious deficiencies in the quality of site investigations particularly with regard to the amount of soil testing, and to the evaluation and reporting of the geotechnical conditions to the foundation engineer. Although it is considered that the standard profiling procedures proposed by Jennings et al [1] should be retained, the current practice of assuming that a soil will display certain characteristics simply because it

404

J.T. Pidgeon and R. Govender / Suggested Improvements

looks similar to a material sampled elsewhere should be viewed with the utmost caution. An on-going study by the writers has identified some of the factors believed to be the cause of the current deficiencies. In order to improve this situation, the following approach is proposed: A standard spreadsheet program should be used to sort and compare all the profile information both visually and numerically. Ternary charts be used to standardize the visual description of each soil type using the US Corps of Engineers textural demarcations [10]. A computer contouring program be utilized to facilitate the display of both the lateral and vertical variations of the relevant soil properties. That the ‘Generalized Heave Equation (GHE)’ [6, 7] be used for the determination of both total and differential heave, as well as the ‘equivalent elastic mound modulus (EEMM)’ And the ‘Steinbrenner’ relationship [8] for differential settlement prediction. Since the profession has been using Van der Merwe’s method of heave prediction [5] almost exclusively for more than 45 years it is felt that there will be continued resistance to the GHE, consequently some modifications have been suggested for this method with regard to the determination of the percentage swell anticipated for each soil layer.

1. Textural charts The textural chart presented in (Figure 1) shows the boundaries between different soil types demarcated by the US Corps of Engineers [10]. Two examples are presented here where the percentages of sand, silt and clay have been plotted on the chart. In the case of samples S24 and S29, both can be seen to classify as clayey silts however in the field logs they were both classified as stiff sandy clays. Similarly, in the case of sample S10 which was given the field description of stiff sandy clay, it actually plots as a clayey sand. Sample S21 however was also described as being a silty sand in the logs whereas the chart plot shows it to be a clayey silt. Thousands of discrepancies similar to the ones above have been observed over the years. It is considered self-evident that where the description of the soil does not relate satisfactorily to the indicator test results, serious doubts may be placed in the mind of the foundation engineer. It is essential that such cross checks be carried out by either the geotechnical engineer or engineering geologist and that everything possible be done to ensure the consistent description of the soils and the accurate correspondence with laboratory results, before carrying out the sorting procedure using the spreadsheet program.

J.T. Pidgeon and R. Govender / Suggested Improvements

405

Figure 1. Soil Classification Chart for soil layer 4 at site X

2. Contouring Data The contouring program provides an alternative means by which any missing soil data for each un-sampled soil layer anywhere on the site can be predicted. In the case where a distinct layer of soil has not been sampled, the program can be used to either interpolate or to a limited extent, with care, to extrapolate using the surrounding available data. The measured PI and percentage passing the 0.425µ sieve (425 fraction) for two identifiably different soil layers on two different sites, were plotted and contoured (see Figures 2a and 2b, and 2c and 2d). The method of contouring used in these figures is known as the ‘Minimum Curvature’ method, and the interpolated surface is regarded as being analogous to a thin, linearly elastic plate passing through each of the data points. Other popular contouring methods include ‘Kriging’, ‘system of linear equations’, ‘inverse distance method’, and multilevel B-Spline A’ could have been used but we consider ‘Minimum Curvature’ to be the more preferred method. The contoured images illustrate a number of issues. Figures 2a and 2b illustrate that from a total of 20 test pits excavated across the site, only 4 samples were retrieved from layer 1 in the profile for testing. In addition to the low frequency of pits sampled, these samples were retrieved only from one corner of the site. For the purposes of accurately predicting the soil properties of the various soil layers across the site it is essential that samples are taken from test pits which are adequately distributed to allow for the mathematical interpolation and, wherever possible, extrapolation should be kept to a minimum.

406

J.T. Pidgeon and R. Govender / Suggested Improvements

Figure 2a. Variation of PI across site X - layer 1

Figure 2b. Variation of 0.425 fraction across site X- layer 1

Figures 2c and 2d illustrate a much better spatial distribution of soil sampling and consequently the mathematical interpolation of data across the site can be used with more confidence than in the case of 2a and 2b. These figures illustrate the discrepancy which often occurs when trying to identify a soil layer as being the same as another based on visual characteristics. For instance, layer 2 in test pits 3 and 13 was described in the soil profile as moist, dark brown, medium dense clayey sand however only test pit 3 was sampled. These soils were given identical soil descriptions but are located quite far apart. The soil in test pit 3 has a PI of 12 and 0.425µ fraction of 62. On the basis of the visual description the soil in test pit 13 should have identical values, however the PI and 0.425µ fraction, interpolated using the program are 19 and 92

J.T. Pidgeon and R. Govender / Suggested Improvements

407

respectively. Although these values differ markedly, we are of the opinion that they are more likely to be reliable than those obtained from visual comparison alone. The large spatial variation in the PI values across the site should be noted. It is also important to note that although Figures 2c and 2d illustrate a better distribution it is by no means perfect. Ideally, the sampling should have been carried out on a more uniformly distributed grid.

Figure 2c. Variation of Plasticity Index at Site Y-layer 2

Figure 2d. Variation of 0.425 fraction across Site Y-layer 2

408

J.T. Pidgeon and R. Govender / Suggested Improvements

With further reference to Figures 2a and 2b, it can be seen that increasing PI’s correspond to decreases in the 0.425µ fraction. The trend illustrated in Figure 2a and 2b often occurs when a low percentage of material passes the 0.425µ sieve, because a larger portion of that material classifies as clay. For example, if 20% of a soil sample is composed of clay and 100% of the sample passes the 0.425µ sieve then one gets a true representation of the soils composition. If only 50% of the soil sample passes the 0.425µ sieve then the percentage of clay increases to 40%, resulting in a significantly higher value of PI. It is therefore suggested that plotting the PI of the whole sample (PI w) against the percentage clay (Van der Merwe, Williams) can significantly underestimate the potential expansiveness of the soil where the percentage passing the 0.425µ sieve is low. Consequently it is recommended that it would be preferable to plot PI of the 0.425µ fraction (PI425), which was Skempton’s original proposal [11], and that the values of PI forming the boundaries between low, medium, high and very high be kept as they are but this time in terms of PI425 and that the resulting predicted swell be multiplied by the 0.425µ fraction. In addition it is felt that the use of linear shrinkage, rather than the combination of PI and the clay fraction, provides a more reliable indicator to the potential expansiveness of a soil. Where any particular heave prediction program makes use of the PI425, this value of PI may be determined by measuring the LS and using equation (1) below, PI 425 = LS 425 x 2.13

(1)

The factor 2.13 has been derived from the BS 1377 [4]. The average factor for most South African soils appears to be about to 2 but may vary markedly depending on the predominant type(s) of clay minerals and on the value of PI especially for PI values of 30 and above. As pointed out above, the determination of the anticipated volume change behavior of the soil using the 0.425µ fraction is often not reliable. It is therefore suggested that the linear shrinkage of the whole sample should be determined and these values be contoured instead. Research is currently underway to develop different types and sizes of shrinkage moulds that would enable the testing of soils with larger grain sizes, as well as to determine the influence of soil fabric and structure by comparing the volume change behaviour of undisturbed soil samples with that of corresponding remolded soil samples. Van der Merwe’s method for heave prediction is widely used by the engineering community, however it should be used with extreme caution as it can seriously under predict the actual heave. This is because it was originally inadequately and /or incorrectly calibrated (see Pidgeon [10]). The use of the plasticity index of the whole sample in Van der Merwe’s heave prediction method may produce misleading heave predictions particularly where the PIw falls in a different swell potential category than PI0.425. It is recommended that PIw be replaced by PI425 on the vertical axis of the chart bringing it into line with Skempton’s original activity chart and that the 0.425µ fraction be used to factor down the heave predicted by this chart. Examples are tabulated in Table 1 below.

409

J.T. Pidgeon and R. Govender / Suggested Improvements

Table 1. Van Der Merwe method alternative derivation of swell potential

^ǁĞůůWŽƚĞŶƚŝĂů;йͿ ^ǁĞůůůĂƐƐ

Ϭ͘ϰϮϱђ

W/;Ϭ͘ϰϮϱђͿ

ϴϬ

ϯϱ ϴ͘ϯϯ sĞƌǇ,ŝŐŚ

W/;tŚŽůĞ ƐĂŵƉůĞͿ Ϯϴ ϰ͘ϭϳ ,ŝŐŚ

Ϭ͘ϰϮϱђ

W/;Ϭ͘ϰϮϱђͿ

ϴϯ

Ϯϱ ϰ͘ϭϳ ,ŝŐŚ

W/;tŚŽůĞ ƐĂŵƉůĞͿ ϭϵ Ϯ͘Ϭϵ DĞĚŝƵŵ

Ϭ͘ϰϮϱђ

W/;Ϭ͘ϰϮϱђͿ

ϳϱ

ϭϰ Ϯ͘Ϭϵ DĞĚŝƵŵ

W/;tŚŽůĞ ƐĂŵƉůĞͿ ϭϬ͘ϱ Ϭ >Žǁ

ƵƌƌĞŶƚůƚĞƌŶĂƚŝǀĞ͗ϴ͘ϯϯǆϴϬсϲ͘ϲϲйƵƌƌĞŶƚůƚĞƌŶĂƚŝǀĞ͗ϰ͘ϭϳǆϴϯсϯ͘ϰϲйƵƌƌĞŶƚůƚĞƌŶĂƚŝǀĞ͗Ϯ͘Ϭϵǆϳϱсϭ͘ϱϳй

Until such time as the above research has been carried out and reported upon, the following modifications to Van der Merwe’s method are suggested: • The PI of the whole sample or of the 0.425µ fraction can be derived from multiplying the linear shrinkage of the whole sample or the linear shrinkage of the 0.425µ fraction by the factor 2.13 respectively • In the case where the linear shrinkage of the whole sample cannot be determined due to the shrinkage moulds not being able to accommodate larger grain sizes, the PI of the 0.425µ fraction should be determined and multiplied by the percentage passing 0.425µ • It is further recommended that the additional step of replacing the Van der Merwe chart with a graph relating swell percent directly to the linear shrinkage of the whole sample as illustrated in Figure 3 below should be used. The values of swell potential (free swell) used in this graph have been derived from Van der Merwe and correspond to the lowest value of PIw in each heave category.

Figure 3: Graph illustrating the relationship between the linear shrinkage of the whole sample and percentage free swell.

410

J.T. Pidgeon and R. Govender / Suggested Improvements

The use of figure 3 above provides a more conservative approach to the use of Van der Merwe’s heave prediction chart. The LS of the whole sample is derived from approximately halving the PI of the whole sample and this is plotted against the corresponding free swell percentage as it appears on the original chart. Instead of using a step wise function to define swell percentage, a gradual curve is used to derive a value for the swell potential corresponding to any value of either PI or LS. Table 2: Free Swell and PI values taken from Van der Merwe [5]

Free S well %

PI (Whole S ample)

Very High

8.33

32

ϭϱ͘ϬϮ

High

4.17

23

ϭϬ͘ϳϵ

Medium

2.09

12

ϱ͘ϲϯ

Low

0

0

Ϭ

>^;tŚŽůĞ^ĂŵƉůĞͿ

Concluding comments The importance of a well-structured geotechnical investigation which produces results that can be used to construct safe, reliable and cost-effective foundations and structures is well appreciated. The above case studies point out a number of problematic issues arising from current practice as dictated by GFSH-2. Analyses carried out on numerous geotechnical reports have revealed that the main failing across the board pertains to inadequate sampling of the soil profiles. This can result in an inordinate waste of professional time, using all manner of techniques, both visual and numerical, in an effort to fill in the missing soil properties for the unsampled layers. In fact it would normally be more cost-effective to sample all the soil layers rather than spending time trying to fill in the missing data. It is currently the norm for geotechnical reports to contain zoning maps which provide only the NHBRC site classifications (H, H1, H2, C, etc.). Information in this form is not adequate for foundation designers to carry out any rational design. It is the responsibility of the engineering geologist/geotechnical engineer to ensure that these investigations are carried out sufficiently well so that the total and differential heave or settlements, the EEMM values in the case of expansive soils and the shape of any soil mound which could form under an impermeable cover can be calculated accurately. There must be a satisfactory correlation between the observed field conditions and the laboratory results. The spatial variation of PI values for the same layer of soil across the site should not be overlooked and it must be stressed that although soil layers may have the same visual descriptions, they can have very different index properties. The simple procedure of cross checking field observations with laboratory determined index properties of the soil is highly recommended. Deriving PI values from the linear shrinkage of the whole sample seems to be the way forward in deriving more reliable heave predictions or alternatively deriving heave prediction equations which are based on the determination of linear shrinkage (0.425µ or whole sample) rather than on the PI or liquid limit. Although the use of the Van der Merwe method is not recommended, its simplicity of use and widespread acceptance by the engineering industry is recognized. As a first

J.T. Pidgeon and R. Govender / Suggested Improvements

411

step it needs to be re-calibrated making use of additional field observations so that it may be used more reliably for site classification purposes. It is recommended that a more sophisticated method such as the GHE proposed by Pidgeon 1987 be used for detailed foundation design. The many issues pertaining to the carrying out of proper geotechnical investigations in order to produce results that one may use with confidence does require a great deal of additional research. CEFE is currently researching standards, methods and guidelines in order to identify areas that may be improved upon, as well as devising new procedures and recommendations to carry out geotechnical investigations in order to achieve meaningful and reliable geotechnical reports. This will enable the professional team to design cost effective and enduring structures with confidence.

References [1]

Jennings, JE, Brink, ABA and Williams AB, Revised Guide to Soil Profiling for Civil Engineering Purposes in Southern Africa. The Civil Engineer in South Africa, January 1973. [2] National Department of Housing. Geotechnical Site Investigations for Housing Developments. Project Linked Greenfield Subsidy Project Developments. Generic Specification GFSH-2. 2002. [3] Bond, A, Harris, A, Eurocode 7. Taylor and Francis Group. 2008. [4] BS 1377:1967, Methods of Testing Soils for Civil Engineering Purposes. British Standards Institution, December 1967. [5] Van der Merwe, DH, Prediction of Heave from the Plasticity Index and Percentage Clay Fraction of Soils, The Civil Engineer in South Africa. June 1964. [6] Pidgeon, JT, The Rational Prediction of Differential Heave. 9th Regional Conference for Africa on Soil Mechanics and Foundation Engineering. Lagos. June 1987. [7] Pidgeon, JT, Williams, AAB, Day, PW, Expansive Soils. Problem Soils in South Africa, State-of –the – Art. The Civil Engineer in South Africa. July 1985. [8] Steinbrenner, W, Tafeln zursetzungsberechanung. Die strasse, Vol. 1, 121-124. 1934 [9] Hough, BK, Basic soils engineering, Ronald Press, U.S. 2nd Revised Edition. 1969 [10] Pidgeon, JT, The responsive of expansive soils to changes in both moisture content and applied load under different degrees of confinement, Proc. Of the Problem Soils Conference in South Africa. Nov 2008. [11] Skempton, AW, The Colloidal Activity of clay’s, Proc. 3rd Inter. Conf. Soil Mech. Found. Eng. Switzerland. Vol. 1. 1953.

This page intentionally left blank

Section 7 Roads

This page intentionally left blank

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 415 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-415

Improvement of Unbound Aggregates in Khartoum State Omer O. G1, Elsharief A. M.2 and Mohamed A. M.2 1. Civil aviation Authority, Sudan 2. Building and road Research Institute, University of Khartoum

ABSTRACT.Due to the scarcity of well graded aggregate base course material within the vicinity of Khartoum state natural aggregates [gravel] of lower quality are commonly used as base material for economic purposes. However from local practice pavements with natural base materials failed to show satisfactory performance under heavy traffic. This paper presents the outcome of an intensive study aimed at stabilization of natural aggregates by various agents such as cement, coarse sand, stone dust and crushed stone. Cement was found to be the best in terms of improvement of the strength of the natural aggregates. Small quantities of cement (less than 3%) gave excellent results. The other stabilizers gave good and acceptable results. The paper discusses different technical and economic aspects relating to the use of each stabilizer.

Key words. Unbound aggregates, blending, natural aggregates, stabilization

Introduction Khartoum, the capital of Sudan has witnessed huge expansion in the first decade of the 21st century. Hundreds of kilometers of roads have been constructed; some of them are part of the national highway grid system. The road construction industry is nowadays challenged with lack of local design and construction codes and specifications. Specifications are borrowed from either AASHTO or Transport Research Laboratory of United Kingdom (TRL). Flexible pavements with hot mix asphalt wearing course are commonly used to carry traffic loads. Natural materials from quarries located around the capital are used for base and subbase layers. The available materials often satisfy the requirements for subbase but they seldom satisfy the strict requirements of unbound base course materials. For important projects contractors usually mix the natural material with sand and/or crushed stone. The specifications are relaxed for small projects or those experiencing low traffic. The overall performance of pavements carrying heavy traffic could be rated as poor. This poor performance may be attributed to the behavior of the natural unbound aggregates often adopted for use as base and subbase materials. In flexible pavements,un-surfaced or thinly surfaced unbound granular layers play an important role in the overall performance of the structure [1]. The main structural function of the base layer is to distribute the stress generated by wheel loads acting on the wearing surface so that the stresses transmitted to the subgrade will not result in excessive deformation or displacement of the foundation soil. Wide range of materials

416

O.G. Omer et al. / Improvement of Unbound Aggregates in Khartoum State

can be used as unbound road bases including crushed quarried rock, mechanically stabilized, modified or naturally occurring gravels (TRL Overseas Road Note 31, [2]). The stabilization process involves the addition of a stabilizing agent (natural sand, cement, and crushed stone) to the gravely materials aiming at improving their strength and deformation characteristics. This paper focuses on assessing different stabilization methods for three natural unbound aggregatesof subbase quality obtained from open quarries in Khartoum state, in an attempt to upgrade theme to be used as road subbase materials.

1.

The Materials and Testing Program

Khartoum state is blessed with various sources of natural graded gravels. Huttab, AlSilate and Umm Ketti are the currently utilized ones.The study area is considered as part of a depression filled with formations of different ages. It is located adjacent to Central Sudan rift basin which is Khartoum basin. The studied formations are colluvial deposits originally conglomerates belonging to Nubian Sandstone Formation. The formation has been deposited by braided streams under semi-dry tropical climate. The natural unbound gravelly materials used in this study are currently been used as base course or subbase material for road construction. Figure 1 shows the three gradations plotted with the Transport Research Laboratory (TRL) standard envelopefor base material. Representative gravels were batched from each quarry and transported to the laboratory. Proper manual mixing was done and the samples were then bagged and stored in plastic barrels. The samples were subjected to the following tests: grain size analysis, Atterberg Limits, linear shrinkage, Modified Proctor, California Bearing Ratio (CBR) and Los Angeles Abrasion (Table 1). The coarse aggregates (gravels) are about rounded or elongated for Huttab and Alsilate materials whereas they are rounded to angular for Umm Ketti (Figure 2).

Figure 1.Grain Size Distribution for the Three Natural Gravely Materials.

O.G. Omer et al. / Improvement of Unbound Aggregates in Khartoum State

417

Table 1.Test Results for Natural Gravel Materials in the Study. Gravel Material Type Umm Ketti Al – Silate Huttab

O.M.C % 6.40 7.20 5.88

MaxDry Density (/cc) 2.25 2.14 2.21

Fines Content % 9.5 18.5 20.8

Liquid Limit % 26 37 36

Plasticity Index % 11 15 19

Linear Shrinkage % 12.5 14.0 14.3

Abrasion (LA) % 32 37 29

C.B.R % 62 56 34

Figure 2.Typical Coarse Aggregate from the Natural Materials(Passing 20mm, Retained Sieve 4.75mm).

Four activities or tasks were carried out on the natural samples in an attempt to improve their engineering properties. The samples were chemically stabilized (Task 1) with ordinary Portland Cement (OPC) and mechanically with natural wadi sand (Task 2), crushed basaltic rock “sand size” (Task 3) and crushed stone of different sizes (Task 4). For Task 1 the cement was added to the natural samples using the following percentages (by weight):1%, 2%, 4% and 6%.The mechanical stabilization (Tasks 2 and 3) was carried out by adding coarse wadi sand and crushed rock with maximum size of 5.0 mm, each separately, using the following percentages: 3%, 6%, 9%, 15%, 25% and 35%. Crushed rock material (19-12 mm, 12 – 9 mm and 5 – 0 mm) was added with suitable percentage of each particle size (Task 4) to attain a gradation falling within the B.S. envelope for base material (Figure 3). The grain size distribution, Atterberg limits and California bearing Ratio were measured or determined for each mix.

Figure 3.Test Results for Natural Gravel Materials Blended with Crushed Stone.

418

2.

O.G. Omer et al. / Improvement of Unbound Aggregates in Khartoum State

The Results

Summary of the results of the tests carried out on the natural samples was given in Table 1. The CBR was determined at the optimum moisture content and maximum dry density. All the samples stabilized with cement measured CBR greater than 100 %. The measured CBR values for the mechanically stabilized samples with wadi sand and crushed sand are given in Table 2 for the added percentages. Table 3 gives the results of the tests performed on the samples mixed with crushed stone. Table 2.Test Results for Umm Ketti Natural, Blended with Wadi Sand and Crushed Rock Dust. Blended with Natural Blended With Rock Dust Coarse Sand (5 – 0 mm) P.P. N.G. B.C.A 15% 25% 35% 6% 15% 25% 35% C.B.R. 62 168 126 79 189 103 Density (g/cc) 2.25 2.31 2.27 2.23 2.25 Fine Cont. % 9.5 7.0 Plasticity Index % 11 11 10 7 7 11 Note: N.G : Natural Gravel ; P.P. : Physical Properties ; B.C.A.: Blend Crush Aggregate

104 2.29 11

115 2.30 12

173 2.32 11

Table 3.Test Results for Natural Gravels Blended with Crushed Stone. Gravel Material Type Umm Ketti Al – Silate Huttab

3.

O.M.C %

MaxDry Density (g/cc)

5.70 6.72 5.60

2.31 2.29 2.29

Fine Content % 7.0 10.3 14.1

Liquid Limit % 25 31 32

Plasticity Index % 11 15 15

Linear Shrinkage % 15.0 14.0 14.6

Abrasion (LA) % 38.7 35.5 20.0

C.B.R. % 168 109 128

Analysis and Discussion

Figure 1 demonstrates the upper and lower ends of TRL gradation envelopeand the gradation of the three natural aggregates and Table 1 gives their engineering properties. The Umm Ketti and Alsilate samples coincide well with the lower end of the TRL gradation curve for the gravel sizes (> 5.0 mm) but lack sand sizes and therefore are gap graded, whereas the sample from Hattab coincides with the upper end of the curve. The sample from Hattab has the highest fines content, plasticity index and linear shrinkage and consequently the highest clay content.The low CBR for Huttab may be attributed to the high fines and clay contents. The PI and linear shrinkage are relatively high for Huttab sample. The three materials are below the required base course strength (CBR >=80%).The samples do not satisfy the gradation and plasticity requirements of TRL unbound aggregate base specifications (fines content < 15% and PI = <6%, respectively). The sample from Umm Ketti performed better than Alsilate mainly because it has better gradation, lower fines content and lower plasticity. The outcome from the four tasks is discussed here-under: Task 1: The chemical stabilization process with cement was carried out on the three samples in question in accordance to B.S. 1924 [3]. It is observed that with a minimum percentage of cement (1% only) the measured CBR is very high and is reported as >100% for the three samples (Table 2). Small quantities of OPC if mixed properly with the quarry materials will remarkably improve their strength. Therefore, cement proved to be very efficient in increasing the strength of the unbound aggregates of Khartoum state.

O.G. Omer et al. / Improvement of Unbound Aggregates in Khartoum State

419

Task 2: Figure 4 plots the results of Task 2 activity, i.e. mechanical stabilization with wadi sand. It is observed that Hattab and Umm Ketti samples when mixed with 15% and 25% attained CBR values close to or above 80%. The material from Umm Ketti required 25% and more of wadi sand to give CBR higher than 80%. The plasticity index of the fines dropped for the natural sand mix and was slightly affected by the addition of the crushedsand; therefore, the increase in strength may be attributed to better or improved gradation.

Figure 4.The Results for Natural Gravels Blended with Wadi Sand.

Task 3: The results from the mechanical stabilization process using crushed basalt rock dust (5 – 0 mm) are displayed graphically in Figure 5. The Figure shows that all mixtures achieved more than 80% CBR when mixed with 15% and 25% crushed dust. Similar to Task 2, the 35% mix for Huttab and Alsilate experienced drop in strength compared to the 25% mix to the extent that Alsilate sample measured CBR value lower than 80%.On the other hand the 35 % mix of Umm Ketti material (Task 2 and Task 3) showed improvement in strength compared to the 25% mix. The similarity in the behavior of the samples with the same mix percentages for Task 2 and Task 3 indicate that the improvement and drop in strength could be attributed to the gradation of the tested samples.Comparing Figures 4 and 5 it is evident that crushed sand is better stabilizer when compared to natural sand.

Figure 5.Variation of CBR with Crushed Stone Sand.

420

O.G. Omer et al. / Improvement of Unbound Aggregates in Khartoum State

Task 4: Mechanical stabilization with manufactured crushed basalt (Task 4) was conducted on the three materials showing gradation within the TRL envelope (Figure 3). The three blends fitted well within the TRL jacket and they all gave high CBR (Table 3). It is observed that Umm Ketti blend which is closer to the middle band produced the highest CBR value (168%). The improvement could be attributed to improved gradation of the mixes. Figure 6 presents the gradation curves for Umm Ketti material mixed with 15%, 25% and 35% wadi sand. All the three blends fall within the middle of the TRL envelope. The three blends satisfied the strength requirement for base material (Figure 4). From this observation and previous discussion it may concluded that gradation of the mechanically stabilized unbound aggregates has great influence on their strength.

Figure 6.Grain Size Distribution for Umm Ketti Natural Gravel Blended with 15%,25% and 35% Wadi Sand.

4.

Financial Evaluation

Table 4 demonstrates the actual cost of blends for one of the three sources stabilized to attain minimum recommended base CBR of 80%. The wadi sand seems to be the cheapest stabilizer for improving a unit volume of base material in Khartoum state. Type of stabilization Stabilizer CBR Cost €/m3

Table 4.Actual Cost of 1.0 m3“ processed”for Umm Ketti Natural Cement Crushed Crushed Gravel Stone Dust 0% 1% 35 % 15 % 62 > 100 168 104 9.08 11.06 15.15 12.2

Wadi Sand 15 % 126 10.6

O.G. Omer et al. / Improvement of Unbound Aggregates in Khartoum State

5.

421

Conclusions

This paper presented the results of a comprehensive testing program aimed at assessing chemical and mechanical stabilization techniques of three unbound aggregates of subbase quality from Khartoum state in an attempt to set guidelines for their improvement to satisfy the strength requirements for base materials. The aggregates were tested in their natural state and after being stabilized with cement, wadi sand, crushed rock (to sand size) and crushed rock. The cement content ranged from 1% to 6%. The wadi sand and crushed rock sand were added by the following percentages 3%, 6%, 9%, 15% 25% and 35%. Crushed stone was added to the natural samples to satisfy the grading requirements of the unbound TRL base course material. The classification, compaction and strength properties were measured for each mix. Cement was found to be very effective in improving the strength of the natural aggregates. Only 1% of cement was needed to increase the CBR to the required value. Sand and crushed stone were also effective if the gradation is improved to fit within the TRL jacket or envelope regardless of the plasticity and content of the fines. Simple economical evaluation has shown that wadi sand is the cheapest stabilizer.

References [1]Dawson A.R Pavements Unbound. University of Nottingham Proceedings of the 6th International Symposium on Pavements Unbound (2004) pp. 51-59. [2] TRL Overseas Road Note 31 – A guide to the structural design of bitumen-surfaced roads in Tropical and Sub-Tropical countries pp. 1-31 [3] B.S. 1924-2: 1990 “Stabilized Materials for Civil engineering Purposes, Methods of tests for Cement Stabilized and Lime Stabilized Materials ”, British Standards Institute.

422 Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-422

Applying the Dynamic Cone Penetrometer (DCP) Design Method to Low Volume Roads Philip PAIGE-GREEN CSIR Built Environment, Pretoria, South Africa

Abstract. The Dynamic Cone Penetrometer (DCP) is a simple, cheap and effective apparatus for assessing the bearing capacity of in situ materials for the design of new roads or the upgrading of unsealed or existing sealed roads. Although a number of methods have been described for the use of the DCP for the design of low volume roads, no comprehensive method has been published. This paper summarizes the investigation, analysis and design techniques for application of this very useful and cost-effective design method taking into account traffic and environmental conditions. Use of the method is illustrated by actual design examples. Keywords. Dynamic Cone Penetrometer, road, pavement, design

Introduction The Dynamic Cone Penetrometer (DCP) [1] has been in use since the 1950s for various applications in pavement investigation [2]. During the 1980s, Kleyn and Van Zyl [3] described a method for upgrading unsealed roads to light sealed road standard based on in situ testing using the DCP. Although this is a simple, cheap and effective method for assessing the bearing capacity of in situ materials for the design of new roads or the upgrading of unsealed or existing sealed roads, a fundamental understanding of the in situ conditions is essential. A number of methods of use have been described and applied for the use of the DCP for the design of low volume roads, but no comprehensive method has been published. This paper summarizes the fundamentals, investigation, analysis and design techniques for application of this very useful and cost-effective design method taking into account the prevailing material, traffic and environmental conditions. Use of the method is illustrated by two actual design examples.

1. Background The DCP is a highly cost effective technique for acquiring large quantities of data on sub-surface material strength and thickness quickly and essentially in a non-destructive process. However, the strength information acquired is related directly to the in situ moisture and density conditions at the time of the investigation. Although the dry

P. Paige-Green / Applying the Dynamic Cone Penetrometer Design Method to Low Volume Roads

423

density of the in situ materials is relatively constant over time, the wet density of the materials beneath unsealed roads varies almost continuously with time and this is manifested in the in situ strength estimated from the DCP data. As the in situ strength is directly and inversely proportional to the density and the moisture content, respectively (i.e. the in situ strength increases with increasing density and decreases with increasing moisture content), it is essential, although difficult, that these relationships are considered during the pavement design process. The designer should preferably be on site during the DCP investigation. 1.1. Moisture content Estimation of the moisture content at the time of testing can be difficult. Although it is recommended that samples are taken for gravimetric moisture determination, this is usually only practicable for the upper and possibly the second 150 mm layer without excavating large holes. Kleyn and Van Zyl [3] described the classification of the overall moisture regime at the time of DCP testing in terms of the expected moisture levels that will prevail during the service of the road. This can be effective as a general classification and the percentile of the determined strengths selected for the analysis will be based on this (see example). This should be considered in the light of the potential equilibrium moisture content of the completed road [4] bearing in mind that the outer edges of the road will be subjected to seasonal fluctuations in moisture content. A method for estimating the material G-class [5] based on the DCP penetration rates and the estimated moisture content was developed in 1985 [6] and improved in 1992 (Table 1) [7]. The use of this method requires a visual estimate of the field moisture content at the time of DCP testing but also has the limitation of assuming that the subgrade moisture content is uniform and is a direct function of the climate. Excavation of holes into the wearing course and underlying layers and extraction of samples for laboratory moisture determination would be highly beneficial. Table 1. Estimate of material G-class from DCP results Material classification

G4 G5 G6 G7 G8 G9 G10

Soaked CBR

80 45 25 15 10 7 3

Approximate field DCP-CBR: gravel roads Subgrade Wearing course Wet Dry Very dry Dry state Moderate climate climate state state 260 205 151 188 148 109 56 66 146 115 85 52 62 137 108 79 39 46 101 80 59 38 44 35 41 -

Damp state 96 69 54 50 37 -

Use of Table 1 without actual moisture content determinations requires an estimate of the moisture content in terms of the optimum moisture content (OMC) for the materials. This can usually be obtained by experienced engineers based on squeezing a sample of the material in one hand and assessing the “cohesion”. At OMC (damp) the material can be squeezed into a “sausage” that remains intact. In the very dry state (less than about 25% of OMC), the material is dusty and loose and has absolutely no cohesion. In the dry state (about 50% of OMC), the material will have no cohesion

424

P. Paige-Green / Applying the Dynamic Cone Penetrometer Design Method to Low Volume Roads

when squeezed into a sausage whereas in the moist state (about 75% of OMC), the material may just be squeezed into a sausage but will be friable and break easily. The expected subgrade equilibrium moisture contents for wet and dry climates are about 95 and 90% of OMC respectively. 1.2. In situ density The in situ density obviously affects the DCP penetration rate considerably. This is a difficult parameter to estimate during the DCP survey, but on an existing unsealed road, it can be assumed that there has been some traffic compaction over time, probably to at least that normally specified for a subgrade or even subbase under a sealed road. It is thus possible to relate the densities to the expected final pavement structure. If the road is to be widened, however, it is usually necessary to carry out testing adjacent to the road to assess the strength of the uncompacted in situ material. This, of course, can also be done to assess the effect of traffic compaction on the material density by comparing DCP penetration rates of obviously un-trafficked material adjacent to the existing road with trafficked material under the road.

2. Design process 2.1. Traffic determination As in any pavement design, the cumulative traffic over the design life of the road should be estimated. This estimate is much more difficult for low volume roads (less than about 300 000 equivalent standard axles (E80s) as the heavy vehicles during short periods of intensive traffic (eg, during agricultural harvesting seasons or temporary delivery seasons for mining produce) are often difficult to estimate accurately. Short periods of heavy traffic also have a disproportionate influence on the overall traffic of low volume roads. As the traffic estimate will directly influence the pavement design, this should be done as carefully as possible, taking into account such issues as the potential for overloading. 2.2. Required pavement structure The required pavement structure will usually be determined from available catalogues [8] [9] [10], or for very low traffic roads (less than about 50 000 E80s) from past experience or comparison with other similar roads. This should provide an indication of the number and strengths of the different layers as well as the individual layer thicknesses at the expected worst moisture condition in the road. In the two cases shown in the examples, the proposed structural design included 100 mm G4 base, 125 mm G6 subbase and 150 mm G9 support. From this, the necessary layer strength diagram (LSD) can be constructed. This relates the individual layer strengths to the CBR for that layer as shown in Fig. 1. The plots are for the standard soaked CBR design as well as the required LSDs for different DCP test moisture conditions based on Table 1.

P. Paige-Green / Applying the Dynamic Cone Penetrometer Design Method to Low Volume Roads

425

CBR (%) 1

10

100

1000

0 50

Depth (mm)

100 150 200 250 300 350 400 450 500 Design (soaked)

Very dry

Dry

Moderate

Damp

Figure 1. Layer strength diagram for road with 30 000 to 100 000 E80 design traffic and different moisture conditions.

2.3. DCP survey 2.3.1. Depth, interval and number of DCP tests The DCP survey will be carried out to a depth of at least 450 mm but preferably to at least 600 mm, the so-called material depth of the pavement [8]. It is recommended that DCP testing is carried out at 200 m intervals with additional testing in any obviously problematic areas (e.g., wet, cracked). In relatively uniform areas, testing at up to 500 m intervals could be accepted. In general a minimum of about 10 tests per uniform section should be carried out. 2.3.2. Moisture conditions The moisture conditions at the time of the DCP survey need to be carefully estimated as discussed in Section 1.1. As the moisture content at the time of testing determines the in situ strength at that time, this needs to be carefully assessed and preferably supported by laboratory determinations of the moisture content. This will relate to which of the curves in Figure 1 will be used for the design. 2.3.3. Uniform sections The road should then be divided into uniform sections based on the DCP results. Various techniques are available for this, but it has been found that the cumulative sum technique [11] is simple and appropriate. This involves determining the average DCP CBR for all of the results (for each 150 mm layer tested), subtracting the individual results from the average and then summing these. A plot of the results will show inflection points where each section changes (see example). Once the uniform sections have been identified, each of these will need a specific pavement design or treatment.

426

P. Paige-Green / Applying the Dynamic Cone Penetrometer Design Method to Low Volume Roads

2.4. Pavement design The pavement design process needs to fit the pavement structure (Fig.1) to the in situ conditions on each uniform section determined from the DCP survey as shown in Section 3 (Figure 2). Where the in situ strength is less than the design strength, improvement of the material needs to be carried out. The method should try and fit the pavement design to the available structure as far as possible, without importing additional material. 2.5. Specific treatments To minimize costs, use of the in situ material in all of the layers should be considered. However, mechanical treatment of the in situ materials such as ripping and recompaction may not always be sufficient for the proposed pavement design and some other form of treatment or stabilization may be necessary. This could range from removal and replacement, heavy compaction or mechanical or chemical stabilization. Indications of the need for treatment will be obtained from the DCP results when particularly poor material properties in the upper layers are identified.

3. Examples Examples of the use of the DCP design technique for the upgrading to sealed standard of two existing gravel roads are discussed. Different approaches were used for each. The first road was through a mountainous area in the Western Cape. The second example was for the upgrading of a local access road in the Eastern Cape Province. The same basic design was proposed for both pavements, which were Category D roads in areas with Weinert N-values of about 8 and 1.8 respectively with a 10 year design life. The estimated traffic was less than 100 000 standard E80 axles for both roads. From TRH 4 [8] this would require a structure of 100mm G4 base over 125 mm G6 subbase. A 150 mm G9 support layer would be required under this. The thickness of the subbase was increased to 150 mm in both cases. 3.1. Western Cape example DCP data from the mountainous pass were all plotted using a computerized DCP code and the mean penetrations for the layers between 0 and 150 mm (proposed subbase), 150 and 300 mm (proposed upper selected) and 300 to 450 mm (proposed in situ subgrade) along the centre-line were determined. It should be noted that only 3 of the 10 penetrations were able to reach more than 450 mm, 4 reached more than 300 mm and 8 reached deeper than 150 mm before reaching refusal. The results obtained are summarized in Table 2 and shown graphically in Figure 2. Although the road is in a dry area and the testing was carried out at the end of the wet season, the 25th percentile value for the CBR was utilized. This decision was based on the frequent presence of springs adjacent to roads in such mountainous areas. The following conclusions were drawn from the data.

P. Paige-Green / Applying the Dynamic Cone Penetrometer Design Method to Low Volume Roads

427

Table 2. Summary of DCP results and statistics Depth and DCP-CBR (%) at depth 150 – 300 mm 300 – 450 mm

Kilometer point 0 – 150 mm 13.4 14.66 17.16 18.16 20.89 21.89 22.89 23.39 23.89 24.7 Mean 25th percentile 75th percentile

65 113 88 98 200 184 65 94 113 84 110 85 113

200 202 200 200 200 200 103 106 115 132 166 119 200

200 200 200 200 200 200 64 45 200 190 170 193 200

Depth of refusal (mm) 120 250 150 180 60 150 800 800 300 500 331 150 450

CBR (% )

1

10

100

1000

0 50 100 Depth (mm)

150 200 250 300 350 400 450 500

Figure 2. Plot of DCP data (25th percentiles dashed red lines).

300 – 450 mm depth No problems exist at this depth where a G9 material would be required (minimum soaked CBR of 7%). The minimum average in situ CBR was 45% with a 25th percentile of 193 for the entire road, and generally the material was impenetrable. 150 – 300 mm depth The proposed upper selected layer requires a G9 (soaked CBR of 7%). The mean CBR was 166 and the 25th percentile 113, indicating adequate materials. The in situ material should, however, be disturbed as little as possible during construction. The cumulative sum analysis (Figure 3) showed two distinct sections (from the start to km 22.0 where refusal was reached, and from there to the end), but both had strengths well in excess of those required (Table 2).

428

P. Paige-Green / Applying the Dynamic Cone Penetrometer Design Method to Low Volume Roads

150 - 300 mm 50 Cum sum

0 -50 -100 -150 -200 -250 10

12

14

16

18

20

22

24

26

Km

Figure 3. Cumulative sum plot for material at depth of 150 to 300 mm.

0 – 150 mm depth The proposed subbase material is included in this layer, which showed much more variation than the other layers. A cumulative sum approach was again used to identify uniform sections. It should be noted that it is based on only 10 results and for better definition of uniform sections it would have been advisable to use DCP data along the centre-line from at least 500 m intervals or preferably 200 m intervals. The results of the cumulative sum analysis (Figure 4) show three distinct uniform sections, probably related to the use of three different borrow pits along the road during recent regravelling operations. 0 - 150 mm

Cum sum

100

Section 1

Section 2

Section 3

50 0 -50 -100 10

12

14

16

18

20

22

24

26

Km

Figure 4. Cumulative sum plot of DCP strengths between 0 and 150 mm depth.

The proposed designs of the three sections were as follows: • Section 1: km 13.4 – 18.16 This section has an average in situ CBR of 91% (25th percentile of 82) indicative of a suitable G6 subbase material. Local ripping and re-compaction of the upper 75 to 100 mm to refusal was advised in some areas shown by the DCP plots. • Section 2: km 18.16 – 21.89 This section has an average in situ CBR of 192% (25th percentile of 188) indicative of a suitable G6 subbase material.

P. Paige-Green / Applying the Dynamic Cone Penetrometer Design Method to Low Volume Roads

429

• Section 3: km 21.89 - 24.7 This section has an average in situ CBR of 89% (25th percentile of 79) indicative of a suitable G6 subbase material, very similar to the first section. Local ripping and re-compaction of the upper 75 to 100 mm to refusal was advised in some areas. Base On top of the sub-structure described above, a base course with a suitable thickness and strength would be necessary. This would typically be 150 mm of G4 or an equivalent layer of stabilized or bitumen treated material. 3.2. Eastern Cape example The length of the road was 20 km and 98 DCP tests were carried out. Although the moisture environment was described as dry during the survey, the testing was carried out during the rainy season and some seepage was seen on bedding planes in the shales and sandstones. Rather to err on the conservative side, the 20th percentile has been used for the CBR estimations (the DCP software used did not allow computation of the 25th percentile). The DCP data was analyzed in 2-km sections. The standard deviation and percentile values were examined to determine whether the sections were uniform. In all cases the sections were surprisingly uniform with very small standard deviations of the DCP determined CBR. Analysis of the average profile was then carried out using an overlay of the proposed pavement structure and the 20th percentile DCP values for each section. The results are summarized in Table 3. Table 3. Suitability of existing road cross-section Chainage

Uniformity

0.2 – 1.8 2.0 – 3.8 4.0 – 5.8 6.0 – 7.8 8.0 – 9.8 10.0 – 11.8 12.0 – 13.8 14.0 – 15.8 16.0 – 17.8 18.0 – 19.8

√ √ √ √ √ √ √ √ √ √

Support layer

√ √ √ √ √ √ X √ √ √

Subbase

√ √ √ √ X √ X X X √

Base

X X X X X X X X X X

An analysis of all of the DCP results confirmed the findings of the individual sections. The Redefined Layer Strengths show a G6 subbase quality layer to 192 mm (20th percentile in situ CBR 32%), and generally good support beneath this (20th percentile CBR of 27 to 30% down to 272 mm). The following designs were thus applied. Km 0 – 8, 10 – 12 and 18 -20: Km 8 – 10 and 14 – 18:

Shape existing wearing course Import 150 mm G4 base Rip and re-compact 150 mm as subbase

430

P. Paige-Green / Applying the Dynamic Cone Penetrometer Design Method to Low Volume Roads

Km 12 – 14:

Import 150 mm G4 base Windrow top 300 mm of material from road Rip and re-compact underlying 150 mm Mix and replace windrowed material in two 300 mm layers Import 150 mm G4 base

4. Conclusions The DCP design technique has been shown to be highly appropriate for the design of low volume roads. Testing is quick, cheap and non-destructive. Two roads designed using this technique have been described, the latter having been successfully built to this design 7 years ago. It is, however, important that the designer has a good understanding of the in situ moisture and density conditions at the time of the DCP testing and understands the relationships between the field (design) and laboratory test (standards and specification) results for the materials involved.

References [1] P. Paige-Green and L. J. du Plessis. Use and interpretation of the dynamic cone penetrometer (DCP) test. Accessed http://researchspace.csir.co.za/dspace/bitstream/10204/3692/1/Paige-Green_2009.pdf 2/01/2011 [2] D.J. van Vuuren, Rapid Determination of CBR with the portable Dynamic Cone Penetrometer. The Rhodesian Engineer, Paper No 105, September 1969. [3] E.G. Kleyn and G.G. Van Zyl, Application of the Dynamic Cone Penetrometer (DCP) to light pavement design, Pretoria: Transvaal Provincial Administration, Laboratory Report L4/87, 1987. [4] S.J. Emery, Prediction of moisture content for use in pavement design. Johannesburg: University of the Witwatersrand, PhD Thesis, 1985. [5] National Institute for Transport and Road Research (NITRR). Guidelines for road construction materials. TRH 14, NITRR, CSIR, Pretoria, 1985. [6] M.C. Shackleton & S.J. Emery. Investigation of CBR versus moisture content relationships for untreated materials. Pretoria: National Institute of Transport and Road Research. Report TS/4/85, 1985. [7] P Paige-Green, J Lea and C Barnado. The relationship between in situ DCP strength and soaked CBR. Pretoria; CSIR Transportek, Technical Report TR-99/003, 1999. [8] Committee of Land Transport Officials (COLTO). Structural design of flexible pavements for interurban and rural roads. Draft TRH 4, Department of Transport, Pretoria, 1996. [9] H. Wolff, G.D. van Zyl, P. Paige-Green & S.J. Emery. The development of a structural design catalogue for low volume roads. Proc Annual Transportation Convention, Vol 4C, Pretoria, 1993. [10] G.D. van Zyl, P.W. Jordaan, H. Wolff, E. de Beer, H.E. Bofinger, D.J. Jones, P.C. Curtayne, P. PaigeGreen, H. Ribbens, P.A. Pienaar and S.J. Emery. Guidelines for low volume roads. Report PR 92/466/001, South African Roads Board, Pretoria, 1993. [11] Ministry of Works, Transport and Communications. Pavement testing, analysis and interpretation of test data. Guideline No 2, International Cooperation Roads Department/NPRA, Oslo, Norway, 2000.

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 431 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-431

The Use of a Sedimentological Technique for Assessing the Engineering Performance of Sands in Roads Philip PAIGE-GREENa and Michael PINARD b CSIR Built Environment, Pretoria, South Africa b InfraAfrica Consultants, Gaborone, Botswana

a

Abstract. Although sands cover vast areas of southern Africa, their use as road materials in an untreated state has been generally avoided. An investigation into the properties of various sands has indicated, however, that they can be used as subbase and even base course materials in low volume roads if carefully selected and tested. The investigation has shown that by expressing the particle size distribution of the sands as the mean particle size and standard deviation around this mean using the sedimentological Phi scale, it is possible to differentiate between sands that are likely to perform well and those that will not. The paper discusses this process and its application to selected sands from Botswana, Namibia, South Africa and Mozambique. Keywords. Sand, road, unbound, Phi-scale

Introduction The need for an improved road network for better access and increased mobility in many rural areas of southern Africa is growing rapidly as the populations increase. The provision of roads in these areas, however, is often constrained by the cost of obtaining suitable construction materials for use in their bases and subbases. This is particularly evident in those areas of southern Africa including localized parts of South Africa and Mozambique overlain by Tertiary and Quaternary sands and specifically widespread areas in Botswana and Namibia with surficial Kalahari sand deposits (Figure 1). These sands have a variety of origins but are predominantly aeolian with some river and beach deposits. Use of these sands as structural layers in roads in an untreated state has been generally avoided in the past. Various local investigations into the properties of a range of sands combined with some past experience have indicated, however, that they can have relatively high strengths when compacted in confined conditions and be used as subbase and even base course materials in low volume roads if carefully selected and tested and when properly constructed. This investigation has shown that by expressing the particle size distribution of the sands as the mean particle size and standard deviation around this mean using the sedimentological Phi-scale [1], it is possible to differentiate between sands that are likely to perform well and those that will not.

432

P. Paige-Green and M. Pinard / The Use of a Sedimentological Technique

The paper discusses this technique and its application to selected sands from Botswana, Namibia, South Africa and Mozambique. The objective of the paper is to introduce the concept of the phi technique and not to relate the results of specific sands to performance.

Figure 1. Distribution of Kalahari sands in southern Africa

1. Background 1.1. Types and origin of sand The sands that predominate in southern Africa were mostly produced by rock weathering after which the constituents have been transported by wind or water. Their composition, shape and properties typically differ from the well-rounded and sorted sands that are normally associated purely with river and beach deposits. It is these unique properties that allow them to perform satisfactorily as road construction materials. Considerable work was carried out in Australia during the early 1980s [2] on the local “sand clays”, which are mostly derived from stranded beach ridges and consist of rounded to sub-angular quartz grains, cemented together and containing some clay and iron staining [3]. Brazilian sands tend to be derived from the weathering and transport of sandstones and consist of sandy quartz with kaolinite and ferruginous oxides [4]. The Kalahari sands of southern Africa (known as the Kgalagadi sands in Botswana) were derived from the erosion of underlying rock and subsequent transport and redistribution. This was carried out by rivers into lakes and by wind. The surficial sands observed today were deposited primarily by wind. Baillieul [5] carried out sedimentological work in Botswana and analyzed samples from the “topmost layer of the sand”. He identified four major sand areas, each having distinct types of sand depending on their mode of formation (aeolian, residual, fluvial

P. Paige-Green and M. Pinard / The Use of a Sedimentological Technique

433

with bioturbation) using the sedimentological phi-scale classification to characterize the sands. 1.2. Previous work The lack of materials and widespread nature of the “sand clays” in Australia led to considerable research in this area in the early 1980s [6]. The importance of the particle size distribution was highlighted in this work and was related to the performance of the materials in specifications published in 1984 [7]. These specifications made use of the traditional sedimentological technique of expressing the particle size (mean) and standard deviation about the mean in terms of Phi units and analyzing the materials in this way similar to the method used by Baillieul [5]. The method is discussed in detail below. Metcalf and Wylde [7] plotted (Figure 2) the mean particle size on the vertical axis and the standard deviation on the abscissa (both in Phi-scale units) and identified a zone into which sand materials suitable for use as base course would fall (B). Materials falling in zone A were described as loamy, boney or puggy, i.e. not enough fines to bind the material and would not perform well as a base course material. Material in zone D is generally too greasy (plastic) for use and although some of the materials in zone C had been used successfully, they had given problems during construction and before sealing.

D 3.0

Mean size (phi)

C B

A

2.0

1.0 1.5

2

2.5

3

Standard Deviation (phi)

Figure 2. Plat of material performance using mean particle size and standard deviation [7].

This paper applies these sedimentological principles to various sands found in southern Africa, where traditional test methods and classification parameters (e.g., grading modulus) fail to differentiate adequately between the sands.

434

P. Paige-Green and M. Pinard / The Use of a Sedimentological Technique

2. Properties of sands 2.1. Fundamental properties Engineering materials are generally described in terms of various physical properties, the particle size distribution and plasticity being the common “classification” or indicator tests. These two properties are suitable for typical soils and aggregates but generally lack adequate discrimination for fine sandy materials. Such sands frequently fall within only a few fine sieve sizes with the majority of the material being between 0.075 and 2 mm in size. In addition, the plasticity as determined from the conventional Atterberg test is usually non- or possibly slightly plastic. When such sands are proposed for use in pavement layers, be it lower support layers or even upper structural layers in low volume roads, more information regarding their properties is necessary. It has become local practice to determine the plasticity index on the fraction passing 0.075 mm (instead of the normal 0.425 mm) and it is not uncommon to measure plasticity indices up to 40 or 50 per cent. These, of course, should not be compared directly with the conventional Atterberg limits but are useful indicators in the context of fine materials. This aspect is not discussed further in this paper. Comparison of grading analyses is always difficult as these are usually represented graphically or by a combination of various values. Parameters such as the grading modulus (GM) and grading coefficient have been used to reduce particle size distributions to a single value for comparative purposes and are useful in their respective contexts. The grading modulus of sands, however, typically lies in a restricted range (0.9 to 1.2) allowing little discrimination between materials. This is the result of one of the properties (percentage retained on the 2.0 mm sieve) frequently being close to zero and the percentage retained on the 0.075 mm sieve frequently being between 95 and 100. This essentially limits discrimination between sands using the GM to changes in the percentage passing 0.425 mm. 2.2. Interpretation and comparison The performance of sands thus cannot be determined from typical grading analyses. However, it is known that their performance is a function of the inclusion of some fines, usually too few and too small to be identified without careful hydrometer analyses and not usually considered in the standard interpretation of grading analyses of fine materials. Discussions on some performance aspects of sands in this regard have been published previously [8][9].

3. Proposed sedimentological method 3.1. The phi (Φ) method Because particle size distribution plots of fine sands using cumulative percentages passing are difficult to compare and quantify in simple terms, sedimentologists [1], [10] have developed and implemented the Phi (Φ)-scale for particle size distribution analysis where

P. Paige-Green and M. Pinard / The Use of a Sedimentological Technique

435

Φ = -log2d d = particle size in mm. (A particle size of 0.5 mm = Φ of 1 and a particle size of 0.125 mm = Φ of 3). The classification of the sands by Baillieul [5] discussed earlier was based on the Phi (Φ)-scale as it allows a simple calculation of the mean particle size (Φmean) and the standard deviation of the particle sizes about the mean (Φsd). These two parameters give a direct indication of the mean particle size of the samples analyzed as well as the degree of sorting, based on the standard deviation. This facilitates the interpretation of sand properties using two simple parameters and simplifies the direct comparison of different sands. The higher the standard deviation, the wider is the grading (less sorting) of the sand and the more material there is available to provide a tighter packing of the sand when compacted to minimize the voids in the material. These fines also contribute to increasing the soil suction as the material dries back from compaction moisture content. Use of the method requires the determination of the particle size at various percentiles of the particle size distribution plotted in terms of the cumulative percentage retained. If there is a significant portion (more than about 6%) of material finer than 0.075 mm, a hydrometer analysis is required in addition to the standard sieve analysis as the 95th and 84th percentiles (P95 and P84 respectively) retained (measures of the fine fractions) are required for the calculations. The mean (1.61 Φ units or 0.385 mm) and standard deviation (1.84 Φ units) of the following grading analysis (Table 1) are calculated as follows and illustrated with the grading curve in Figure 3 [11].

Φmean = (P84 + P50 + P16)/3 Φsd = ((P84 – P16)/4 ) + ((P95 – P5)/6.6) Table 1. Grading (sieve and hydrometer) analysis results in phi terms Sieve size (mm) 4.75 2.0 1.18 0.425 0.25 0.15 0.075 0.07 0.048 0.029 0.019 0.012 0.008 0.006 0.003 0.002

Sieve size (Φ) -2.25 -1.00 -0.24 1.23 2.00 2.74 3.74 3.84 4.38 5.11 5.72 6.38 6.97 7.38 8.38 8.97

% passing 99.0 98.2 97.9 50.3 40.4 27.8 9.5 8.7 8.1 7.5 6.8 5.9 4.7 4.5 4.0 3.2

% retained 1.0 1.8 2.1 49.7 59.6 72.2 90.5 91.3 91.9 92.5 93.2 94.1 95.3 95.5 96.0 96.8

436

P. Paige-Green and M. Pinard / The Use of a Sedimentological Technique

Percent retained

φsd = 1.61 ± 1.84 φmean = 1.61

100 90 80 70 60 50 40 30 20 10 0

P95 P84

P50

P16 P5

-7

-5

-3

-1

1

3

5

7

9

Sieve size (phi)

Figure 3. Cumulative percentage retained plot and percentile values.

4. Application to southern African sands 4.1. Botswana During an investigation into the use of various Kgalagadi sands in structural layers in low volume roads, the technique was applied to a wide range of sand materials from Botswana. Those sands that performed well as base course materials were clearly identified using this approach [9]. The results of all of the analyses are shown in Figure 4. It can be seen that there is a wide spread of results allowing separation between different materials that was not possible using the grading modulus. Those materials that plot with low Φmeans and high Φsds indicate slightly coarser sands with a wider particle size distribution, providing higher strengths and better filling of voids. 4.2. Namibia A similar investigation to that described above was carried out in Namibia, in which materials that performed well as unsealed wearing course gravels were compared with those that performed poorly. The results of the analyses are also plotted on Figure 4, where the materials with the higher standard deviations performed markedly better than those with lower standard deviations. 4.3. Mozambique Testing of various aeolian/river sands for use in road construction has been carried out in Mozambique as well. It is interesting to note that these sands contained a high proportion of heavy mineral particles (e.g. ilmenite and zircon), which obviously affects a particle size distribution analysis based on gravimetrically determined size fractions. These materials, however, were generally quite coarse grained (average mean particle diameter = 0.6 mm) and had wide gradings (Figure 4). High densities and good CBR strengths were obtained from those materials that had the higher standard deviations.

P. Paige-Green and M. Pinard / The Use of a Sedimentological Technique

437

4.4. South Africa The South African data is restricted to a number of test results from an investigation carried out on a road rehabilitation project in northern KwaZulu-Natal - KZN (Figure 3). The materials were relatively fine (mean particle size 0. 25 mm) but those with the higher standard deviations gave higher densities and were markedly stronger (CBRs > 30%) 4.50 4.00

Mean particle size (phi)

3.50 3.00 Botswana

2.50

Namibia KZN

2.00

Mozambique

1.50 1.00 0.50 0.00 0.00

0.50

1.00

1.50

2.00

2.50

3.00

3.50

Standard deviation (phi)

Figure 4. Plots of Φmean versus Φsd for various southern African sands

It is clear that the distribution and spread of the data plotted show marked differences between the sands from the various regions as well as significant differences within sands from any one region. This illustrates the potential usefulness of the Phi-scale technique.

5. Conclusion The comparison of fine sands using conventional engineering test methods is very difficult. Recent research making use of the sedimentological Phi (Φ)-scale has allowed direct comparison of different sands based on a simple sieve analysis together with a hydrometer analysis when necessary. The results allow much better discrimination between the different materials and have been correlated with higher strengths and improved field performance. It is recommended that more use should be made of this simple technique when sandy materials are being investigated in engineering projects, particularly for investigations related to the performance (good or bad) of sandy materials.

References [1] W.C. Krumbein, and L.L. Sloss, Stratigraphy and sedimentation. Freeman and Co San Francisco, 1951. [2] L.J. Wylde, Marginal quality aggregates used in Australia. Report ARR No 97, ARRB, Melbourne, 1979. [3] R. Sandman, R. Wall and N. Wilson, N. Use of natural materials for pavement construction in South Australia. Report MS 26, Highways Department, South Australia, 1974.

438

P. Paige-Green and M. Pinard / The Use of a Sedimentological Technique

[4] L.A.S. Aranovich, and A.T. Heyn, Performance of low cost pavements in Parana, Brazil. Proc Int Conf Roads and Development, Paris, (1984), 2, 761-766. [5] T.A. Baillieul, A reconnaissance survey of the cover sands in the republic of Botswana. J Sedi Petrology 45 (1975), 494-503. [6] L.J. Wylde, Personal Communication, Adelaide, Australia, 1982. [7] J.B. Metcalf, and L.J. Wylde, A re-examination of specification parameters for sandy soil roadbase materials. Bull Int Ass Engng Geol 30 (1984) 435-437. [8] M.B. Mgangira. Microstructural pavement material characterization: some examples. Partnership for research and progress in Transportation. 27th Southern African Transport Conference (SATC), Pretoria, South Africa, 2008, pp 12 [9] P. Paige-Green, M. Pinard and M.B. Mgangira. The use of aeolian sands in the provision of low volume roads, Paper accepted for publication at 10th International Conference on Low-Volume Roads, Orlando, Florida, July 2011. [10] D. Marsal, Statistics for geoscientists. Pergamon Press, New York, 1987. [11] R.L. Folk & W.C. Ward. Brazos river bar: a study of significance of grain size parameters. J. Sediment. Petrol., 1957, 27, 3-26.

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 439 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-439

Characterization of Pozzolanic Geomaterial for the Construction of Pavement Structures of Songwe Airport in Tanzania Paul OMINDOa,1, John MUKABIb, Prosper TESHAc, Vincent SIDAId, Sylvester KOTHEKIe and Leonard NGIGI f a Kundan Singh Construction Co. Ltd, Mbeya, Tanzania b,e,f Kensetsu Kaihatsu Consultants, Nairobi, Kenya c Tanzania Airports Authority, Dar es Salaam, Tanzania d Bamburi Cement-The Lafarge Group, Nairobi, Kenya Abstract. Ǥ ͸ͳͺǦͻͶ ǡ ǡ Ǥ ǡ Ȁ Ǥ ǡ Ǧ ǡ ǯ Ǧ Ǥ ǡ ȋȌǤ ǡ Ǥ Keywords. Pozzolana, geomaterial, OPMC, pavement, design, MRR, SCDR

1

John MUKABI: CEO, Kensetsu Kaihatsu Limited, A6 Mac Apts., Lavington, P. O. Box 3524600200, Nairobi, KENYA; E-mail: [email protected]

440

P. Omindo et al. / Characterization of Pozzolanic Geomaterial

Introduction The idea of construction of a new airport in Mbeya was first conceived in 1977, with a followed up in 1987. The Feasibility Study (F/S) recommended locating the new airport outside the city due to the extent of development in the downtown area. The initial design was to cater for a Fokker F50 as the design aircraft with provision for future expansion to accommodate a Boeing 737. In May 2004, however, a redesign of the runway, taxiway and aprons was undertaken in order to facilitate for B737 operations. In order to confirm, more precisely, the engineering properties of the existing soils as well as the behavior of the existing ground and pavement structure, monitoring, technical evaluation and geotechnical engineering investigations were undertaken. The preliminary results indicated that the existing ground and pavement structure exhibited much higher bearing capacity and strength responses in comparison to the values that may have been considered in the existing design. As a consequence, the Tanzania Airports Authority (TAA) made a decision to embark on further and more detailed laboratory and in-situ experimental testing, technical evaluation, geotechnical engineering investigations and analyses. This was also in consideration of the fact that it was most likely that the existing design did not take into account the pozzolanic cementetious nature of the natural geomaterial in existence and its immediate response to compaction and the effects of time related consolidation, thixotropy and creep (secondary consolidation). In this Study, naturally existing geomaterial constitute of pozzolanic properties is characterized through the combination of silica and calcium hydroxide in the presence of water to form stable calcium silicates which have cementitious properties. This chemical reaction is further enhanced and observed through the physical addition of varying percentages of factory produced Pozzolanic Portland Cement (PPC) of similar raw material extruded from the vicinity of the construction site of Songwe International Airport in the Mbeya Region of Tanzania. This geomaterial is considered to be calcined diatomaceous soil originating from volcanic ash.

1. Testing and Geotechnical Investigation Regimes Details of the standard and innovative testing regimes are discussed in the Airport Pavement Design Review Engineering Report No. SAT09T1 of March 2010, the Pavement Structural Design Engineering Report Based on Analysis of Trial Sections of April 2010 and Innovative Laboratory and In-situ Methods of Testing in Geotechnical Engineering [1].

2. Relevant Scientific and Engineering Concepts Applied for Analysis The relevant scientific and engineering theories and concepts adopted for analysis in this study are reported in the Songwe Airport Pavement Design Review Engineering Report No. SAT09T1, March 2010 [2].

441

P. Omindo et al. / Characterization of Pozzolanic Geomaterial

3. Materials Characterization and Analysis of Test Results The importance of studying consolidation properties was considered for three main reasons; 1) to analyze the effect of chemical stabilization on consolidation properties since consolidation is one of the methods commonly applied for ground improvement, 2) to evaluate whether or not and to what extent water infiltration or groundwater seepage would affect the consolidation properties of the chemically stabilized geomaterials associated with settlement and reduction in magnitude of shear stress as well as resistance to deformation, 3) to evaluate whether further secondary consolidation is likely to occur to a detrimental extent that would cause settlement particularly for the lower layers under surcharge and dynamic air traffic loading. Typical consolidation results, which were derived on the basis of CSSR functions [3], are presented in Table 1. In general, it can be noted that chemical stabilization enhances the vital consolidation parameters such as δCSR, CSR and Δφ,. From this Study, it was also observed that the degree of influence of the chemical stabilization on the vital consolidation parameters depends on the type of geomaterial, and that the correlation of the vital consolidation parameters is quite consistent notwithstanding the curing or soaking conditions. Table 1 Summary of Consolidation Stress Parameters Derived from In-situ Tests ^ĞƌŝĂů EŽ͘

>ŽĐĂƚŝŽŶ

ϭ Ϯ ϯ ϰ ϱ ϲ ϳ ϴ ϵ ϭϬ

ϮнϬϬϬ > ϮнϬϬϬ Z,^ ϮнϭϬϬ Z,^ ϮнϭϬϬ >,^ ϮнϮϬϬ Z,^ ϮнϮϬϬ >,^ ϮнϯϬϬ Z,^ ϮнϯϬϬ >,^ ϮнϰϬϬ Z,^ ϮнϰϬϬ >,^

h^ƋƵ ;DWĂͿ

ϱ͘Ϯϯ ϴ͘ϲϯ ϰ͘ϰϴ ϯ͘Ϯϭ ϭϴ͘ϵϲ ϭϱ͘ϲϰ Ϯ͘Ϭϱ ϭϯ͘ϭϭ ϳ͘ϱϰ ϭϬ͘ϳϲ

δCSR ϱϲ͘ϯϭϳ ϲϮ͘ϰϰϵ ϱϰ͘ϵϱϲ ϱϮ͘ϲϱϵ ϴϭ͘ϭϭϳ ϳϱ͘ϭϮϯ ϱϬ͘ϱϲϳ ϳϬ͘ϱϰϰ ϲϬ͘ϰϴϲ ϲϲ͘ϯϭϬ

CSR ϭ͘ϯϰϰ ϭ͘ϯϳϴ ϭ͘ϯϯϲ ϭ͘ϯϮϭ ϭ͘ϰϲϱ ϭ͘ϰϰϬ ϭ͘ϯϬϴ ϭ͘ϰϭϵ ϭ͘ϯϲϳ ϭ͘ϯϵϴ

Δφ ϭϳ͘ϭϯϭ ϭϴ͘ϭϯϬ ϭϲ͘ϵϭϬ ϭϲ͘ϱϯϲ Ϯϭ͘ϭϲϵ ϮϬ͘ϭϵϯ ϭϲ͘ϭϵϱ ϭϵ͘ϰϰϴ ϭϳ͘ϴϭϬ ϭϴ͘ϳϱϴ

1 1 Δφ/δCS ψ 1 (ΔSR)C KO K C σ ac σ rc ;DWĂͿ ;DWĂͿ

Ϭ͘ϯϬϰ Ϭ͘ϮϵϬ Ϭ͘ϯϬϴ Ϭ͘ϯϭϰ Ϭ͘Ϯϲϭ Ϭ͘Ϯϲϵ Ϭ͘ϯϮϬ Ϭ͘Ϯϳϲ Ϭ͘Ϯϵϰ Ϭ͘Ϯϴϯ

Ϭ͘ϵϴϰ Ϭ͘ϵϳϬ Ϭ͘ϵϴϴ Ϭ͘ϵϵϱ Ϭ͘ϵϰϮ Ϭ͘ϵϰϵ ϭ͘ϬϬϭ Ϭ͘ϵϱϲ Ϭ͘ϵϳϰ Ϭ͘ϵϲϯ

ϭ͘ϭϬϳ ϭ͘Ϯϯϱ ϭ͘Ϭϳϵ ϭ͘ϬϯϮ ϭ͘ϲϯϲ ϭ͘ϱϬϱ Ϭ͘ϵϴϵ ϭ͘ϰϬϳ ϭ͘ϭϵϰ ϭ͘ϯϭϳ

Ϭ͘ϯϲϯ Ϭ͘ϯϮϯ Ϭ͘ϯϳϮ Ϭ͘ϯϴϵ Ϭ͘Ϯϭϴ Ϭ͘Ϯϰϵ Ϭ͘ϰϬϰ Ϭ͘Ϯϳϰ Ϭ͘ϯϯϱ Ϭ͘Ϯϵϵ

Ϭ͘ϯϱϱ Ϭ͘ϯϯϵ Ϭ͘ϯϱϵ Ϭ͘ϯϲϲ Ϭ͘Ϯϵϳ Ϭ͘ϯϬϵ Ϭ͘ϯϳϮ Ϭ͘ϯϭϵ Ϭ͘ϯϰϰ Ϭ͘ϯϮϵ

ϭϬ͘ϭϯ ϭϲ͘ϰϮ ϴ͘ϳϭ ϲ͘Ϯϴ ϯϰ͘ϱϴ Ϯϴ͘ϴϵ ϰ͘Ϭϰ Ϯϰ͘ϰϲ ϭϰ͘ϰϯ ϮϬ͘Ϯϵ

ϭ͘ϳϲ Ϯ͘ϲϮ ϭ͘ϱϰ ϭ͘ϭϱ ϰ͘Ϯϰ ϯ͘ϴϲ Ϭ͘ϳϲ ϯ͘ϰϴ Ϯ͘ϯϳ ϯ͘Ϭϳ

qC

pC

’

(MPa) (MPa) ϴ͘ϯϳ ϭϭ͘ϯϬ ϭϯ͘ϴϬ ϭϴ͘ϭϳ ϳ͘ϭϲ ϵ͘ϳϯ ϱ͘ϭϯ ϳ͘Ϭϰ ϯϬ͘ϯϰ ϯϳ͘ϰϭ Ϯϱ͘Ϭϯ ϯϭ͘ϰϲ ϯ͘Ϯϴ ϰ͘ϱϰ ϮϬ͘ϵϳ Ϯϲ͘ϳϴ ϭϮ͘Ϭϲ ϭϲ͘Ϭϭ ϭϳ͘ϮϮ ϮϮ͘ϯϰ

One of the typical results of ground and geomaterial characteristics subjected to in-situ dynamic loading adopting the Dynamic Cone Penetration (DCP) is graphically presented in Figure 1. The fact that the existing ground is very sound having undergone Long-Term Consolidation (LTC) can be derived from the very high bearing capacity and strength magnitudes that it exhibits attributable to its’ pozzolanic nature.

442

P. Omindo et al. / Characterization of Pozzolanic Geomaterial

h^;ŬŐĨͬĐŵϮͿ Ϭ

ϱϬ

ϭϬϬ

ϭϱϬ

ĞƉƚŚ;ŵŵͿ

ϮϬϬ ϮϱϬ ϯϬϬ ϯϱϬ ϰϬϬ ϰϱϬ Figure 1. Effect of Long-Term Consolidation on the pozzolanic nature of existing ground

Table 2 Summary of Shear Stress Parameters Derived from In-situ Tests UCS qu (M Pa)

h ^͕ƋƵ > d

Ϯ н Ϭ Ϭ Ϭ Ͳ> , ^

ϭ Ϯ ϯ ϰ ϱ ϲ ϳ ϴ

h ^͕ƋƵ

S/N Location

ϯ͘ϲϮ ϳ͘Ϯϱ ϳ͘ϱϴ ϴ͘ϲϵ ϱ͘ϭϳ ϭϭ͘ϳϰ ϭϮ͘Ϯϵ ϭϰ͘ϭϱ

P

'

Cu

q max

(kgf/cm 2)

(kgf/cm 2)

(kgf/cm 2)

ϭ͘ϴϭ ϯ͘ϲϮ ϯ͘ϳϵ ϰ͘ϯϱ Ϯ͘ϱϴ ϱ͘ϴϳ ϲ͘ϭϱ ϳ͘Ϭϴ

ϱ͘ϴϬ ϭϭ͘ϱϵ ϭϮ͘ϭϯ ϭϯ͘ϵϭ ϴ͘Ϯϳ ϭϴ͘ϳϴ ϭϵ͘ϲϳ ϮϮ͘ϲϰ

ϰ͘ϵϮ ϵ͘ϮϮ ϵ͘ϱϵ ϭϬ͘ϳϵ ϲ͘ϴϮ ϭϯ͘ϴϰ ϭϰ͘ϯϳ ϭϲ͘Ϭϲ

f

'A Average

Ϯϵ͘ϱϬϮ ϯϭ͘ϯϮϳ ϯϭ͘ϰϵϳ ϯϮ͘Ϭϱϳ ϯϬ͘ϮϴϬ ϯϯ͘ϱϵϭ ϯϯ͘ϴϳϬ ϯϰ͘ϴϬϴ

σ

' a

σ

' r

(kgf/cm 2)

(kgf/cm 2)

ϳ͘Ϭϴ ϭϯ͘ϴϵ ϭϰ͘ϱϭ ϭϲ͘ϱϱ ϭϬ͘Ϭϭ ϮϮ͘Ϭϯ Ϯϯ͘Ϭϭ Ϯϲ͘Ϯϵ

ϭ͘Ϯϴ Ϯ͘Ϯϵ Ϯ͘ϯϴ Ϯ͘ϲϰ ϭ͘ϳϰ ϯ͘Ϯϱ ϯ͘ϯϱ ϯ͘ϲϱ

Δ SR ϭ͘ϭϵϵϱ ϭ͘Ϯϳϲϱ ϭ͘Ϯϴϯϳ ϭ͘ϯϬϳϯ ϭ͘ϮϯϮϯ ϭ͘ϯϳϮϬ ϭ͘ϯϴϯϴ ϭ͘ϰϮϯϰ

Emax (M Pa)

ϱϱϭϭ ϳϭϳϭ ϳϮϵϳ ϳϲϴϲ ϲϯϬϳ ϴϲϭϰ ϴϳϲϳ ϵϮϰϵ

Gmax (M Pa)

ϭϴϯϳ ϮϯϵϬ ϮϰϯϮ ϮϱϲϮ ϮϭϬϮ Ϯϴϳϭ ϮϵϮϮ ϯϬϴϯ

The results in Table 2 demonstrate very high shearing and deformation resistance properties of the pozzolanic material in its natural state due to LTC as also shown in Figure 2 . The effects of LTC were computed by applying Eq. 1[3]. LTC = q max

[

K 0STC

STC K 0STC • qmax − (1n t / t 0 )Aφ '•CSR STC

]

(1)

where, Superscript LTC and STC denote Long-Term and Short-Term Consolidation respectively whereas t : LTC time and to : STC time.; for OC conditions (Δεa/Δt)fcSTC=1. The results basically indicate that, for stiff geomaterial such as the one tested, the shearing strength increases, to a large extent, directly proportional to the deformation resistance. The results also showed that, for pozzolanic geomaterial, as the shearing strength increases with the deformation resistance, the linear elastic range is immensely enhanced. Figure 2 indicates that a difference of 1% cement addition significantly enhances the durability of the OPMC-CC (Optimum Mechanical and Chemical Cement Concentrate) stabilized pozzolanic geomaterials tested in this Study.

P. Omindo et al. / Characterization of Pozzolanic Geomaterial

443

Figure 2 Durability Characteristics of OPMC-CC Stabilized Pozzolanic Materials

4. Pavement Structural Analysis computed The pavement structural analysis was carried out based on the resulting from Eq. (2), which defines the elastic stiffness of the composite pavement structure, derived from one of the modules of the GECPRO model [4] and applied in section 5. (2) where, all thicknesses are expressed in cm and, = Thickness of Asphalt Concrete, = Thickness of Asphalt Treated Base Course, = Thickness of Crushed Aggregate Base Course, = Thickness of Cement Treated Base Course, = Thickness of Granular Subbase, = Thickness of Existing LTC Subbase, = (100- )=Thickness of Subgrade. = Thickness of Composite Pavement.

5. Comparative Analysis Figures 3(a) and (b) make a comparison of the original and reviewed design. It can be observed that, due to the consideration of the intrinsic nature of the existing pozzolana geomaterial and the OPMC method of stabilization, the reviewed (proposed) design exhibits superior geotechnical engineering properties in comparison to the original one.

444

P. Omindo et al. / Characterization of Pozzolanic Geomaterial

Figure 3 Schematic Cross-section of varying Layers of (a) ORIGINAL (Existing) Design and, (b) Proposed Design Option

As a result, not only did the Reviewed Design realize enormous cost and time savings, but it also enhanced the structural capacity (bearing capacity, strength, serviceability) and deformation resistance of the pavement structure as demonstrated in Figure 4. The reviewed design was implemented within a very short period (approximately 65% less construction time), whilst a substantial cost savings of approximately 30% was realized in comparison to the Original Design.

6. Analysis of Time Dependent Structural Soundness The time dependent structural depreciation was analyzed using the Structural Capacity and Deformation Resistance (SCDR) model. The comparative results for various options are presented in Figure 4. It can be noted that the proposed option 1, which represents the reviewed design, exhibits the highest resistance to structural capacity deterioration and does not approach the critical zone during the design life. Consequently, the Maintenance Requirement Ratio (MRR) was computed to be 0.2 in comparison to 1.29 of the original design.

Figure 4 Graphical depiction of depreciated structural capacity in terms of Result.

for Varying Options

P. Omindo et al. / Characterization of Pozzolanic Geomaterial

445

7. Conclusions This Study has shown that naturally existing pozzolanic geomaterials exhibit extremely high geotechnical engineering properties. Based on the results reported in this paper, it can be concluded that the structural capacity, strength and deformation resistance properties are significantly enhanced when this material is subjected to OPMC-CC stabilization and Long-Term Consolidation. In this case, only minimal addition of factory produced Pozzolanic Portland Cement (PPC) was found to suffice.

References [1] Mukabi J.N (2013). Innovative Laboratory and In-situ Methods of Testing in Geotechnical Engineering, to be published in the Proceedings of the International Conference of the Institute of Engineers Kenya. [2] Kensetsu Kaihatsu Limited (2010). Songwe Airport Pavement Design Review Engineering Report No. SAT09T1. Submitted to Tanzania Airports Authority, The United Republic of Tanzania. [3] Mukabi J.N., Kotheki S. (2010a). Mathematical Derivative of the Modified Critical State Theory and its Application in Soil Mechanics. Proceedings of the 2nd International Conference on Applied Physics & Mathematics, 2010 IACSIT, Kuala Lumpur, Malaysia. [4] Mukabi J.N., (2011). Derived Empirical Relations and Models of Vital Geotechnical Engineering Parameters Based on Geophysical and Mechanical Methods of Testing, to be published in Proc. of 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering, 2011 ARCSMGE, Maputo.

446 Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-446

Correlation between the dynamic cone penetration index and the Falling Weight Deflectometer-determined subgrade resilient modulus a

Samuel I.K. Ampadu a,1 and Emmanuel Klu Okang b Kwame Nkrumah University of Science and Technology, Kumasi, Ghana b Ghana Highway Authority, Accra Ghana

Abstract. Modern pavement design characterizes the subgrade soil in terms of the resilient modulus. This parameter is determined in a non-destructive in-situ method using the falling weight deflectometer (FWD) and in the laboratory from the repeated load triaxial test. These methods however are both expensive and time consuming and are not readily available in most highway departments in developing countries. On the other hand, the simple and inexpensive dynamic cone penetrometer has been extensively used for pavement in-situ subgrade characterization. This paper reports on an attempt at a correlation between the results of the Dynamic Cone Penetrometer test and the output of the Falling Weight Deflectometer (FWD) test for purposes of estimating the subgrade resilient modulus. Fifty-two FWD deflections and Dynamic Cone Penetrometer field tests were conducted on sections of an urban arterial road in Accra, with varying terrain. Soil samples were also recovered from trial pits sunk at selected locations and subjected to standard laboratory tests for purposes of identification and classification. The results of the output of the FWD sensors were analyzed and correlated with the DCP penetration index. The results suggest that the DCPI may be used to predict the subgrade resilient modulus. Keywords. subgrade resilient modulus, Falling weight deflectometer, dynamic cone penetrometer, correlation

1. Introduction The resilient modulus is an important parameter used to characterize the subgrade soil for the design and rehabilitation of roads. Currently, it may be obtained directly by either performing the repeated triaxial test on undisturbed subgrade soils in the laboratory or from the Falling Weight Deflection (FWD) test directly on the design road. Being a field test, the FWD is preferred since it better simulates the dynamic loading conditions due to traffic loading as observed in the field. However, the use of FWD to evaluate pavements is limited to Road Agencies that can afford the cost of acquiring and operating the equipment. The Dynamic Cone Penetration (DCP) test however, is cheap to own and simple to operate but it is increasingly becoming an 1

Corresponding Author.

S.I.K. Ampadu and E.K. Okang / DCP Index and FWD-Determined Subgrade Resilient Modulus

447

important instrument for characterizing subgrade soils by road engineers for pavement design and rehabilitation. This is because reliable correlations between the DCP and the California Bearing Ratio (CBR) have been developed for subgrade soils [1][2][3]. There have been various studies to extend the use of the DCP into compaction verification for example, [4] and into estimation of bearing capacity [5][6]. There is however no generally accepted correlation between the DCP penetration and the Modulus. There have been some attempts [7][8][9] at using the DCP to estimate the modulus through conversion of the CBR to the modulus. This paper presents an attempt to develop a correlation for estimating the FWD determined subgrade resilient modulus from the DCP penetration index on subgrade soils underneath lateritic base and subbase layers.

2. Methodology 2.1. Equipment Description The FWD used in this study is the trailer mounted Dynatest 8000E shown in Fig. 1 and consists of the loading, the deflection measuring and of the processing systems. The loading system consists of a 155kN drop weight mounted on a vertical shaft, hydraulically controlled to fall freely through a distance of 510mm to strike a 300mm diameter loading plate resting on a 5.6mm thick rubber buffer. Embedded in the loading plate is a load cell that measures the load applied during the test. The deflection measuring system consists of seven geophones arranged on a raise/lower bar at predetermined distances of 0, 305, 457, 610, 914, 1219, and 1524mm away from the centre of the loading plate. The signals from the geophones are fed into the Dynatest 8600 System Processing unit which controls the FWD operation, performs scanning, conditioning and further processing of the geophone signals. The FWD load application is remotely controlled by the operator in the tow vehicle. The essential features of the DCP equipment used in this study have been described in [5] and it consists of an 8kg metal hammer falling through a vertical distance of 575mm to strike a metal anvil to drive a 20mm diameter, 60 o cone at the end of a 16mm rod into the formation. The penetration is measured using a scale along the rod

Fig. 1 Equipment required for FWD test.

448

S.I.K. Ampadu and E.K. Okang / DCP Index and FWD-Determined Subgrade Resilient Modulus

2.2. Test Site The study road is the Pantang-Abokobi urban arterial road branching off N4 at Pantang just outside of Accra. The road is 5.3km long and lies in a rolling terrain. It carries a daily vehicular traffic of about 200 private cars, mini buses and occasional trucks. The study road was divided into three sections according to the terrain as shown in Fig 2. Whereas sections 1 and 3 are in a valley, section 2 is on higher ground. The test line was the middle of the Abokobi-bound lane. White paint was used to mark out the 52 investigation points at 100m intervals along the test line.

Fig. 2 Layout of Study sections on Study Road

2.3. Test Procedure The FWD equipment was towed into position with the loading system over the investigation point. Traffic in both lanes was then stopped some distance from the test location in order to eliminate interfering vibration from passing traffic. The geophone bar was then lowered into position such that the loading plate is over the test point mark and the geophones were in contact with the road surface. By command from the control computer, the test load was lifted hydraulically to a height of 510mm and allowed to fall freely unto the rubber bumpers. This transmitted an approximately half sinusoidal wave with a loading time of 25 microseconds into the pavement. The generated pulses were picked up by the geophones and transmitted to the computer for storage. The test was repeated at each test point. After the test, the lower/riser bar was lifted, locked into position and towed into position at the next test point. After all the 52 points have been tested, the DCP test was also performed at the same test locations. The readings were recorded after predetermined number of blows of the hammer in accordance with [11]. In order to determine the pavement structure, five trial pits were excavated approximately 900m apart along the project road at the locations shown in Fig. 1 and to depths of between 0.30m to 0.82m depending on the location. The profiles exposed on the sides of the excavation were logged and disturbed and undisturbed samples of the subgrade material were collected for further testing in the laboratory. In the laboratory, the samples were subjected to water content determination, Atterberg limit tests, grading analysis and density measurement.

S.I.K. Ampadu and E.K. Okang / DCP Index and FWD-Determined Subgrade Resilient Modulus

449

3. Discussion of Results 3.1. Subgrade Material Characteristics The logs of the five trial pits show that the study road has a double surface dressing 25mm thick. In sections 1 and 3, the pavement consists of a reddish brown, medium dense lateritic gravel sub-base 0.15-0.20m thick lying on top of a subgrade of grey stiff silty to clayey sand. In section 2, the pavement is a dense brown lateritic gravel base 0.2m thick on top of a medium dense reddish brown lateritic gravel subbase also 0.2m thick overlying the sandy gravel subgrade. The grading curves of the subgrade material are shown in Fig. 3 while a summary of the index properties of the subgrade material including the in-situ dry density (γdry) values is shown in Table 1. It can be seen that the subgrade in sections 1 and 3 being in a valley have higher natural water contents (wn), higher liquid limit and Plasticity Index values. Effectively therefore, sections 1 and 3 may be modeled as similar pavement subgrade.

Fig. 3 Grading of subgrade material Table 1 Summary of Subgrade material properties Clay Content (%)

Liquid

Plasticity

wn

γdry

AASHTO

Limit

Index

(%)

Mg/m3

Class

Test Section

Trial Pit Location

Section 1

0+600

8.7

32.0

19.0

4.3

1.936

A-6

Section 2

1+500

8.7

24.3

11.8

3.5

1.834

A-2-7

(Hill)

2+400

8.2

23.1

11.0

3.6

1.833

A-2-7

Section 3

3+600

17.7

43.9

20.6

5.5

1.836

A-6

(Valley)

4+500

21.2

32.0

15.2

4.8

1.995

A-6

3.2. Analysis of FWD Results Table 2 is a typical output of the FWD sensors. At each FWD location two load drop tests were conducted on the road pavement. The first drop test was the seating blow.

450

S.I.K. Ampadu and E.K. Okang / DCP Index and FWD-Determined Subgrade Resilient Modulus

Table 2 Output of FWD sensor measurements dĂƐƚ ŚĂŝŶĂŐĞ ϬнϬϬ ϬнϬϬϬ ϬнϭϬϬ ϬнϭϬϬ ϬнϮϬϬ ϬнϮϬϬ

ƌŽƉ ^ƚƌĞƐƐ͕σϬ ŶĞƌŐǇ͕ EŽ ;ŬEŵͿ ;DWĂͿ ϭ Ϯ ϯ ϰ ϱ ϲ

ϭϬϴϰ ϭϬϴϴ ϭϬϵϬ ϭϬϵϬ ϭϬϳϵ ϭϬϴϴ

ϳϲ͘ϲϮ ϳϲ͘ϵϭ ϳϳ͘Ϭϯ ϳϳ͘Ϭϱ ϳϲ͘Ϯϯ ϳϲ͘ϴϵ

Ͳϯ

'ĞŽƉŚŽŶĞƐĞŶƐŽƌĚĞĨůĞĐƚŝŽŶǆϭϬ ŵŵ Ěϯ Ěϰ Ěϱ Ěϲ

Ěϭ

ĚϮ

ϲϭϯ͘ϭϬ ϲϭϯ͘ϬϬ ϵϭϰ͘ϳϬ ϴϴϴ͘ϭϬ ϳϮϯ͘ϵϬ ϳϭϳ͘ϯϬ

ϰϲϳ͘ϰϬ ϰϳϮ͘ϱϬ ϲϮϬ͘ϬϬ ϲϭϮ͘ϳϬ ϰϴϮ͘ϱϬ ϰϳϮ͘ϴϬ

ϯϬϱ͘ϵϬ ϯϬϰ͘ϳϬ ϰϮϰ͘ϯϬ ϰϮϮ͘ϰϬ Ϯϵϵ͘ϲϬ Ϯϵϰ͘ϭϬ

ϮϬϭ͘ϲϬ ϮϬϭ͘ϳϬ Ϯϲϰ͘ϭϬ Ϯϲϱ͘ϮϬ ϭϳϰ͘ϲϬ ϭϳϭ͘ϴϬ

ϭϱϮ͘ϭϬ ϭϱϬ͘ϵϬ ϭϳϯ͘ϲϬ ϭϳϱ͘ϲϬ ϭϮϬ͘ϯϬ ϭϭϵ͘ϭϬ

Ěϳ

ϴϴ͘ϴϬ ϴϵ͘ϯϬ ϴϬ͘ϯϬ ϴϬ͘ϰϬ ϳϭ͘ϴϬ ϳϬ͘ϯϬ

ϱϯ͘ϳϬ ϱϯ͘ϬϬ ϯϬ͘ϴϬ ϯϬ͘ϲϬ ϯϯ͘ϭϬ ϯϮ͘ϲϬ

For the analysis, the pavement is modeled as a two-layered system consisting of the base and/or sub-base modeled as a single layer of thickness 150mm, 400mm and 200mm for sections 1, 2 and 3 respectively on top of the subgrade. The surface dressing is neglected. The ELMOD5 program was used for the back-calculation of the layer moduli from the FWD deflections. The program uses the method of equivalent thicknesses for the back-calculation. This method is based on Boussinesq’s solution for a semi-infinite half space, combined with Odemark’s transformation of a two layered system to a semi-infinite space [10]. The solution has been further modified to accommodate stress dependent non-linearity. A typical output of ELMOD5 is shown in Table 3 and it includes the non-linear parameters. In this investigation only the resilient modulus for the subgrade MR2 is discussed. The variation of MR along the study road is shown in Fig. 4. Table 3 Typical output of ELMOD5 Layer Thickness (mm)

Resilient Modulus (MPa)

Non-Linear Properties

H1

MR1

MR2

C0

n

0+00

150

1906

141

164

-0.151

0+100

150

1619

69

91

-0.352

0+200

150

1454

105

144

-0.00

Test Chainage

Fig. 4 Variation of Subgrade Resilient Modulus with location along study road

3.3. Analysis of DCP Results The DCP test data was also modeled as a two-layered system as was used in the FWD analysis and the Penetration Indices DCPI1 and DCPI2 for the first and second layers respectively were determined by linear regression analysis of the data for the layer. No correction for confinement was applied to the DCP data.

S.I.K. Ampadu and E.K. Okang / DCP Index and FWD-Determined Subgrade Resilient Modulus

451

3.4. DCP and FWD Correlations for the Subgrade The MR values for the subgrade were plotted against the equivalent DCPI values for sections 1 and 3 combined and for section 2 separately in Fig. 6. The resulting loglog equations and the regression statistics are also shown in the figure. The data from all 3 sections was also modeled together and the resulting model is shown in Equation (1) with a coefficient of regression, r of -0.4678. The regression statistics show relatively lower coefficient of correlation values. This may be due to the relatively smaller number of data points involved in this study and the variation especially in the DCP test data. Despite this, the results though preliminary suggest strongly that a relationship as shown in Equation (1) exists for local subgrade soils. The overall model obtained for the subgrade in this study is compared with those of [8] and [9] in Fig. 6. The comparison shows that this study gives lower values that both models.

Log(MR ) = 2.5874 − 0.7956 Log(DCPI2 )

Equation (1)

Fig. 5 Plot of subgrade resilient modulus versus DCP penetration index, DCPI2

Fig. 6 Comparison of Correlation Models

452

S.I.K. Ampadu and E.K. Okang / DCP Index and FWD-Determined Subgrade Resilient Modulus

4. Conclusions The results of this study though preliminary strongly suggest that there exists a correlation between the FWD subgrade resilient modulus and DCP penetration index, DCPI, in the form of logarithmic linear relation for the subgrade soils beneath lateritic sub-base and base layers.

References [1] Kleyn, E.G. and van Heerden H.J.J., 1983, "Using DCP soundings to optimize pavement rehabilitation," Annual Transportation Convention, Transvaal Roads Department, Johannesburg [2] Transport Road Research Laboratory, 1990 A user’s manual for a programme to analyze dynamic cone penetrometer data, Overseas Road Note 8, Crawthorne. [3] Webster, S.L, Grau R.H. and Williams R.P. 1992 Description and application of dual mass dynamic cone penetrometer, US Army Engineer Waterways Experimental Station, Instruction Report No GL 92-3 [4] Ampadu, S.I.K. and Arthur, D.T. (2006): “The Dynamic Cone Penetrometer in Compaction Verification on a Model Road Pavement”. Geotechnical Testing Journal, Volume 29, No. 1, GTJ 12306, pp 70-79, January 2006 [5] Ampadu, S.I.K. (2005), “A correlation between the Dynamic Cone Penetrometer and the bearing capacity of a local soil formation”. Proceedings of the 16th ICSMGE, September 12-16th Osaka, Japan. Millpress Science Publishers, Rotterdam, Netherlands [6] Ampadu, S.I.K. and Ditze-Awuku, D., (2009), Model tests for bearing capacity in a lateritic soil and implications for the use of the dynamic cone penetrometer, 17th ICSMGE, Alexandria, Egypt, October 5-9th. Hamza (Eds) pp 1095-1099 [7] Heukelom, W. & Klomp, A.J.G. (1962). “Dynamic Testing as a Means of Controlling Pavements During and After Construction”. Proceedings of the 1st International Conference on the Structural Design of Asphalt Pavements (pp. 667-685). [8] Powell, WD Potter JF, Mayhew HC and Nunn ME 1984, The structural design of bituminous roads TRRL Report LR 1132, 62pp. [9] Chen D Lin D Liau P. Bilyeu J 2005, A correlation between dynamic cone penetrometer values and pavement layer moduli, Geotechnical Testing Journal, Vol 28, Issue 1 pp8 [10] Ullidtz P. Pavement analysis, Elsevier Science, Amsterdam 1987 [11] American Society of Testing Materials (2003). “Standard Test Method for Use of the Dynamic Cone Penetrometer in Shallow Pavement Applications” ASTM D 6951-03, ASTM International, West Conshohocken, PA.

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 453 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-453

Fundamental Theory of the ReCap Technique and its’ Application in the Construction of Pavement Structures within Problematic Soils John N. MUKABIa,1 Bernard NJOROGEb,Tilahun ZELALEM c, Samuel KOGId, Maurice NDEDAe and David KAMAUf a Kensetsu Kaihatsu Consultants, Nairobi, Kenya b College of Architecture and Engineering, University of Nairobi, Nairobi, Kenya c NDC Consulting Co. Ltd, Addis Ababa, Ethiopia d&e Materials Testing and Research Department, Ministry of Roads, Nairobi, Kenya f Kenya Airports Authority, Nairobi, Kenya Abstract. ȋ ȌǤ ǡ Ǧ Ǥ ǡ λ ͷ Ȁ Ǥ Ǥ ǡ ǡ Ǥ ȋȌ ǡ ǡ ǡ ȋȌ ǡ ǡ ǦǤ ȋȌ ǡǡǦǤ Keywords. pavement, theory, ReCap, technique, problematic, geotechnical

Introduction The design and construction of the 240km Addis Ababa ~ Debre Markos International Trunk road was predominantly based on research oriented techniques. This road, traversing mountainous topography through the 1200m deep Abay Gorge crossing the Blue Nile, provides the only access to the Sudan and Eritrea. It is the most vital link for oil importation and trade in various products for the Government and people of 1

John MUKABI: CEO, Kensetsu Kaihatsu Limited, A6 Mac Apts., Lavington, P. O. Box 3524600200, Nairobi, KENYA; E-mail: [email protected]

454

J.N. Mukabi et al. / Fundamental Theory of the ReCap Technique and Its’ Application

Ethiopia. The project was grant aid funded by the Government of Japan through its’ Implementing Agency, the Japan International Cooperation Agency (JICA). Almost 80% of the total stretch is underlain by black cotton soil, which is a highly problematic geomaterial [1]. As a result, quick impact Value Engineering solutions giving careful consideration to the environmental and financial constraints were to be sought. In order to provide the appropriate and suitable long-term solutions as well as appreciably sustainable countermeasures, comprehensive geotechnical investigations, studies and research were undertaken. Consequently, the Optimum Mechanical and Chemical (OPMC) stabilization method, which has been reported [2], the ReCap technique and suction-stress methods developed whilst undertaking research oriented design were implemented over most of the sections along this road culminating in enormous cost savings of 30 ~ 46% for the composite pavement structure and 60 ~ 70% in earthworks. Consequently, significant time savings were also realized. The fundamental concepts of most of the solutions reported in this Study were developed during this research.

1. Innovative Testing Techniques and Geotechnical Investigation Regimes Details of the innovative testing and geotechnical investigation regimes are discussed in the Comprehensive Engineering Report on Proposed Countermeasures to Slope, Embankment and Pavement Structure Failure, Volume 1 of March, 2004 and Innovative Laboratory and In-situ Methods of Testing in Geotechnical Engineering [3].

2. Basic Concepts for Developing Fundamental Theory 2.1. Generalized Conceptual Equations In order to develop a guiding theory and geotechnical engineering concept that would culminate in the development of the ReCap technique, it was considered vital to model pavement structures of various layer quality and configurations as well as structural thicknesses which would be constructed on and within problematic soil subgrades. Full-scale field investigations and comprehensive laboratory testing were innovatively designed and implemented [3]. The definitive generalized equation delineating and simulating the road conditions under this conceptual framework is expressed as a function of loading conditions, pavement type (structurally), pavement layer quality, structural thickness as well as intrinsic material properties as depicted in Eq. (1).

[

v Rc = f φ df , ∂t i , Pc , Pe , Δte, Δα ms

]

(1)

where, R c represents road condition, φdf is the dynamic load factor, ∂ t i defines the response mode factor of layer of the pavement structure, Pc is the pavement configuration, Pe is the pavement layer quality, Δt e is effective structural thickness and v = parameter delineating moisture ~ suction variation. Δα ms

J.N. Mukabi et al. / Fundamental Theory of the ReCap Technique and Its’ Application

455

On the other hand, the extent of deformation as a result of problematic subgrade can be derived by carrying out back analysis of the deformation history of an existing pavement structure supported by soils of a similar nature. In a generalized state, this can be expressed as shown in Eq. (2).

[

ε dh = f φ ' ,ψ ' , p'ocf , q ocf , φ 'ocf , Σf yi , δ ijo

]

(2)

where, ε dh represents the parameter delineating deformation history, φ ' is the consolidation stress ratio, ψ ' is the modifier between Isotropic and Anisotropic stress paths, p ' ocf , q ocf are the invariant stresses under over consolidation conditions, φ the Angle of Internal Friction within the failure zone.

, f

is

2.2. Concepts Applied for Analyzing Impact of Environmental Factors Most tropical problematic soils are known to be highly sensitive and susceptible to changes in environmental factors. Development of methods that can quantify the magnitude of the impact of such variations on the performance of these soils is therefore of great essence. In developing the theory of the ReCap Technique, some of the main representative equations, which were developed within this research regime, are presented. •

Effect of swelling

Δ sc = ϑ sc ln λ sc + Β sc (%)

(3)

where, Δ sc represents swell in relation to surcharge pressure, ϑ sc = 12.9; logarithmic gradient constant for tropical geo-materials , λsc is the surcharge pressure in Kpa, BSC =36.5; logarithmic intercept for most fine grained tropical problematic geo-materials. •

Seasonal effects on bearing capacity and resilient modulus The effects of seasonal changes on the bearing capacity and resilient modulus of some tropical problematic geo-materials is presented in Eqs. (4) and (5).

Ψ wdr = ι gl ln CBR w + ι gi

(4)

where, Ψwdr represents the wet to dry season bearing capacity ratio,

ιgl = 0.0022; logarithmic CBR gradient constant for tropical problematic soils, ιgi = 0.54; logarithmic CBR intercept constant for tropical problematic soils

ζ wMr = α gl ln M r + α gi where, ζ wMr represents the wet to dry Season resilient modulus (Mr ) ratio, α gl

(5) =

0.0022 and α gi = 0.54 are logarithmic Mr intercept constants for tropical geomaterials. •

Effect of moisture ~ suction variation

456

J.N. Mukabi et al. / Fundamental Theory of the ReCap Technique and Its’ Application

The impact of the moisture ~ suction variation is analysed by considering the correlation between the change in moisture content as a function of the plasticity ratio as demonstrated in Eqs. (6) and (7) proposed by Mukabi et al. [4] for the resilient modulus, .

= [− × Δ

+

]×

×

(6)

where, Awc=0.092 and Bwc=1 are resilient modulus related gradient and intercept constants, Δwc is the change in moisture content expressed as a percentage, PIR=16 is the Reference Plasticity Index, while PIm is that determined for the tested material and is the initial resilient modulus. The initial resilient modulus can be determined from the modified empirical equation proposed as:

[

= ×

]

͵

× ͳͲ

−͸

[

−

]

ʹ

[

+

]−

(7)

where, AMr=2, BMr=0.0012, CMr=0.623 and DMr=0.775 are average constants determined for tropical geomaterials and =10.3×CBR, is an equation proposed by the American Asphalt Institute. • Evaluation of Variation in Quality of Pavement Layer Materials

ǀŝĚĞŶĐĞŽĨůĂĐŬŽƚƚŽŶ ^Žŝů /ŶƚƌƵƐŝŽŶŝŶƚŽhƉƉĞƌ WĂǀĞŵĞŶƚ>ĂǇĞƌƐƵĞƚŽ >ŽŶŐͲdĞƌŵǇŶĂŵŝĐdƌĂĨĨŝĐ >ŽĂĚŝŶŐ

Figure 1 Impact of Inferior Black Cotton Soil Intrusion into SuperiorUpper Pavement Layer Materials

Depending on the nature of the subgrade, topography of environment and seasonal changes, intrusion of native subgrade material into overlying layers of the pavement

J.N. Mukabi et al. / Fundamental Theory of the ReCap Technique and Its’ Application

457

structure can be rampant and extremely detrimental. Intrusion of problematic soils has even more detrimental effects as can be derived from Figure 1. The consequences of such a physical action are the deterioration in bearing capacity, cohesion intercept (c), internal friction (φ), mechanical stability, strength and deformation resistance. The quantitative assessment of deficiency in the physical properties of pavement materials with time through the intrusion of fines to upper pavement layers is carried out by employing the following equation, which defines cumulative intrusion. 3 2 (8) CBR ult = Aγ − Bγ + Cγ + CBR init

∫

∫

BCi

where, CBR A γ = 0.057,

∫

BCi

∫

BCi

is the ultimate CBR, CBR init . is the initial CBR = 2.8, and C γ = 15, are material constants and,

ult .

Bγ

= Problematic soil cumulative intrusion content expressed in (%).

3. The ReCap Mathematical Relations The Replacement and Capping (ReCap) technique quantitatively determines the optimum quantity of problematic soil to be replaced depending on its nature, properties of the capping geomaterial, surcharge pressure and environmental conditions. In determining the necessary thickness tCL of the Capping Layer to replace the problematic soil, the following equations were derived from the tests, theories and concepts briefly introduced in the preceding sections and reported in other publications.

{

}

t CL = T P − t bp xS

SP

(9)

The total pavement thickness TP is expressed as: T P = t Pb xR

f

+tv

(10)

The coefficient of subgrade structural performance S SP is computed from the relation between design CBR (CBRd) and the ae parameter defined in Eq. (13).

[

S SP = e 1 / CBR

d

]

/ α e 0 .5

(11) b

On the other hand, the basic pavement thickness tP from Equation (9) is computed from the following equation.

[

]

= − ( ) + ( ) [ Ȁ ] ʹ

(12)

where, the resilience factor R f = [2Ri (Rt − Ri )]0.25 : Ri is the initial resilience factor and Rt is the terminal resilience factor; tV in Eq. (10) is the positive value of the specified tolerance for pavement thickness, N is the No. of ESA, AP=219, BP=211, CP=58 and DP=120 are material related constants. The parameter αe in Equation (11) is defined as:

α e = Aee(Be −CeVe )

LL M cn

(13) where Ae=0.23, Be=0.54, Ce=0.08 are constants and Ve=Annual Average Evapotranspiration in m/year, LL=Liquid Limit in percentage and Mcn=natural moisture

458

J.N. Mukabi et al. / Fundamental Theory of the ReCap Technique and Its’ Application

content of the problematic subgrade material expressed in percentage form. All thicknesses are calculated in mm. The equation for determining the required capping layer thickness Tcl in mm, is then derived as follows:

= ×

− ×

(14)

where Acl=2253 and Bcl=0.455 are capping layer material constants, while CBRsg=subgrade bearing capacity measured from the California Bearing Ratio (CBR).

4. Example of Mode of Specification and Nomographs An example of the mode of stipulating specifications derived from the ReCap technique is summarized in Table 1, whilst Figure 2 is a depiction of some of the Nomographs developed for this method. Table 1. Method of determining Required Thickness in (cm) for Different Subgrade Bearing Capacity

W/хϰϱ ϯϱфW/фϰϱ W/фϯϱ

ϰфZфϳ

Zсϰ

Zсϯ

ZсϮ

Zсϭ

^ǁĞůů;йͿ^ŵ

WůĂƐƚŝĐŝƚǇ/ŶĚĞǆ

ŽĚŝŶŐKƉƚŝŽŶ

WůĂƐƚŝĐŝƚǇ ĂŶĚ ZĞƋƵŝƌĞĚ dŚŝĐŬŶĞƐƐ ĨŽƌ ^ǁĞůůŽŶĚŝƚŝŽŶ ŝĨĨĞƌĞŶƚ ^ƵďŐƌĂĚĞ ĞĂƌŝŶŐĂƉĂĐŝƚǇ

^ŵхϭϬ ϭϰϬ ϵϬ ϳϬ ϲϬ ϯϬ ^ŵфϭϬ ϭϭϬ ϳϱ ϲϬ ϱϬ ϮϬ ^ŵфϱ ϳϬ ϲϱ ϱϱ ϱϬ ϮϬ

Figure 2 Relation between Optimum Capping Layer Thickness and Design CBR for Varying Subgrade Classes Derived from the ReCap Technique

J.N. Mukabi et al. / Fundamental Theory of the ReCap Technique and Its’ Application

459

5. Case Examples of Application The ReCap Technique has been widely applied in the East and Central African Region in countries such as Ethiopia, South Sudan, Burundi, Tanzania and Kenya. Case Examples of its’ application have been reported in various publications [2][4]. Figure 3 shows one of the typical cross-sections of the Addis Ababa ~ Debre Markos International Trunk Road where both the ReCap and the OPMC stabilization methods were applied. In this case it had been specified, in the original design, that 1400mm of the black cotton soil subgrade class S1 (refer to legend in Figure 2) be replaced by a capping layer constituent of material with CBR>15 and PI<20. Through the application of the ReCap technique it was determined that a replacement thickness of 450mm would suffice. As a consequence, a reduction of approximately 70% (approximately 1.2 million cubic metres) of the capping layer volume was realized.

ϰϱϬŵŵ ůĂĐŬ ŽƚƚŽŶ ^Žŝů>ĂǇĞƌZĞƉůĂĐĞĚ Figure 3 Application of the ReCap Technique in Determining the Optimum Capping Layer Thickness

6. Conclusions Introduction of the fundamental theory of the ReCap technique and its’ application has been made in this paper. It has been demonstrated that this technique is not only essential in determining the optimum quantities of replacing problematic and/or unsuitable subgrade material, but is also an effective means of realizing significant cost and time savings in comparison to the conventional approach where the replacement depth is usually decided arbitrarily and based on non-scientific considerations.

References [1] Gono, K., Mukabi, J.N., Koishikawa, K., Hatekayama, R., Feleke G., Demoze W., Zelalem A., (2003). Characterization of Some Engineering Aspects of Black Cotton Soils as Pavement Foundation Materials, Procs. of the International Civil Eng. Conf. on Sustainable development in the 21st Century. [2] Mukabi J.N (2007). Unique Methods of Enhancing Engineering Properties of Geomaterials for Slopes, Embankments and Pavement Structures. Proceedings of the 23rd World Road Congress. Paris, France. [3] Mukabi J.N (2013). Innovative Laboratory and In-situ Methods of Testing in Geotechnical Engineering, to be published in the International conference of the Institute of Engineers of Kenya. Nairobi, Kenya. [4] Mukabi J.N., Kotheki S., Ngigi A., Gono K., Njoroge B.N., Murunga P.A., Sidai V. (2010) – Characterization of Black Cotton Soil under static and dynamic loading. Published in Proceedings of the International Conference on Geotechnical Engineering (ICGE), Tunis, Tunisia.

460 Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-460

Utilisation des Bétons Compactés au Rouleau (BCR) a

I.K CISSE. a, and A. SALL b Ecole Polytechnique, Thiès, Sénégal b SONATEL, Dakar, Sénégal

Abstract. The construction of flexible pavements is very expensive for countries like Senegal. Moreover these counries aren’t producing petroleum which is a raw material for asphalt used in coating. So, it is necessary to find a method of construction that combine fiability, durability and economy. One of these methods is Roller Compacted Concrete or RCC. The modeling of the RCC structure was done with Ecoroute program that allows for a check of allowable stress and strain on all pavement layers. The application of RCC in the stretch Diamniadio-Mbour has shown interesting results. The crushing test made in laboratory yield on the average, 33 MPa as compressive strength and 4 MPa as tensile strength, at 28 days. On the other hand, for a pavement with a thickness of 30 cm, using RCC, 47 percent of economy was achieved relatively to flexible pavement. Keywords: RCC, pavement, coating, Ecoroute, economy.

Introduction Au Sénégal, la tradition a toujours voulu que les routes soient en chaussées souples ; or celles-ci présentent certes des avantages mais, nécessitent un entretien plus coûteux pendant leur durée de service relativement court. De plus, le bitume, sous- produit du pétrole dont le Sénégal n’est pas producteur est donc importé, ce qui constitue un grand manque à gagner du point de vue économique. C’est dans ce cadre que rentre cette étude du Béton Compacté au Rouleau (BCR) comme revêtement des chaussées. Après avoir décrit le BCR, nous procéderons à sa formulation à l’aide d’une des différentes méthodes (théoriques, empiriques ou semi-empiriques) qui existent. Ensuite, tenant compte des critères de dimensionnement d’une part et des caractéristiques des matériaux d’autre part, nous déterminerons l’épaisseur de la chaussée sur un tronçon choisi sur lequel une étude économique comparative entre les chaussées en BCR et les chaussées en revêtement hydrocarboné sera faite.

1. Généralités sur le BCR Du point de vue formulation, le BCR est un béton sec qui nécessite l’apport d’une énergie de compactage pour être bien consolidé. Du point de vue structural, un revêtement en BCR est un ouvrage rigide au même titre que toute autre dalle de béton, et il est soumis aux mêmes critères de conception. Cependant, le BCR comporte

I.K. Cisse and A. Sall / Utilisation des Bétons Compactés au Rouleau (BCR)

461

plusieurs avantages (durabilité, économie) par rapport d’une part aux autres variantes de chaussées rigides et d’autre part aux chaussées souples généralement utilisées. Il a été prouvé qu’un BCR dosé à 300 kg/m3 de ciment avec un rapport E/C d’environ 0,35 peut développer une résistance en compression à 28 jours de 40 MPa et une résistance en flexion à 3 jours de 5 MPa [1].

2. Caractéristiques d’un BCR Le ciment Le liant est de type hydraulique et peut être du ciment Portland de type CPA-CEM I ou CPJ-CEM II dont la classe varie en fonction des performances voulues. Le dosage classique d’un mélange de BCR est de 200 à 350 kg/m3, soit une teneur en liant de 10 à 15 % de la masse totale des constituants secs. Les granulats Les granulats sont les mêmes que ceux des bétons classiques (roulés ou concassés). La dimension maximale du granulat D doit être inférieure à 20mm [3]. Le fuseau granulométrique doit être divisé en plusieurs fractions (exemple : 0/3-3/8-8/16). Toutes les fractions granulométriques doivent avoir un indice de concassage supérieur à 30 % dans le cas d’un trafic trop faible et environ 100% dans le cas d’un trafic fort. L’eau Le BCR étant un béton sec à affaissement nul, sa teneur en eau doit donc être faible. Elle est de l’ordre de 4 à 6 %, fixé par le laboratoire lors de l’essai Proctor modifié. Quant à la teneur en eau au chantier, elle doit tenir compte des conditions atmosphériques et de transport. Les adjuvants Les adjuvants sont utilisés dans le même but que pour les bétons classiques. Toutefois, dans le cas du BCR, la durée courte de malaxage et la faible quantité d’eau font qu’il y a une atténuation de l’effet des adjuvants, d’où une augmentation du dosage en adjuvant par rapport au béton plastique afin d’accroître leur efficacité.

3. Principales Propriétés d’un BCR 3.1 Maniabilité La qualité du BCR est étroitement liée à sa maniabilité; elle ne doit être ni trop faible, ni trop élevée [3]. Contrairement à l’affaissement au cône d’Abrams du béton ordinaire, la maniabilité du BCR est déterminée à l’aide de l’appareil Vebe [2] du fait de sa consistance trop sèche. Elle est exprimée en temps Vebe et la plage optimale est de 40 à 90 secondes. 3.2 Résistance à la compression La résistance à la compression fait partie des qualités les plus représentatives du BCR. En effet, la capacité du BCR à supporter des charges lourdes concentrées résulte de sa forte résistance en compression. Un BCR bien formulé peut avoir une résistance à la compression à 28 jours variant de 40 à 60 MPa ; pour cela, il faudrait une bonne optimisation du squelette granulaire [2].

462

I.K. Cisse and A. Sall / Utilisation des Bétons Compactés au Rouleau (BCR)

4. Méthodes de Dosage Utilisées Différentes méthodes (théoriques, empiriques et semi-empiriques) ont été développées [3] et [5]. Un exemple de méthode semi-empirique est celle du volume de pâte optimal développée par Richard GAGNE [5]. Cette méthode se base sur l’hypothèse que la quantité de pâte contenue dans le mélange de BCR est celle nécessaire pour combler les vides inter-granulaires subsistant après un bon compactage. Nous utiliserons donc la méthode de Richard GAGNE utilisée assez fréquemment dans les projets de BCR. 4.1 Formulation du BCR 4.1.1 Caractéristiques des constituants Pour le ciment utilisé, il est de classe 32,5 avec une masse spécifique de 3,15 g/cm 3; quant aux granulats, on a utilisé du sable de dune roulé 0/1 et des graviers concassés de basalte de classes 3/8 et 8 /16 de la région de Thiès de masses volumiques respectives 1075 kg/cm3 pour le sable et 1485 kg/cm3 pour les graviers et de poids spécifiques respectifs 2,626 et 2,970. Les teneurs en eau initiales sont de 0,27 % pour le sable, 0,24 % pour le gravier 3/8 et 0,13 % pour le gravier 8/16. 4.1.2.Formulation proprement dite Pour le choix de la granulométrie optimale du squelette, la courbe de référence de type Talbot-Fuller-Thompson (T-F-T) pour D = 20 mm et n = 0,45 servira de courbe de référence pour la détermination des différentes proportions de basalte et de sable [3]. La Figure 1 illustre le calcul et on a ainsi 26 % de sable, 39 % de gravier 3/8 et 35 % de gravier 8/16. série de courbes granulométriques 120,00

100,00

passant(%)

80,00

sable dune 3/8

60,00

8/16 40,00

20,00

0,00 0,01

Fuller-Thomson n = 0,45

0,1

1

10

100

diametre (mm)

Figure 1 : Détermination de la granulométrie optimale avec la courbe de référence de Fuller-Thompson.

Pour le choix du dosage volumique de la pâte, le volume des vides des granulats compactés est généralement compris entre 180 et 200 l/m3. En l’absence de données, on prend Vvc = 190 l/m3 [2]. La plage optimale de la maniabilité pour un mélange de BCR est de 40 à 90 secondes de l’appareil Vebe. Pour nous mettre dans le cas le plus

I.K. Cisse and A. Sall / Utilisation des Bétons Compactés au Rouleau (BCR)

463

défavorable, on prend la valeur de 40 secondes (en effet, plus le béton est consistant, plus le temps Vebe est grand). Ainsi, pour cette valeur de la maniabilité, on a (Figure 2) :

Figure 2 : Relation expérimentale entre la maniabilité et le rapport Vp/Vvc pour des mélanges de BCR sans air entraîné [2].

Pour le choix du rapport E/C, en visant une résistance de 40MPa et en considérant un BCR contenant 0 à 10 % de fumée de silice, on obtient un E/C de 0,45 à partir de la Figure 3[2].

Figure 3 : Relation entre le rapport E/C et la résistance à la compression à 28 jours de différents mélanges de BCR (sans air entraîné) [2].

A partir de la relation Vpate=Vciment+Veau [3], on calcule les différentes proportions, et compte tenu des corrections dues à la teneur en eau initiale, nous avons finalement le dosage suivant :

464

I.K. Cisse and A. Sall / Utilisation des Bétons Compactés au Rouleau (BCR)

Ciment : 260 kg/m3

Eau : 117 litre/m3

Basalte 3/8 : 899 kg/m3

Basalte 8/16 : 806 kg/m3

Sable dune : 559 kg/m3

4.2 Confection des éprouvettes et caractérisation mécanique Les mélanges seront confectionnés dans des éprouvettes de diamètre 11 cm et de hauteur H = 22 cm, soit un élancement de 2 et un volume de 2090,73 cm 3. On appliquera une énergie équivalente à celle de l’essai Proctor modifié. On calcule ainsi le nombre de coups par couche suivant la formule.

Les caractéristiques mesurées sont données au Tableau 1. Tableau 1: Valeurs moyennes de la masse volumique et des résistances du BCR à 28j

Grandeur moyenne

Notation

Unité

Valeur

Masse volumique humide

ρh

kg/m³

2767,77

Masse volumique à l'état durci

ρ

kg/m³

2758,41

Résistance en compression

Rc28

MPa

33

Résistance en traction par fendage

Rt28

MPa

4

4.3 Analyse et interprétation des résultats

On constate que ces éprouvettes ont une masse volumique de l’ordre de 2700 kg/m3. Cela s’explique par le fait même de la densification du squelette granulaire minimisant le volume des vides. Il faut signaler que la masse volumique du béton courant est de l’ordre de 2300 à 2400 kg/m3.

La résistance en compression moyenne à 28 jours qui est de 33 MPa est un peu inférieure à celle qui était visée lors de la formulation. Cette différence était prévisible dans la mesure où la courbe donnant la résistance à la compression en fonction du rapport E/C est établie pour des mélanges de BCR avec ajout cimentaire. Dans notre cas, nous n’avons pas eu à utiliser d’ajouts vu leur non disponibilité et leur coût.

La résistance en traction obtenue à partir de l’essai brésilien sur un ensemble d’éprouvettes donne une résistance moyenne de 4 MPa. Cette valeur est jugée bonne pour une dalle destinée au revêtement de chaussée où les sollicitations en traction sont importantes.

5. Dimensionnement de la Chaussée en Revêtement de BCR 5.1 Hypothèses de calcul •

Estimation du trafic

465

I.K. Cisse and A. Sall / Utilisation des Bétons Compactés au Rouleau (BCR)

Le trafic équivalent (N) en nombre d’essieux de 13 tonnes au bout de n années de service est donné par la relation : N = 365 *TJMA*A*C

(1)

Tous calculs faits sur la base des hypothèses décrites à la référence [3] conduisent à N = 2.2.107 essieux de 13 tonnes. •

Caractéristiques du sol de Plate-forme

Le CBR sur l’ensemble des tronçons varie de 27 à 70 [3]. Afin d’être plus sécuritaire, on prend pour le dimensionnement un CBR = 27. Le module de rigidité est : E = K*CBR avec 50 ≤ K ≤ 100 ; on prend K = 50, d’où : E = 50 * 27, E2 = 1350 bars Le coefficient de Poisson est de : ν2 = 0,35 •

Choix de la structure

Nous pouvons adopter comme structure, une dalle sur couche drainante de 10cm ; dans une telle structure (Figure 4), c’est essentiellement le revêtement qui supporte les charges. E0, υ0

Revêtement en BCR

ε0 Couche drainante de 10 cm

Sol de plate-forme

σt0

σz0

E1, υ1

ε1

σ

σt1

E2, υ2

Figure 4 : Structure de la chaussée en BCR adoptée

•

Caractéristiques de la couche drainante

Son module de rigidité est lié à celui du sol de plate-forme par la relation suivante : 2 E2 ≤ E1 ≤ 4 E2. Nous pouvons prendre : E1 = 3 E2 = 3*1350 = 4050 bars ν1=0,35 H1 = 10 cm 5.2 Vérification des contraintes et déformations 5.2.1 Dans le sol support •

Contrainte admissible

Pour C variant de 0.006 à 0.008, E étant le module de Young de la plate-forme et N, le nombre d’essieux équivalent, on a :

466

I.K. Cisse and A. Sall / Utilisation des Bétons Compactés au Rouleau (BCR)

•

Déformation admissible

Formule de Dormon :

5.2.2 Dans le revêtement •

Contrainte de traction

Il vient, après développement [3] :

σ

t ,adm

= σ

6

(N E

/106

(

σ t , adm = 1.85 2.2.107 /106

)

)

b

−0.067

10

10

− ub ⎡SN ⎣

2

+ (c2 / b2 ) Sh 2 ⎤ ⎦

0 .5

kd kc

− ( −1.911)( −0.067) ⎡⎣12 + (0.022 / 0.0672 )0.032 ⎤⎦

0.5

1 1.5 1.7

=0,99 MPa =9,9 bars

5.3 Calcul des contraintes sur Ecoroute Pour le calcul des contraintes sur Ecoroute, on se fixe une épaisseur de 10 cm pour la couche 1 (couche drainante). Le calcul sera fait pour des épaisseurs de revêtement variables, ce qui nous permettra ainsi de déterminer l’épaisseur optimale. Ainsi, pour des revêtements de 15, 20 et 25 cm, les contraintes et déformations engendrées au niveau du sol de plate-forme sont inférieures à celles admissibles. Cependant, la contrainte de traction à la base du revêtement reste supérieure à la contrainte de traction admissible. Donc l’épaisseur de BCR convenable est de 30 cm [3].

6. Calcul Économique Un calcul économique [3] prenant en compte le coût d’investissement, les coûts d’entretien a permis de comparer les structures routières en BCR et en revêtement hydrocarboné. Cette étude s’est faite sur la base des données indiquées à la référence [3] et des hypothèses qui y sont faites. L’analyse comparative révèle un large avantage de la chaussée en BCR à celle en liant hydrocarboné ; soit 47 % d’économie.

I.K. Cisse and A. Sall / Utilisation des Bétons Compactés au Rouleau (BCR)

467

Conclusion Ce travail avait pour objectif de valoriser l’utilisation du Béton Compacté au Rouleau (BCR) comme revêtement des chaussées. Les résultats expérimentaux ont démontré que le choix d’une distribution granulométrique optimale permet de diminuer très significativement le volume des vides d’un mélange granulaire et par conséquent de minimiser le volume de pâte nécessaire pour produire un BCR pour revêtement routier, possédant des caractéristiques optimales, nécessitant des coûts de fabrication moins élevés. Par ailleurs, l’application du BCR au tronçon Diamniadio-Mbour montre qu’un revêtement d’épaisseur 30 cm reposant sur une couche drainante de 10 cm en graveleux latéritique cru, permet de reprendre toutes les sollicitations imposées par les pneumatiques, et ce, avec une réduction du coût global de construction de 47 % par rapport à l’option revêtement en béton bitumineux sur couches de base et de fondation. Pour assurer un confort des usagers, à un coût assez abordable, l’étude préconise le recours à une mince couche de bitume gravillonné 0/3 dosé à 20 l/m 2 et qui sera enduit sur le revêtement en BCR.

Références [1] Association Internationale Permanente des Congrès de la Route (AIPCR) : Emploi du béton compacté dans les chaussées, 1993. [2] GAGNE, Richard : Les bétons compactés au rouleau- principes, application et nouveau développement BCR, CRIB 2004. [3] SALL, A., NDIAYE D.: Conception, Formulation et Mise en œuvre du BCR comme revêtement des chaussées ; projet de fin d’étude à l’Ecole Polytechnique de Thiès, 2007. [4] GABRIEL J. Assaf : Catalogue de dimensionnement des aires de circulations et de chargement en béton compacté au rouleau. [5] GAGNE, Richard : Méthode de formulation et d’optimisation des mélanges de BCR, CRIB 2004.

468 Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-468

Pavement Rehabilitation Options for Developing Countries with Marginal Roadbuilding Materials Khaimane M.D. DE DEUSa and Wynand Jvd STEYNb Millennium Challenge Account, Maputo, Mozambique b University of Pretoria, Pretoria, South Africa

a

Abstract. Pavement rehabilitation involves measures used to restore, improve, strengthen or salvage existing deficient pavements so that these may continue, with routine maintenance, to carry traffic with adequate speed, safety and comfort. Developing countries around the world are facing serious difficulties to maintain their road networks in a good condition. These difficulties include scarcity of suitable road-building materials, lack of funds for maintenance and construction of new roads and road agencies with insufficient well-qualified technical staff. This paper evaluates possible suggestions for developing countries with marginal roadbuilding materials on pavement rehabilitation options in order to maintain the existing road network in a sound condition. The paper evaluates the different types of rehabilitation methods available internationally, as well as their pros and cons. This is followed by an evaluation of the typical process involved in these methods, and analysis of which aspects are affected by scarcity of good road building materials. The possible effects of usage of marginal road building materials on rehabilitation options are evaluated, and recommendations made as to a procedure for adequate evaluation and rehabilitation of such pavements. Keywords. Marginal road-building materials, developing countries, pavement rehabilitation, rehabilitation methods

Introduction This paper evaluates possible techniques for developing countries with marginal roadbuilding materials on pavement rehabilitation options in order to maintain the existing road network in a sound condition. The paper evaluates the different types of rehabilitation methods available internationally, followed by an evaluation of the possible effects of usage of marginal road building materials on rehabilitation methods. Recommendations are made regarding a procedure for adequate evaluation and rehabilitation of such pavements.

1. Pavement Rehabilitation Six pavement rehabilitation design methods are typically used for flexible pavements in southern Africa, namely, the Asphalt Institute Method, the Dynamic Cone Penetrometer (DCP) Method, the Transportation and Road Research Laboratory

K.M.D. de Deus and W.J. Steyn / Pavement Rehabilitation Options for Developing Countries

469

(TRRL) Deflection Method, the SHELL Overlay Design Method, the American Association of State Highway and Transportation Officials (AASHTO) Design Method and the South African Mechanistic Design Method. 1.1. The Asphalt Institute Method The aim of this method is to provide adequate protection to the sub-grade of the pavement, similar to the California Bearing Ratio (CBR) approach where the aim is to determine the layer thicknesses needed to protect the lower layers. Table 1 summarizes the main characteristics of the Asphalt Institute method [1]. Table 1 – Main Characteristics of the Asphalt Institute Pavement Rehabilitation Design Applicability - Developed with data from pavements with granular sublayers and thin surfacing; - Pavement component analysis method incorporates all types of material using conversion factors and subjective condition ratings

Limitations (main) - Empirically based with limitations to applicability - Design based on protection of sub-grade only - No seasonal adjustment factors given - AlI curve applicable to traffic loading up to 7,3x106 E80s - Component analysis procedure applicable up to 36,5x106 E80s

Advantages (main) - Based on easy-to-apply and well-known concepts

1.2. The DCP Method The DCP is used to measure the shear strength of the various layers. It is widely used in southern Africa due to cost effectiveness and simplicity. In the development of the method the first objective was to improve the utilization of DCP tests as a measurement of the structural capacity of pavements. Many of the concepts used originated from practical experience with the use of the instruments [1]. The main characteristics of the method are summarized in Table 2. Table 2 – Main Characteristics of the DCP for Pavement Rehabilitation Design Applicability Limitations (main) Advantages - Based on Non - Empirically based with limitations - Light pavement with Destructive Testing associated with the component analysis granular sub-layers and (NDT) in-situ pavement approach thin surfacing test - Does not take seasonal variations into - Pavements with lightly - Allows for the account cemented layers (further assessment of individual - Some concepts have not been properly research) pavement layers - Developed with data verified in practice - Based on the optimum - For use on balanced pavements only from balanced utilization of in-situ pavements only - Design curves applicable to traffic between pavement layer 0,2x106 and 10x106 E80s properties

The method is widely applicable in developing countries and provides important information, mainly regarding the pavement strength, that could be used on other pavement rehabilitation methods. It optimizes the utilization of the in-situ pavement material strength, in order to determine the most economical rehabilitation method, taking into account the existing pavement structure and the materials locally available (including marginal materials).

470

K.M.D. de Deus and W.J. Steyn / Pavement Rehabilitation Options for Developing Countries

1.3. The TRRL Surface Deflection Method The method allows the designer to predict the remaining life of a pavement before reaching the critical condition and also to design the thickness of overlay required to extend the life of the pavement to carry a given design traffic. It is based on the evaluation of elastic surface deflection data which means that it is dependent on deflection measurement devices [1]. The main characteristics of the TRRL deflection method are summarized in Table 3. Table 3 – Main Characteristics of the TRRL Surface Deflection Method Applicability Limitations (main) - Design charts developed - Empirically based with limitations in for main types of flexible applicability pavement - No seasonal adjustment given - Some limitations of use - Applicable for traffic loading up to for pavements with 10x106 E80s cemented layers - Design (except for pavement with cemented layers) based on deformation originating from the sub-grade only

Advantages (main) - Based on easy NDT testing - Easy to use - Distinguishes between the types of pavement - Some limitation in the use of the method on pavements with cemented layers - Adjust deflections to take the effect of temperature variations into account

1.4. The SHELL Overlay Design Method The method uses a number of design charts to determine the required thickness of overlays for the rehabilitation of a pavement. The design charts were derived from the results of analyses where the pavement was assumed to be adequately represented by a three-layered model consisting of a top layer (surfacing) of asphaltic material, a middle layer (base) of a granular cemented material and a bottom layer (sub-grade) of semiinfinite dimensions [1]. In the development of the method, the pavement stresses and strains were calculated and the results used to compile design charts where the primary design criteria used were related to the compressive vertical strain in the surface, the horizontal tensile strain in the asphalt and the horizontal tensile strain in any cementitious base layer. Table 4 summarizes the main characteristics of the Shell overlay design method. 1.5. The AASHTO Design Method: Guide for Mechanistic – Empirical Design of New and Rehabilitated Pavement Structures This is a mechanistic-empirical iterative method (for new pavement and rehabilitation) that considers site conditions (traffic, climate, sub-grade, existing pavement condition for rehabilitation) and construction conditions. The method considers different environmental conditions and materials. It is easy to apply and combines mechanistic and empirical procedures. It is suitable for developing countries conditions, as it allows the use of available materials (including marginal materials) [1].

K.M.D. de Deus and W.J. Steyn / Pavement Rehabilitation Options for Developing Countries

471

Table 4 – Main Characteristics of the SHELL Overlay Design Method Applicability Limitations (main) Advantages (main) - Incorporates only temperature - Based on NDT tests All types of flexible pavements (special changes as climate effect - Takes overlay mix properties into account - Does not assess unbound provision for - Incorporates and allows for temperature cementitious layers) pavement layers gradients in the asphalt layer - Charts based on asphalt fatigue - Allows for the checking of deformation in and sub-grade deformation asphalt layers only - Incorporates different climates through wMAAT factor

1.6. The South African Mechanistic Design Method This is a mechanistic method based on the theory of linear elasticity, and it is applied to pavement analysis through the use of a catalogue of behaviour states or, alternatively, through a computer-aided simulation of the response of the pavement under loading. The theory of linear elasticity as well as results and experience gained over many years of HVS testing have been combined to compile the catalogue [1]. The main characteristics of the South African mechanistic design method are given in Table 5. Table 5 – Main characteristics of the South African Mechanistic Design Method Applicability Limitations (main) All flexible and - Catalogue includes a limited number of analyses rigid types of covering only the main trends in behaviour pavement - No direct procedure of material characterisation is given - Does not take into account variations in environmental conditions - Does not take into account past life

Advantages (main) Allows for analysis of all types of material in all type of layers Allows for analysis of pavement with thin asphalt layers

In order to simulate pavement response using an available and suitable computer based theoretical model, the characteristics of the pavement must be accurately determined as an input into the model. The modelling of the pavement should enable the theoretical model (in this case based on the linear elasticity theory) to accurately simulate the response of the pavement when subjected to a static standard dual wheel load of 40 kN (tire inflation pressure 520 kPa) at a spacing of 350 mm [2].

2. Availability of Road Building Materials Three criteria should be used to select materials to be used in the structural layers of the pavement namely, availability, economic factors and previous experience. All these criteria are important, but knowing the previous experience of materials could lead to reduction in costs if the materials can be utilized effectively. In the northern region of Mozambique there are large quantities of lateritic soils that have been used with relative success on past projects (e.g. Chitima – Mágoè road in Tete Province; Nametil - Angoche road in Nampula Province, etc). As no information regarding its behaviour has been properly recorded, foreign consultants are not confident of its applicability and performance. This typically leads to a situation where the specified materials to be applied on projects need to be imported from long distances thus increasing the hauling cost. In southern Africa, the material types mostly used are natural soils and gravels,

472

K.M.D. de Deus and W.J. Steyn / Pavement Rehabilitation Options for Developing Countries

processed gravels and rock, bituminous and cementitious-treated materials and Portland cement concrete. 2.1. Use of Fine Lateritic Soils Studies carried out in Brazil indicated excellent behaviour of experimental sections with fine lateritic sand bases in economical pavement structures replacing conventional bases (mostly constituted by crushed stone material, stone or soil cement). Numerous pavements with this type of base are being used in urban roads, runways, and parking lots [3]. The main particularities on fine lateritic sand pavement behaviour in São Paulo (Brazil) include no base failure evidence, low elastic deflection, good structural contribution for base layer, high resilient modulus, limited deformation (rutting) and low rehabilitation cost. From the experience gained on more than 35 years of construction and performance monitoring in Brazil of lateritic soil bases, it was concluded that the successes obtained are directly linked to the correct use of construction technique, technological control and edge break control. It was shown that the optimal economical behaviour of pavements is linked to optimal construction of the base and correct application of prime and surface layer. The main economical consequences are increases in the availability of soils applicable for base layers (considered inadequate by conventional criteria) and decreases in the amount of pavement defects with a resultant decrease in pavement conservation costs. These positive results should be imported to southern Africa to be used as a guide as there are similarities between Brazil and southern Africa in terms of the climate (tropical) and environmental conditions. It should, however, be appreciated that laterites are notoriously variable and conclusions from one region should be treated carefully when applying to other regions. In another study related to the use of lateritic gravels as road base in southern Africa by TRRL on two roads in Malawi, the roads were constructed and monitored over a number of years and have given a better than expected performance. From the study, a structural design chart was developed to enable better use to be made of the lateritic gravels available in the region [4]. The chart is based primarily on the performance of low to medium volume rural roads constructed with marginal quality materials throughout the southern Africa region including the lateritic gravel sections in Malawi. The study concluded that the evidence indicates that marginal quality (in the traditional sense) base course materials, including the lateritic gravels, have performed satisfactorily for low volume rural roads carrying typical rural road traffic, and where lateritic gravels are used in the base, such as those available in Malawi, the field evidence has shown that these materials can sustain very heavy axle loads. From the study, it was also concluded that the differential between the cost of crushed stone base and the 1 km lateritic gravel base trial section was about 4:1 [4]. 2.2. Use of Marginal Materials in Low Volume Roads in Southern Africa Marginal materials are materials with inferior properties to those required by traditional specifications, but that can be used under favourable circumstances [5]. Sustainable transport requires the sustainable supply and use of construction materials. This

K.M.D. de Deus and W.J. Steyn / Pavement Rehabilitation Options for Developing Countries

473

includes the use of marginal materials, waste materials, novel/innovative materials and reuse of existing materials. Construction materials obtained from quarries and borrow pits are non-renewable resources and their continued use results in depletion of a natural capital [6]. The use of a “non-renewable material” could in many instances make the use of alternative materials considerably more attractive in terms of real cost. An investigation into the performance of 57 sections in South Africa identified as being constructed using marginal quality base course materials concluded that materials not complying with the traditional requirements should not be rejected for use in lightly trafficked roads (<0.5 million E80s) without careful consideration. Aspects such as presence of large stones or high plasticity (which will make construction difficult or affect the surfacing adhesion) should, however, be considered. Similarly, very fine materials with little aggregate appear to detrimentally affect the surfacing/base adhesion [5]. Under the same study it was recommended that roads using marginal materials are raised above the natural ground level so that the bottom of the base course is at least 0.75 m thick in order to facilitate the retention of low moisture contents in the critical pavement layers. The main problem using marginal materials is the risk of premature failure through unexpected environmental conditions. The only way to manage this risk is to ensure that the construction quality is well controlled and drainage measures are implemented and maintained. It becomes difficult in developing countries, such as Mozambique, where the quality of maintenance is questionable. Preventative maintenance in terms of timely resealing is also critical to ensure that the pavement does not reach a stage where routine maintenance becomes excessive. The performance of marginal quality materials in lightly trafficked roads depends to a far greater extent on the drainage and seasonal moisture variations than on the quality of the material [5]. 2.3. Use of Locally Available Materials in Low Volume Roads The material being used in low volume roads, especially roads constructed with local materials is often quite variable in its properties, which has resulted in many of these roads failing prematurely since the material was either not strong enough or was not compacted properly [7]. Layer works can make full use of local materials but are not particularly cost-effective in terms of labour-enhanced construction methods. It has been shown that high densities, low material variability and good construction control are necessary for successful performance of low volume roads. These are all difficult (but not always insurmountable) problems to overcome using labour-enhanced techniques. Naturally occurring soils and gravels are an important source of material for use in the construction of Low Volume Sealed Roads (LVSR). This is because these materials are relatively cheap to exploit compared, for example, to processed materials such as crushed rock. Moreover, in many southern Africa countries, they are often the only source of material within a reasonable haul distance of the road. Thus, because of the substantial influence that naturally occurring materials exert on the cost of a LVSR, typically of the order of about 70 per cent, it is essential that the benefits of using them is exploited in road construction [7]. Unfortunately, many of the naturally occurring road building materials in the southern Africa region is disparagingly described as being “non-standard”, “marginal”, “low-cost”, or even “sub-standard”. This is because such materials are often unable to

474

K.M.D. de Deus and W.J. Steyn / Pavement Rehabilitation Options for Developing Countries

meet the required construction specification based on traditional high volume pavement design method requirements. However, there are many examples of naturally occurring materials, such as laterite and calcrete, which have performed satisfactorily despite being “sub-standard” with respect to their grading, plasticity or strength [8]. Specifications are meant to exclude most unsatisfactory materials for use in roads by placing limits on their various properties such as grading, plasticity and strength. The derivation of appropriate limits requires an intimate knowledge of local material performance in a specific environment (climate and drainage measures) and for specific traffic loading. The challenge is to relate the materials’ physical properties with performance in a particular environment. A national inventory of materials would prove useful and the need for changes of specifications and testing methods for marginal materials should be considered.

2.4. Combination of Marginal Materials and Rehabilitation Design Methods Among the six rehabilitation design methods earlier mentioned in this paper, three can be used with marginal materials. The DCP, the AASHTO, and the SAMDM allow the use of the available pavement materials (including marginal materials). The abovementioned pavement design methods are based on the optimum utilization of in-situ pavement layers properties. Actually, these are the three most used methods in the southern Africa region, including Mozambique. The remaining three methods namely, the Asphalt Institute, the TRRL, and the SHELL are better combined with materials which meets the specifications already defined by the existing standards.

3. Conclusions and Recommendations There are several types of rehabilitation methods around the world, some being more applied for developing countries conditions. The best rehabilitation method to be chosen shall be done after an economical assessment of the various applicable options and strategies. More than one method should be used so that it is possible to assure that the final result is optimized. It should also be determined whether the specific method is compatible with local materials, especially when these are marginal materials. Another conclusion that could be drawn in this paper is that developing countries (mainly in southern Africa region) have several materials that may be suitable for use in road construction and rehabilitation. However, these marginal materials should only be used after carrying out tests to understand their behaviour and should also be used taking into account the past experience. Drainage is one of the main aspects that should be taken into account. It follows that empirical pavement rehabilitation design method where good knowledge of local material behaviour is available, and the relationships in the model has been calibrated for this local material knowledge should provide good rehabilitation designs. The use of local or marginal materials for road works should be supported. Investigations to review test methods and to provide standards based on past experience with marginal material should also increase to reach a stage that one can use marginal materials with total confidence. Road agencies should also consider the use of construction materials information system.

K.M.D. de Deus and W.J. Steyn / Pavement Rehabilitation Options for Developing Countries

475

References [1] Jordaan, G. J., Practical approach to pavement rehabilitation investigations and design, Pretoria, South Africa. 2006. [2] Jordaan, G.J., The South African mechanistic pavement rehabilitation design method, Research Report RR 91/242, CSIR, South Africa, 1994. [3] Villibor, D.F. and Nogami, J.S., Pavimentos económicos – Tecnologia do uso dos solos finos lateríticos: Arte e Ciência, São Paulo, Brazil, 2009. [4] Gourley, C.S. and Greening, P.A.K, Use of ‘substandard lateritic’ gravels as roadbase materials in Southern Africa, TRL, United Kingdom, 1997. [5] Paige-Green, P., Recommendations on the use of marginal base course materials in low volume roads in South Africa, Research Report RR 91/201, CSIR, South Africa, 1996. [6] Steyn, WJVDM and Paige-Green, P., Evaluation of issues around materials for sustainable transport, In Proceedings of 28th South African Transport Conference, Pretoria, South Africa. 2008. [7] SATCC, SADC Guidelines low volume sealed roads, SADC, 2003. [8] Overseas Road Note 31, A guide to the structural design of bitumen-surfaced roads in tropical and subtropical countries, TRL, United Kingdom, 1993.

476 Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-476

Applications of Participatory Road Maintenance using “Do-nou” Technology in Kenya a

Makoto KIMURAa,1 and Yoshinori FUKUBAYASHI b Kyoto University, Graduate school of Engineering, Japan b Community Road Empowerment (NGO), Japan

Abstract. The Do-nou (soilbags) technology, which uses geotextile materials readily available in rural areas, has been applied in maintenance of rural access roads in Kenya. Damaged sections of unpaved rural roads, where vehicles would have otherwise got stuck during the rainy season, were repaired and maintained voluntarily by organized local communities. Training on use of this technology was conducted with the aim of empowering the communities to be able constantly repair and maintain sections of many of their roads on their own. The geotechnical technology is being transferred to the communities and contributing the poverty reduction practically in the rural area of Kenya. Keywords. Rural access road, Do-nou , geotextile, trafficability, community

Introduction It is often said that the people in rural areas of developing countries rely on rural access roads for their lifelihood; this is because the roads connect them to basic social services, markets, aid posts, schools among others. Due to the poor conditions of the rural roads, the people are often struggling to rush patients to hospital as evidenced in Figure.1.

Figure 1. The conditions of the rural access roads and the stuck vehicle on the roads 1

Corresponding Author: Professor, C cluster, Kyotodaigaku katsura, Nishikyo-ku, Kyoto, Japan; E-mail: [email protected]

M. Kimura and Y. Fukubayashi / Applications of Participatory Road Maintenance

477

Most engineers are targeting to design and build new trunk roads and maintain the main road networks which bring a big impact to nation-wide economy. The authors aim at maintaining rural roads hence providing a geotechnical engineering poverty reduction solution. The authors took notice of the gunny bags (made from polypropylene) used for harvesting and storing grain found easily in the rural areas of the developing countries tested and proposed as a cheap geotextile material to repair and maintain rural roads. In Japan, the gunny bags, which are called “Do-nou” in Japanese, are commonly used for raising embankments during times of inundation and as temporary structures during reconstruction after disasters. However it is revealed that the soil wrapped inside a gunny bag and compacted has five to ten times higher bearing capacity than that of the soil itself[1]. The high bearing capacity of Do-nou has motivated the development of a novel and effective earth reinforcement method in which the bearing capacity of soft foundations can be greatly improved. The maintenance of the unpaved roads using Do-nou has been developed[2]. The Figure 2. shows the standard cross section of the maintained roads with Do-nou. The mud and soft spots are replaced with Do-nou, which is the in situ soil wrapped with gunny bags. The road base formed with Do-nou has high bearing capacity to reduce the settlement of the road surface after the traffic has passed on the roads. As shown in Figure 3., which is the result of the full scale driving test, it is found that the Do-nou reduce the settlement of the road surface after the traffic has passed on the road by 33% compared with the settlement of that without the reinforcement of the gunny bags. If the granular material which has larger internal friction angle are available then used as the material inside the Do-nou, the settlement of the road surface become smaller. This has found through the full scale driving test [2].

Figure 2. The standard cross section of the road maintained using Do-nou technology

Settlement (mm)

-100

Road surface reinforced with Do-nou

0

Road surface without reinforcement

100

200

0

1000

2000

3000

Distance from the measuring point (mm)

Figure 3. The settlement of the road surface after 10 passes of the vehicle

478

M. Kimura and Y. Fukubayashi / Applications of Participatory Road Maintenance

After the effectiveness of the maintenance of the unpaved road with Do-nou has been proven, the maintenance has been demonstrated in several countries, such as Kenya, Uganda, Tanzania, Congo, Zambia and some countries in Asia. In this paper the applications of Do-nou for the maintenance of rural access roads are reported. Through the interview to the communities who participated in the road maintenance using Do-nou, the impact and benefits to the communities along the maintained roads has been revealed.

1. Rural road maintenance with the incentive of the communities The approach which was taken in this study to keep the trafficability of the rural access roads was that the maintenance should be conducted by the communities with the incentive of the self development rather than for payment. The question is for the communities to be motivated to implement the road maintenance voluntarily and continuously. The purpose of this study is to examine the applicability of Do-nou technology, geotechnical engineering, to the road maintenance conducted with the incentive of the communities and the impact on the life of rural people. Therefore the authors set the strategies to make the participatory road maintenance practical and effective as itemised below; 1. Only the damaged portions of the road are identified and repaired by the communities themselves. 2. To render the road repairs effective and long lasting provision of drainage system is prioritized. 3. The road base is reinforced with Do-nou technology. The characteristics of Do-nou technology are summarized as follows; a. The required material is locally available material. b. All the processes including compaction can be done using human labour with minimized use of machines and equipment. c. It is cheap and simple but effective.

2. Applications of the road maintenance using Do-nou About 10 km of rural roads have been repaired through the community participatory road maintenance technique using Do-nou technology from 2007 to 2010. Through the maintenance conducted by the communities, it was found that the problematic (damaged) sections of rural roads can be categorized into three considering the cause of damage: 1. Flat terrain, 2. Steep slope and 3. Sag (valley). When these portions are intended to be maintained, it is most important that the drainage system should be managed properly. In this paper the maintenance at the sag (valley terrain) is reported. 2.1. Maintenance at the sag Rain water is accumulated at the sag, where the gradient of the road is changing from downward to upward. Due to high water contents, the soil becomes plastic and liquid.

M. Kimura and Y. Fukubayashi / Applications of Participatory Road Maintenance

(a) The condition before maintenance

479

(b) The condition after maintenance

Figure 4. The road maintenance at the sag

Figure 5. The cross section of the maintained road at the sag using Do-nou technology

As the soil is disturbed by the traffic, the unconfined compressive strength becomes low. It results in the reduction of the bearing capacity of the roadbed and appearing deep ruts (Figure 4. (a)). At the lowest portion of the sag, the culvert crossing the road is installed to drain water collected from adjoining lands. However, it is often seen that the existing culvert is blocked with mud. It is better to replace old culverts with new larger diameter culverts to avoid blockage. Due to lack of funds to procure new culverts it is difficult to replace. The authors then suggested to clean up the blocked culvert, then reinforce the roadbase around the culvert to prevent scouring even if the road surface is submerged under water. In the event that the community is not able to install culverts, alternatives like use of drifts or log bridges should be considered. Figure 5. shows the cross section diagram of the maintained road using Do-nou in longitudinal direction. Figure 4. (b) is the condition of the road at the same location as Figure 4. (a) 6 months after the maintenance.

3. Response of the communities to the participatory road maintenance In this study, the road maintenance using Do-nou technology has been transferred to farmers who were willing to participate in voluntarily work for improvement of their own living standards.

480

M. Kimura and Y. Fukubayashi / Applications of Participatory Road Maintenance

In a village of Kenya, the number of the participants in the work was increasing as shown in Figure 6 as the road was improved. At first they did not believe in effectiveness of the technology, therefore only a few villagers joined the trainings. As the problematic portions have been improved applying Do-nou technology, the road users realized the effectiveness and started to feel interested in the activities. As a result, the number of the participants was increasing and on the final day of the work thirty one people attended, which was more than three times the number of the participants on the first day. Till the end of the construction, a hundred and fifty metres of the road have been improved and become passable. The demonstration of the road maintenance using Do-nou makes the communities motivated to continue the maintenance and apply to the several structures by themselves. A farmers’ group utilized the second hand gunny bags for the seed of maize as the Do-nou bags for the maintenance of the roads as shown in Figure 7. Figure 8. shows the embankment of a water harvesting dam built using Do-nou.

4. Impact of the participatory road maintenance using Do-nou technology 35

Number of the participants

30 25 20 15 10 5 0 1

2

3

4

5 6 7 Working day

8

9

10

11

Figure 6. The number of the participants to the work

Figure 7. The second hand gunny bags utilized as Do-nou bags

Figure 8. The Do-nou applied to the embankment of the water harvest

M. Kimura and Y. Fukubayashi / Applications of Participatory Road Maintenance

481

The impact and benefits to the communities after the road maintenance have been surveyed through the interview to the people who participated in the work after a year has passed from the maintenance. Table 1. The result of the interview to the communities who participated in the work

Category Traffic

Agribuisness

1. 2. 3. 4. 1. 2. 3. 4.

Group activities

1. 2.

3. Life

1. 2. 3.

Impact and benefits Number of the bike taxi on the road has increased. Time taken to reach the town has become shorter. The improved road enabled the motorcyclists increase the fuel mileage. The fee of the bike taxi has become cheap. The frequency of the visits of the buyers has increased. The farmers become able to arrive at the market earlier in the morning , then their products command good prices. The planting area of the farmers have increased, since the most of the products can be transported and sold. The extention officers have visited the village more frequently because of the improved road conditions. The number of the member of the group has increased. The group became more cohesive and started the new selfdevelopment project, such as the new crops, fish pond and nursery. The group started to transfer the Do-nou technology to the neighbors who got interested in the road maintenance. The patient could be transferred to the hospital in time. A new kindergarten has built. Some people started to commute to the town near the village, since the commute time has reduced.

5. Conclusion The maintenance of the rural access roads using Do-nou technology has been developed and transferred aiming to reduce the poverty in rural developing countries. The simplicity and effectiveness motivated the communities to provide their labour for the improvement of rural infrastructures. Then, with the initiative of the communities for self development, the rural road is maintained continuously. The impact and benefits to the communities have revealed through the interview. It can be said that this is the case the geotechnical engineering is put into practice by the community leading to poverty reduction.

References [1] Matsuoka, H. and Liu, S., A New Earth Reinforcement Method using Soilbags, Taylor & Francis Group㧘 London, 2006. [2] Fukubayashi, Y. and Kimura, M., Maintenance of Unpaved Road on Problematic Soil using Labor Based Technology in East Africa, Proc. of the 14th African Regional Conference for Soil Mechanics and Geotechnical Engineering, Yaounde, (2007), .253-260.

482 Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-482

Modélisation Numérique du Renforcement des Chaussées non Revêtues par Géogrille Mohamed Saddek REMADNA, Sadok BENMEBAREK1 et Lamine BELOUNAR Laboratoire de Génie Civil, Université de Biskra, BP 145 Biskra, Algérie

Résumé. Le présent travail s’intéresse à la simulation numérique à l’aide du logiciel FLAC, de l’amélioration de la portance du corps d’une chaussée non revêtue renforcée reposant sur un sol support de faible portance. L’étude permet de déterminer le comportement pression–déplacement en petite et en grande déformation pour les chaussées avec ou sans renforcement. Il en est déduit l’amélioration apportée par le renforcement. L’analyse des contraintes tangentielles et normales sur l’interface sol–base permet d’expliquer le rôle joué par le renforcement dans l’amélioration de la portance ou ce qui est communément appelé mécanisme de renforcement. L’étude permet aussi de montrer l’effet de la raideur et de la longueur d’ancrage de la géogrille sur le renforcement. Mots-clés. Modélisation, Chaussée non revêtue, Renforcement, Géogrille, Flac

Introduction Cette communication s’intéresse au comportement pression-déplacement d’une chaussée composée d’un bicouche constitué d’une couche de base en grave sélectionnée reposant sur un sol de faible résistance avec ou sans renforcement en géogrille interposée entre le sol et la couche de base. La chaussée étudiée est soumise à l'application d'une charge statique unique. L’examen du travail significatif concernant les méthodes de conception, indique quatre travaux originaux qui ont contribué considérablement à une meilleure compréhension des géosynthétiques utilisés dans des applications de chaussée [1,2,3,4,5,6]. Dans les méthodes analytiques, il est supposé que toute la profondeur d'ornière est développée dans le sol de fondation et que la base se déplace comme un bloc. Cette hypothèse est, pour tous les cas pratiques, correcte, où les sols de fondation sont de faible résistance et l’épaisseur de couche de base est mince. S’appuyant sur la théorie d'équilibre plastique, la capacité portante ultime q lim pour les sols en cette condition est : q lim = (2 + π ) c (pour une épaisseur de base nulle). Cependant les déformations plastiques localisées qui peuvent causer de quelque manière la rupture localisée commencent vers la limite élastique q lim = π c (pour une épaisseur de base nulle). Le mécanisme de rupture du sol argileux, supposé à 45° dans la zone plastique, est reproduit dans la Figure 1. 1

Corresponding Author. Sadok Benmebarek, Laboratoire de Génie Civil, Université Mohamed khidervBiskra, BP 145 Biskra, Algérie, E-mail: [email protected]

483

M.S. Remadna et al. / Modélisation Numérique du Renforcement des Chaussées non Revêtues

q lim chaussée > q lim sol β

Base

β

q lim sol = mNcCu

Sol (Cu)

Figure 1. Mécanisme de Rupture

Pour une épaisseur de base donnée, la pression limite sur le sol support est donnée par : qlim sol = mNcCu. Où m représente pour les auteurs, le facteur de mobilisation de capacité portante. Nc est le facteur de capacité portante. C u est la cohésion non drainée. Les auteurs prennent pour Nc, tanβ, et, m, les valeurs mentionnées dans le Tableau 1. Tableau 1. Valeur de Nc, tanβ, m selon différents auteurs m

tanβ

Valeur de Nc Auteurs Sans Renforcement.

Avec géotextile

Avec géogrille

[1]

3,3

6,00

-

Selon Boussinesq

1

[2]

2,8

5,0

-

Selon Boussinesq

1

[3]

3,14

5,14

-

0,6

1

[4]

3,07

5,69

-

A fixer arbitrairement

1

[5,6]

3,14

5,14

5,71

Calculable

≤1

Perkins et Ismeik [7] fournissent une vue d'ensemble de la majorité des études d’expérimentation et d'analyse numérique qui ont été conduites sur les chaussées renforcées. Les travaux expérimentaux grandeur nature et en laboratoire effectués jusqu’aujourd’hui indiquent une amélioration notamment dans la profondeur d’ornière et gain substantiel dans l’épaisseur du corps de chaussée. Cependant ces résultats expérimentaux pris par eux-mêmes semblent être insuffisants pour le développement d'un procédé reconnu de conception dû aux nombreuses variables dépendantes influant le problème. Par ailleurs des degrés variables de succès ont été réalisés dans le développement de modèles d’éléments finis pour prévoir la réponse des chaussées souples renforcées. Le présent travail s’intéresse à la simulation numérique en petite et en grande déformation, à l’aide du logiciel FLAC, de l’amélioration de la portance du corps d’une chaussée non revêtue renforcée reposant sur un sol support de faible portance.

484

M.S. Remadna et al. / Modélisation Numérique du Renforcement des Chaussées non Revêtues

1. Simulation Numérique avec Flac 1.1 Présentation du Cas Étudié Le cas étudié consiste à analyser le comportement d’une chaussée sous l’effet d’une charge unique statique. La chaussée considérée ici est une chaussée non revêtue qui peut accepter des déformations sous forme d’ornières qui peuvent atteindre 100 mm et plus. La chaussée se présente comme un bicouche composé d’un corps de chaussée, appelé aussi dans ce travail couche de base, en matériau sélectionné, reposant sur un sol de faible portance. L’étude du comportement d’une chaussée qui peut être renforcée ou non par géogrilles, se présente comme étant un problème, de déformation plane, de détermination de la capacité portante d’une fondation superficielle continue reposant sur un bicouche. En effet la déformation permanente de la chaussée sous forme d’une ornière permet de justifier l’hypothèse de déformation plane. B

B : demi largeur de Fondation Déplacement imposé de la fondation Couche de base

β

Df

Renforcement

B’ Sol peu portant

Dc

Wc Figure 2. Cas étudié

Le problème sera formulé en grande déformation pour représenter les grandes profondeurs d'ornière qui peuvent se développer, et sont admissibles, sur les routes non revêtues. Le contact sol-géogrille et base-géogrille sont régis par une interface ayant un comportement élastique parfaitement plastique de Mohr Coulomb. Etant donné la symétrie par rapport à l’axe vertical, et en considérant une demilargeur de fondation, B = ½ a = 0.159 m, les conditions aux frontières peuvent être présentées comme indiqué sur la Figure 2. Le chargement du corps de chaussée est réalisé par déplacement imposé de la charge jusqu’à atteindre un déplacement final, appelé ornière, limité dans ce travail à δ = 0.8 B = 0.127 m. On admet qu’au-delà de cette profondeur d’ornière la chaussée devient impraticable. Par conséquent la pression de fondation requise pour atteindre ce déplacement est considérée comme la pression ultime. Les propriétés physiques et mécaniques des matériaux utilisés sont comme suit :

M.S. Remadna et al. / Modélisation Numérique du Renforcement des Chaussées non Revêtues

485

Sol support : E=10MPa, ν=0.33, ρ=1900 kg/m3, cu=30 kPa, Dc=2.54, m=16B, Wc=3.18, m=20B Base : E=50MPa, ν=0.25, ρ=2200 kg/m3, ϕ=40°, c=0, ψ=20°, Df=0.212 m Géogrille : E=146 MPa, ν=0.33 Interface Sol/Géogrille et Base/Géogrille : k n = ks = 5 x109 N/m3, ϕ=35°, c=0, kn et ks étant respectivement la raideur normale et la raideur de cisaillement de l’élément d’interface. E, ν, γ, cu, ϕ, ψ, c ont les significations habituelles, à savoir respectivement : module d'élasticité, coefficient de Poisson, poids volumique, cohésion non drainée, angle de frottement, dilatance et cohésion. 1.2 Analyse Numérique avec Flac L’analyse en déformation plane est élaborée en utilisant le logiciel Flac (Fast Lagrangian Analysis of Continua) [8]. Le maillage correspondant est présenté dans la Figure 3. Pour aboutir à une ornière (déplacement) finale de δ=0.8B=0.8x0.159=0.127 m, un déplacement vertical descendant est imposé aux 4 points (gridpoints) représentant la semelle, selon une vitesse de déplacement constante égale à -2.5 10-6 m/pas de calcul pour le cas de chaussée non renforcée et une vitesse de -1. 10-6 m/pas pour une chaussée renforcée. Ces vitesses ont été arrêtées après plusieurs simulations préliminaires. La géogrille est modélisée comme un élément structurel poutre, défini par Flac. La poutre considérée a une inertie nulle pour caractériser l’effet membranaire de la géogrille. JOB TITLE : CHAUSSEE BICOUCHE NON REVETUE RENFORCEE PAR GEOGRILLE

FLAC (Version 5.00)

3.000

LEGEND 2.500

19-Jan-07 12:25 step 127202 -2.473E-01 <x< 3.427E+00 -4.593E-01
2.000

Boundary plot 0

1E 0 1.500

Grid plot 0

1E 0

Beam plot

1.000

0.500

0.000

0.250

0.750

1.250

1.750

2.250

2.750

3.250

Figure 3. Géométrie du Maillage

1.3 Résultats des Simulations et Analyse en Petite et Grande Déformations On présente sur la Figure 4 les résultats des simulations charge-déplacement des 4 cas possibles, chaussée non renforcée en petite déformation, non renforcée en grande déformation, renforcée en petite déformation et renforcée en grande déformation. En ce qui concerne l’amélioration dans la capacité portante de la structure, apportée par le renforcement, elle est selon les résultats que nous avons obtenus des simulations

486

M.S. Remadna et al. / Modélisation Numérique du Renforcement des Chaussées non Revêtues

avec Flac, de 29% pour l’analyse en petite déformation, simulations BF40C30S et BF40C30RS. L’analyse en grande déformation équivalente, simulations BF40C30 et BF40C30R, montre une amélioration de 46% dans la capacité portante. Ceci montre que le renforcement a un effet meilleur sur l’augmentation de la capacité portante d’un bicouche en large déplacement.

pression de la charge q/cu

12 10 8 6 4 2

BF40C30S BF40C30RS

BF40C30 BF40C30R

0 0% 10% 20% 30% 40% 50% 60% 70% 80%

déplacement vertical δ/B Figure 4. Comportement Pression-Déplacement (BF40C30 = ϕ=40° pour la base ; cu=30 kpa pour le sol. S = non renforcée petites déformations ; R = renforcée grandes déformations ; RS = renforcée en petites déformations)

1.4 Étude Paramétrique 1.4.1 Influence de la Raideur du Renforcement sur la Capacité Portante La Figure 5 présente la variation de la pression ultime avec la variation de la raideur de la géogrille. On remarque l’évolution de la capacité portante avec l’augmentation de la raideur de la géogrille. Mais cette évolution atteint une limite pour les raideurs dépassant J = 1000 kN/m.

1.4.2 Influence de la Raideur du Renforcement sur la Tension Maximale dans la Géogrille La Figure 6 présente la variation de la tension maximale dans la géogrille avec la variation de la raideur de la géogrille. On remarque l’évolution de la tension maximale avec l’augmentation de la raideur de la géogrille. Il vient donc que, au-delà d’une raideur J=1000 kN/m la tension maximale continue à augmenter sans contrepartie en capacité portante. Par conséquent, on peut conclure qu'il y a intérêt à étudier le rapport raideur/capacité portante pour arrêter le choix d’une géogrille optimale.

M.S. Remadna et al. / Modélisation Numérique du Renforcement des Chaussées non Revêtues

487

Pression limite qu / cu

14 12 10

pression limite pour un orniérage de 0,8 B

8 6 0

1000

2000

3000

4000

5000

Raideur J de la géogrille (KN/m) Figure 5. Influence de la Raideur J de la géogrille sur la capacité portante

Tension max. dans la géogrille (KN/m)

60

40

Tension max.

20

0 0

1000

2000

3000

4000

5000

Raideur J de la géogrille (KN/m)

Figure 6. Influence de la Raideur J de la géogrille sur la Tension maximale dans la géogrille

1.4.3 Influence de la Cohésion Cu du Sol support sur la Capacité Portante Dans la Figure 7 on a voulu exprimer le rapport de la pression limite avec renforcement sur la même pression sans renforcement. On remarque que l’amélioration de la capacité portante est plus importante pour les sols à faible résistance. En effet l’amélioration de portance est de 58% pour un sol de cu=15 kPa mais de 35% seulement pour un sol de cu=60 kPa.

M.S. Remadna et al. / Modélisation Numérique du Renforcement des Chaussées non Revêtues 1.7

ratio qu Chaus. Renf. / qu Chaus. Non Renf.

1.6

Chaus. Non Renf.

Ratio q u Chaus. Renf. / q u

488

1.5 1.4 1.3 1.2 0

15

30

45

60

75

Cohésion non drainée du sol cu (kpa) Figure 7. Influence de la résistance du sol (cu) sur l’amélioration de la capacité portante

2. Conclusions L’analyse des résultats de simulation de la présente recherche à l’aide du logiciel FLAC permet de conclure les points suivants : • Les simulations en grandes déformations témoignent plus d’amélioration de la portance et suivent mieux le comportement réel. • L’étude paramétrique sur l’influence de la raideur du renforcement permet de distinguer deux zones : une zone de faible raideur caractérisée par une forte sensibilité de la portance et une zone de forte raideur caractérisée par une sensibilité atténuante avec l’accroissement de la raideur. Cette étude explique les avis contradictoires dans la littérature sur ce point. • La tension maximale dans la géogrille continue à augmenter proportionnellement avec la raideur sans contrepartie en capacité portante. Ce qui pose l’intérêt de l’étude du rapport raideur/capacité portante pour le choix d’une géogrille optimale. • L’étude montre que l'amélioration de la pression limite est inversement proportionnelle à la cohésion non drainée du sol. References [1]

[2]

[3] [4]

[5] [6] [7] [8]

E.J. Barenberg, J. Halesand and J. Dowland, Evaluation of Soil-Aggregate Systems with MIRAFI Fabric. University of Illinois Report No. UILU-ENG-75-2020, prepared for Celanese Fibers Marketing Company, 1975. J.E. Steward, R. Williamson and J. Mohney, Guidelines for the use of Fabrics in Construction of lowVolume Roads, Report N° FHWA – IS – 78 – 205. Pacific Northwest Region Forest Service, US. Departement of Agriculture, Washington, DC, USA, 1977, 172 p. J. P. Giroud and L. Noiray, Geotextile Reinforced Unpaved Road Design. Journal of the Geotechnical Engineering Division, ASCE, Vol. 107, No. GT9, (1981), 1233-1254. G. T. Houlsby, and R. A. Jewell, Design of Reinforced Unpaved Roads for Small Rut Depths, Proceedings of the Fourth International Conference on Geotextiles, Geomembranes, and Related Products, Balkema, Vol. 1, The Hague, Netherlands, . (1990), 171-176. J.P. Giroud, and J. Han, Design method for geogrid-reinforced unpaved roads. I: Development of design method. J. Geotech. Geoenviron. Eng., 130(8), (2004), 775–786. J. P.Giroud and J. Han, Design method for geogrid-reinforced unpaved roads. II: Calibration and applications. J. Geotech. Geoenviron. Eng., 130(8), (2004), 787–797. S.W. Perkins, and M. A Ismeik, Synthesis and Evaluation of Geosynthetic-Reinforced Base Layers in flexible Pavements: Part I and Part II. Geosynthetics International, Vol. 4, No. 6 (1997),. 549-621. FLAC– Fast Lagrangian analysis of continua, Itasca Consulting Group, Inc., Minneapolis, MN, 2000.

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 489 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-489

Reducing the cost of road construction through targeted geotechnical and geophysical investigations – a case study of road section re-design in the Hwereso valley of Ghana a

C. F. A. AKAYULIa, S.O. NYAKOa, J. A. YENDAWb Building and Road Research Institute, CSIR, Kumasi, Ghana b University of Mines and Technology, Tarkwa, Ghana

ABSTRACT / RÉSUMÉ: The Hwereso valley that is crossed by a section of the main highway linking the north and south of Ghana, is a 2.0km wide flood plain formed by tributaries of the Hwere River in the Ashanti Region of Ghana. Prior to its reconstruction in 2004, as part of the general upgrading of the Accra-Kumasi highway, this section of the road was subjected to perennial flooding and pavement failures as a result of an incompetent clay sub grade. The original road upgrade design recommended the construction of a bridge with piled foundation across the Hwere and an embankment across the valley. Through detailed geotechnical investigation targeted at this section of the road during the construction stage, the soil profile along the centerline of the road was defined in detail and a complete characterization of the subsoil soil materials undertaken. The depth to bedrock was determined from seismic refraction surveys. Based on the results of these detailed investigations, the drainage structures and foundation design were revised to a large, single bay box culvert with 3No helper pipe culverts. The idea of a 2km long embankment across the valley was abandoned and subsoil drainage system consisting of French drains was adopted to eliminate accumulation of water around the culverts inlets. The redesign enabled the project to avoid the expensive piling exercise and large volume earthworks of the embankment construction recommended in the initial upgrade design thereby making savings in the project cost and construction time KEY WORDS: embankment, seismic refraction survey, subgrade, geotechnical site investigation.

Introduction Hwereso valley is a 2.0km flood plain formed by the Hwere River and its tributaries that is traversed by the Kumasi-Accra Highway in Ghana. This road is the main highway that links the northern and southern sectors of the country. Prior to its rehabilitation, this section of the road was mainly a bituminous surface dressed road that had developed several potholes due to increased volume of traffic, increased use of a,1

C.F.A. AKAYULI: Building and Road Research Institute, CSIR, P.O. Box UP40, Kumasi, Ghana. Email: [email protected]

490

C.F.A. Akayuli et al. / Reducing the Cost of Road Construction

heavy vehicles, perennial flooding of the valley etc, thereby making it dangerous for road users with the resultant economic implications. The rehabilitation of the Kumasi to Konongo section of the Accra – Kumasi highway which includes the Hwereso valley was given out on contract as part of the general upgrade of the Accra-Kumasi Road into a first class highway. Hwereso valley is located in the Ashanti Region of Ghana about 20km south east of Kumasi the Regional Capital. The area is generally low lying and prone to flooding. The region has a climate, vegetation and rainfall pattern that is characteristic of the wet semi-equatorial zone. A bimodal rainfall pattern with average annual rainfall ranging between 1250 and 2000mm is typical of the area. The original design of the road across the Hwereso valley proposed the construction of a single span piled Foundation Bridge over the Hwere River and an embankment across the valley. To implement this design the contractor would necessarily have had to subcontract a piling company from outside Ghana to undertake the bridge foundation work since no piling company existed in the country during the time of the road construction. This coupled with the additional earthworks involved in the 2km embankment construction called for a detailed geotechnical and geophysical investigation along the road centre line and more specifically at the banks of the river to completely characterize the subsoil materials and determine the depth to the rock head to enable a redesign of the pavement and river crossing structure. 1.

Methodology

The study methodology involved geotechnical and geophysical investigations. 1.1 Geotechnical investigations This comprised field work and laboratory testing. Fieldwork included cable percussion drilling, trial pitting, in-situ testing and sampling. At the bridge site, two boreholes were drilled to depths of 16.5m and 12.5m each on the left and right banks respectively. In addition, 1.5-2m deep test pits were excavated at 100m intervals across the valley. Disturbed and undisturbed soil samples were taken for laboratory testing which comprised moisture content, Atterberg Limits, triaxial compression and one dimensional consolidation tests. Standard penetration tests (SPT) were carried out in accordance with BS 1377 [1] specifications at 1.5m intervals in each borehole. 1.2 Seismic Refraction survey Four seismic spreads were run on traverses across the valley with a single channel FS3 portable Seismograph. Geophone positions were located at the beginning of each traverse line, two such positions being located about 5m from the bridge approaches. Hammer positions were located 1.5m apart for the first 6m at the beginning of each traverse and later increased to 3m separations for the next four hammer positions, then increased to 6m separations for the remaining length of the traverse. By gradually increasing the hammer positions from 1.5m interval to 6m intervals it was possible to increase the depth of investigation. Elastic waves were generated from the 10-pound

491

C.F.A. Akayuli et al. / Reducing the Cost of Road Construction

(20.5kg) hammer on a plate and picked by the geophones and recorded on the seismogram. 2.

Discussion of Results

All the laboratory test results were tabulated. Figure 1 shows some grading curves obtained from the complete grading analysis. The natural water content (NMC), liquid limit (LL) and plastic limit (PL) of samples taken were recorded and the plasticity index calculated. Table 1 shows the average values of NMC, PL and LL for the generalised soil types at the site 100

% PASSING

80 60 40 20 0 0.001

0.01

0.1

1

10

SIEVE SIZE (mm) Figure 1: Grading curves of soils at site

2.1 Soil Profile Geologically, the valley is underlain by the Birrimian Supergroup rocks comprising metamorphosed lavas, pyroclastics, hyperbasal basic intrusives, phyllites, schists and greywackes. Soils underlying the valley are mostly residual; derived from the weathering of the underlying parent rocks. The borehole logs and grading curves established that there exists a topsoil of mainly loose silty SAND that is mostly alluvial in origin and increases in thickness from the upstream to the downstream. Specifically at the bridge site, the upstream had 1.5m thick alluvium while at the downstream 2.7m thickness of topsoil was encountered. Underlying the loose sand is a thick layer (about 15m) of soft to moderately stiff SILT with varying amounts of clay and sand. Table 1 shows a generalized soil profile at the bridge site. 2.2 Water Content and Atterberg Limits The natural water content of the loose sandy topsoil was found to be higher than the liquid limit indicating the potential for liquefaction or quick condition. This zone is therefore not suitable for any foundation. The average values of the natural moisture content, liquid limit and plastic limit suggest that the underlying residual soil could be some - to – heavily over consolidated (Bowles [2]). The average liquidity index (L L)

492

C.F.A. Akayuli et al. / Reducing the Cost of Road Construction

for these strata was found to be less than 1 which gives an indication that there is no possibility of liquefaction below the sandy topsoil. Table 1: Generalized soil profile at the bridge site DETH, LEGEND STRATA DESCRIPTION m Light brown, wet, loose, gravelly SAND. 1-3 NMC=18.7%, LL=17,5%, PL=14.6 Yellowish brown (mottled), wet, soft, SILT with inclusions of greenish gray clay and 3 – 15 varying proportions of sand and gravel NMC=27.2%, LL=41.0%, PL=11.4% Dark brown, moist, moderately dense, sandy 15 - 17 GRAVEL. (Moderately weathered phyllite.). NMC=11.1%, LL=42.4%, PL = 32.4%

N VALUES No readings 11 - 52

60-80

2.3 SPT and Bearing Capacity calculations The N values obtained from the SPT are shown in Table 2. Generally, high N values were recorded in BH2 than in BH1. Refusal (N>50) was recorded at 10m in BH2 while in BH1 refusal was at about 15.5m. N values were not recorded in the first 4.0m because the soil had turned to a liquid mass that did not support SPT. Table 2: Variation of N values with depth BH1 BH2 Depth

N

Depth

N

5.7

11

4.5

41

8

15

6.5

36

9.2

15

8

35

11

28

10.5

52

15

44

12.5

60

16.5

80

The N values were used to calculate the presumed bearing capacity using the appropriate set of equations developed by Meyerhof [3], [4] and adjusted by Bowles (1988). 2.4 Seismic refraction data analysis The depth and velocities of the refractors on site were calculated on the assumptions that: • Stratum to be investigated possesses higher velocity than overlying strata. • Velocity reversals do not occur above the layer of interest • Each consecutively deeper layer must possess a certain finite thickness related to velocity contrast and depth in order to be detectable • Each succeeding layer must be thicker than the one above For 2-layered strata, the thickness of the upper layer can be determined by the following relationship

C.F.A. Akayuli et al. / Reducing the Cost of Road Construction

݀ଵ ൌ

௑೎ ଶ

௏ ି௏

ൌ ට௏మ ା௏భ మ

493

(4)

భ

Xc = critical distance which is the distance from the hammer position to the geophone position where the travel times of the reflected and the refracted waves are equal, d1 = the thickness of upper layer, V1 = velocity of seismic wave in the upper layer, V2 = velocity of the seismic wave in the lower layer From the travel time curves the depth to the rock head which is the thickness of the soil stratum was recorded at both banks of the river and shown in Table 3. The seismic analysis indicates an overburden soil stratum of 7.6m thick resting on a variably hard stratum with a seismic velocity of 1565m/s. Borehole drilling confirmed this underlying material to be boulders that were used to fill the excavations for the bridge foundations before backfilling with soil. The seismic tests indicated the rockhead to be at a depth of about 14m below ground level with a margin of error of ±10%. This was confirmed by the percussion drilling. Table 3. Computed Depths and Velocities of Horizontal Strata at Hwereso Bridge Site Traverse ID G1 G2 G3 G4

No. of Layers 1 2 1 2 1 2 1 2

Layer Thickness (m) 7.6 4.6 5 -

Layer Velocity (m/sec) 1,565 6,000 2,909 324 5,000 750 3,600

3. Upgrade Redesign Considerations Due to the high N values recorded during the in situ testing, the original design of the bridge was reviewed to take advantage of the resulting high bearing capacities. From 4.5m depth and below, the N values increased steadily to refusal (N> 50 blows) at depths between 10 - 15m. From the N values it could also be said that the scour depth is at 4.5m as suggested by Kuhn and Williams [5]. The original road upgrade design provided for a bridge whose foundation would be on piles. However by this targeted geotechnical investigation it was seen that the top soil which was about 3.0m thick and composed of transported loose silty SAND and prone to liquefaction could be replaced with a more suitable material to provide a competent foundation for the bridge abutments without piling to reduce cost. The hydraulic redesign conducted by Cowi and Conterra [6], the design consultants, also showed that a box culvert could efficiently replace a bridge at the site. Based on the combination of the Geotechnical and Geophysical investigation results, it was recommended that the soil within the footprint of the box culvert be excavated to about 4.0m where the presumed bearing capacity was calculated as

494

C.F.A. Akayuli et al. / Reducing the Cost of Road Construction

681kPa. The void would then be backfilled with 150-300mm rock boulders and cobbles to a thickness of about 1.5m. The boulder foundation would then be overlain with rock aggregates of varying sizes and compacted before placing a 300mm mass concrete blinding over the subsoil rock fill to form a strong free draining foundation for the construction of the proposed three-bay box culvert. The redesign was easy to construct since the bridge was replaced by a culvert and piling was avoided. The cost was reduced considerably since culvert construction did not require any specialists Engineers from outside the country. Generally, also, the construction time for the road and hydraulic structures along this section was reduced by about 6 months with resultant savings in overall cost of the project. 4. Conclusion By carrying out a detailed geotechnical and geophysical investigation at the Hwereso valley, the soil profile to bedrock was clearly defined and the in-situ strength and liquefaction potential of the soil at the proposed bridge footprint were also determined. This information enabled the bearing capacity to be determined and soil improvements adopted for a redesign of the foundation to enable the construction of a 3-bay box culvert to replace a bridge that was proposed in the original upgrade design. This targeted geotechnical and geophysical investigations enabled the contractor to reduce cost in earthworks and piling and save construction time. References [1] Bowles, J. E.: Foundation Analysis and Design, 4th Edition McGraw-Hill Books Co. Singapore. 1004pp, 1988 [2] British Standards Institution BS1377: Methods of Test for Soils for Civil Engineering Purpose (Engineering Analysis and Modelling Vol. 1), 1975 [3] Kuhn S.H and Williams A. B: Scour depth and soil profile determination in river beds 5th ICSMFE vol. 1 pp 487-490 [4] Meyerhof G. G.: Penetration Tests and Bearing Capacity of Cohesionless Soils JSMFD, ASCE, vol 82, SM 1, pp 1-19, 1961 [5] Meyerhof G. G.: General Report: Outside Europe Proceedings. Conference on Penetration Testing, Stockholm, vol. 21 pp 40-48, 1956 [6] Cowi and Conterra: Engineering Report on consulting services for the rehabilitation of the Konongo – Kumasi Road. (Chapter 4: Hydrological studies and drainage design) January, 2003

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 495 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-495

Appropriate Engineering Solutions for Rural Roads in Mozambique Luis FERNANDES and Irene SIMOES Administração Nacional de Estradas, Maputo, Moçambique

Abstract. This paper presents the National Road Administration (ANE) Strategy for upgrading of Rural Roads in Mozambique, with the use of durable and definitive engineering solutions, for the provision of low-volume roads with special emphasis on the use of locally available materials and application of appropriate surfacings. The main aim of this programme is the development of documentation which includes a Guideline on Specifications and Work norms for Low-Volume Roads. The paper gives information on what has been trialled up to now and the expected outputs and outcomes. Keywords. Otta Seal, Gravel, Prime.

Introduction Mozambique has serious shortage of road building materials. Materials for the construction of bases are very poor in most areas. The bulk of eastern Mozambique is covered in fine coastal sands and natural base materials are hard to obtain. It is even more difficult to find surfacing aggregate and this is because rock is hard to find and when it is found it is usually too weak to be used for surfacing. Shortage of surfacing stone has caused serious financial constraints in road provision. Surfacing aggregate is hauled for long distances to projects and in some instances the haul distances are in excess of 500km. The landed costs are huge owing to the transportation costs which are more than five times the purchase costs. This may be necessary for the national roads but the same may not be viable for secondary and tertiary roads. It has become apparent that it is necessary to develop alternative solutions. Most of the quarry stone produced in Mozambique cannot be used successfully because it crushes and polishes under traffic loading and when found the haul distances are prohibitive. However, in the western parts some natural gravel can be found though in small quantities. In some areas in the southern part of Mozambique deposits of calcrete can also be found. The Rural Road Investment Programme (RRIP) is providing the country with an opportunity to use marginal materials for construction of various surfacing seal and to develop home grown specifications and work norms which will enable practitioners to use these materials effectively and sustainably.

496

L. Fernandes and I. Simoes / Appropriate Engineering Solutions for Rural Roads in Mozambique

1.

Prioritisation Criteria

The Technical Strategy for rehabilitating and upgrading regional roads was developed with the aim of gradually bringing ‘problematic’ roads towards a stable condition allowing the application of affordable routine maintenance regimes. The idea is to improve the impassable (or difficult to pass) sections of the road (so called ‘P1’ sections) and applying ‘targeted interventions’ (durable engineering solutions), in order to avoid the actual tendency to carry out repeated routine or periodic maintenance works on these sections, without producing a long-lasting solution. The selection of sites for interventions was based on the criteria given in table 1. Table 1. Passability Criteria

Passability Criteria

Description

Challenges

P1

Not passable even with a four wheel drive vehicle

To provide all weather access through sustainable interventions

P2

Passable with a four wheel drive but with difficulty

Improve the condition so that smaller vehicles can pass throughout the year, especially during the rainy season

P3

Passable throughout the year but in poor condition

Carry out normal routine maintenance

The prioritisation criteria have been broadly set as follows: ▪ ▪ ▪ ▪

2.

Important connectivity to the province, according to the ranking in the MultiCriteria Analysis (MCA); Must have P1 or P2 sections, where the road becomes intransitable in no more than 15% of its extension; Be in maintainable conditions through a normal routine maintenance regime; Having consumed large amount of money in the last years through a routine maintenance program, without achieving a substantial impact.

The Pilot Projects

In order to clearly define the appropriate engineering solutions for rural roads in the country, the Rural Road Investment Programme is being implemented through the execution of pilot project, divided into three distinct phases: Phase I, Phase II and Phase III. Up to now, about 26 projects with different solutions have been constructed. This programme has a research component, where data of the performance of the various sections built under the three phases will be monitored for a period of about 3 years, in the first phase, and the performance data will be collected. Obviously this monitoring period is not long enough but the data will be extrapolated for the longer term predictions and the monitoring will continue in order to verify and correct the predictions.

L. Fernandes and I. Simoes / Appropriate Engineering Solutions for Rural Roads in Mozambique

497

The data will be used to develop performance trends and deterioration characteristics and these will be co-opted in the development of specifications and work norms based on the outputs of the research. Under the Rural Road Investment Programme, one of the options being carried out is the design and application of Otta seal, so, in the next chapters will be given an overview on what has transpired on some projects where this surfacing option has been applied. 2.1

Inhacufera-Machaze Project

Inhacufera Machaze Project has been carried out in phases and details are given in the case studies of each phase given below. The project is remote, a distance of approximately 130km from the National Road N1. This section was targeted because it was for vehicles to pass due to high roughness resulting from the big boulders in the wearing course. The boulders were damaging vehicle tyres and transporters shunned this route. In order to minimise this problem it was decided by ANE to improve this section by reworking the surface, crush the boulders with a grid roller and recompact. For sustainability, it was decided to put a surfacing on top so that the intervention could provide a long lasting solution. There were sources with small quantities of quartzitic gravel but not enough to build a wearing course. It was decide that this gravel be processed and used for construction of Otta seals. The grid roller failed to crush the boulders effectively because they were very hard quartzite and the result was that the surface of the processed layer had protrusions of quartz boulders. This was not good for the surfacing because it meant that some of the Otta seal aggregate would sit on top of the boulders with no embedment. Such aggregate would not have much binder to hold it and would most likely strip off. It should be noted that the aggregate for the Otta seal on Inhacufera Machaze is coarse with nominal maximum particle size of 19mm. 2.1.1

Project scope

The project was executed in two phases. The first started in September 2008 and finished in March 2009 while the second one started on August 2009 and finished on June 2010. The section had previously been built as a gravel road with very coarse quartzitic gravel. The oversize material was very big and some stones were about 200 to 300mm. The road was very rough and almost impassable to small vehicles. The design involved reprocessing of the existing base and surfacing using Otta seal with MC3000 and natural graded quartzite sieved on 20mm and 5mm sieves. The excavation and sieving were carried out using labour based methods using local labour force. During phase 2 the design included scarification and regularisation of the existing road surface, a 150mm natural gravel layer consisting of quartzitic gravel from gravel pits located within the vicinity of the project site. The construction of the base and surfacing was carried out using machine based methods. The acquisition and processing of Otta seal aggregate was carried out using labour based methods. Some experimental sections have been incorporated which include:

498

L. Fernandes and I. Simoes / Appropriate Engineering Solutions for Rural Roads in Mozambique

1.

2.

2.1.2

Phase 1: a. single Otta seal, b. double Otta seal, c. single Otta seal with sand seal on a steep slope, d. 1 ½ Otta seal e. single Otta seal on primed base f. single Otta seal on unprimed base. Phase 2: a. single Otta seal at 2.0 l/m2 binder application rate, P0.075 < 10% b. single Otta seal at 2.0 l/m2 binder application rate, P0.075 ~ 13% c. single Otta seal at 1.8 l/m2 binder application rate, with washed aggregate d. single Otta seal at 1.9 l/m2 binder application rate, P0.075 < 10% e. single Otta seal at 1.8 l/m2 binder application rate, P0.075 < 10% f. single Otta seal at 1.7 l/m2 binder application rate, P0.075 < 10% g. single Otta seal at 1.6 l/m2 binder application rate, P0.075 < 10% h. single Otta seal at 1.4 l/m2 binder application rate, P0.075 < 10% i. sand seal at 1.0 l/m2 binder application rate. Observations

After four months of construction, the bulk of the section constructed during phase 1, was in good condition. However, there were short sections on the single Otta seal where stripping has occurred. The sections with 1 ½ and double Otta seal and single Otta seal with sand seal on top were intact. On the other hand, observations undertaken on the sections constructed during phase 2, showed that no defects have been noticed after 4 months of trafficking. The first 1.2 km was not trafficked for about 2 weeks. This section remained closed because there was a number of water crossing structure which had not been completed. The rest of the sections have been opened to traffic and to date the performance is very good on all sections. Particular attention was given to the section where the binder application rate was reduced to 1.4l/m2. There was not observed differences in visual appearance compared to sections where higher application rates of the binder were used. A crude test was carried out on this section and it involved hard braking a Toyota Land Cruiser at high speed and no rolling or stripping of the stone was observed. This section covers a steep slope, a gentle curve and a straight and flat stretch of road. It is anticipated that further monitoring may show that with high level of quality assurance it is possible to lower application rates of bitumen without adversely affecting the overall performance of the Otta seal surfacing. An additional experimental section of sand seal was incorporated. Sand seal has been used before in Mozambique. However, the quality of sand obtained in this area was poor. It was very fine, like fine beach sand. The sand was available locally in streams that are close to site. Both sections are still under curing process. The sand seal section will look black once all loose sand is removed from the surface.

L. Fernandes and I. Simoes / Appropriate Engineering Solutions for Rural Roads in Mozambique

2.2

499

Zero-Mopeia Project

Zero Mopeia Project is located in the Zambezi Valley in Zambezia Province. In this area the soils are naturally sandy but there are deposits of fine quartzitic gravels. The gravel was found in sufficient quantities for base construction and Otta seal surfacing. The nominal maximum size of the aggregate was between 9mm and 13mm. The content of fines was approximately 13% which slightly exceeded the specification of a maximum of 10%. Both projects are on low volume roads and the variation in aggregate size provided an opportunity to see how the Otta seals would behave. In addition, the application rates of the binder were varied on each of the projects in order to determine the minimum allowable application rate for these low volume roads. The project details are given below. 2.2.1

Project Scope

At the initial stage of the project under phase 1, the road was reshaped and gravelled. Some low lying sections which flooded during the rainy season were raised through construction of embankments. The whole 40km was covered under this phase. Under phase 2, the project involves the construction of base using locally available quartzitic gravel and Otta seal surfacing using sieved natural quartz aggregate which is also available locally. The aggregate for the Otta seal surfacing was prepared using labour based methods, i.e. the excavation and the sieving. The construction of the base and the surfacing was carried out using machine based methods. The design was based on the work norms [2] which are the referral Manual used by ANE. The manual stipulates application rates of 1.0 l/m 2 for the prime, 2.0 l/m2 for the binder and 1.6l/m2 for the application of aggregate. Some experimental sections have been incorporated which include: a. single Otta seal at 1.8 l/m2 binder application rate, P0.075 ~ 13% b. single Otta seal at 1.7 l/m2 binder application rate, P0.075 ~ 13% c. single Otta seal at 1.6 l/m2 binder application rate, P0.075 ~ 13% d. single Otta seal at 1.5 l/m2 binder application rate, P0.075 ~ 13% e. single Otta seal at 1.4 l/m2 binder application rate, P0.075 ~ 13% f. single Otta seal at 1.3 l/m2 binder application rate, P0.075 ~ 13% g. single Otta seal at 1.2 l/m2 binder application rate, P0.075 ~ 13% The nominal maximum size of the aggregate was between 9 and 13mm compared to 19mm given in the specifications. The percentage of fines P0.075 was greater than the maximum figure allowed in the specifications of 10%. Generally the aggregate was fine and did not meet standard specifications. The application rate for the prime was 0.6 l/m2. Most importantly, there exists a myth that single Otta seal doesn’t work in Mozambique and therefore most sections built previously are either double Otta seal or single Otta seal with a sand seal capping. This is a costly design for low volume roads. The original design prepared involved a single Otta seal but this had been changed to double Otta seal based on that myth. However, during construction, and in order to verify and monitor its performance, some sections have been constructed with single Otta seal.

500

L. Fernandes and I. Simoes / Appropriate Engineering Solutions for Rural Roads in Mozambique

Observations

2.2.2

There was no need for specific observations for phase 1, as on this phase the road was only reshaped and gravelled. On the other hand, observations undertaken on the sections constructed during phase 2, after 4 months of trafficking showed that no defects had developed. The bitumen has migrated to the surface and carriageway is beginning to look black especially on the wheel tracks. The same has occurred on all sections including the last kilometre where the application rate of the binder was 1.2 l/m2. There is currently no noticeable difference in appearance or performance between sections with high and low binder application rates. Some sections along the cross-sections of the carriageway have not been trafficked at all and this is owing to the low volume of traffic. Traffic has concentrated more on the centre of the road and on the inside of curves. The road is looking very good and the surfacing is intact. However, the centreline is a little crooked and the jointing was not very neat though the sealing is intact. There are some longitudinal lines on the surfacing on some sections which are an indication of uneven spray of binder across the width of the carriageway. It is not anticipated that this will have any significant effect on the performance of the surfacing. This project is likely to be influential in the review of specifications for Otta seal for low volume roads in Mozambique. A binder application rate of 1.5 l/m2 was used for most of the phase 2 sections and further reduction of application rates to 1.2 l/m2 led to significant savings. Particular attention was given to the section where the binder application rate was reduced to 1.2 l/m2 during inspection. It is actually difficult to distinguish this section from the rest. However, further monitoring will reveal performance aspects which will be instrumental in the development of Otta seal specifications for Mozambique.

3.

Conclusion

A few conclusions can be drawn from the observations described above. 1. 2. 3. 4.

It is apparent that finer aggregate leads to more rapid curing of the Otta seal than coarse aggregate at low traffic volumes. There are opportunities to use Otta seal more extensively in Mozambique and more work need to be done to find other sources and materials. The research component will help Mozambique to develop specifications which suit materials found in the country. Otta seal can reduce construction costs significantly if proper specifications and work norms are developed for the Mozambique situation.

References [1] [2] [3] [4]

O. Charles, A Guide to the Use of Otta Seals, Norwegian Public Roads Administration, Oslo, 1999 Administraçao Nacional de Estradas, Normas de Execução, Maputo, 2007 Overby& Pinard, The Otta Seal Surfacing - An Economic and Practical Alternative to Traditional Bituminous Surface Treatments, Botswana, 2007 Division of Roads and Transport Technology, The Design, Construction and Maintenance of Low Volume Roads and Bridges in Developing Areas, CSIR, Pretória, 1990

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 501 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-501

Preliminary Studies on the Utilization of Sand Treated With Emulsion Luis FERNANDES, Irene SIMOES and Hilário TAYOB Administração Nacional de Estradas, Maputo, Moçambique

Abstract. The objective of this paper is to present the preliminary studies undertaken to perform an appropriate mix and structural design of sands treated with emulsion. Its intention that the models developed in this study will be validated and provide an interim guideline for the design of emulsion treated sands as pavement layers, knowing that the main purpose of structural design is to obtain an objective rational estimate of the capacity of the pavement, with a certain level of confidence, to provide an acceptable service level without major structural distresses. For that purpose a project has been commissioned under the Rural Road Investment Programme (RRIP) where the use of emulsion is being trialled and researched for future use in low volume roads. Keywords. Soil treated with emulsion, sands.

Introduction Conventionally the treatment or stabilisation of bases has been carried out using cement or lime. Recently, the use of soil treated with emulsion (STE) has gone on the spotlight. STE involves the mixing of road bases with emulsion either cationic or anionic depending on the type of base. Some bases are acidic and require cationic emulsions and others are basic and require anionic emulsions. Mozambique has a long coastline which is about 2.000km and most areas close to the coast are covered in coastal sand. In many areas the coastal sands extend well inland and this is mainly the only soil found in these areas. This makes road provision a serious challenge for the National Road Administration (ANE), the practitioners and the financiers. Eight out of ten provinces are affected. Road construction materials are scarce and in most areas they are actually non-existent. ANE is left with very little option other than using the locally available sand for the construction of roads. Most unpaved roads are built with a sand wearing course which in most cases only lasts less than six months except in cases where traffic is very low where regravelling cycles could reach 1 to 1.5 years. This is brings into question the sustainability of Mozambique’s rural road network. Research was carried out by Transport Research Laboratory (TRL) in collaboration with ANE for a period of 4 years on Engineering Standards and LifeCycle Costing for Low Volume Unpaved Roads. This project revealed a serious performance deficiency of the unpaved rural road network. In brief the performance was found to be very poor with average gravel loss of about 70mm per year. Some roads had gravel loss figures above 100mm per year. This has affected ANE’s capacity to keep these roads in a maintainable condition. As a result the financing of

502

L. Fernandes et al. / Preliminary Studies on the Utilization of Sand Treated with Emulsion

maintenance has been a big challenge as well. The Road Fund which is the main source of funding for maintenance in Mozambique cannot cope with this maintenance demand. As part of the recommendations which emanated from the above mentioned research, it was necessary to surface rural roads in Mozambique with low cost surfacing. However, the question still remained regarding the source of base materials. Sources of good natural bases tend to be very far and haul distances are thus prohibitive. One of the viable options was to treat the sand to form the road bases. Cement stabilisation is the most common practice but with this comes with many challenges as well. Firstly, cement stabilisation of sand requires a high cement content to be effective, mostly about 5% and at time up the 7%. This is expensive but that’s not all. High cement content causes high prevalence of shrinkage cracking which in most cases leads to block cracking and lamination. Treatment with lime is mainly for clayey materials and this is not applicable in the Mozambique scenario. The focus nowadays is on STE as a viable alternative in the treatment of sand bases and the research is now being intensified to develop specifications for low volume roads.

1. Background The RRIP is a programme that was commissioned to design and implement targeted interventions of sections of road that posed passability bottlenecks and the objective was to provide all-weather passability to the rural communities. Many different interventions have been designed and implemented but the programme has provided an opportunity to carry out further research on innovative designs and work methods or work norms which is being spearheaded by TRL. One of the roads targeted for intervention is Beira Savane Road. Beira is a coastal town which is situated in the central region of Mozambique and it is also a vital port for Mozambique and neighbouring countries in the region like Zimbabwe, Zambia and Malawi. Beira Savane Road traverses the coastal flood plains from Beira Town to Savane Resort and services some rural communities along the way. Tourism is one of the major sources of income in Mozambique hence all weather passability is a necessity. The coastal plains flood every rainy season and drainage is a big problem. The water in the flood plains just rises and more so when the sand become saturated. There is no straightforward design or even estimation of the flooding because it depends on the amount of rainfall received just like filling up a jar. The sections of the road that are more prone to flooding were identifies and it was decided the raise the existing road through construction of high embankments. The project was carried out in two phases. Phase 1 involved the design and construction of embankments including the construction of culverts. Apparently the design of culverts was a little unique in that the inlets and outlets are designed to be at the same level so that water can flow either way depending on which side of the road receives more rain. Phase 2 involves the design and construction of further increases in the height of embankments, untreated sand subbase and emulsion treated sand base and low cost surfacing.

L. Fernandes et al. / Preliminary Studies on the Utilization of Sand Treated with Emulsion

1.1

503

Project Scope

The project was selected to be a major focus for the RRIP and especially the research component. Its proximity to the central city of Mozambique makes it a project of interest for ANE and decision makers of interested parties. The main objective of this project is to develop specifications and work norms for use of coastal sands in road building. The project involves a number of aspects. 1. 2. 3. 4.

Evaluation of the targeted sections Selection of type and method of intervention Assessment of materials and subsequent laboratory tests Geometric design of the road structure i.e. the embankment, the slopes and the carriageway 5. Preparation of drawings 6. STE design 7. Selection of surfacing options for each section 8. Design of surfacing 9. Preparatory works including tendering 10. Earthworks 11. Construction of the emulsion treated sand base 12. Construction of untreated sand base 13. Construction of surfacing (slurry seal, sand seal, double surface dressing, single surface dressing) 14. Monitoring for performance assessment and evaluation 15. Preparation of specifications and work norms 16. Development of maintenance strategy It was difficult to prepare a design to mitigate the problem of flooding because all common methods of estimating flooding do not apply in this situation where water simply rises in the flood plain. It is mainly dependent on the amount of rainfall and the duration of the rainy season. Local knowledge influenced the decisions on the design of the embankments and subsequently the geometric design. The design for the STE is a little complex. There are two approaches to it. a. The maintenance layer design b. The structural design The first design option assumes adequate structural strength is provided for in the pavement layers underlying the STE and that the STE provides a capping layer for the pavement that is less sensitive to moisture. The second approach assumes the proper functionality of a base layer that it should have adequate strength to disperse the load. More details of the STE design are shown in figure 1.

504

L. Fernandes et al. / Preliminary Studies on the Utilization of Sand Treated with Emulsion

Figure 1. Schematic Design of STE

2. STE Design Construction on Beira-Savane mainly involves the construction of test sections of STE. The whole project involves raising the existing embankment to prevent flooding and the construction of subbase with sandy soil. The base is designed to be an emulsion treated sand base (STE). Untreated sand base will be used on one of the sections. The designs include the ‘maintenance layer’ design and the ‘structural layer’ design. The surfacing will include slurry seal, sand seal, single surface dressing and double surface dressing on both the treated base and untreated base.

3. STE Construction There are two main approaches in the construction of STE especially for low volume roads. 1. In-situ mixing 2. Plant mixing The preparation of STE involves the mixing of emulsion and base material. Emulsion is a suspension of bitumen droplets in water aided by an emulsifying agent. It is important to ensure that the bitumen droplets are well dispersed in the base material to province an effective stabilisation. Like any normal compaction where the compaction moisture plays a critical role in the construction process emulsion treated bases behave in the same manner. During the lab design of the STE, the optimum moisture content of the unstabilised material is

L. Fernandes et al. / Preliminary Studies on the Utilization of Sand Treated with Emulsion

505

determined using the mod AASHTO test. During construction of STE it should be assumed that the optimum fluid content (OFC) should be equal to or slightly less than the OMC. It is important that the OFC is controlled on site otherwise good compaction of the STE may not be achieved. 3.1

In-situ mixing

In-situ mixing is generally problematic in that it takes great effort to achieve a homogeneous mix and sometimes it’s never good enough. Usually this can be carried out using a grader and disc harrows. It is not an easy process as emulsion becomes sticky after it breaks i.e. the bitumen separates with the water. The process involves haulage and dumping of the soil, spreading and mixing. The emulsion is then applied and mixing with disc harrow follows. The final mixing and spreading is carried out by the grader and then rolling follows. A vibratory steel roller or a combination of the vibratory steel roller and a pneumatic roller may be used the latter being a better option especially for if the base is predominantly sand. Curing follows and it is recommended to leave the section closed to traffic for a period of 3 to 7 days but in cases where detours are not provided for and the soil is granular trafficking of the completed section can be immediate. 3.2

Plant Mixing

Plant mixing is the most recommended approach. Plant mixing produces a homogenous mix which leads to better performance. For large scale operations large automated mixing plants can be used. For small works concrete mixers can be adequate. The process involves feeding the soil and the emulsion into the plant and setting the mix proportion stipulated to achieve the required bitumen content (usually between 4% and 6%). After mixing, the STE is transported and dumped, spread and compacted. It is important to determine the breaking time of the emulsion after mixing and use of stable grades of emulsion are recommended for situations where the process is likely to take long. If the bitumen content is high the layer behaves in the same manner as cement stabilised bases and application of prime should be lighter than normal. A fog spray or a maximum of 0.6l/m2 should be applied. Surfacing is applied normally. The STE cannot be used as a wearing course. On the Beira-Savane Project the STE will be produced through plant mixing. The contractor has a concrete mixing plant in Beira which is approximately 20 km from site. 8 m3 trucks are available for the transportation and the mixer can produce 8m 3 every 20 minutes. Every load will cover approximately 13 meters. SS60 anionic emulsion will be used for the STE mix design. The coastal sand that will be stabilised has a very fine grading and it’s foreseen that the amount of stabiliser will be higher than usually used in Southern Africa. Construction is underway and a substantial part of the earthworks have been completed up to subbase level. The laboratory design of the STE is also underway. In the mix design process, the bitumen content will be varied from 2% to 6% and the amount of cement will not be more than 1%.

506

L. Fernandes et al. / Preliminary Studies on the Utilization of Sand Treated with Emulsion

4. Conclusion The ongoing research is very important to ANE and the road sector in general and the outputs are eagerly awaited. If successfully accomplished with the necessary specifications and work norms properly documented in a manual, the benefit to the road sector will be significant even for the high volume roads.

References [1] [2] [3]

Administração Nacional de Estradas, Normas de Execução, Maputo, 2007 Overby& Pinard, The Otta Seal Surfacing - An Economic and Practical Alternative to Traditional Bituminous Surface Treatments, Botswana, 2007 Division of Roads and Transport Technology, The Design, Construction and Maintenance of Low Volume Roads and Bridges in Developing Areas, CSIR, Pretoria, 1990

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 507 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-507

Geosynthetics in Road Pavement Reinforcement Applications Garth JAMES, Kaytech, Durban, South Africa

Abstract: This paper references recent research and highlights applications of geosynthetics for reinforcement of road pavements. The applications include subgrade stabilisation and basal reinforcement where either or both of the separation and reinforcement function of the geosynthetic are fulfilled. The primary mechanisms of reinforcement in operation for sub-grade stabilisation include lateral restraint or confinement, increased bearing capacity and the tensioned membrane type support. Pavement performance and benefit from the reinforcement layer in full-scale tests are mentioned. Basal reinforcement is used for permanent paved roads and is typically applicable for low volume roads founded on weak sub-grade where the geosynthetic is placed between or within the sub-base or base layers of the pavement structure provides lateral restraint. The behaviour and benefits of geosynthetics, particularly geogrids, in these applications is mentioned. For purposes of design and specification, geosynthetic properties and the tests needed to define those properties are of great importance. Some of these properties and test methods are presented with recent innovations in geosynthetic products taking design to an approach of performance based specifications rather than the standard property based specifications.

Keywords: geosynthetics, reinforcement, separation, roads, pavements, sub-grade, basal, stabilisation, reinforced separation

Introduction The importance of the properties of geosynthetics and the standard tests available to define their properties for design, specification and application cannot be underestimated. The engineer has to understand the functions the geosynthetic needs to perform, particularly reinforcement and separation for road pavement reinforcement. These properties, mechanisms and test methods are reviewed, with some specifically developed to suit the intended application. This paper covers sub-grade stabilisation and basal reinforcement. Innovations in reinforcing geosynthetics are noted. This stateof-the-art review is based primarily on the collaborative work done by Perkins et al, 2010, [1]. 1. Sub-grade stabilisation To provide a stable working platform for construction of unpaved roads or more flexible or rigid pavement systems geosynthetics are often placed on the weak, often

508

G. James / Geosynthetics in Road Pavement Reinforcement Applications

saturated sub-grades and covered with a granular layer to stabilize the sub-grade. This granular layer may form the base or sub-base component of the road pavement system. These soft, sub-grade soils may have some strength in compression but are weak in tension thereby providing very little lateral restraint. When a granular layer is allowed to shove laterally rutting of the aggregate surface occurs. To provide tensile resistance to the lateral movement a geogrid with good interlocking characteristics or a geotextile with good frictional capabilities may be used to provide reinforcement. More recently these characteristics are combined in the form of composite reinforcement geotextiles or geogrids. Another important function of the geosynthetic playing an equally critical role in this application is separation to prevent the mixing of fines from the sub-grade into the granular material. Just a small fraction of fines intruding into the granular layer will have a negative effect on its structural integrity (i.e., reduced shear strength and lower permeability). Geotextiles or composite geosynthetics provide this separation function. Geogrids have a more open lattice structure and although they prevent the penetration of the coarser aggregate into the sub-grade layer they cannot prevent the intrusion of fines into the granular layer. A layer of geotextile in combination with a geogrid or a composite reinforcing geosynthetic offer greater benefit. Because the sub-grade materials in this application comprise fine grained soil with a high water content the geotextile, whether the primary stabilisation geosynthetic or a separation layer used with a geogrid, must also provide filtration to allow excess pore water pressure to dissipate into the aggregate base course and, in cases of poor-quality aggregate, through the geotextile plane itself. In the case of geogrids without a geotextile separation layer the granular layer supported by the geogrid must have a grading that is compatible with the sub-grade, based on standard granular filter criteria. It is the reinforcement, separation, filtration and drainage functions that combine to provide the mechanical stabilisation for weak, sub-grade soils [2]. Geosynthetics are primarily used in sub-grade stabilisation applications to facilitate construction. Geosynthetics offer a cost-effective alternative to other expensive foundation stabilisation methods such as dewatering, excavation and replacement with select granular materials, utilization of thicker stabilisation aggregate layers, or chemical stabilisation. The stabilisation application is primarily used for initial construction, but geosynthetics also provide long-term benefits and improve the performance of the road over its design life. The geosynthetic continues to perform by maintaining the base course material integrity by preventing the aggregate from penetrating the sub-grade. In addition, the separation function provided by geotextiles, geogrid/geotextile geocomposites or geogrids with appropriately designed filter aggregate prevent the migration of fines into base/sub-base materials, especially into open graded bases, maintaining the support and drainage characteristics of the base over the life of a pavement system. 1.1 Mechanisms of reinforcement The two primary mechanisms for sub-grade stabilisation are increased bearing capacity and lateral restraint, both effectively contributing to the load carrying capacity of the reinforced layer. When this aggregate layer is loaded by a vehicle, the aggregate is inclined to shove laterally and is restrained by the sub-grade (if it has strength) or

G. James / Geosynthetics in Road Pavement Reinforcement Applications

509

geosynthetic reinforcement, a pavement reinforcement mechanism is thus induced (Figure 1a). The components of this mechanism include restraint of lateral movement of the base aggregate (i.e. confinement); an increase in modulus of base aggregate due to the confinement; an improved vertical stress distribution on the sub-grade due to the increased base or sub-base modulus; and reduced shear strain along the top of the subgrade. Also, the geosynthetic reinforcement forces the potential bearing capacity failure surface below the wheel load, which is just like a footing foundation on soil, to follow an alternate higher strength path as shown in Figure 1b. This tends to increase the bearing capacity of the sub-grade soil. A third possible geosynthetic reinforcement function is a tensioned membrane-type support of the wheel loads, as shown conceptually in Figure 1c. For this to occur, the wheel load stresses must be large enough to cause significant rutting through plastic deformation of the sub-grade. By having a sufficiently high tensile modulus, tensile stresses will develop in the geosynthetic reinforcement, the vertical component of this tensioned membrane stress will help support the applied wheel loads. Tensile stress can only develop within the geosynthetic reinforcement when some elongation takes place therefore wheel path rutting (in excess of 100 mm as determined by Giroud and Noiray method) is required to develop membrane-type support [3]. Less than 75 mm the effect is negligible giving reason to state that the tensile properties of the geosynthetic reinforcement are not the governing properties. If they were then high strength woven geotextiles would outperform geogrids. Properties other than tensile strength should be used in assessment and design [4]. This mechanism is generally limited, therefore, to temporary roads or the first aggregate lift in permanent roads, where significant rutting can be tolerated. Where the sub-grade conditions are stronger or where additional aggregate layers are placed the influence of the geosynthetic reinforcement decreases. Likewise, the effect of the reinforcement increases with increasing acceptable deformation (rutting). As the thickness of the aggregate layer increases or stiffer components of the pavement section are added, the stress at the geosynthetic decreases to a point where there is little or no geosynthetic deformation and correspondingly little or no reinforcement is required. The actual thickness where this occurs is related to the subgrade strength, the type and magnitude of the wheel load and the number of vehicle passes. Thus design solutions should evaluate each of these elements. Research indicates that stabilisation for construction is generally less likely to be required for sub-grade soils with a soaked CBR value > 3 to 4, shear strengths > 90 to 120 kPa, and resilient modulus > 30 to 40 MPa. From a foundation engineering point of view, clay soils with undrained shear strengths of 90 kPa or higher are considered to be stiff clays [5] and are generally quite good foundation materials and will readily support reasonable truckloads and tyre pressures, even under relatively thin granular bases. Note: the reinforcing function can be compromised if separation and filtration are not provided. Case histories have documented poor performance of geosynthetic reinforcements where the separation function was not achieved [6].

510

G. James / Geosynthetics in Road Pavement Reinforcement Applications

Wheel Load

Lateral Restraint of Geosynthetic

Geosynthetic

(a): LATERAL RESTRAINT

Wheel Load

Hypothetical Shear Surface with Geosynthetic

ǭ

Shear Surface Probable Geosynthetic without

Geosynthetic

(b) BEARING CAPACITY INCREASE

Wheel Path Rut

Geosynthetic

Wheel Load

Membrane tension in

Geosynthetic Vertical support Component of Membrane (c) MEMBRANE TENSION SUPPORT Figure 1: Possible reinforcement functions provided by geosynthetics in roadways. (a) lateral restraint, (b) bearing capacity increase, and (c) membrane tension support [2], [8].

1.2 Full Scale Performance

The use of geotextiles in sub-grade stabilisation to solve problems encountered when constructing unpaved and paved (both flexible and rigid pavements) roads over wet, soft, sub-grades was already well established internationally in the 1970s. The performance of geosynthetics used in stabilisation applications in low volume roads have been well documented in numerous case histories, full-scale laboratory experiments, and instrumented field studies, [7], [8], [9], [10], and [11]. Summaries of some of this research are contained in references [12], [13] and [14]. The results of these studies vary in terms of the performance of different geosynthetic types. For example, in some studies geogrids were found to perform better than geotextiles [15]; in some studies, geotextile and geogrid performance has been

G. James / Geosynthetics in Road Pavement Reinforcement Applications

511

found to be essentially the same [8]; in others, geotextiles were found to perform better than geogrids [10]; and in all cases, where composite geogrid/geotextile systems were used, they always performed the best [8], [10], and [11]. These varying results may be due to pore water pressure development in the sub-grade and the ability for the pore water pressure to dissipate during loading [16]. Full scale laboratory tests performed to evaluate geosynthetics used in both stabilisation and basal reinforcements on a number of different geosynthetics in several separate studies [10], [16], and [17] have observed the development and increase of pore water pressure measured in very wet, nearly saturated sub-grade during cyclic loading. The pore water pressure measurements in most of the tests were found to directly correspond to the performance of the geosynthetic. The largest amount of deformation per cycle was found to occur in the control tests with the highest developed pore pressure and the best performing tests were in the sections with the lowest measured pore pressure. The full scale studies found that the reinforcement action of an open geogrid positively results in lower pore water pressure development than measured in control tests (i.e., with no geosynthetics) performed on the same subgrade. The addition of a nonwoven geotextile to the reinforcement geogrid provides additional separation and filtration features that further limit the development of excess pore water pressure and significantly reduce rutting. These results also indicate that geotextiles with better filtration and drainage characteristics tend to perform better with wet silt and clay type soils than geotextiles with low permeability or permittivity as well as open geogrids, where separation performance was questionable. Geosynthetics influence the development and magnitude of pore water pressure in the sub-grade through a reduction in stress in the sub-grade [12]; separation, reducing point loading and corresponding pore pressure developed from granular material penetration into the sub-grade layers [10]; and/or, pore pressure dissipation in the plane of composite geosynthetics when the in plane permeability thereof is greater than the permeability of the adjacent sub-base or base layer (e.g., poorly draining base layers containing fine grained soils) [2]. 1.3 Geosynthetic Material Properties and Tests The geosynthetic material properties required for design are based on the properties required to perform the primary and secondary function(s) of reinforcement, separation, filtration and drainage for sub-grade stabilisation over the life of the system, and the properties required to survive installation. The separation and filtration functions are related to the opening size characteristics and are based on the gradation of the adjacent layers. Strength is required, of course, for the reinforcing function, which is based on the requirements in the specific design approach. For correct design and for the geosynthetic reinforced road pavement to work the stress at the top of the sub-grade due to the weight of the granular layers and the traffic loading should be less than the bearing capacity of that soil times a safety factor. The stresses and strains applied to the sub-grade and the reinforcing geosynthetic during construction are generally greater than those applied during in-service life, therefore, the strength of the geotextile or geogrid in road pavement and embankment applications is usually governed by the anticipated construction stresses and the

512

G. James / Geosynthetics in Road Pavement Reinforcement Applications

required level of performance i.e. survivability – the geosynthetic must survive the construction operations for it to perform its intended function. The geosynthetic survivability tends to control the strength requirements and not the reinforcement function. In the US, the Federal Highway Administration (FHWA) [2] and the American Association of Highway and State Transportation Officials (AAHSTO) [18] provide tables specifically for stabilisation applications that relate geosynthetic index properties defined by ASTM test methods (i.e., grab strength, CBR puncture resistance, and tear resistance for geotextiles; and, wide width strength and strength for geogrids) to survivability of geosynthetics to match specific installation conditions (low, moderate and high). Other properties, such as stiffness, aperture size and interlock effect, may be required for the specific design method. 2. Basal Reinforcement Basal reinforcement refers to the placement of reinforcing geosynthetics within the unbound aggregate sub-base and/or base layers of a paved flexible pavement for the purpose of improving the permanent deformation (rutting) and fatigue cracking performance of the pavement during its operational life. Traffic load applied to a flexible pavement results in dynamic stresses within the various pavement layers. The stiffness of these layers dictates the magnitude of the dynamic strains and displacements, which are expected to be small in a well designed pavement. The application of repeated traffic loads induces permanent strain which accumulates as traffic passes grow, leading to rutting of the pavement surface. Low-strain stiffness is important for characterizing the geosynthetic reinforcement. 2.1 Mechanisms of Reinforcement The overriding mechanism for basal reinforcement is where the reinforcement prevents lateral movement of the base aggregate through shear interaction between the aggregate layer and the geosynthetic. The effect is an increase in the confinement or mean stress in the aggregate adjacent to the geosynthetic. Aggregate materials have a resilient modulus that is mean stress dependent, meaning that as the confinement increases, the stiffness of the aggregate increases. Lateral restraint of the aggregate occurs during both construction of the road pavement layers and during traffic loading. During construction, heavy compaction equipment applies large compressive stresses to the granular layers. When geosynthetic reinforcement is present, lateral restraint of aggregate occurs through shear interaction between the aggregate and the geosynthetic as the aggregate experiences lateral extensional strain which occurs principally in the longitudinal direction during compaction, therefore the stiffness of the reinforcement in this same direction is of most importance. During traffic loading when extensional strains are greatest in the transverse direction, reinforcement stiffness in this direction will be more important (Figure 2). Reinforcement may also be effective in preventing the release of locked-in horizontal stress during vibratory compaction.

G. James / Geosynthetics in Road Pavement Reinforcement Applications

513

Figure 2. Mechanisms of reinforcement [1]

Extensional lateral strains (εh) are created in the base below the applied load as base material moves down and out away from the load centreline. Once lateral restraint of the aggregate occurs and a stiffening of the aggregate ( σh) is observed, the pavement layers respond by mechanisms further illustrated in Figure 2: (i) Compressive stress (σv) is reduced on the sub-grade. (ii) Shear stress (τ) transmitted from the base course to the sub-grade decreases as shearing of the base transmits tensile load to the reinforcement. (iii) Less shear stress, coupled with less vertical stress, results in a less severe state of loading [19] leading to lower vertical strain (εv) in the sub-grade. (iv) Finally, reduced vertical strain (εv) in the base and sub-grade results in less surface deflection, which in turn results in less dynamic tensile strain in the bottom of the asphalt overlay thereby giving a greater fatigue life. 2.2 Full-Scale Performance Construction, traffic loading and monitoring of full scale base-reinforced test sections have been on-going for nearly 30 years taking many forms ranging from laboratorybased test sections to demonstration projects constructed on public roads [1]. Several synthesis reports have been prepared [12] and [13] which describe work performed in this area. Paved roads are considered to have failed once large surface deformations are evident. A number of studies have demonstrated that the service life of the pavement, as defined by the number of load repetitions carried by the pavement to reach a particular permanent surface deformation, can be increased by a factor ranging from just over one to in excess of 100 by the inclusion of a geosynthetic reinforcement in the base aggregate layer. Studies have also shown that base course thickness can be reduced by up to 50 % by the inclusion of geosynthetic reinforcement. Most studies have quantified benefit in terms of rutting.

514

G. James / Geosynthetics in Road Pavement Reinforcement Applications

2.3 Geosynthetic Material Properties and Tests Geosynthetic material properties and their corresponding test methods are critical for design and specification of reinforcement geosynthetics for base reinforcement. This is especially true for generic specifications and non-proprietary design methods. It is generally accepted that the tensile stress-strain properties of the geosynthetic itself and geosynthetic-aggregate shear interaction properties are important for assessing the performance of the basal reinforced pavement. This implies that good interaction properties are necessary to transfer load from the aggregate to the geosynthetic reinforcement and that good stress-strain properties of the geosynthetic are required to limit lateral movement of the aggregate. Stress-strain material properties and interaction behaviour are important but to correlate a single material property with reinforcement performance has its problems and no single index property can define performance. Properties work together or against each other in determining how the reinforcement will benefit the pavement. For this application small strains are involved. Material properties may be broadly classified as those needed for design and those needed for specification. The advent of multi-axial grids has brought in properties other than tensile strength into assessment and design. Interlocking with the aggregate is the main mechanism of unpaved and paved road improvement (Figure 3). More complete assessment may include other geogrid properties in addition to or instead of aperture stability modulus, e.g., junction strength, aperture size and aspect ratio, rib thickness and profile and rib stiffness. For triaxial geogrids interlocking plays an important role because the triangular shape of the apertures is the best possible match for the hexagonal arrangement of dense aggregate. Therefore interlocking is of greater importance in triaxial geogrids compared to biaxial geogrids. For triaxial geogrids emphasis in design and calibration should be on aperture and rib properties, i.e. mechanical properties are of less importance in triaxial geogrids than in biaxial geogrids.

Figure 3: Interlock of aggregate for triaxial geogrids giving radial stiffness In current design methods the two dimensional situation is often assumed. This is realistic if the traffic loading is channelized and deep ruts have developed. If this is not

G. James / Geosynthetics in Road Pavement Reinforcement Applications

515

the case then the situation is closer to axisymmetrical. In the axisymmetrical situation geogrids perform better if they are isotropic. Triaxial geogrids have radial stiffness as opposed to biaxial geogrids thereby fulfilling a greater degree of isotropy (Figure 3). Two dimensional or biaxial geogrids have flatter, wider ribs thus a high surface area thereby maximising interface friction, but limiting potential interlock between the aggregate and the biaxial geogrid. Biaxial geogrids also exhibit little or no interaction between the aggregate and the sides or profile of the ribs, due to the ribs low lateral surface area. 3. Specification The currently accepted method-based or property specification is a ‘recipe book’ approach for materials, dimensions and installation requirements for most aspects of pavement construction. This approach has imposed consistency and protected construction standards in the past. However, in their Performance Specifications Strategic Roadmap: a Vision for the Future (2004) the US Federal Highways Administration points out that method-based specification ‘could not deal with rewarding a contractor for “better-than-minimum practice”;’ and ‘...could not consistently deal with work that was outside the bounds of “reasonably close conformance”.’ Method specification ‘inhibited innovation’ and there is no incentive for a project manager to consider ‘departures from standards’ and ‘departures from procedure’. If the proposed construction method lies outside the standard ‘recipe’, then performance data based on in-situ performance testing by accredited third parties and actual on-site trials must be used to verify the expectations of the design or method under scrutiny, (commonly pavement layer surface modulus and performance under construction traffic loading) [20].

4. Conclusion The use of reinforcing geosynthetics in the stabilisation of weak, soft sub-grades for unpaved roads and reinforcement of sub-base and base layers is well established in civil engineering, yet a full understanding of the mechanisms of reinforcement for both applications need to be fully understood. The main benefits of using geogrids in basal reinforcement include a reduction in layer thickness; increased bearing capacity providing an increase in the load carrying capacity of roads; increased service life of the road; the control of differential settlements and spanning voids or very weak deposits in areas prone to internal erosion, dissolution, collapse or subsidence [21]. This paper has drawn heavily on the experiences of Perkins et al [1] and the references listed below to give the reader the best possible review of these mechanisms of reinforcement and best practice going forward. Many case studies, full-scale tests and trials conducted to endorse this best practice are referenced. It is important to understand the behaviour of these reinforcing geosynthetics in relation to the layers or pavements they are intended to reinforce to be able to design and specify the best fit reinforcing geosynthetic to the intended project application.

516

G. James / Geosynthetics in Road Pavement Reinforcement Applications

New triaxial, reinforcing geosynthetics are taking the normal 2-D design approaches to new levels pushing the engineer into a more performance driven specification based on full-scale trials and tests [20]. References [1]

[2]

[3] [4] [5] [6]

[7]

[8] [9]

[10] [11]

[12]

[13] [14]

[15]

[16]

[17]

[18]

[19]

Perkins, S.W. (presenter), Christopher, B.R., Thom, N., Montestruque, G., Korkiala-Tanttu, L. and Want, N., Geosynthetics in Pavement Reinforcement Applications, Keynote lecture, Proceedings of the 9th International Conference on Geosynthetics, 2010, Guarüja, Brazil, pp 165-192. Holtz, R.D., Christopher, B.R., and Berg, R.R. 2008. Geosynthetic Design and Construction Guidelines Participant Notebook, NHI Course No. 13213; FHWA Publication No. FHWA HI-95-038 (revised), Federal Highway Administration, Washington, DC. Giroud, J.P. & Noiray, L., 1981. Geotextile-Reinforced Unpaved Roads, Journal of the Geotechnical Engineering Division, American Society of Civil Engineers, Vol. 107(9), pp. 1233-1254. Giroud, JP, 2010, Reinforcing Pavement Layers in Roads using Triaxial Grid, Training Lecture, 9th International Conference on Geosynthetics, Guarüja, Brazil. Terzaghi, K. & Peck, R.B. 1967. Soil Mechanics in Engineering Practice. John Wiley and Sons, New York. US Army Corps of Engineers 1999, Test Pilot Study – Chemical Demilitarization Alternative Technology Program (Alt-Tech) Aberdeen Proving Grounds (Edgewood Area), Maryland, U.S. Army Engineer District, Baltimore, February, 93p. Steward, J., Williamson, R. & Mohney, J. 1977. Guidelines for Use of Fabrics in Construction and Maintenance of Low-Volume Roads, USDA, Forest Service, Portland, OR. Also reprinted as Report No. FHWA-TS-78-205. Fannin, R.J. & Sigurdsson, O. 1996. Field Observations on Stabilization of Unpaved Roads with Geosynthetics, Journal of Geotechnical Engineering, Vol. 122(7), pp. 544-553. Watts, G.R.A., Blackman, D.I. & Jenner, C.G. 2004. The Performance of Reinforced Unpaved Subbases Subjected to Trafficking, Proceedings of the Third European Geosynthetics Conference, Munich, Germany, pp. 261 – 266. Christopher, B.R. & Lacina, B. 2008. Roadway Subgrade Stabilization Study, Proceedings of the Conference GeoAmericas 2008, Cancun, Mexico, pp. 1013 -1021. Christopher, B.R. and Perkins, S.W. 2008. Full-scale testing of geogrids to evaluate junction strength requirements for reinforced roadway base design, Proceedings of the Fourth European Geosynthetics Conference, Edinburgh, United Kingdom, International Geosynthetics Society. Berg, R.R, Christopher, B.R. & Perkins, S.W. 2000. Geosynthetic Reinforcement of the Aggregate Base/Sub-base Courses of Pavement Structures, GMA White Paper II, Geosynthetic Materials Association, Roseville, Minnesota, USA, 176 p. Christopher, B.R., Berg, R.R & Perkins, S.W. 2001. Geosynthetic Reinforcement in Roadway Sections. NCHRP Project 20-7, Task 112, TRB, National Research Council, Washington DC. Watn, A., Eiksund, G., Jenner, C. & Rathmayer, H. 2005. Geosynthetic Reinforcement for Pavement Systems: European Perspectives. Geotechnical Special Publication, 130-142, Geo-Frontiers 2005, American Society of Civil Engineers, Reston, VA, pp. 3019-3037. Barksdale, R.D., Brown, S.F. & Chan, F. 1989. Potential Benefits of Geosynthetics in Flexible Pavement Systems, National Cooperative Highway Research Program Report 315, Transportation Research Board, Washington, D.C., 56 p. Christopher, B.R., Perkins, S.W., Lacina, B.A. and Marr, W.A. 2009. Pore Water Pressure Influence on Geosynthetic Stabilized Subgrade Performance, Proceedings of the Conference Geosynthetics 2009, Salt Lake City, Utah, USA. Perkins, S.W., Christopher, B.R., Cuelho, E.L., Eiksund, G.R., Hoff, I., Schwartz, C.W., Svanø, G., & Watn, A. 2004. Development of Design Methods for Geosynthetic Reinforced Flexible Pavements. U.S. Department of Transportation, Federal Highway Administration, Washington, DC, FHWA Report Reference Number DTFH61-01-X-00068, 263p. AASHTO 2006. Standard Specifications for Geotextiles – M 288, Standard Specifications for Transportation Materials and Methods of Sampling and Testing, 26th Edition, American Association of State Transportation and Highway Officials, Washington, D.C. Houlsby, G.T. & Jewell, R.A. 1990. Design of Reinforced Unpaved Roads for Small Rut Depths. Proceedings of the Fourth International Conference on Geotextiles, Geomembranes and Related Products, Balkema, The Hague, The Netherlands, Vol. 1, pp. 171-176.

G. James / Geosynthetics in Road Pavement Reinforcement Applications

517

[20] Hall, C. and Cashman, D., Plotting the Road Map for Performance Specifications, Tensar International Publication, Version 2, 2010. [21] Jaros, M.B., James, G.M. & Gewanlal, C. 2009. Multi-layer Geosynthetic-reinforced Embankment over Potential Sinkholes for a Rapid Rail Link in South Africa, GeoAfrica, The First African Regional Conference of the International Geosynthetics Society, Cape Town, South Africa, Paper 336.

518 Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-518

Treatment and Stabilization of the National Road E.N. 379-1 Hillsides, Between Outão and Portinho da Arrábida Jorge DINISa , João PINAa, Baldomiro XAVIERa a Teixeira Duarte, S.A. Abstract. The increase in the occurrence of rock mass destabilization throughout national road EN 379-1’s hillsides, between Outão and Portinho da Arrábida, led to its closing to traffic. In order to confer the necessary safety conditions to the road, it was necessary to implement stabilization and protection solutions. The aim of this presentation is to relate the construction solutions adopted on the road and hillsides and to describe the difficulties that occurred during the execution of the works. On the road, concrete structures were essentially executed, such as stabilization walls and protection false tunnels. On the hillsides, the work mostly consisted in the installation of dynamic protection barriers against falling rocks and protection nets, sometimes reinforced with steel cables and soil nailing.

Introduction As a result of forest fires in Arrábida Natural Park, the conditions for stability of the hillsides adjacent to National Road EN 379-1, between Outão and Portinho da Arrábida (Setubal-Portugal) were strongly affected in an extension of nearly 4 km. The detachment of rocks has intensified (Fig. 1). Since this placed in jeopardy the safety of people and assets, it led to the closure of this section. Furthermore, all efforts were made in order to provide the road with the safety conditions necessary for its use.

Figure 1 –Some of the risk situations that led to the enclosure of the road: (a) falling blocks, (b) overhanging massifs.

The management of the entire process was entrusted to E.P. – Estradas de Portugal, E.P.E., which put up for tender a Design and Construction Contract Job,

J. Dinis et al. / Treatment and Stabilization of the National Road E.N. 379-1 Hillsides

519

sustained by a Basic Design carried out by COBA. As this is a road with heavy traffic during the summer season, the Site Owner required that within 330 days - the beginning the next bathing season - the safety conditions necessary for transports to circulate be guaranteed between Outão and Praia da Figueirinha and that all that works be completed. Since this intervention is to take place in a classified, protected and highly sensitive area, the tender proposal required that the solutions provided and implemented on-site ensured the preservation of its natural heritage, namely, with regards to ecological and landscape aspects. This condition required the direct involvement of the Park Regulatory and Management Entities, in order to approve the solutions. We refer specifically to the Arrábida Natural Park (ANP), the Instituto de Conservação da Natureza (Institute for Conservation of Nature) (ICN), and the Ministry of the Environment, amongst others. Additionally, both the solutions and the works inherent to their implementation had to be adapted to the extremely rugged terrain and the existing geological and geotechnical reality. Therefore, it was necessary to develop measures that would maximize safety and minimize visual impact by preserving the hillsides in their natural state.

1. Proposed Solution 1.1. General Given the constraints, and using as support the land survey, the geotechnical report and other elements within the Basic Design, was developed the Execution Project. In its preparation, was maintained the Basic Design’s philosophy and structure, mainly with regards to the design of the structural solutions. The project options were broadly divided into two groups of structural solutions: those applied to the road and those applied to the hillsides. 1.2. Solutions applied to the road On the road, in order to create artificial ditches for block retention, stabilize the hillsides that showed signs of instability at the base, and, in some cases, to withstand the impact of falling blocks, retaining and protection walls were designed in reinforced concrete, with an "L" section, in a reversed "T" and sloped with nails (Fig. 2a), and cyclopean concrete walls. Additionally, whenever there was a possibility that smallsize stones could be projected, rigid barriers were installed at the crowning and the back walls. In order to include the walls in the landscape, the entire exposed surface was coated with natural limestone (Fig. 2b).

520

J. Dinis et al. / Treatment and Stabilization of the National Road E.N. 379-1 Hillsides

Figure 2 - Walls: (a) retaining/protection reinforced concrete wall to avoid falling blocks, (b) retaining wall in cyclopean concrete with a rigid barrier at the crowning.

In situations in which the escarpment was over the road or presented over breaks (overscouring) at the foot of the hillside that could potentiate instability and the risk of detachment of rocks onto the road, false tunnels were designed with a portico section (Fig. 3). The false tunnels with the portico section proposed for three different areas of the road layout, characterized by a reinforced concrete slab, were sustained on the hillside by a continuous vertical wall along the entire length of the tunnel, with a direct foundation in the bedrock. On the ocean side, they were sustained by a row of pillars in a ∅ 600 mm circular section, 5.00 and 9.00 m apart, founded through micropiles headed by a longitudinal beam or through ∅ 600 mm piles, respectively. The top slab was designed to withstand, in addition to the required regulatory overloads, the impact of a 500 kg block, which could eventually loosen from the hillside from a height of 100 m. In terms of calculation, this occurrence was simulated by applying a 750 kN static force. The damping of the fall of this block was assured by a backfill layer (sandy material) placed over the portico slab.

Figure 3 - Protection Tunnels

1.3. Solutions applied to the hillsides In the areas where the hillsides were stable but where there was a probability of local detachment of insignificant masses and in order to prevent the blocks from gaining excessive speed, protection nets were applied, suspended at the crest of the hillside. This was carried out using a system made up by support cables nailed to the hillside through rockbolting. When the massif was next to the road and showed significant weathering and erosion, with significant fracturing, and signs of potential detaching or sliding of rocks onto the road, a restraining system using protection nets, reinforced with high-

J. Dinis et al. / Treatment and Stabilization of the National Road E.N. 379-1 Hillsides

521

resistance steel cable panels was applied (Fig. 4 a). The strengthening of the embankments and sustaining of the steel cable panels was foreseen through rockbolting, distributed over 2.00 and 2.50 m square mesh from the side, and between 6.00 and 10.00 m long. When the embankments and hillsides were further away from the road and there was the possibility that multiple detachments of significant masses could hit the road or surrounding areas, in order to retain these masses, we resorted to installing medium energy (1500 kJ) and high energy (3000 kJ) dynamic barriers. These were placed in predetermined areas downstream of the escarpments (Fig. 4 b). These systems were formed by highly-resistant steel reinforced nets, sustained by 5.00 m high metal posts, bolted to the ground and joined through a load-bearing cable system, associated to a braking system. This consisted of a perforated plate through which a set of cables was woven to dissipate the impact energy of the blocks.

Figure 4 - (a) Nets bolted and reinforced with steel cables panels (b) Dynamic protection barriers against falling blocks

Finally, in situations where none of the above solutions provided an answer to the required needs, the rocky masses that showed signs of instability was dismantled. To this end, we resorted to various methodologies, of which we highlight demolition using explosives.

2. Difficulties and unforeseeable situations in the execution of the contract job 2.1. Preparation of the works After the awarding of the contract job and in order to allow the works to begin, it was necessary to define the strategy for all the works. Given all the conditions set forth, including those of an environmental, orographic and geological and geotechnical order, which don’t permit the execution of an alternative access to EN379-1, along with the requirement to carry out numerous activities at once, both on the road and on the overlying hillsides, the development of an effective plan, to adequately respond to the demands of all the entities involved, proved to be a lengthy and extremely difficult process.

522

J. Dinis et al. / Treatment and Stabilization of the National Road E.N. 379-1 Hillsides

2.2. Works on the road After the beginning of the works, a complementary geotechnical reconnaissance survey was carried out throughout the road. It focused on the execution of additional boreholes with dynamic penetration tests on sections where tunnels and investigation wells were to be executed in the area of the walls. Following these studies, it was ascertained that many of the assumptions considered in the preparation of the project were not valid, particularly those related to the foundation conditions and the layout of the structural elements. It then became necessary to set the details for all the walls and tunnels in the Execution Project. In most tunnels, throughout the longitudinal wall to be executed close to the hillside, where the foundations were expected to be direct, through continuous ground plates, colluvium was found down to approximately 20.00 m deep. This fact led to the need to execute indirect foundations on the hillside using piles. On a structural level, this change led to a loss of overall stiffness that, essentially, given issues of seismic behaviour, led to the replacement of the solution of a foundation through micropiles (foreseen on the seaward side of one of the tunnels) with piles and the execution of braking cross-section foundation beams in all the tunnels. Additionally, due to construction reasons, including the need to cross the colluvium deposits with blocks of various sizes involved in sandy matrixes of reduced compactness, the diameters of the piles were changed to 800 mm. These changes, besides the obvious consequences, as far as execution cost is concerned, resulted in major constraints to the entire plan foreseen. This was essentially due to the fact that the road circulation had to be interrupted during the execution of the cross-section foundation beams. This latest setback was overcome through the placement of temporary metal bridges over the beam’s area. 2.3. Works on the hillsides Besides all the difficulties inherent to this type of intervention, primarily with regard to the safety of the intervening parties in virtually all activities and the safety of the road where other works were being carried out, during an initial phase it was necessary to carry out a exhaustive survey on-site. This was required in order to adapt the solutions proposed to the actual geomorphology. It was then necessary to define how to place the equipment on-site, in some cases, at more than 50 m high. In this case, given the rugged terrain and the inability to create access ways, large cranes were used. This activity, although seemingly simple, had as a direct consequence the interruption of the road circulation, requiring an additional effort in what was already the difficult management of all the work fronts. With regard to the development of the works, given the need to execute rockbolting on 40-50 m high sub-vertical hillsides, it was necessary to develop small diameter drilling equipment in order to be operated suspended by steel cables and to be moved by hydraulic tirfor’s. Since we are in the presence of limestone massif, very fractured and, although apart, with very wide cracks as a result of karstification phenomena, during the works to seal the rockbolting there was an overconsumption of cement grout which was much greater than foreseen. Given these circumstances and in order for this activity not to become critical, it was necessary to significantly strengthen the means of injection.

J. Dinis et al. / Treatment and Stabilization of the National Road E.N. 379-1 Hillsides

523

This led to significant extra costs, both in terms of material and of manpower and equipment. 2.4. Implementation of additional structural solutions In one of the fronts that had a higher degree of difficulty and concentration of net and barrier works located at the beginning of the section and through which the critical path of the contract job passed through, a 40.00 m horizontal crack was found, 14.00m deep and with an average gap of 0.30 m. Given this situation, it was necessary to implement a solution that consisted, initially, in precisely identifying the size and characteristics of the crack. Subsequently, it was filled with fluid concrete, in phases, and controlled by a predetermined instrumentation plan. Afterwards, 56 permanent anchors, pre-stressed to 600 kN and ranging from 16.00 to 18.00 m long were executed, also in phases and monitored (Fig. 5). During the execution of the anchors, due to the high degree of karstification of the formations, nearly 650 ton of cement were injected. This represents almost 12 ton per anchor, an extremely high amount considering the type of works carried out.

Figure 5 - Anchors in the area where a crack was found.

3. Conclusions The size of the area subject to intervention, coupled with the high heterogeneity of the formations, led to changes in many of the assumptions considered in the Basic Design. As is usual in this type of intervention, it was necessary to develop at the same time as the works, all the details of the Execution Project. The implementation of the various solutions - insofar as to the hillsides - within the deadline, the workload, the lack of access, the distance from the road, as well as the difficult working conditions on the hillsides, which generally had high pendency, led to the mobilization and training of a large number of skilled workers. They mostly worked at heights and used a large number of light equipment, specifically designed for the conditions under which these works took place. With the creation of multidisciplinary teams, it was possible to carry out - in only 330 days - about 25.0 km of rockbolting and 950.0 m of permanent anchors, inject 1800 ton of cement, install 35,000 m2 of nets and 2.7 km of dynamic barriers and execute over 1.0 km of walls and 285.0 m of tunnels, founded over a total of about 1.0 km of reinforced concrete piles.

524 Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-524

Contraintes géotechniques associées à la Construction de la Deuxième Piste de l’Aérodrome d’Oran en Algérie a

Vicente RODRIGUES a1, Mário ROLDÃO b et António SILVA, c Ingénieur Géologue, COBA, SA b Ingénieur Génie Civil, COBA, SA c Ingénieur Génie Civil, ZAGOPE, SA

Résumé: L’emprise de construction de la deuxième piste de l’aérodrome d’Oran fut essentiellement composée par la réalisation d’une piste de 3000 m de long, le prolongement ouest de 600 m de la piste existante, ainsi que par la construction de plusieurs zones de parking et de voies de circulation d’accès aux pistes. Les terrains de l’aérodrome avoisinant le grand lac d’Oran présentent des conditions géotechniques tout à fait particulières dans la mesure où la nappe phréatique est très proche de la surface et les dépôts quaternaires qui composent la fondation sont essentiellement des argiles gypseuses, sols qui présentent de nombreuses cavités qui se forment en raison de la dissolution du gypse suite au contact avec l’eau. Il s’agit d’un phénomène endémique de formations de cavités, quelques-unes ayant de grandes dimensions et avec plusieurs ramifications. Un total d’environ 200 cavités fut rencontré, ce qui a obligé à que le projet d’exécution et les travaux adoptent la mise en place de plusieurs dispositions constructives pour assurer d’adéquates conditions de fondation. Une particulière mise en évidence est donnée à la stratégie de détection de cavités, à son comblement et scellage, ainsi qu’au dimensionnement d’une structure de chaussée adaptée aux sollicitations aéronautiques. Dans le cadre du marché, l’Entreprise était également tenue d’assurer un contrôle moyennant la réalisation d’essais et de mesures de qualité en cours de chantier. Un aperçu est donné sur l’activité de l’unité d’auto-contrôle qui fut la responsable par le suivi et réalisation de plus de 17000 essais de contrôle sur les sols, agrégats et couches bitumineuses. Mots clés. Argiles gypseuses, nappe phréatique, cavités, auto-contrôle

Introduction Le contrat de prestation de services signé entre COBA et ZAGOPE, Chef de File du Groupement ZAS – ZAGOPE, Andrade Gutierrez & Sahraoui, incluait initialement le projet d’exécution, l’assistance technique ainsi que la coordination d’une Unité d’Auto-contrôle, responsable par le contrôle de qualité des travaux, étant la mise en place de cette unité à la charge des entreprises en raison du marché passé avec le Maître d’Ouvrage (la Direction des Travaux Publics d’Oran) la prévoir, fonction de l’importance stratégique des infrastructures à construire et de la nécessité en garantir la qualité de toutes les réalisations. 1

Vicente RODRIGUES. Chef du Département Ouvrages Géotechniques, COBA, SA, Av. 5 de Outubro, 323, 1649-011 Lisbonne, Portugal; E-mail: [email protected].

V. Rodrigues et al. / La Construction de la Deuxième Piste de l’Aérodrome d’Oran en Algérie

525

Ultérieurement, entre Février 2008 et Octobre de 2009, à la demande de ZAGOPE, COBA a renforcé sa participation dans l’Unité d’Auto-Contrôle avec l’installation dans la base-vie d’un laboratoire équipé avec le but de réaliser tous les essais sur sols, agrégats et mélanges bitumineux. Deux Techniciens de laboratoire ont également renforcé l’équipe résidente.

1. Formation endémique de cavités Les terrains sur lesquels était prévue la construction de la deuxième piste de l’aérodrome d’Oran sont des dépôts quaternaires composés essentiellement par des argiles gypseuses, avec formation en surface de croutes gréso-gypsifères. Il s’agit d’une formation géologique résultante du dépôt de matériaux fins appartenant en général du démantèlement des massifs avoisinants. Sur le plan hydrogéologique, la région se manifeste par un régime d’écoulement endoréique; les eaux de ruissellement aboutissent dans les dépressions formées par la grande Sebkha d’Oran, le lac situé à proximité des terrains de l’aérodrome. Dans la région d’Es-Senia, banlieue d’Oran, le sol présente de nombreuses cavités qui se forment en raison de la dissolution du gypse suite au contact avec l’eau de la nappe phréatique. Il y a un phénomène endémique de formation de cavités, quelquesunes à grandes dimensions et avec plusieurs ramifications (Photo 1).

2. Stratégie de détection de cavités L’Avant Projet Détaillé prévoyait la réalisation d’une reconnaissance moyennant l’adoption du géoradar pour la totalité des surfaces concernées par la construction de l’emprise de la deuxième piste. Cette campagne était contractuellement à la charge des entreprises. Cette investigation indirecte (le géoradar), essentiellement prévue pour la détection de l’occurrence de cavités et pour valider le scénario de fondation des ouvrages, devrait être complétée avec la réalisation d’une campagne de prospection géotechnique directe à exécuter par le Maître d’Ouvrage. Plusieurs difficultés vécues par l’emprise ont obligé à que COBA recommande une stratégie alternative pour la détection de cavités. D’une part, on a constaté que le modèle géologique existant pour les terrains en question était incompatible avec ce qui avait été suggéré par l’APD, vue l’inefficacité du géoradar dans des tels terrains meubles avec une nappe phréatique assez proche de la surface du terrain naturel. D’autre part, entre-temps, le Maître d’Ouvrage n’a pas pu effectuer la campagne de prospection directe prévue (54 pénétromètres dynamiques et 48 sondages carottés, avec recueil d’échantillons pour une ultérieure caractérisation de laboratoire), étant les premières activités des terrassements, décapage et excavations, sur le chemin critique de l’avancement des travaux et du délai global de l’emprise. Courts délais de réalisation de l’emprise, impossibilité d’exécution d’investigations directes sur une assez vaste surface, les travaux étant en cours et la nécessité de progresser avec les terrassements ont obligé à la définition d’une stratégie alternative pour la détection de cavités. Ayant comme base l’avion de projet, le Boeing 747-400, une analyse de répartition de contraintes fut effectuée en profondeur à travers la structure de chaussée considérée.

526

V. Rodrigues et al. / La Construction de la Deuxième Piste de l’Aérodrome d’Oran en Algérie

Les charges résiduelles enregistrées au niveau de la fondation, à 1,35 m de profondeur, ont été de l’ordre de 0,2 à 0,8 bars (calculs effectués avec les logiciels PLAXIS et ECOROUTEW95), ayant COBA, dans un tel cadre géotechnique, recommandé l’adoption d’un compactage lourd au niveau de l’arase terrassements, avec des lignes de travail écartées de 5 m. Ainsi, 12 rangées furent définies pour les pistes principales et 4 rangées pour les voies d’accès et plaques de stationnement, ayant chaque ligne de travail subi un total de 24 passes de compacteur distribuées de la façon suivante (Photo 2): 10 passes avec un compacteur à pneus de 35 tonnes, suivies de 4 passes avec un rouleau vibrant de 20 tonnes et à la fin 10 passes supplémentaires avec le premier compacteur à pneus de 35 tonnes. Un total d’environ 200 cavités furent détectées, 60 desquelles au niveau de la deuxième piste et prolongement de la piste principal, ayant les restantes 140 été retrouvées dans la zone du parking-fret.

3. Traitement et scellage de cavités Les cavités rencontrées par l’emprise au niveau des terrains de fondation de la future plateforme de l’aéroport étaient classifiées en trois types: •

Type I : superficielles, avec une profondeur et diamètre allant jusqu’à environ 1 mètre; cavité à détecter et à éliminer, étant son comblement effectué avec des couches de tuf calcaire de 20 cm;

•

Type II: cavités avec un diamètre d’environ 3 m, rencontrées jusqu’à une profondeur de 6 m, devant être détectées et traitées; celles-ci se situaient déjà dans la zone d’oscillation saisonnière de la nappe phréatique (située entre 2,5 à 3,5 m) ;

•

Type III : cavités situées entre 6 et 20 m de profondeur, ne faisant pas l’objet de traitement, sauf si elles présentaient des communications avec celles de type II.

L’occurrence de cavités dans le sol est un phénomène assez fréquent dans cette région, cadre confirmé en 1984 lors de la reconstruction de la piste principale. Selon les informations recueillies, lors de l’exécution des travaux dans la première piste de l’aérodrome d’Oran, beaucoup de cavités furent à l’époque rencontrées, ayant les plus importantes, avec plusieurs mètres de diamètre et profondeur, fait l’objet d’un comblement avec béton cyclopéen. L’APD prévoyait pour cette emprise que les cavités des types II et III, une fois taillées selon une forme géométrique et son fond énergiquement compacté, elles seraient comblées par couches de 20 cm de tuf calcaire 0/40; pour le tuf calcaire de comblement des cavités, le Cahier des Charges spécifiait un compactage relatif de 100% du Proctor normal. La faible profondeur de la nappe phréatique impliquerait que beaucoup de cavités se développent et/ou se prolongent au-dessous du toit de la nappe, en étroite relation avec le lac d’Oran, situé à côté des terrains de l’aérodrome, ce qui contraignait beaucoup l’obtention d’un adéquat comblement des vides occurrents dans la mesure où

V. Rodrigues et al. / La Construction de la Deuxième Piste de l’Aérodrome d’Oran en Algérie

527

l’efficacité du compactage des sols fins au-dessous de la nappe phréatique deviendrait particulièrement difficile. Dans ce contexte, COBA a préconisé l’adoption d’une solution alternative pour le traitement de cavités des types II et III, solution qui fut approuvée par l’Assistance Technique et qui, en bref, une fois fait le taillage des parois latérales et du fond, prévoyait, le rabattement de la nappe phréatique de l’intérieur des cavités par pompage spécifique, le déversement gravitaire de matériaux granulaires (D máx avec 3 à 5 cm) compactés par couches de 0,5 m d’épaisseur, et le scellage avec un mortier de ciment. Une fois scellées les cavités, les terrassements courants pourraient être repris avec la mise en place et compactage des couches de tuf calcaire 0/40 jusqu’à l’atteinte des cotes de travail correspondantes à une profondeur d’environ 1,35 m, niveau retenu pour la fondation de la chaussée.

Photo 1. Cavités – Inspection initiale

Photo 2. Détection de cavités par compactage lourd (écart de 5 m entre lignes d’investigation)

4. Dimensionnement de la chaussée Le calcul structurel de la chaussée fut effectué ayant comme base la prévision et définition du trafic à considérer, ainsi que les caractéristiques des matériaux a utiliser pour les différentes couches. On a recouru à la méthode de la Fédéral Aviation Administration (FAA) des EUA, suivant les consignes du manuel AC 150/5320-6D, en utilisant le programme de calcul automatique LEDFAA et en suite nous avons procédé à une vérification selon les “Instructions du Service des Bases Aériennes Françaises, 1983”. La surface de chaussée fut divisée en zones, selon les charges prévues; trois types de chaussée ont été définis: •

Chaussée type I – Bretelles O, C et D, zone d’élargissement du parking avions. Dans ces zones on a considéré un nombre de mouvements correspondants au trafic total de l’aérodrome.

•

Chaussée type II – Prolongement de la piste principale, deuxième piste, bretelles A, B et E. Dans ces zones on a considéré un nombre de mouvements correspondants à 60% du trafic total de l’aérodrome.

528

V. Rodrigues et al. / La Construction de la Deuxième Piste de l’Aérodrome d’Oran en Algérie

•

Chaussée type III – Parking-fret. Dans cette zone on a considéré un nombre de mouvements correspondants à 60% du trafic d’avions correspondants aux classes ICAO C et D, en raison de cette plaque ne desservir pas les avions de la classe E.

Le trafic fut extrapolé pour l’année 2020, horizon de projet défini par les autorités algériennes, considérant un facteur de croissance annuel de 4%. Aux effets de dimensionnement on a considéré pour tous les avions son poids maximal lors du décollage. La caractérisation des matériaux considérés dans la structure de chaussée fut basée sur les résultats obtenus aux essais de laboratoire disponibles. Le sol de fondation est un silt ou un sable silteux à faible ou nulle plasticité, mais très sensible à l’eau, démontrant une grande réduction de capacité de support en présence d’eau. D’après les résultats enregistrés aux essais CBR et de charge avec plaque, compte ténue de son comportement évolutif, on a adopté pour ce sol de fondation une valeur de CBR de 5. Les matériaux spécifiés pour la chaussée, en conformité avec l’APD, ont été les suivants: couche de roulement en béton bitumineux, couche de base en grave bitume, couche de fondation en grave concassé et couche de forme en tuf calcaire. Le béton bitumineux et le grave bitume ont une qualité concordante avec la spécification P401 de la FAA, dont on a adopté des valeurs standard de la norme FAA pour sa caractérisation. Pour la couche de fondation en grave concassé, des roches calcaires de bonne qualité furent sélectionnées qui, moyennant un adéquat concassage, assurent un standard de qualité encadré dans la norme FAA P209. La couche de forme en tuf calcaire de bonne qualité non concassé et à granulométrie étendue, compacte, est encadrée dans la norme P154 de la FAA. Le module d’élasticité considéré, de 180 MPa, est concordant avec les essais de laboratoire disponibles. Le dimensionnement de la chaussée ainsi que les principales décisions adoptées, en conformité avec le cadre géotechnique en présence, ont ténue compte d’autres situations: •

L’épaisseur totale de chaussée fut définie avec 115 cm, valeur à laquelle il faut ajouter une couche de forme en tuf calcaire avec 20 cm comme mesure constructive destinée à contrarier la possible formation de cavités dans la fondation; ainsi, l’épaisseur de 115 cm fut considérée comme étant le minimum à adopter.

•

Pour éviter des désordres causées par la possible formation de cavités au niveau de la fondation, vue qu’il s’agit d’un phénomène endémique, on a prévu l’installation d’un géotextile de renforcement entre la structure de chaussée et la couche de forme, composé par deux couches croisées de géotextile tissu avec 400 KN/m de résistance à la traction (superposition transversale de 0,3 m des rouleaux). Selon les calculs présentés en APD ainsi que par le fabricant, ce géotextile permet de maintenir l’intégrité structurelle de la chaussée, même avec la formation d’une cavité dans la fondation ne dépassant un mètre de diamètre (Photo 3).

V. Rodrigues et al. / La Construction de la Deuxième Piste de l’Aérodrome d’Oran en Algérie

•

529

De mode à envisager aussi une plus efficace intégrité structurelle à la chaussée, compte tenue de la possible formation de cavités, l’APD prévoyait la mise en place d’une géogrille 30/30 sous la couche de fondation en grave concassé (Photo 4). Cette solution fut maintenue par le projet d’exécution, ayant été prise en compte la contribution de cet élément pour la capacité de support de la chaussée. Avec cette disposition constructive, nous avons ajouté une deuxième ligne de défense supplémentaire dans la structure de chaussée contre la formation de cavités (au-delà de l’adoption du géotextile croisé).

Compte tenue d’une l’épaisseur minimale de 115 cm, la composition finale de la chaussée qui fut recommandée par le projet développé par COBA est présenté au Tableau 1. Tableau 1. Structure finale de chaussée préconisée par COBA Types de chaussée Types I et II

Type III

Couche de roulement en béton bitumineux

10 cm

10 cm

Couche de base en grave bitume

20 cm

15 cm

Couche de fondation en grave concassé

35 cm

35 cm

Deuxième couche de fondation en tuf calcaire

50 cm

55 cm

Épaisseur totale

115 cm

115 cm

Couche de forme en tuf calcaire

20 cm

20 cm

Épaisseur totale du décaissement

135 cm

135 cm

Photo 3. Première couche de tuf calcaire appliquée sur le géotextile

Photo 4. Épandage et mise en place du grave concassée sur la géogrille

5. Auto-contrôle En bref, voici les principales activités développées par l’Unité d’Auto-Contrôle: •

Suivi de la campagne de prospection géotechnique initialement à la charge de l’entreprise LTPO

530

V. Rodrigues et al. / La Construction de la Deuxième Piste de l’Aérodrome d’Oran en Algérie

•

Repérage, description et classification de toutes les cavités rencontrées

•

Appui au Projet d’Exécution et à l’Assistance Technique aux travaux

•

Contrôle de qualité des essais de laboratoire

•

Sélection d’échantillons dans les emprunts, carrières et centrales bitumineuses

•

Prescriptions d’exécution et suivi des remblais d’essais des matériaux de remblai, agrégats et mélanges bitumineux

•

Suivi de l’exploitation des zones d’emprunt et carrières

•

Contrôle de qualité des terrassements et des bitumineux

•

Formulation des mélanges bitumineux

•

Affinage des centrales bitumineuses

•

Suivi des travaux et de la conformité des études d’exécution

•

Assistance à la Direction Technique du Groupement dans la coordination de plusieurs activités.

L’Unité d’auto-Contrôle a réalisé un total de 17024 essais pendant la durée du chantier, avec la répartition suivante : •

Sols : 616 essais (analyses granulométriques, limites d’Atterberg, valeur en bleu de méthylène, équivalent de sable, Proctor modifié, CBR direct et imbibé, poids volumique, absorption, CACO3 et teneur en matière organique)

•

Agrégats : 601 essais (analyses granulométriques, limites d’Atterberg, valeur en bleu de méthylène, équivalent de sable, Proctor modifié, CBR imbibé, poids volumique, absorption, aplatissement et Los Angeles)

•

Grave bitume : 2115 essais (analyses granulométriques, pourcentage et teneur en bitume, densité maximale théorique, densité réelle, stabilité Marshall, déformation, porosité, teneur volumétrique en bitume, relation filler/bitume, volume de vides de l’agrégat et degré de saturation en bitume)

•

Carottes de grave bitume : 235 carottes (hauteur et degré de compactage)

•

Béton bitumineux : 2490 essais (analyses granulométriques, pourcentage et teneur en bitume, densité maximale théorique, densité réelle, stabilité Marshall, déformation, porosité, teneur volumétrique en bitume, relation filler/bitume, volume de vides de l’agrégat et degré de saturation en bitume)

•

Carottes de béton bitumineux : 217 carottes (hauteur et degré de compactage)

•

Caractérisation mécanique du béton bitumineux : 2 essais (flexion et résistance à la déformation permanente)

•

Contrôle de la compacité à source radioactive : 6834 essais Troxler

•

Contrôle de compactage : 292 essais de charge à la plaque

•

Taux d’imprégnation cut-back et émulsion : 131 essais

V. Rodrigues et al. / La Construction de la Deuxième Piste de l’Aérodrome d’Oran en Algérie

531

•

Rugosité des bitumes : 390 essais à la tache de sable

•

Règle de 3 m : 1366 essais pour le grave bitume et 1737 essais pour le béton bitumineux.

Photo 5. Essai de charge avec plaque de 600 mm

Photo 6. Signalisation horizontale de la deuxième piste

6. Observations finales Les travaux ont vécu plusieurs contraintes parmi lesquelles il faut souligner des conditions météorologiques assez difficiles, l’épuisement de plusieurs carrières et emprunts de la région, ainsi que la nécessité en assurer la circulation aéronautique dans la piste principale de l’aéroport pendant toute la durée du projet. L’emprise de construction de la deuxième piste de l’aérodrome d’Oran fut certainement une des réalisations algériennes de ce genre avec un plus important nombre d’essais de contrôle en raison d’un cadre géotechnique tout à fait particulier et dangereux, compte tenue d’une tendance endémique de formations de cavités dans les terrains d’assise. Les solutions de projet et les dispositions constructives adoptées sont le résultat du contexte géotechnique en présence, en pondérant dans la mesure du possible l’aléa géologique occurrent.

Références [1] COBA. Études d’Exécution de la Deuxième Piste de l’Aérodrome d’Oran en Algérie. Mars – Décembre 2007. Lisbonne, Portugal. [2] COBA. Construction de la Deuxième Piste de l’Aérodrome d’Oran en Algérie. Rapport final. Février 2011. Lisbonne, Portugal.

532 Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-532

Effect of Geosynthetic on the Performance of Road Embankment on Algeria Sabkha Soils Sadok BENMEBAREK1, Naima BENMEBAREK and Lamine BELOUNAR Civil Engineering Laboratory, Biskra University, Biskra, Algeria

Abstract. With the help of geosynthetics innovative, economic and durable solutions can be offered to several situations where standard soil improvement techniques are still extensively used. This paper is interested by the design and the construction of the road with reinforced embankment crossing the sabkha flat of Chott El Hodna on 11 km length in the north middle of Algeria. Sabkha soils are associated with many geotechnical problems, due to the presence of salt crystals of different sizes, shapes and compositions; and the highly saline and shallow ground water table. Due to the poor bearing capacity of the present sabkha surface and the arising water table over the soil surface serious difficulties were faced during the investigation of the subsurface soil and the construction of the road embankment. After the project description and soil investigation, the need of reinforced embankment by geosynthetic was well highlighted by numerical modelling and confirmed by the difficulty reencountered in the placement of the two first embankment lift and the compaction performance. The in situ observation indicated also that the use of geotextile separation has a beneficial effect on sabkha soils, especially under wet conditions and shallow ground water table. Keywords. Embankment, geosynthetic, numerical modeling, reinforcement

Introduction Sabkha is originally an Arabic name for saline flats that are characterised by very low bearing capacities and underlain by sand, silt and clay, and often encrusted with salt. Sabkha soils are very sensitive to moisture whereby complete collapse and large reduction in the bearing capacity are anticipated when these soils are in contact with water [1]. Such behaviour is attributed to the fact that some of the cementing materials that bond the mineral grains of sabkha together, such as halite, are highly soluble in water, while others, such as gypsum, aragonite, and calcite are less soluble. The work done by [2] on different sabkha soils confirms the acute water sensitivity and chemical aggressiveness of sabkhas. Road engineers in Algeria like Tunisia, Arabi Saoudit, USA, India and Australia often face the challenge to design a solid road foundation on top of very soft soils which are characterized by sabkha soils. 1

Corresponding Author. Sadok Benmebarek, Civil Engineering Laboratory, Biskra University, BP 145 Biskra, Algeria, E-mail :[email protected]

S. Benmebarek et al. / Effect of Geosynthetic on the Performance of Road Embankment

533

In the present paper, the proposed road with reinforced embankment which crosses the sabkha of Chott El Hodna on 11 km is located in the M’Sila department in the north middle of Algeria (Figure 1). This road reduces the current distance from two towns by 140 km and improves considerably the commercial and agriculture activities The in situ observations show that in summer surface soil is partially dry and soft enough where only a very small weight vehicles can cross the Sabkha. However, in winter the sabkha is inundated where water table may arise up to 60 cm over soil surface. Figure 2 presents photograph showing the sabkha surface state taken in July 2005. In this paper, after the description of the project and soil investigation, the need of reinforced embankment by geosynthetic was well highlighted by numerical modelling using an explicit finite difference code and confirmed by the difficulty reencountered in the placement of the first embankment lift and the compaction performance. Ain El Khadra

M’Cif

Figure 1. Site project

Figure 2. Subsurface state of the project road

1. Hydrological and subsurface soil investigations The sabkha of Chott El-hodna in the middle north of Algeria is a large closed flat of 26000 km2 developed where surface runoff converges from the Saharian Atlas in the South and the Tellien Atlas in the North and also by soil infiltration. In summer, the surface is encrusted with salt. The embankment road devises the sabkha in two parts. The hydrological study shows the maximum water level which may reach 1.40 m over soil surface for a one period of thousand-year-old return. The program of sabkha subsurface investigations contains boring hole, cone penetration test and vane shear test every 300 m of the embankment length. Due to the poor bearing capacity sabkha surface and the arising water table over the soil surface serious difficulties were faced during the investigation of the subsurface soil. Therefore, the subsurface investigations were accomplished with the advancement of the two first lifts reinforced by geosynthetic of the embankment. Subsurface state conditions at the middle of the Sabkha consists of a brown muddy clay layer with thickness varying from 3m to 5m, underlain by grey muddy marl and gypsum concretions with traces fine sand with thickness varying from 5m to 7m. Near the edges of the sabkha the thickness of the soft layers decreases. The partially laboratory testing results show that the compression index C c varying

534

S. Benmebarek et al. / Effect of Geosynthetic on the Performance of Road Embankment

from 0.31 to 0.56, the plasticity index I p varying from 27.5 to 48.5 and the dry density varying between 1.38 and 1.64 indicating high soil compressibility. The undrained shear strength of the layers brown muddy clay and grey muddy marl reaches 9 kPa. In the sabkha centre, the thickness of the very soft layers may reach 10 m. These results are in good agreement with the static cone penetration test results showing no point resistance for this depth.

2. Reinforced embankment and slope protection In the present case the geotextile is used to separate the subsoil and the embankment aggregate, while the reinforcement by geogrid layer is used to increase the stiffness of the foundation and to increase the compaction quality. For the embankment, a sandy gravel material was chosen to allow free drainage of the foundation soils and reducing the pore pressure build-up below the embankment; The construction steps used as showing in Figure 3 can be summarised as following: • Laying directly over sabkha surface corresponding to embankment base a nonwoven geotextile layer as separator/filter to prevent contamination of embankment material (Figure 4).

Figure 3. Reinforcement and protection of the embankment

• Construction of the first lift of 30 cm thickness compacted to obtain plane surface; • Laying the geogrid over the surface to uplift the tensile strength to the embankment base (Figure 5); • Construction the embankment layer by sub-layers with compaction control; • After reaching the embankment height 1.70m, the hydraulic PEHD reinforced tubes were installed (Figure 6). These tubes are flexible and inert to sabkha soil aggressively; • Protection of embankment slopes with separate geotextile GT 2 placed under rock ripraps (Figure 7).

S. Benmebarek et al. / Effect of Geosynthetic on the Performance of Road Embankment

535

Figure 4. Laying directly over sabkha surface a nonwoven geotextile GT 1

Figure 5. Laying the geogrid over the first lift of the embankment

Figure 6. PEHD reinforced tubes installation

Figure 7. Protection of embankment slopes

3. Numerical analysis of the embankment reinforcement For reinforcement applications, solutions have been proposed for situations where the tensioned-membrane reinforcement function will be realized and for situations where the lateral base course restraint mechanism is appropriate. Since separation is typically an integral part of the tensioned-membrane reinforcement function, design solutions for this geosynthetic function generally lump these two functions together. Designs incorporating the tensioned-membrane reinforcement function are applicable for unpaved roads and situations where relatively large rut depths in the roadway can be tolerated and where the traffic is mainly canalized. This approach was recommended by [3] for temporary unpaved roads. However, incorporation of the lateral base course restraint mechanism is applicable for roadways where rut depth needs to be limited to 25 mm. The performance of the road and embankment base reinforcement over soft subsurface depends on several factors particularly the geogrid stiffness, characteristic of the subsurface and parameters of the interface groundreinforcement ([4],[5]). This area of research is very favourable to numerical computations. The present work interests with the numerical simulation of reinforced embankment base over soft subsurface in order to improve the bearing capacity. The improvement of the bearing capacity is evaluated by comparing the wheel loaddisplacement response corresponding. The analysis was carried out using the computer code FLAC-2D (Fast Lagrangian Analysis of Continua) [6] which is a commercially available finite difference explicit program.

536

S. Benmebarek et al. / Effect of Geosynthetic on the Performance of Road Embankment

The embankment material and the soil behavior were modeled by the elasticperfectly plastic Mohr-Coulomb model encoded in this code. The embankment material was characterized by a unit weight γ=20 kN/m3 angle of friction ϕ=35° and a null cohesion. The sabkha soil was characterized by the undrained cohesion Cu =9kPa. The geogrid, modelled by beam element without flexural strength, is connected to embankment material via interface elements obeying the criterion of Mohr-Coulomb and characterized by a null cohesion and a friction angle δ representing the angle of friction of the contact geogrid-embankment material. For the reason of the lack of the laboratory tests, the friction angle δ was taken equal to the 2/3 of the friction angle of the embankment (ϕ=35°). The wheel load-displacement response was determined by this study in large strain analysis for embankment first layers with and without reinforcement. Using a FISH function, the bearing capacity can be calculated as the integral of stress components for all soil zones in contact with the footing area or by the reaction force resultant in the vertical direction at footing nodes. From these simulations it was deduced the improvement made by the reinforcement. A cross section of the embankment was modelled in two dimensions assuming plane strain conditions. The procedure of simulation used in the present analysis was based on the two following steps: • A mechanical calculation of the geostatic stresses : These were computed assuming the material to be elastic; • A mechanical calculation of the improvement of the bearing capacity: the bearing capacity was modeled by a downward velocity applied to the area representing the wheel load until obtaining tolerable rut. The value of the velocity applied to the footing area was 2.5 × 10−6 m/step for this analysis. This value was sufficiently small to minimize any inertial effects in the present conditions. Figure 8 visualizes the vectors of displacement (yellow vectors) and tensile effort (red curve) mobilized in the biaxial geogrid with tensile strength 58 KN/m in the two directions for a lengthening of 12% for the simulation of the bearing capacity for the two first lifts of the embankment by indenting the tires of an axle.

Figure 8. Visualisation of the field displacement and the geogrid tensile due to wheel indentation.

A typical plot of the load-displacement curve is shown in Figure Figure 9-b with reinforcement by this geogrid. The asymptotic limiting value corresponds to the ultimate bearing capacity for the first case (Figure 9-a). However the bearing capacity

537

S. Benmebarek et al. / Effect of Geosynthetic on the Performance of Road Embankment

increases with displacement for the second case (reinforced by geogrid Figure 9-b). These simulations show an improvement of the bearing capacity about 60% for a tolerable rut limited to 10 cm (Figure 9). (10

05

)

1.600 1.400

(10

05

1.200 )

1.000 1.000

0.800 0.800

0.600 0.600

0.400

0.400

0.200

0.200

4 20

30

40

50

60

8

12

16

20

70

(10 (10

03

04

)

)

(a) Without reinforcement (b) With reinforcement Figure 9. Reinforcement Influence on soft subsurface bearing capacity

4. Conclusions In light of the work observations and the numerical computation results, the following conclusions may be drawn: • From soil investigation the present sabkha subsurface is dominated by a muddy clay very sensitive in wet conditions; • In the present project, without separating geotextile it was not possible to prevent the mixing of the first aggregate lift and the soft subgrade; • The need of reinforced embankment by geotextile separation and geogrid was well highlighted by numerical computation and confirmed by the difficulty reencountered in the placement of the first embankment lift and the amelioration of the compaction performance; • Numerical computations of the present project show an improvement about 60% of the bearing capacity of reinforced embankment. References [1] [2]

[3]

[4] [5] [6]

O.S.B. Al-Amoudi, S.N. Abduljauwad, Z.R. El-Naggar and Rasheeduzzafar, Response of Sabkha to Laboratory Tests: A Case Study, Engineering Geology, Vol. 33 (1992), 111-125. S.A. Aiban, O.S.B. Al-Amoudi, I.S. Ahmed and Al-A bdul H.I. Wahhab, Reinforcement of a Saudi Sabkha Soil Using Geotextiles, Proceedings of the Sixth International Conference on Geosynthetics, IFAI, Vol. 1, Atlanta, Georgia, USA, March (1998), 805-810. R.D. Holtz, B.R. Christopher and R.R. Berg, Geosynthetic Design and Construction Guidelines: Participant Notebook, FHWA Publication No. FHWA-HI-95-038, Federal Highway administration, (1995), p 417. R.K. Rowe and S.K. Ho, Continuous panel reinforced soil walls on rigid foundation”, Journal of Geotechnical and Geoenviromental Engineering, ASCE, 123(10) (1997), 912-920. M.C. Alfaro, S.Hayashi, N. Miura and D.T. Bergado, Deformation of reinforced soil wallembankment system on soft clay foundation, Soils and Foundations 37(4) (1997), 33-46. Itasca Consulting Group, FLAC - Fast Lagrangian Analysis of Continua, v 5.00, Itasca Consulting Group Inc., Minneapolis, MN, USA, 2005.

This page intentionally left blank

Section 8 Site Characterisation

This page intentionally left blank

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 541 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-541

Geotechnical characteristics of the Portuguese Triassic mudstones Mário QUINTA-FERREIRA Dep. of Earth Sciences, University of Coimbra, Portugal

Abstract. The Portuguese Triassic mudstones show frequently severe geotechnical problems, changing very rapidly from a soft rock to a soil, when decompressed and in the presence of water. They are an unusual soft rock, without expansive clays, extremely weatherable after wetting, due to the very fine equidimensional network of pores, developing high capillary stress. The water turns this soft rock into a muddy soil. The more problematic situations are related to foundations and slopes following intense rainfall. Keywords. Mudstones, strength, weatherability

Introduction The Portuguese Triassic mudstones are responsible for a large number of construction problems. The more problematic situations are related to foundations and slopes, which suffer degradation after intense rainfall, following the exposure or remobilization of the mudstones. They change very rapidly from a soft rock to a soil, when decompressed and in the presence of water. Field work in the area of Coimbra and laboratory tests were executed, being the results presented and discussed.

1. Geological setting In the Meso-Cenozoic west zone of Portugal, in the top of the Triassic sediments, in the transition to the Jurassic, outcrop a narrow and irregular strip of mudstones, up to a few hundred meters wide, and along approximately 120 km, between Aveiro and Tomar, These mudstones are mainly constituted by silt, being most of the times laminated, showing predominantly a grey colour. The identification of the mineralogical composition of the mudstones was done by X Rays, allowing to recognize illite (or muscovite), kaolinite, chlorite and quartz. Calcite and dolomite can also be found, but they don’t have expansive clay minerals (smectites).

2. Porosity, unit weight and grain size The rapid degradation suffered by these mudstones when immersed in water, prevented using the procedure requiring the vacuum saturation and hydrostatic weighing of the

542

M. Quinta-Ferreira / Geotechnical Characteristics of the Portuguese Triassic Mudstones

sample [1]. Alternatively the mercury porosimeter was used, overcoming the disadvantages of the interaction between the water and the surface of the minerals. The mercury porosimeter allows determining both the porosity and the variation of the pore size, through the relationship between the pressure required to introduce a certain amount of mercury and the corresponding pore size of the sample. In the intact rock, the porosity is around 15% to 20%, with very fine pores, presenting a unimodal distribution, with an average dimension of the pores close to 0.025 micrometre. The total area of the pores is estimated between 10 and 14m 2/g. The apparent unit weight of the intact mudstones is around 22kN/m3. After manual disaggregation and immersion in water, the particles size distribution of the mudstones was determined, showing that they have around 10% of clay, 60% of silt and 30% of sand.

3. Strength evaluation and slake durability The uniaxial compression strength test over regular samples was not executed because this soft material is quite difficult to cut, cracking and slaking very easily during the preparation. Alternatively, the point load strength was determined, using irregular lumps, following the procedures recommended by the ISRM [2]. The point load strength obtained using dry mudstone lumps reached values up to 1.27MPa, reducing drastically even below 0.1MPa after a few minutes of submersion, corresponding to a water content between 5% and 15%. The results of the point load strength test, versus the water content, confirm that the saturation of the mudstones causes an abrupt decrease of strength, to values one fifth to one-twentieth of the strength of the dry mudstones. The slake durability test was executed according to the ISRM procedures [1]. The number of cycles was extended up to six, going beyond the standard two cycles proposed in the test methodology. After de second cycle of the “Slake Durability Test”, up to 60% of the mudstone is lost, being totally disintegrated until the sixth cycle. According to the values mentioned for the loss of the material after the 2 nd cycle, the mudstones are classified by the ISRM [1] as low durability materials. During the field study of the mudstones, seeking to evaluate their in situ properties, dynamic penetrometers super heavy (DPSH) tests were executed, according to the international procedure [3]. The results showed null or very low penetration strength at the surface, in the muddy weathered material, increasing the strength very rapidly until the unweathered material was reached at a depth between 1 and 2 meters, stopping the test due to the high strength of the in situ mudstones.

4. Expansibility The expansibility was studied using two different techniques: the Lambe test [4] and the unconfined swelling test [5]. The Lambe test aims to evaluate the soils that can present swelling problems. The method consists of measuring the expansion of a specimen of compacted soil, with known moisture content [4]. The tested samples were either soil resulting from the manual disaggregation of the mudstones and also rock samples that were hand carved with a knife. The tests on rock samples aimed to understand the field behaviour of the

M. Quinta-Ferreira / Geotechnical Characteristics of the Portuguese Triassic Mudstones

543

intact mudstones when exposed to the water in excavations. The tests on soil samples tried to reproduce the behaviour of the excavated and remobilized mudstones, presenting a soil like behaviour. The swelling is more intense in the first minutes of wetting, tending to stabilize after ten to twenty minutes. The peak expansion stress was close to 10 kPa in the intact rock, and up to 35 kPa in the soils derived from the laboratory disintegration of the mudstones. Tests to determine the unconfined linear expansion were performed [5]. After the onset of wetting, the highest swelling rate was observed during the first 5 to 30 minutes, reducing till stabilization, around two hours later. The maximum unconfined linear expansion is from 13% to 16%.

5. Classification The in situ and intact mudstones are a soft rock, with low durability, presenting a point load strength up to 1,27 MPa. The weathered mud like material is always present at the surface of the ground after rainfall, causing serious difficulties in the execution of engineering works. Thus it was considered necessary to determine the Atterberg limits [6] and the methylene blue value of the soils [7]. The results showed that these soils have a Plasticity Limit (PL) arround 24% and a liquid Limit (LL) arround 34%. The methylene blue value (VBS) using the spot test, is between 0.5 and 1.2 g/100g. Based on the results of the tests executed on the soils derived from the disintegrated mudstones, three classification systems were used. According to the unified soil classification [8], they are ML-MI (silt of low to intermediate plasticity) or CL (low plasticity clay). Based on the classification for road construction purposes [9], these soils are A1 (silty soil) to A6 (clay soil). Considering the LCPC/SETRA classification [10] the soils are A1 (low plasticity silt).

6. Discussion The Triassic mudstones are an unusual soft rock material, presenting serious geotechnical problems on exposed ground surfaces without any confinement. After exposure to persistent rain, the surface of the mudstones is transformed into a muddy ground turning very difficult the movement of vehicles and the development of the construction tasks, due to weathering and loss of strength. Despite the intense weathering and the geotechnical problems associated with the mudstones, that motivated their study, they don't possess expansive clay minerals (smectites). The strong weatherability after wetting, results mainly form the fine equidimensional network of pores, developing high capillary stress, driving the water into the interior of the material. The absorbed water destroys the cohesion forces between the uncemented silt particles, transforming rapidly the soft rock into a muddy soil. This strength reduction is sufficient to justify the occurrence of the failure in many slopes that failed during the first rains that were able to saturate the mudstones. As observed in the field, the depth of the superficial weathering is usually low, typically around 1 metre, usually not deeper than two metres. In vertical cuts with moderate heights in the periphery of the work areas, the behaviour of the in situ and undisturbed mudstone is quite reasonable, without significant degradation or destabilization of the cut face, even when wetted. This abrupt transition showed that a reasonable way to prevent the degradation of the mudstone is to maintain a moderate vertical stress, above the

544

M. Quinta-Ferreira / Geotechnical Characteristics of the Portuguese Triassic Mudstones

expansion pressure, and to avoid wetting. The overburden stress of the superficial weathered materials is sufficient to prevent the development of the weathering process in depth. These muddy soils have lower permeability than the intact rock, reducing the progression of weathering process.

7. Conclusions With the wetting of excavated surfaces, occurs the softening of the mudstones. This degraded superficial layer must be removed in an appropriate thickness to ensure safe conditions for the foundation, allowing a good contact with the undisturbed ground, and avoiding future problems in the structure, mainly due to differential settlements. An efficient drainage of the foundations and ground floors is also required, minimizing the process of expansion and shrinkage of these materials. Whenever possible the ground floor should be a suspended slab. Another solution could be the use of a layer of gravel, with a thickness around 0.5 meters, placed over a geotextile blanket, under the pavement structure, ensuring that the final thickness would reach around one metre above the excavated ground surface. To improve the construction efficiency it is recommended that the excavations should be executed with dry weather and the exposure time of the mudstones should be as low as possible.

Acknowledgments To the Town Hall of Coimbra and to Instituto Pedro Nunes (IPN) for the opportunity and facilities to develop this study. The work was executed in collaboration with the research project POCTI/ECM/38444/2001, financed by Fundação para a Ciência e Tecnologia (FCT), and by the European Fund For Regional Development (EFRD) .

References [1] ISRM, Rock characterization testing & monitoring - ISRM Suggested methods, E.T. Brown, Pergamon Press, 1981. [2] ISRM, Suggested method for determining point load strength, Commission on Testing Methods. Working Group on Revision of the Point Load Test Method, 1985. [3] EN ISO 22476-2. Geotechnical Engineering – Field Testing. Part 2: “Dynamic Probing”. 2005. [4] UNE 103 600. Determination of expansibility in a soil in the Lambe apparatus, AENOR, 1996. [5] E 200, Soil swelling test. LNEC, Portugal, 1967. [6] D4318 – 10, Standard Test Methods for Liquid Limit, Plastic Limit, and Plasticity Index of Soils, ASTM. [7] NF P 94-068, Determination of the methylene blue value of a soil by means of the stain test, AFNOR, France, 1993. [8] D 2487 – 93, Classification of soils for engineering purposes (Unified Soil Classification System), ASTM. [9] E 240, Soils. Classification for road construction purposes. LNEC, Portugal, 1970. [10] NF P 11-300, Classification of materials for use in the construction of embankments and capping layers of road infrastructures, AFNOR, France, 1992.

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 545 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-545

Hydraulic Conductivity of Compacted Foundry Sand Treated with Bagasse Ash Kolawole OSINUBI.a,1 and George MOSESb a Ahmadu Bello University, Zaria, Nigeria. b Nigerian Defense Academy, Kaduna, Nigeria. Abstract.The hydraulic conductivity of compacted material is the most important parameter in assessing its suitability for use in engineered waste containment facility. Foundry sand mixed with up to 8% bagasse ash by weight of dry soil was evaluated for use as a suitable hydraulic barrier material. Tests were carried out on the foundry sand – bagasse ash mixtures to determine the index properties and compaction characteristics. The relationship between between hydraulic conductivity of the mixtures compacted using the British Standard light (BSL) energy with moulding water content, bagasse ash content and unit weight were determined. Test results show that the regulatory minimum hydraulic conductivity (k) value of 1 x 10-9 m/s or lower required for a material to be used in waste containment application can be achieved when foundry sand is treated with 4 % bagasse ash, prepared at molding water content in the range 11.2 – 15.4% and compacted to a unit weight of at least 17.56 kN/m3. Keywords. Bagasse ash, Compaction, Foundry sand, Hydraulic conductivity, Unit weight.

Introduction Solid waste consists of materials, which are classified according to their physical and chemical properties as garbage, rubbish, trash, junk and ashes [1]. Waste containment in engineered landfills is not yet enforced in Nigeria and there is no legislation on the subject. Compacted clay liners are normally used as an integral component of the lining system, to impede the transport of contaminants, to cover landfills, municipal and hazardous waste impoundments, and also to cap new or old waste disposal units [2]. Studies have been carried on the use of compacted lateritic soil as liners and cover in waste containment application [3, 4]. Foundry green sand had also been used with other additives such as bentonite in waste containment structures etc. [5]. In Nigeria, there are 50 commercial foundries with a total consumption capacity of about 4,000 tonnes of silica sand, bentonite and charcoal [6]. Large quantities of waste materials from mineral, agricultural, domestic and industrial sources are generated daily and the safe disposal of these wastes are increasingly becoming a major concern around the world. Sugar cane is grown on 25 – 30,000 hectares in Nigeria with a production rate of about 80 tons/hectare [7]. Bagasse is the fibrous residue obtained from sugar cane after the extraction of sugar juice at sugar cane mills, while bagasse ash is the residue obtained from the incineration of bagasse in sugar producing factories. Bagasse ash possesses pozzolanic properties. Research work has been carried out on the improvement of geotechnical characteristics of soils using bagasse ash [8]. However, no work has been done on the use of bagasse ash treated foundry sand in waste containment applications. The study was aimed at the evaluation of the hydraulic conductivity of compacted foundry sand treated with bagasse ash for use in waste containment applications.

546

K. Osinubi and G. Moses / Hydraulic Conductivity of Compacted Foundry Sand

1. Materials and Methods 1.1.Materials The foundry sand used in this study was obtained from Defense Industries Corporation of Nigeria (DICON), Kaduna (Latitude 10°30’N and Longitude 7°27’E), Nigeria. The bagasse ash utilized in this work was reported by (8) to be pozzolanic based on its oxide composition. The ash passed through BS No. 200 sieve (75 μm aperture) was mixed with foundry sand to form four different foundry sand – bagasse ash mixtures in stepped increment of 2% up to 8% by weight of foundry sand. 1.2.Methods • Index properties of the foundry sand and foundry sand – bagasse ash mixtures were determined in accordance with [9, 10]. Moisture - density relationship and hydraulic conductivity tests were carried out using air dried soil samples passed through a 4.76 mm aperture sieve. The BSL compactive effort that is easily achieved in the field was used to prepare specimens. • Hydraulic conductivity was measured using the rigid wall permeameter under falling head condition as recommended by [11]. A relatively short sample was connected to a standpipe, which provided the head of water flowing through the sample. Foundry sand - bagasse ash samples at the different bagasse ash contents (0, 2, 4, 6 and 8%) and different moulding water contents (-2, 0, +2 and +4%) of the OMC, respectively, were compacted using the BSL energy. The fully saturated test specimen was then connected to a permeant liquid (tap water). Hydraulic gradient ranged from 5 to 15. 2. Results and Discussion 2.1.Index Properties The index properties and compactions of the untreated and treated foundry sand are shown in Table 1. The non-plastic sand is classified as A-3 according to AASHTO shrinkage was not significantly affected since the soil is predominantly sand. Table 1: Physical properties of foundry sand and bagasse ash treated foundry sand

Bagasse Ash Content, %

Property

Liquid Limit, % Plastic Limit, % Linear Shrinkage, % % Passing BS No. 200 Sieve. AASHTO Classification USCS Classification Specific Gravity MDD, Mg/m3 OMC, % pH Value

0

2

4

6

8

19.0 NP. 0.9 31 A-3 SC 2.64 1.96 11.5 8.9

18.0 NP 1.0 26.5 A-3 SC 2.65 1.89 11.6 9.9

23.3 NP 0.0 27 A-3 SC 2.66 1.89 11.7 10.2

19.4 NP 0.9 27.5 A-3 SC 2.60 1.88 12.0 10.6

18.8 NP 0.7 26.5 A-3 SC 2.56 1.89 12.2 10.8

NP= Non-plastic

K. Osinubi and G. Moses / Hydraulic Conductivity of Compacted Foundry Sand

547

classification system [12] and SC according to the Unified Soil Classification System [13]. The particle size distribution curves are shown in Figure 1. The liquid limit initially slightly decreased in value from 19 to 18% and later increased to a peak value of 23.3% at 4% bagasse ash treatment. This increase can be attributed to the increase in water absorption or changes in the particle packing of the mixture. Beyond 4% bagasse ash content the liquid limit reduced in value. Foundry sand has been reported by [14] as not possessing plasticity, this largely due to the presence of a high percentage of fine sand and also the bentonite that was subjected to high temperature. Treatment of foundry sand with bagasse ash did not improve its plasticity,

2.2.Compaction Characteristics The effect of bagasse ash on the maximum dry density (MDD) and optimum moisture content (OMC) of the foundry sand - bagasse ash mixtures are shown in Figure 2. The

548

K. Osinubi and G. Moses / Hydraulic Conductivity of Compacted Foundry Sand

MDD generally decreased with higher bagasse ash treatment up to 8%. This was probably due to the comparative low specific gravity value of 2.20 of bagasse ash (12) compared to that of foundry sand which is 2.64. The OMC increased with higher bagasse ash treatment up to 8%. The OMC ranged from 11.5 to 12.2%. This was due to the increase in fines content because of inclusion of bagasse ash with larger surface area that required more water to react. It also could be due to the larger amounts of water required for the hydration of bagasse ash. These results are in agreement with those reported by [15]. 2.3.Effect of Moulding Water Content The variation of hydraulic conductivity with compaction molding water content for foundry sand with different bagasse ash contents is shown in Figure 3. Generally, the

trend is that of an initial decrease to minimum values and subsequent increases in hydraulic conductivity values. Hydraulic conductivity generally decreased with higher molding water content. Compaction with higher molding water contents resulted in foundry sand - bagasse ash mixtures that were devoid if macro pores which conduct flow. The arrangement of individual particles influenced by molding water content controlled the hydraulic conductivity [16]. Furthermore, soft wet clods of soil are easily remolded resulting in smaller inter-clod voids and hence lower hydraulic conductivity [17]. This result is consistent with those reported by [17]l and [4]. The foundry sand attained the maximum regulatory hydraulic conductivity value of 1 x 10-9m/s at 0 and 4% bagasse ash content. However, beyond 4% bagasse ash treatment there was an increase in hydraulic conductivity values possibly due to the presence of excess bagasse ash that would have changed the soil matrix [4]. 2.4.Effect of Bagasse Ash Content The effect of bagasse ash on foundry sand (see Figure 4) attained a minimum limiting hydraulic conductivity value at 4% bagasse and thereafter increased possibly due to the

K. Osinubi and G. Moses / Hydraulic Conductivity of Compacted Foundry Sand

549

presence of excess bagasse ash that would have changed the soil matric leading to increased flocculation [18].

2.5.Effect of Dry Unit Weight The variation of hydraulic conductivity with dry unit weight for foundry sand with different bagasse ash contents is shown in Figure 5. Generally, the trend is that of

decreasing hydraulic conductivity value to a minimum value with higher dry unit weight for all bagasse ash treatments and thereafter increased with reduction in dry unit weight. This trend occurs as samples are compacted from the dry to the wet side of optimum moisture content. The foundry sand attained the minimum regulatory hydraulic conductivity value of 1 x 10-9m/s at 4% bagasse ash treatment when compacted to a dry unit weight of 17.56 kN/m3. However, beyond 4% bagasse ash treatment hydraulic conductivity value

550

K. Osinubi and G. Moses / Hydraulic Conductivity of Compacted Foundry Sand

increased and did not meet the minimum regulatory requirement. This is in agreement with the findings reported by [4]. 3. Conclusion The hydraulic performance of foundry sand treated with up to 8% bagasse ash was studied. Specimens were prepared at moulding water contents -2, 0 +2 and +4% of optimum moisture content and compacted using the energy of the British Standard light. The hydraulic conductivity of specimens decreased to a minimum at 4% bagasse ash content and thereafter increased. The range of moulding moisture content required to achieve the regulatory hydraulic conductivity value was achieved at the moulding water content range of 11.2 – 15.4% at a compacted unit weight of 17.56 kN/m 3. This finding will assist in providing an economic means of disposal of foundry sand and bagasse generated by foundries and the sugar industry, respectively, that pollute the environment. References [1] T. D. Hagerty, A. Maigan, and E. Epstein, Waste disposal and resources recovery, Proceedings of the Seminar on Solid Waste Management, Bangkok, Thailand, 25th – 30th September, 1973. [2] C. H. Benson and D. E. Daniel, “Influence of clods on hydraulic conductivity of compacted clay, J. Geotech. Engrg., ASCE Vol. 116, No. 8, (1990). pp. 1231 – 1248. [3] K. J. Osinubi and C. M. Nwaiwu, Hydraulic conductivity of compacted lateritic soils, Journal of Geotechnical and Geoenvironmental Engineering, ASCE, Vol. 131, No. 8, (2005). pp. 1034 – 1041. [4] K. J. Osinubi and A. O. Eberemu, Compatibility and attenuative properties of laterite-blast furnace slag mixtures, Journal of Waste Technology and Management, Vol. 35, No. 1, (2009). pp. 7 – 16. [5] T. Abichou, C.H. Benson and G. T. Edil, Foundry green sand as hydraulic barriers laboratory studies, J. of Geotech. and Geoenvironmental Engrg. ASCE, Vol. 126. (2000). pp 1174 – 1183. [6] I. B. Okeke, and E. C. Sadjere, Foundry technology operating experience at Delta Steel Company (DSC) foundry, Proceedings of Nigerian Metallurgical Society Annual Conference, (1991). pp. 6 – 15. [7] S.M. Misari, L. A. Busari, and S. Agboire, Current status of sugar cane research and development in Nigeria. National coordinated Research Program on Sugarcane. (NCRP – SC), Proceedings of the Inaugural Meeting and Planning Workshop for Collaborators, 17 – 18 Aug. (1998), pp. 2 – 12. [8] K. J. Osinubi and T. S. Ijimdya,. ‘Laboratory investigation of engineering use of bagasse ash. Nigerian Society of Engineers Technical Transactions, Vol. 43, No. 1, (2008). pp. 1-17. [9] BS 1377, Methods of Testing Soils for Civil Engineering Purposes. BSIInstitute, London 1990. [10] BS 1924, Methods of Tests for Stabilized Soils. British Standard Institute, London 1990. [11] K. H. Head, Manual of Soil Laboratory Testing, Vol. 2. ,Permeability, Shear Strength and Compressibility Tests. Pentech Press, London 1992. [12] AASHTO, Standard specification for transportation materials and methods of sampling and testing, 14th Ed., Washington, D.C. 1986. [13] ASTM, Annual book of ASTM standards, Vol. 04. 08, Philadephia, 1992. [14] C.K. Johnson, Phenols in Foundry Waste Sand, Modern Casting, American Foundrymen’s Society, 1981. [15] P. G. Nicholson, and V. Kashyap, Fly-ash stabilization of tropical Hawaiian soils, In: Fly Ash for Soil Improvement. Ed. By Kevan D. Sharp. Geot. Spec. Pub. No. 36, pp. 15 – 29, 1993. [16] Y. Acar and I. Oliveri, Pore fluid effects on the fabric and hydraulic conductivity of laboratory compacted clay, Transportation Research Record, Vol. 1219, pp. 144 – 159, 1989. [17] J. P. Prashanti , P. V. Sivepullaiahand A. Sridharan, Pozzolanic flyash as a hydraulic barrier in landfills.” Engr. Geology, Vol. 60. Issue 1 – 4 June, (2001), pp. 245-252. [18] K. J. Osinubi and A. O. Eberemu , Desiccation-induced Shrinkage of Compacted Lateritic Soil treated with bagasse ash.” The 24th Inter. Conference on Solid Waste Technology and Management CD-ROM, 1518 March, Philidelphia, PA, U.S.A. Session 5C: Bio-reactors and Innovative Landfills, 2009, pp.856-867.

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 551 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-551

Subsurface Conditions in Central Khartoum Eisa A. Mohameda and Ahmed M. Elshariefb a University of Juba, Sudan b BRRI, University of Khartoum, Sudan

Abstract. Khartoum city lies between the Blue Nile and White Nile at the confluence of the two rivers. The city is nowadays experiencing huge development projects. Several structures of all types are currently under construction e.g. towers, bridges and underground structures. The understanding of the subsurface conditions is fundamental for good planning and design of these structures. Central Khartoum is dominated by thick layer of Miocene-Quaternary sediments geologically recognized as Upper Gezira Formation, underlain with sandstone (Lower Omdurman Formation). Geotechnical data from several service reports carried out in Central Khartoum were collected and analyzed. Variations in layering and geotechnical properties of the Upper Gezira Formation were statistically examined. It has been found that the Gezira Formation is constituted of an upper thin crust of hard silty sandy clay of medium to high plasticity overlying deposits of silt, silty sand and poorly graded sand. Contours are given for the depth of the upper clay crust, depth of the Nubian Sandstone Formation and depth of the water table. Important geotechnical factors governing and controlling the design of different foundation systems on the encountered soils were highlighted. These factors include presence of pockets and/or areas dominated by potentially expansive soils and loose silty sands.

Key words. Nubian Sandstone, Khartoum, Gezira Formation

Introduction The recent upsurge in the economic environment in Sudan joined with growing willingness to finance the construction activities places more demand on the engineering and building construction community. Greater Khartoum, with population more than 6 million, is the largest and the most important city in the Sudan. Accordingly great development and expansion in construction has taken place in Khartoum. This large urban development has occurred over a considerable area in three cities (Khartoum, Khartoum North and Omdurman) separated by three Niles, namely the Blue, White and River Niles. The geology and geomorphology of the cities are greatly influenced by the Niles. Thick alluvial deposits, mainly deposited by the Niles, lie conformably on Nubian Sandstone Formation (NSF) in Khartoum and parts of Khartoum North and Omdurman. The alluvial deposits are known for their

552

E.A. Mohamed and A.M. Elsharief / Subsurface Conditions in Central Khartoum

heterogeneity caused by their random nature of deposition. The challenge of founding buildings on the soils of greater Khartoum is always there. In this study geotechnical data has been collected from service reports carried out in Khartoum city for better understanding of geotechnical factors which influence the selection and design of foundations of light to heavy buildings in Khartoum. The study focuses on the area bounded to the north, east and west by the White and Blue Niles and to the south by latitude 1714000 m (15˚ 30΄ 15˝) N. The area at the confluence of the two Niles is particularly covered in more details in this study.

1.

General Geology

The Albian- Cenomanion age sedimentary bedrock of Khartoum area, referred to as "Lower Omdurman formation” is mainly covered with Miocene-Quaternary age, alluvial subsoil layers known as “Upper Gezira Formation”. The thickness of the alluvial subsoil layers varies within the limits, 10 to 30m.The major geological units in the study area can be distinguished in the following chronological order from bottom to top (Figure 1): basement complex, lower Omdurman formation, Wad Medani and older formations, lower Gezira formation, upper Gezira formation and superficial deposits ([1],[2] and [3]). More comprehensive hydrogeological studies in Khartoum area has shown that the depth to the static water level ranges from few meters to more than 100m ([4] and [5]). The aquifer system is developed mainly in the “Nubian Sandstone Formation” and partly in the Gezira Formation. It is recharged almost exclusively from the Nile Rivers, and divided into upper and lower aquifer zones.

Figure ͘1͘ĂGeneral geological map of Khartoum

E.A. Mohamed and A.M. Elsharief / Subsurface Conditions in Central Khartoum

553

Figure .1͘b Sketch block diagram of northern part of Khartoum basin[5]

2.

Data Collection and Analysis

Data from over hundreds of foundation investigation boreholes, representative of different sites, drilled in the region were utilized in the present study. The borehole depths range between 5 to 50m. They include mainly: location, depth, soil groups and soil parameters - liquid limit (LL), plastic limit (PL), plasticity index (PI), natural moisture content, bulk density (BD), dry density (DD), standard penetration test Nvalue, shear strength parameters (cohesion C and angle of internal friction Φ), the percentage passing #200 sieve ( 0.075 mm), the ground water level (GWL) and depth to rock. The spatial distribution of these boreholes was random and was dictated mainly by the availability of the data. To locate the investigated borehole sites, digital map of the Khartoum City was used as a base map. Global Positioning System (GPS) was used to read the East-North (EN) co-ordinates of the sites studied and their approximate altitudes. Observed borehole stratigraphy has been drawn and then generalized profile was selected for each site as a representative log. The following generalized trend of stratification and setting (starting from the ground surface downward) can be inferred from boreholes: • An upper crust of very stiff to hard silty clay of low to high plasticity (CL/CH) is identified in most of the boreholes. It extends down to about 10m from

554

E.A. Mohamed and A.M. Elsharief / Subsurface Conditions in Central Khartoum

ground surface in some boreholes. The clay fraction is dominated with montmorillonite clay mineral [6]. The highly plastic clay is stiff to very stiff and highly desiccated. The liquid limit varies from 50% to 159%, the plasticity index from 19% to 119% and the natural moisture content is generally close to the plastic limit at the crust[7]. The highly plastic clay has high to very high tendency for swelling. The low plastic clay (CL) is of moderate to high potential for swelling. It was encountered at the ground surface in some boreholes and extends down to a depth of 15 m. This soil too is desiccated and generally very stiff to hard. A contour for clay depth is given in Figure 2. • The low to medium plastic silt (ML) with rarely intermittent highly plastic clayey silt (MH) is identified below the highly to low plastic clays and extends to a depth of 16 m below the ground surface in some areas. It contains seams of silty sands or sands, and is generally medium stiff to hard. The SPT-N values measured in this zone varied between 4 to greater than 50 blows per foot . • Silty sand or clayey sand with few gravels are encountered beneath the clayey silt/silty clay stratum. The sand generally consists of silty sand, clayey sand and poorly graded sand (SC, SM, and SP) and extends down to a depth of 30 m from ground surface and generally starts as medium dense and becomes dense to very dense with depth. Some loose sands are also identified, but wellgraded sands are non-existent in the boreholes investigated. The measured SPT-N values vary between 8 to greater than 50 blows / foot. • The Nubian Sandstone (the lower Omdurman formation) is encountered below the sand deposits in most of the boreholes at depths generally ranging from 10m to 30m below the ground surface (Figure 3). The formation is quite variable in terms of type and strength (or quality). It is generally dominated by Sandstone with frequent interbedded layers of Mudstone occurrences. The lower Omdurman formation starts as highly weathered in some boreholes and becomes slightly to moderately weathered with depth. • The ground water table is strongly dependent on the water levels in the Blue and White Niles and varies during the different seasons of the year; it ranges from 3m near the White Nile to 14 m at the southern border of the study area (Figure 4). It can be clearly observed from the data collected and displayed that the encountered formations generally grade from fine material to coarse material with depth. Contour line maps for clay depth; depth to Nubian sandstone formation, and groundwater level were developed using Surfer Program (Version 1994) (Figures 2, 3 and 4). The clay depth has a general trend of decrease southwards whereas the depth to rock increases as one moves far from the White Nile and the water table depth increases southwards.

E.A. Mohamed and A.M. Elsharief / Subsurface Conditions in Central Khartoum

Figure 2: Clay depth contour

Figure 3: Nubian sandstone formation depth contour

555

556

E.A. Mohamed and A.M. Elsharief / Subsurface Conditions in Central Khartoum

Figure 4: Ground water depth contour

3.

Foundation Alternatives

Several types of structures such as residential and commercial buildings, towers, bridges, warehouses and industrial plants have been constructed in greater Khartoum during the last two decades. The buildings range in height from single story residential houses or offices to multi-story buildings rising to more than 30 floors. They were founded on strip, pad, raft and pile foundations depending on the geotechnical and structural factors governing each project. The geotechnical factors which govern the choice of a suitable foundation alternative in Khartoum city are ͗ •

The upper clay crust covers almost all of the study area except a small area south east of Khartoum airport, where the upper soil is clayey sand (SC). It is interesting to note that the extremely highly plastic clay (LL > 100) was encountered below 3.0 m depth in the area covered by SC and extends down to depth 7.0 m. The upper clay crust is mostly of high plasticity (CH) and extends down to about 10.0 m depth in some areas (Figure 2). It is highly desiccated and potentially expansive. Reports of damages to buildings caused by the upheaval of foundations have been reported for strip and pad foundations of light buildings founded on the CH, CL and even SC soils. Wherever the clay crust is shallow (less than 3.0 m depth) foundations for light buildings may be located on the stable soil right below the plastic clay. Short bored concrete piles are the best and safest alternative for light

E.A. Mohamed and A.M. Elsharief / Subsurface Conditions in Central Khartoum

557

structures where the clay is deep (more than depth 4.5 m). Suspended slabs are also adopted, wherever possible, to guard against floor movements.

ϰ͘

•

The SM/SP or SC formation occasionally found below the upper crust is medium dense and is partly above and partly below the water table. It becomes denser and coarser with lesser amount of fines with depth. It is considered as good support for raft foundation especially when basement floor(s) is considered. Dewatering problems may be envisaged for raft foundation option where the building is close to the Niles and double basement is needed.

•

The dense to very dense poorly graded sand SP usually encountered below the medium dense SM/SP and above the Nubian Sandstone is good base support (end bearing stratum) for piles. Bored concrete piles are often adopted when column loads are heavy. They are either extended to rest on the very dense SP or extended to socket in the Nubian Sandstone formation when very heavy column or pier loads are to be supported (e.g. high rise buildings and bridges across the Niles).

•

Challenges faced when constructing raft foundation are dewatering when the structure is located near the Nile(s) and support of the sides if excavations are deep.

•

Side collapse and boiling of sand inside the boreholes are problems facing pile contractors when construction of bored concrete piles is close to the Blue Nile in central Khartoum. Driven piles could be adopted if driving is carried out from levels below the upper hard clay crust. Failures to drive precast concrete piles from ground surface through the upper clay crust were realized.

•

The soils which are considered as problematic in the study area are the potentially expansive upper clay crust (CH/CL) and the probably liquefiable loose to medium dense saturated fine silty sand found close to the Blue Nile in Khartoum center.

Conclusion

The alluvial deposits in Khartoum area generally, start as clay near the ground surface and grade into silt and sand with depth. They lie conformably on the well known Nubian sandstone Formation. This paper gives summary of the analysis made on data collected from over 70 geotechnical service reports carried out in Khartoum. Contour maps were prepared for the depth of the upper clay crust, the depth to the Nubian Sandstone formation and the depth to bedrock. The maps show that the upper clay zone extends to a maximum depth of 10.0 m and is generally underlain by medium dense SM/SC becoming coarser and denser with depth down to the Nubian Sandstone Formation. The factors controlling or affecting the choice of foundation alternatives for structures in the study area have been outlined.

558

E.A. Mohamed and A.M. Elsharief / Subsurface Conditions in Central Khartoum

References [1] [2]

[3] [4] [5] [6] [7]

S. E. Ali, The effects of intrinsic Properties of Expansive Soils on their swelling and shrinkage behavior, M. Sc. Thesis, Building and Road Research Institute, University of Khartoum, Sudan, 2003 I.M. El Boushi and Y, Abdelsalam, Stratigraphy and Groundwater Geology of the Gezira Plain, Central Sudan, in Williams and Adamson (Eds.), 1982, A Land between Two Niles, Rotterdam, Balkema, pp 246 E. A. Farah, Groundwater Geology of the Northern Part of the Khartoum Basin -Central Sudan, M.Sc. Thesis, Department of Geology, University of Khartoum, Sudan, 1994 E.A. Mohamed , Khartoum City Subsoil Analysis, M.Sc. Thesis, Department of Civil Engineering , University of Khartoum, Sudan, 2001 M.K. Omer, The Geology of the Nubian Sandstone Formation in Sudan, Geol. and Min. Resources Department, Sudan , 1983, pp 225 E.M. Saeed, Hydrology of Khartoum Province and Northern Gezira Area, Geological Minerals Resources Department, Bulletin 29, 1976 A.G. Whiteman, The Geology of the Sudan Republic. Claredon Press, Oxford, Britain, 1971.

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 559 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-559

An Alternative to the Re-drive for determining Rod Friction exerted in DPSH Testing Charles MacROBERT a, Denis KALUMBA b, Patrick BEALES c a Anglo Technical Service, Johannesburg, South Africa b University of Cape Town, Cape Town, South Africa c Kantey & Templer Consulting Engineers, Cape Town, South Africa

Abstract: An uncertainty in Dynamic Probe Super Heavy (DPSH) penetration resistance values is rod friction during probing. The re-drive is a common method used in southern Africa to determine rod friction, however there is no reported use of torque readings on the rods. An investigation into the effectiveness of the two methods was carried out in a deep homogeneous sand deposit, with a shallow water table, at Capricorn in Cape Town, South Africa. The re-drive was undertaken at refusal by withdrawing the rods approximately 1 meter. The re-drive blow counts increased dramatically over the 1 meter suggesting wet sand collapsed into the void ahead of the probe. The torque readings were either taken each time a new rod was added or only once refusal had been reached. The incremental torque readings showed that the rod friction increased with depth below the water table. No great difference between single torque readings taken at refusal with those taken as part of determining the entire torque profile was found. From a comparison with SPT N values a method of determining the rod friction is suggested based on a single torque reading at refusal. Keywords: Geotechnical Site Investigations, Dynamic Probe Super Heavy (DPSH), Rod Friction, Re-drive correction, Torque correction

Introduction The Dynamic Probe Super Heavy (DPSH) test is a simple and cost effective test that can be used to investigate the soil strata below a proposed development. However, poor understanding of rod friction in the DPSH test has limited its acceptance as a cheap and practical design tool by engineers [1]. This paper presents a study into two methods of determining rod friction; re-drive corrections and torque corrections. The re-drive is a commonly used method in southern Africa while torque readings are commonly used in Europe. To investigate the practicality of the methods, full scale field testing was undertaken at Capricorn in Cape Town, South Africa. 1. Method Testing was carried out adjacent to a previous site that was investigated by Kantey & Templer [2]. A borehole probe from the study described the subsoil conditions as follows: “deep Quaternary age transported sand overlies the site to roughly a depth of

560

C. MacRobert et al. / An Alternative to the Re-Drive for Determining Rod Friction

50 m beneath which is soft rock of the Cape Granite Suite. The upper 3 m comprises of a loose to medium dense consistency soil increasing to dense at a depth of 9 m. A fluctuating ground water table exists at the site. A Fairbrother Geotechnical Engineering DPSH rig, illustrated in Figure 1, was used for the testing. The rig consisted of a trailer mounted A frame from which a continuous driven hammer (drive provided by a small petrol engine and hydraulic motors) with the drop height mechanically controlled was suspended. Extension rods (A rods, diameter 41.3 mm) with a disposable cone (60º, diameter 50.5 mm) were attached to the anvil and percussed into the ground at a rate of 25 blows/minute. In this study the DPSH blow counts were recorded over 100 mm then multiplied by 3 giving the equivalent value over 300 mm, the standard depth increment for both the SPT N and N30SB in southern Africa, resulting in a more continuous profile than adding 3 successive 100 mm values. Probing was carried out to a depth of 6 m as the consistency of the sand below this depth inhibited further progression due to time constraints.

Figure 1: DPSH Rig

A torque coupling was fabricated to enable a torque wrench to be attached to the drive rods each time a new rod was added. As illustrated in Figure 2, the coupling had a male end which screwed into the female end of an extension rod and a square socket on the other to which a torque wrench was attached. When machining the square nut onto the coupling care was taken to ensure that it was placed centrally to minimise any torsion effects. To calibrate the torque wrench the torque coupling was secured between the platens of a vice grip. Known weights at set distances were hung from the torque wrench and the corresponding reading recorded. Calibration was done before and after field testing to ensure accuracy. The torque wrench was used to measure the torque required to rotate the rods. Torque readings between 20 Nm and 200 Nm could be measured. The highest torque measured in this study was 105 Nm.

Figure 2: Profile view and field use of torque coupling and torque wrench

After probing to a depth of 6 m the rods were withdrawn 1 m (the length of each extension rod) and re-driven with the number of blows required recorded in the same manner as the initial drive.

561

C. MacRobert et al. / An Alternative to the Re-Drive for Determining Rod Friction

A total of 15 DPSH tests were carried out over a roughly 20 m 2 footprint. Table 1 describes the various test procedures followed to investigate the build up of rod friction. Table 1. Test Schedule Test Procedure

Qty

Torque Profile Test. Each time an additional extension rod was added to the string, a torque measurement was taken to determine the torque profile with depth. A re-drive at refusal was then undertaken to investigate the effects of rod rotation† on the re-drive values.

7

Re-drive Test. Testing involved percussing rods with minimal rod rotation† and undertaking a redrive at refusal. This was to investigate the effect of not rotating the rods on the re-drive values. (A break down during one re-drive tests resulting in a 2 hour suspension of probing at depth of 5.2 m)

4

Suspended Test. After percussing the rods with minimal rod rotation† a 30 min break was taken (to simulate a lunch break) after which the peak torque on the rods was determined and a re-drive undertaken. To investigate the effects of taking breaks on both the re-drive and torque values.

4

A perforated PVC pipe inserted down the probe holes enabled the determination of the water table level. † Various authors [3] [4] have suggested that rod rotation can relieve rod friction. During the torque profile tests the peak torque required on average 1.5 turns of the rods to determine. For the other tests rod rotation was kept to a minimum by gripping the percussed rods with a monkey wrench while adding the extension. However for all the tests any rotations either during rod addition or torque testing were recorded.

2. Results The consistency of the sand increased rapidly from very loose to medium dense in the first 1 m (Figure 3). After this it remained medium dense to just above the water table when it decreased to a loose state, a consequence of the fluctuating water table. Below the water table the blow counts increased rapidly. It is proposed that this was not a consequence of an increase in consistency but rather wet sand falling onto and gripping the rods and exerting a frictional force. It is illustrated in Figure 4 that the torque values increased below the water table in a similar manner to the blow counts. Above the water table very few torque readings were registered as less friction was exerted on the rods. It was found that single torque readings taken at refusal did not vary considerably from those measured from the entire torque profile. The re-drive profile showed a rapid increase in blow counts over the 1 m re-drive depth, see Figure 5. It is possible that wet sand collapsed into the void ahead of the rods and so this is not an adequate means of measuring rod friction. The sum of all the rotations was compared to the sum of all blow counts over each individual profile to determine the effect of rotation on the build up of rod friction. Hypothesis testing was done to determine the probability of a relationship between the rotations (dependent variable) and blow counts (independent variable). The probability of a zero intercept was extremely low at 3.9E-14. However, the probability of a zero slope was very high at 0.7. This implies that the average blow count is the best fit over all the rotations indicating that rotation has very little impact on the build up of friction.

562

C. MacRobert et al. / An Alternative to the Re-Drive for Determining Rod Friction

SPT N/N30SB 0

20

40

60

T orque (Nm) 80

100 0

1

1

2 3 4

150

GWT

4 5

6

6

7

7 M e a n DP S H

100

3

5

All DP S H

50

2

GWT Depth (m)

Depth (m)

0

0

Raw

SP T

Figure 3: N30SB Profile

Trend

Simplified

Figure 4: Torque Profile

Re-drive N30SB 0

10

20

30

40

4.5

Depth (m)

Raw

Mean

5 5.5

6 Figure 5: Re-drive Profile

Torque readings at refusal for the tests which were suspended were compared with the tests in which no breaks were taken. A two sample t-test was used to determine if the two samples were significantly different. The P-value was found to be 0.9 thus the null hypothesis was accepted implying the population means are equal and breaks have little effect on friction build up. Research by Cearns and McKenzie [5] however did show that probing at a reduced rate in clay leads to an increase in rod friction. 3. Torque Correction In a saturated sandy material it is proposed that a single torque measurement taken at refusal is sufficeint to determine the rod friction. Dahlberg and Bergdahl [6] proposed Equation 1 to determine the rod friction blows based on torque measurements.

C. MacRobert et al. / An Alternative to the Re-Drive for Determining Rod Friction

563

However further research by Bergdahl [7], Scarff [3] and Butcher et al [4] showed that this formula underestimates rod friction to varying degrees in different soils. Therefore a corrective constant, k, is required:

Rod Friction N 30 SB = k ×

2M v e DM o gH

(1)

Where Mv, torque (Nm); e, standard depth increment (300 mm); D, rod diameter (41.3 mm); Mo, hammer mass (63.5 kg), g, acceleration due to gravity (9.81 m/s 2) and H, hammer drop height (0.762 m). By comparing the torque corrected blow counts to blow counts corrected by the empirical method proposed by MacRobert et al [8] it was found that the corrective constant, k, varied with depth below the water table. To account for this variation the torque distribution below the ground water table has been simplified so a single corrective constant can be applied. Equation 2 below describes the distribution and Figure 4 illustrates the simplified distribution based on the collected data.

⎛d− G⎞ ⎟⎟ M v = ⎜⎜ − R G ⎝ ⎠

2.7

× TR

G ≤d ≤R

(2)

SPT N/N30SB 0

20

40

60

80

100

0 Raw DPSH

1

To r. Co r. DPSH Emp . Co r. DPSH

Depth (m)

2

SPT

GWT

3 4 5 6 7

Figure 6: Comparison of Corrected N30SB Profiles

Where d, depth below ground level (m); G, depth to ground water table (m); R, depth to refusal (m) and Tr, torque at refusal (Nm). With this distribution the value of the corrective constant is:

564

C. MacRobert et al. / An Alternative to the Re-Drive for Determining Rod Friction

k =12

(3)

Figure 6 compares the above proposed method to the blow counts corrected by the empirical method showing the resulting close correlation. 4. Conclusions • •

•

•

For the sandy soils that typically overly the Cape Flat area rod rotation did not appear to relieve rod friction. Breaks did not appear to increase rod friction. The re-drive method of determining rod friction was found to be time consuming and results suggested that wet sand fell into the void ahead of the probe with the blow counts measuring the resulting resistance. Increasing blow counts below the ground water table in the soil profile investigated corresponded to increasing torque values. A method requiring a single torque measurement at refusal is proposed to quantify the rod friction. Further testing would be required to determine how the corrective constant varies for different soils, as rod friction is not only developed in saturated sandy profiles as was the case with this study.

Acknowledgments The authors wish to acknowledge Fairbrother Geotechnical Engineering for the provision of a DPSH rig and MSP Developments for granting access to the test site. REFERENCES [1] C.J. MacRobert, D. Kalumba, P. Beales, Penetration testing: test procedures and design use in South Africa. Civil Engineering 18(3) 29-38, 2010. [2] Kantey & Templer, Borehole log sheet. Fibrecorp Capricorn Park. Job Number 80609T, Kantey & Templer archives, Cape Town, 2002. [3] R.D. Scarff, Factors governing the use of continuous dynamic probing in UK ground investigation. Penetration Testing in the UK, Thomas Telford, London, 29-132, 1988. [4] A.P. Butcher, K. McElmeel, and J.J. Powell. 1995. Dynamic probing and its use in clay soils, Advances in site investigation practice, Thomas Telford, London, 383-395, 1995. [5] P.J. Cearns, and A. McKenzie, Application of dynamic cone penetrometer testing in East Anglia. Penetration Testing in the UK, Thomas Telford, London, 123-127, 1988. [6] R. Dahlberg and U Bergdahl, Investigations of the Swedish ram sounding method. Proc. of the European Symposium on Penetration Testing (2.2). Stockholm, 93-101, 1974 [7] U. Bergdahl, Development of the dynamic probing test method, Proc. of the 7th European Conference on Soil Mechanics and Foundation Engineering Vol. 2, British Geotechnical Society, 201-206, 1979. [8] C.J. MacRobert, D. Kalumba, and P. Beales, Empirical equivalence between SPT and DPSH Penetration Resistance Values. Proc. of the African Regional Conference on Soil Mechanics and Geotechnical Engineering, Maputo 2011.

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 565 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-565

Empirical Equivalence between SPT and DPSH Penetration Resistance Values Charles MacROBERT a, Denis KALUMBA b, Patrick BEALES c a Anglo Technical Service, Johannesburg, South Africa b University of Cape Town, Cape Town, South Africa c Kantey & Templer Consulting Engineers, Cape Town, South Africa

Abstract: It is often necessary to determine design parameters from Dynamic Probe Super Heavy n values with correlations based on Standard Penetration Test N values. This is often done by assuming equivalence, as both tests use the same drive energy. In this study, an empirical equivalence was developed in sandy material from different geological depositional and weathering environments. Data collected from previous geotechnical investigations from across southern Africa was analysed. It was apparent that energy losses were greater in the DPSH test than in the SPT, leading to higher resistance values in the former. The SPT is carried out within a borehole whereas the DPSH is continuously percussed into the soil. The dynamic whipping of the DPSH causes soil to fill the small air annulus around the rods exerting a frictional resistance. The different geological settings of the test sites revealed that although various factors impact friction differently, the equivalence varied in a similar manner. Hence an empirical equivalence function is suggested to determine equivalent SPT N values from raw N30SB values. Keywords: Standard Penetration Test (SPT), Dynamic Probe Super Heavy (DPSH), Empirical Equivalence, Geotechnical Site Investigations, Dynamic Penetration Testing, Field Tests

Introduction A major requirement of geotechnical engineering is an understanding of the soil and rock profile below a proposed development. Two commonly used in-situ methods in southern Africa are the Standard Penetration Test (SPT) and Dynamic Probe Super Heavy test (DPSH). Both tests are dynamic in that a 63.5 kg hammer is repetitively dropped 760 mm along a guide rail onto an anvil driving a string of rods with a probe attached at the end. However the tests are different in the procedure followed and the probe shape. To undertake the SPT a small diameter (±76 mm) borehole is sunk and at set intervals (±1 m) a 50.5 mm diameter split spoon sampler is lowered into the borehole and then driven into the ground. The number of blows required to penetrate 450 mm is recorded. The initial blows penetrating the first 150 mm are disregarded to account for material disturbed by the drilling process. Hence, the SPT N value is the number of blows required to penetrate 300 mm. In the case of the DPSH a cone with a diameter of 50.5 mm and apex angle of 90º is driven continuously from ground level. The number of blows required to penetrate each successive 300 mm is recorded as the N30SB value [1] The “Franki Blue Book”, a widely used geotechnical engineering

566

C. MacRobert et al. / Empirical Equivalence

manual in southern Africa, suggests that the two tests should be taken as roughly equivalent in initial design [2]. This equivalence is based on the fact that the same drive energy is imparted to the probe per blow and assumes that energy losses due to friction along the drive rods are the same. The limited understanding of friction effects in the DPSH test limited its acceptance as a cheap and practical design tool by engineers [3]. This paper develops an empirical equivalence between the two tests so that correlations between design parameters and the respective penetration resistance values can be used interchangeably.

Figure 1: DPSH rig on left and borehole drilling for SPT testing on right

1. Method and Test sites To correlate the data between the DPSH and SPT, it was necessary to collect data in which both tests had been carried out within 5 m of each other. This was assumed to be close enough proximity for similarity of geological conditions. Seven data sets from different geological depositional and weathering environments across southern Africa were analysed. Data was obtained from geotechnical site investigations undertaken out by various practitioners, as carrying out special field tests on the scale required for this study was not financially feasible. Table 1 below describes various aspects of the tests sites from which data was obtained.

Table 1. Test Sites Test Site

Investigation Description

Geology of site

Bellville Site, Cape Town, South Africa.

Kantey & Templer conducted an investigation on a 0.1 ha site for a proposed office development.

The entire probed profile consisted of a transported fine to medium grained locally calcareous sand with the water table roughly between 1.5 to 2 m below ground level. [3]

C. MacRobert et al. / Empirical Equivalence

567

Test Site

Investigation Description

Geology of site

Gope, Central Kalahari Desert, Botswana.

Extensive investigations on a 450 ha site by Anglo American Civil Engineering Department for a proposed diamond mine.

Tertiary and Quaternary aeolian and fluviolucastrine sands cover the site with a variable poorly developed layer of calcified pedogenic material. Very low moisture content characterized the entire soil profile, any moisture was limited to the zone above the pedogenic horizon. [5]

Beluluane Industrial Park, Matola, Mozambique.

Extensive investigation by Golder Associates Africa on an 11ha site for a proposed steel mill.

The site comprises extensive silty sand dune deposits which becomes progressively consistent with depth. An open-voided structure results in a medium to high collapse potential. [6]

Milnerton, Cape Town, South Africa.

Kantey & Templer conducted an investigation on a 0.3 ha site for a proposed ground flare.

This site was characterised by a cover of transported loose to medium, slightly silty fine sand and residual very stiff clayey silt underlain by completely to highly weathered and fractured very soft Greywacke rock. [7]

Parow, Cape Town, South Africa.

Kantey & Templer conducted an investigation on a 0.4 ha site for a proposed office development.

The site was characterised by dipping layers fill (layerworks) transported material (medium dense to dense variably silty fine sand) and residual material. The water table was at a depth of between 1.2 and 1.25 m. [8]

Umdloti, Durban, South Africa.

Franki Africa carried out an investigation on a 0.4 ha site for a proposed retaining wall behind an apartment block.

A 6 to 7.5 m thick slightly moist to moist fine through to medium through to coarse grained sand underlined the site with the water table sitting at a depth of between 5.3 and 8.5 m. [9]

The data was correlated in a method similar to that used by Cearns and McKenzie [10] for DPSH and SPT results in sand and gravel material in East Anglia, England. Cearns and McKenzie plotted the SPT N/ N30SB ratio against the respective N30SB value. Plotting the data in this manner produced an inverse relationship that was then fitted with a curve to determine the correlation factor at each N30SB value. However in this study the ratio was reversed and so the N30SB /SPT N ratio, KDPSH value, was plotted on the y axis against the respective N30SB value on the x axis [11]. This resulted in linear relationship, defined by Equation 1, with a and b the slope and intercept respectively: KDPSH = a × N30SB + b

(1)

The Equivalent SPT N is then given by dividing the respective N 30SB value by the KDPSH ratio: Equivalent SPT N =

N30SB KDPSH

(2)

The relationship between N30SB values and KDPSH values defined by Equation 1 is then substituted into Equation 2 to give a non dimensional empirical equivalence function, Equation 3: N30SB Equivalent SPT N = a × N 30SB + b

(3)

568

C. MacRobert et al. / Empirical Equivalence

2. Empirical Equivalence

Correlation factors for each site, KDPSH values, were determined and plotted on the y axis against the respective N30SB values on the x axis as illustrated in Figure 2. During the analysis of trends from each individual site it was envisioned that each site would result in site specific correlation factors due to the different depositional environments. However the trends obtained appeared to be very consistent with one another. Using the values for the slope and intercept from for the combined data the empirical equivalence function, Equation 3, could now be fully defined giving Equation 4: N30SB Equivalent SPT N = 0.02 × N 30SB + 0.8

(4)

12 KDPSH = 0.02 × N30SB + 0.8

10

2

R = 0.5

KDPSH

8 6 4 2 0 0

20

40

60

80

100

120

140

160

N30SB Bellville

Gope

Matola

Milnerton

Parow

Umdloti

Regression

Figure 2: Correlation Graph

Cearns and McKenzie [10] as well as Warren [12] presented graphical correlations based on smaller data sets in sand and gravel and fine clayey sands with clayey layers respectively. Although these tests sites were from different geological depositional and weathering environments to southern Africa the results can be compared to the results obtained in this study. The comparison is presented in Table 2. It is clear that the correlation factors obtained with the formula fit closer to the correlation presented by Warren over the entire range of N30SB values. The Cearns and McKenzie’s correlation factors up to a N30SB value of 45 are similar to the ones obtained by the equation. Above this the correlation factors diverge however very few data points above a N 30SB value of 30 were presented by Cearns and McKenzie. Reading values from the graphs could have introduced errors as could the different geologies and testing procedures. The research by Cearns and McKenzie and by Warren appears to validate the research reported.

C. MacRobert et al. / Empirical Equivalence

569

Table 2. Comparison of Correlation Factors Average Correlation Factor over N30SB range N30SB

Cearns & McKenzie §

Warren§

Proposed Function†

0 – 15

0.8

0.9

1.0

15 – 30

1.5

1.1

1.3

30 – 45

2.0

1.7

1.6

45 – 60

3.0

2.1

1.9

60 – 75

4.5

2.3

2.2

§Interpreted graphically †Determined by the Mean Value Theorem

3. DISCUSSION

Analysis showed that N30SB values below 20 were roughly equivalent to the respective SPT N values however with increasing depth and at certain layers the N30SB values became larger. It was proposed that because the SPT is carried out in a borehole, the extension rods do not come into contact with the soil. Although the DPSH is carried out with rods of a smaller diameter than the cone, material falls into the annulus surrounding the rods due to the dynamic nature of the test and ensuing vibration of the rods during impact. This material then exerts a rod friction resulting in larger energy loses in the DPSH than in the SPT. Probing in sandy material from a number of different depositional and weathering environments was analysed and different mechanisms are proposed that can lead to the rod friction: • •

•

Poorly developed pedogenic layers in dry sandy profiles resulted in increased fines and weak cemented gravel fractions which can impede the DPSH probe and increase friction along the rods. Collapsible sands with considerable fines content can fall into the annulus around the DPSH rods as a result of vibrations and exert a friction. (It must be kept in mind that collapsible sands can loss strength when saturated. The DPSH and SPT tests are both likely to over estimate soil strength when probing in low moisture conditions. The proposed correction does not take into account any moisture induced collapse on penetration values; as compared values were from tests done under dry conditions) Partially saturated and saturated sands appeared to also fall into the annulus around the DPSH rods and exert a frictional force leading to increased blow counts.

Although the proposed mechanisms varied the statistical correlation showed a close relationship. However a site with an abundance of shell fragments in the profile appeared to result in excessively high N30SB values. This data set was excluded from the

570

C. MacRobert et al. / Empirical Equivalence

analysis due to limited information for the site and hence knowledge of the local geology is essential for any interpretation of penetration resistance values.

4. Conclusions • •

•

This study analysed geotechnical site investigation data gathered from across southern Africa with the aim of developing an empirical equivalence between N30SB and SPT N values. Considering the general discontinuity of soil properties a close statistical relationship was found between the majority of the SPT N and N30SB data from the different depositional and weathering environments analysed. However it was highlighted that certain anomalies can exist and knowledge of the local geology is essential. This relationship allowed for the development of an empirical equivalence function given by Equation 4. This function allows for equivalent SPT N values, which can be used in design, to be obtained from N30SB values, making this cheap and easy test even more practical.

Acknowledgments The authors wish to acknowledge Anglo Technical Services, Golder Associates Africa, Kantey & Templer Consulting Engineers and Franki Africa for kindly providing the analysed data.

References [1] EN1997-2:2007. 2007. Eurocode 7: Geotechnical design. Part 2: Ground investigation and testing [2] I.H. Braatvedt, et al, A Guide to Practical Geotechnical Engineering in Southern Africa 3rd Edition, Johannesburg: Frankipile South Africa, 1995. [3] C.J. MacRobert, D. Kalumba, P. Beales, Penetration testing: test procedures and design use in South Africa, Civil Engineering 18 (3), 29-38, 2010. [4] Kantey & Templer, Report on Geotechnical Investigation for Senate Building at CPUT Bellville, Kantey & Templer archives, Cape Town, 2007. [5] P.M. Fourie, and A. M. Copeland, Geotechnical investigation report for mining infrastructure at the Gope prospect in central Botswana, Anglo American Civil Engineering Dept., Johannesburg, 1998. [6] Golder Associates Africa, Geotechnical Investigation proposed Mittal steel plant Beluluane Park, Matola, Maputo, Mozambique, Golder Associates Africa archives, Johannesburg, 2008. [7] Kantey & Templer, Report on Geotechnical Investigation for New Ground Flare, Milnerton Refinery, Kantey & Templer archives, Cape Town, 2008. [8] Kantey & Templer, Report on Geotechnical Investigation for UNISA Phase 2 Development, Parow, Kantey & Templer archives, Cape Town, 2009. [9] Franki Africa, Raw SPT and DPSH records for project name: Apartments Umdloti (T06/1205), Franki archives, Johannesburg, 2003. [10] P.J. Cearns, and A. McKenzie, Application of dynamic cone penetrometer testing in East Anglia. Penetration Testing in the UK, Thomas Telford, London, 123-127, 1988. [11] C.J. MacRobert, D. Kalumba, P. Beales, Correlating Standard Penetration Test and Dynamic Probe Super Heavy penetration values in sandy soils, Journal of the South African Institute of Civil Engineering 53 (1), 46-54, 2011. [12] G. Warren. Heavy Weight. Ground Engineering, 40(3): 36-39, 2007.

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 571 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-571

The Dynamic Probe Super Heavy penetrometer and its correlation with the Standard Penetration Test Brian HARRISONa and Tony A’BEARb Inroads Consulting cc, Johannesburg, South Africa b Bear GeoConsultants (Pty) Ltd, Johannesburg, South Africa a

Abstract. The Dynamic Probe Super Heavy (DPSH) is widely used in Southern Africa as an aid for geotechnical investigations. Little information, however, is available on the interpretation of data recovered from it for quantitative design purposes. The Standard Penetration Test (SPT), on the other hand, has been extensively researched and much literature is available both locally and internationally for geotechnical design and analysis employing data from it. This paper attempts to establish a relationship between the DPSH and SPT tests in an attempt to make better use of the information obtained from the DPSH penetrometer. Keywords. Dynamic Probe Super Heavy, Standard Penetration Test

Introduction Continuous dynamic penetration testing is often carried out in Southern Africa as part of routine geotechnical investigations. It generally supplements the information obtained from test pits excavated by means of a mechanical backhoe that seldom excavate to depths of much more than 5 m. Two of the more common tests include the Dynamic Cone Penetrometer (DCP), sometimes referred to as the Dynamic Probe Light (DPL), and the Dynamic Probe Super Heavy (DPSH). Much information is available for utilising the data obtained from the DCP, [1][2][3] but the same cannot be said of the DPSH. In fact, both locally and internationally, there is a dearth of literature available on the manner in which the DPSH should be interpreted. The Standard Penetration Test (SPT) is also widely used as an internationally accepted dynamic penetration test method for carrying out geotechnical investigations. It is because of its widespread use that numerous papers and technical literature have been written on the test and interpretation of the data obtained from it. In light of this, much benefit would be obtained from the DPSH if some relationship existed between it and the SPT.

572

B. Harrison and T. A’Bear / DPSH Penetrometer and its Correlation with SPT

1. The DPSH Test Due to the wide variety of continuous dynamic penetration testing in use, Stefannof and others [4], at the ISSMFE 1st International Symposium on Penetration Testing, presented a reference test procedure for penetrometers. Details of the DPL and DPSH specifications are summarised in Table 1. Table 1. ISSMFE dynamic probe classifications Test type

SPT

DPL

DPSH

Hammer mass - kg

63,5

10

63,5

Hammer fall - m

0,76

0,5

0,75

Rod length - m

3

1

1-2

Maximum rod mass - kg/m

8

3

8

Rod OD - mm

70

22

32

Apex angle - degrees

-

90

90

Nominal area of cone - cm2

-

10

20

Cone diameter - mm

(51)

35,7

50,5

Mantle length of cone - mm

-

35,7

50,5

Standard range of blows

5 - 100

3 - 50

5 – 100

Specific work per blow - kJ/m2

241

50

238

The DPSH test employed in Southern Africa utilises a 50 mm diameter, 60° disposable steel cone that is driven into the ground by means of a 63,5 kg hammer. The hammer falls through a height of 762 mm onto an anvil which is attached to flushcoupled “EW” rods that butt up against each other when driving. The rods are typically 1 m long and 33 mm outside diameter which is less than that of the cone attached to them. The number of blows required to drive the cone 300 mm into the ground is recorded as the DPSH “n” number and this is plotted against the depth. Once the required penetration depth, or refusal depth, has been reached the rods are withdrawn leaving the disposable cone at the bottom of the hole. “Re-drives” are sometimes carried out by recording the number of blows required to penetrate the rods after withdrawing them 600 mm from the cone. This is meant to provide an indication of the resistance to penetration offered by the soil on the rods. From the data acquired a continuous record of penetration resistance with depth is obtained, similar to that illustrated in Figure 1. Typically, and depending on the size of the site, about 6 to 7 tests can be carried out per day for penetration depths of the order of 10 m.

B. Harrison and T. A’Bear / DPSH Penetrometer and its Correlation with SPT

573

0 2

Depth - m

4 6 8 10 12 14 0

10 20 30 40 50 60 DPSH Penetration 'n' -blows per 300 mm

70

Figure 1. Typical DPSH penetration plot

As illustrated in Table 1, the ISSMFE standard differs to that of Southern Africa in that the 50 mm diameter cone has a 90° apex and a 50 mm long mantle. Also, the number of blows to penetrate 100 or 200 mm increments is frequently recorded and presented as the DPSH “n” number. Direct DPSH relationships between tests from Southern Africa and those from the Europe and the UK should therefore be used with these factors in mind.

2. Use of the DPSH Test Data from the DPSH test have been applied mainly in determining the consistency, or change in consistency, of the underlying soil horizons with depth. This application is particularly useful in non-cohesive sand and gravel horizons where rod friction is minimal. Other qualitative uses include establishing variations in subsoil conditions between boreholes, or test pits, and boundaries between different soil types. The test is also often used as an aid in establishing the founding depth of driven piles. Notwithstanding the fact that quantative uses for determining soil parameters from penetration testing, in general, are considered with much scepticism by many engineers and geologists, the technique employed for estimating soil parameters from the DPSH test is largely confined to converting the DPSH penetration for 300 mm, or “n” number, to the SPT “N” number from which soil parameters can be inferred and used for design.

3. A Review of DPSH and SPT Relationships The hammer mass and drop height, or energy, utilised for the DPSH test is similar to that of the SPT and so, by implication, a relationship should exist between the two tests. In Southern Africa it is standard practice to employ a one-to-one correlation given by Byrne & Berry [5], although no support of this relationship is provided in the reference. MacRobert et al [6] also refer to a one-to-one correlation employed by some geotechnical practitioners, but in addition they mention a rule of thumb association of SPT-N = 1,2 DPSH-n.

574

B. Harrison and T. A’Bear / DPSH Penetrometer and its Correlation with SPT

Some work done in the United Kingdom by Warren [7] has shown that linear relationships occur within a range of DPSH “n” values (n being the blow count for 100 mm penetration). For example on a site underlain by fine clayey sands with clay layers for n = 0 to 5, SPT-N = 1,67 DPSH-n, and with n > 13, SPT-N = 1,33 DPSH-n. In the latter relationships the DPSH-n values have been “converted” by the authors of this paper to blows counts for 300 mm penetration. Spagnoli [8] reviewed numerous correlations between the SPT and various penetrometers. He refers to work in Japan where correlations of SPT-N = 0,87 DPSHn was found at one site, where no mention of the soil type was given. Other correlations in Italy include SPT-N = 1,67 DPSH-n for sandy-silty gravels, and SPT-N = 2,0 DPSH-n for fluvio-lacustral deposits of sandy silt with gravels. Tolia [9] refers to work done by Golder and Desai & Roy who drove a 51 mm diameter and 60° cone attached to A-rods by means of a 63,5 kg hammer falling 76 cm. Golder found the approximate relationship of SPT-N = 0,67 DPSH-n for fine sand up to a depth of 7,5 m. Desai & Roy established a direct correlation between the two tests of SPT-N = DPSH-n up to a depth of 6 m.

4. DPSH and SPT Testing Programme In an attempt to determine whether any relationship exists between the DPSH and SPT for soils in Southern Africa, close on 100 comparative tests were carried out within a variety of soil types. The equipment used for this comprised a mobile track mounted Dando Terrier 2002 drilling rig fitted with both window sampler and penetration testing equipment. Window samples from 65 to 85 mm were taken with the equipment by driving a sampling barrel, containing an inner sleeve comprising a 1 m long plastic tube, into the soil. After recovering a sample, the Raymond Spoon comprising the SPT probe was lowered into the bottom of the hole and the test carried out in the standard manner employing an automatic trip hammer. Once the test was completed a 1 m long window sample was again taken and the procedure repeated until the driving shoe of the sampler could no longer penetrate the soil. This process provided SPT test data at 1 m intervals with a soil profile between each test. On completing the window sampling and SPT testing, a DPSH test was carried out using the same machine and trip hammer over the entire depth of the sampling and testing and in close proximity to it. DPSH-n counts were recorded over 300 mm intervals in this instance as is typical of the South African approach. A plot of the data is presented in Figure 2 from where it can be seen that a tenuous trend exists between the two tests.

B. Harrison and T. A’Bear / DPSH Penetrometer and its Correlation with SPT

575

80

N=1,2n

70

SPT - N

60

N=n

50 40 30 20 10 0 0

10

20

30

40 50 DPSH - n

60

70

80

Figure 2. DPSH (n) vs. SPT (N).

The relationship between the two tests is given by: SPT-N = 1.20 DPSH-n with a correlation coefficient r2 of 0.672. Upper and lower bound correlations equate to: SPT-N = 0.50 DPSH-n (lower bound) SPT-N = 2.00 DPSH-n (upper bound) Figure 2 does not differentiate between cohesive and non-cohesive soil and with a coefficient suggesting relatively poor correlation, two further plots were generated to determine whether a better relationship exists for non-cohesive and cohesive soils separately, and these are presented in Figures 3 and 4 respectively. 80 70

SPT-N

60

N=n

50 40 30 20 10 0 0

10

20

30

40 50 DPSH-n

60

Figure 3. DPSH vs. SPT for non-cohesive soil.

70

80

576

B. Harrison and T. A’Bear / DPSH Penetrometer and its Correlation with SPT

80 70

N=n

SPT-N

60 50 40 30 20 10 0 0

10

20

30

40 50 DPSH-n

60

70

80

Figure 4. DPSH vs. SPT for cohesive soil.

The relationships obtained from the data for the two soil types have been determined as: Non-cohesive soils SPT-N = 1.17 DPSH-n with a correlation coefficient r2 of 0.601. Upper and lower bound correlations equate to: SPT-N = 0.70 DPSH-n (lower bound) SPT-N = 2.00 DPSH-n (upper bound) Cohesive soils SPT N = 1.25 DPSH n with a correlation coefficient r2 of 0.626. Upper and lower bound correlations equate to: SPT-N = 0.57 DPSH-n (lower bound) SPT-N = 2.27 DPSH-n (upper bound)

5. Comments on the Test Results All of the tests carried out in the course of undertaking this investigation utilised the same Dando Terrier 2002 machine, and the automatic trip hammer ensured that the energy imparted for both the SPT and DPSH tests were the same, only the soil conditions changed. A few possible reasons for the significant variation in penetration rates and correlation for the two penetrometers are proposed and discussed further below. 5.1. Whip Whip, or bowing, during driving has been given as a cause of variability in correlation between the test types. It has been suggested by MacRobert et al. [6] that whip can cause the rods to jam in the hole during driving, and also generate skin friction through lateral vibration, thereby reflecting higher penetration rates. However, the reference procedure of ISSMFE [4] recognise the influence of skin friction in cohesive soils and suggest that in order to limit such problems the cone should have a

B. Harrison and T. A’Bear / DPSH Penetrometer and its Correlation with SPT

577

diameter of at least 30 % greater than that of the rods. The 50 mm cone and 33 mm diameter EW rods used for the DPSH test meet this criterion and thus it must be assumed that the effect of whip is not too significant, although its effect should not be ignored when scrutinising data. 5.2. Verticality It is virtually impossible to maintain a vertical 33 mm diameter rod over any significant depth within a soil mass. More so when driving a cone through a soil with inclusions such as roots and gravels, since they tend to deflect the tapered cone off the vertical. The effect of an inclined rod is to generate friction or adhesion against the soil reflecting high penetration. In an attempt to assess the degree and extent to which this occurs, the ratio of SPT-N and DPSH-n recorded at various depths were plotted with depth. The results are presented as Figure 5 from where it is evident that there is no trend to a uniform correlation at shallow depth and a more variable one at greater depth. This tendency would be expected if rod friction or adhesion played a uniform or consistent role in the recorded penetration of the DPSH, since its effect would be more significant at greater depths.

3.0 2.5 SPT-N/DPSH-n

2.0 1.5 1.0 0.5 0.0 0

1

2

3

4

5 6 Depth (m)

7

8

9

10

Figure 5. SPT/DPSH correlation with depth.

Re-drives are sometimes carried out at the bottom of the hole in order to assess the rod friction, or adhesion in the case of plastic soil, developed during driving. Unfortunately not many re-drives were done during the course of carrying out the DPSH tests in this investigation. What is of interest, however, is the re-drive values of five tests carried out in the same soil, namely a residual diabase in which the clay and silt fractions comprised 70 % of the soil and which had a plasticity index of 15 and grading modulus of 0,38. In-situ water contents and profiles of the window samples reflected a moist to very moist soil. Re-drive values for the bottom 600 mm of the rods at depths ranging from 5,5 m to 8,0 m were recorded at between 64 % and 91 % and averaging 78 % of the original penetration rates over the same bottom 600 mm. Such wide ranges in re-drives for the same soil can again be attributed to non-verticality of the rods, which when detached from the cone rest against the side of the hole, and

578

B. Harrison and T. A’Bear / DPSH Penetrometer and its Correlation with SPT

when re-driven penetrate into the soil at the sidewall of the hole registering high penetration rates. Similar results were noted in deeply weathered residual andesite found on a site in Braamfontein, Johannesburg. In light of the above, the value of carrying out re-drives to estimate rod friction is questionable, and it is perhaps better practice to eliminate, or at least reduce, friction and adhesion by regularly rotating the rods as they are added to the string during extension. 5.3. Sidewall Collapse Probably the most significant cause of high penetration rates is the collapse of soft plastic soil into the hole formed by the cone as it advances. This would typically occur under very moist or wet conditions where low strength cohesive soil is unable to support itself, and adheres to the rod causing it to be “held up” and reflect high blow counts. First hand experience of this was encountered when DPSH tests “refused” at 6 m in a low lying depression underlain by very moist, soft, clayey silt. Piles were subsequently installed and these were founded at 18 m below surface as the soft soil extended some 9 m more than indicated by “refusal” of the DPSH tests. 5.4. Inclusions and Layering Inclusions such as roots and gravels can significantly affect correlation. This is more apparent in gravels since resistance fluctuates significantly in these soils, particularly if loosely packed and the average particle size is of the same order or greater than that of the cone. Under these conditions penetration is governed by whether the cone strikes the gravels centrally, reflecting high penetration, or strikes it on the edge so as to cause it to be pushed aside thereby recording a relatively lower penetration. Random gravels in a homogeneous profile may also reflect high penetrations when stuck and not replicate the conditions of the surrounding matrix. It has also been noted that SPT results, ostensibly at the same depth as an adjacent DPSH probe, reflect the horizon either just above or just below a more, or less, competent but thin horizon registered on the DPSH results. In this instance there may be a significant difference in result primarily due to a small vertical shift in a thin horizon. 5.5. Repeatability Repeatability of both the DPSH and SPT was investigated by MacRobert et al [6] to establish whether one of the test methods is more precise and hence more readily replicated than the other. Numerous tests carried out at a site underlain by silty sand found far more scatter of the DPSH than the SPT tests. At another site underlain by fill, transported silty sand and residual clayey silt, the SPT tests showed more scatter than the DPSH tests. From this it would appear that repeatability of the two tests is dependant on the soil type and its consistency.

B. Harrison and T. A’Bear / DPSH Penetrometer and its Correlation with SPT

579

6. Conclusions The equipment used to carry out the DPSH test is compact and mobile which, together with the ease and speed with which tests can be undertaken, makes the method attractive, economic and popular for use as part of routine geotechnical investigations. Based on correlations between close to 100 penetration tests, relationships between the SPT-N and the DPSH-n have been determined for a range of soil types. These, however, have poor correlations and it is suggested that this is due largely, but not only, to non-verticality of the driving rods of the DPSH equipment. The effect of an inclined rod is to generate friction or adhesion in non-cohesive and cohesive soils respectively, which results in penetration rates higher than would otherwise be the case. In light of the findings and assumptions presented in this paper, a one-to-one relationship between the SPT-N the DPSH-n is considered sufficiently adequate, but perhaps marginally conservative, for estimating the SPT “N” number from the DPSH “n” penetration. In view of the poor correlation for these relationships, parameter determination from the DPSH probe should be treated with circumspection and is not recommended. It is best used as a means of extrapolating data obtained from more reliable test procedures.

Acknowledgments The authors wish to acknowledge the enthusiasm and help of the Enviro Geotech Drilling Services personnel whose Dando Terrier rig was used in all of the tests described in this paper.

References [1] Kleyn, E. G. 1975. The use of the Dynamic Cone Penetrometer (DCP). Transvaal Roads Department. Report L2/74 Pretoria. [2] De Beer, M. 1991 Use of the Dynamic Cone Penetrometer (DCP) in the design of road structures. Proc. 10th regional conf. for Africa on SM&FE & 3rd Int. conf. on tropical & residual soils. Maseru 23-27. Sept 1991. [3] Burnham, T. & Johnson, D., 1993 In situ foundation characterisation using the Dynamic Cone Penetrometer. Report No 9PR3001 Minnesota Department of Transportation May 1993. [4] Stefanoff, G., Sanglerat, G.,Bergdahl, U., and Melzer, K. 1988 (Members of ISSMFE Technical Comm. on Penetration Testing), "Dynamic Probing (DP): International Reference Test Procedure," Penetration Testing 1988, ISOP -1, De Ruiter (ed.), Balkema, Rotterdam, ISBN 90 6191 801 4. [5] Byrne, G & Berry, A.D. 2008. A Guide to Practical Geotechnical Engineering in Southern Africa. Frankipile South Africa 4th Edition revised and updated. [6] MacRobert, C. Kalumba, D. & Beales P. 2010. Penetration testing: test procedures and design use in South Africa. South African Institute of Civil Engineering Civil Engineering April, Vol 18 No 3 pp 2938. [7] Warren, G. 2007. Heavy Weight. Ground Engineering March, Vol 40 No 3 pgs 36 – 39. [8] Spagnoli, G. 2007 An empirical correlation between different dynamic penetrometers. Electronic Journal of Geotechnical Engineering (EJGE) Vol 12. [9] Tolia D. S., 1977. Interpretation of dynamic cone penetration tests with particular reference to Terzaghi and Peck’s chart. Ground Engineering, October 1977, pp 37-41.

580 Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-580

The Potential of Using Artificial Neural Networks for Prediction of Blue Nile Soil Profile in Khartoum State H. ELARABI and M. MOHAMED University of Khartoum, Khartoum, Sudan

Abstract. Artificial Neural Networks (ANNs) are an Artificial Intelligence technique. In this study, ANNs are used for prediction of soil classification in specified locations within the flood plain of the Blue Nile in Khartoum at different depths. The study was based on the available site investigation data collected from specific areas in Sudan. About 38% of the total data collected has been used as input data. This data applied directly to neural network and the remaining percentage of the total data (about 62 % of total) has been used as tested data. Thirteen models of Neural Networks were constructed and developed to predict soil layers in specified locations in Khartoum Blue Nile area. The results were then compared with data brought from actual boreholes to check the ANN model’s validity. The results indicated that Neural Networks is a useful technique for predicting the soil profile in the studied areas. Keywords. Artificial Neural Networks, Blue Nile, Prediction, Sudan

Introduction The behavior of every foundation depends primarily on the engineering characteristics of underlying deposits of soil and rock. Therefore, the foundation engineer must be able to distinguish among the various deposits of different character, to identify their principal constituents, and to determine their physical properties [1]. The objective of this paper is to develop complex relationship among the design parameters to build up and to investigate a powerful network capable of predicting soil profile based on previously investigated site conditions.

1. Development of Neural Network Model The steps for developing ANN models in this work include the determination of model inputs and outputs, division and preprocessing of the available data, the determination of appropriate network architecture, optimization of the connection weights (training) and model validation. The computer software program used is NEUROSHELL2 Version 4.0. The data used to calibrate and validate the neural network model were obtained from field measurements of soil investigation done by Building and Road Research Institute. The data cover a wide range of Blue Nile area in Khartoum state. The database comprises a

H. Elarabi and M. Mohamed / The Potential of Using ANNs for Prediction of Blue Nile Soil Proﬁle 581

total data of 56 individual cases; data of 21 cases were used as input data and the remaining data (data of 34 cases) has been used for prediction as rested data. 1.1. Inputs and Outputs Models The number of input and output for each model can be simplified in Table 1. Function approximation is one of the most powerful uses of neural networks. Typically, a two or three layer network is sufficient to approximate any function with a finite number of discontinuities. In order to gain an insight as to how topology affects the outputs, tangent-sigmoid, logarithmic-sigmoid and pure linear neuron (transfer) functions were selected for further investigation.

Table 1. Input and output for classifier and parameters models Model’s Name Model(1) Model(2) Model(3) Model(4) Model(5) Model(6) Model(7) Model(8) Model(9) Model(10) Model(11) Model(12) Model(13)

Input parameters - E,N coordinates -Descending depth from the altitude - E,N coordinates -Descending depth from the altitude - E,N coordinates -Descending depth from the altitude - E,N coordinates -Descending depth from the altitude - E,N coordinates -Descending depth from the altitude - E,N coordinates -Descending depth from the altitude - E,N coordinates -Descending depth from the altitude - E,N coordinates -Descending depth from the altitude - E,N coordinates -Descending depth from the altitude - E,N coordinates -Descending depth from the altitude - E,N coordinates -Descending depth from the altitude - E,N coordinates -Descending depth from the altitude - E,N coordinates -Descending depth from the altitude

Output parameters Sand Clay\silt Fine sand Grade of sand SC SM SW SP Clayey layers Silty layers Fine clayey Coarse clayey layers Silty clay Clay Sandy clay Sandy silty clay CL CH Fine silty layers Coarse silty layers Clayey silt Silt Sandy silt Sandy clayey slit ML MH

In the design stage the actual output values for these models represented by “1” or “0” according to the occurrences of model’s parameters. For example, the value of the actual output in column node 1 and node 2 will be “1” if there is occurrence of each output parameters in the Table 1 and will be “0” if no occurrence. In other words, for model “1”, the actual output for column node “1” indicates occurrence of sand if it’s value is 1, and no occurrence of sand if it is 0, and the actual output column “node 2” indicates occurrence of clay/silt if it’s value is 1, and no occurrence of clay/silt if it is 0. In which sand and clay/silt classifier units is the final figure of both columns used to classify the soil in this network.

582 H. Elarabi and M. Mohamed / The Potential of Using ANNs for Prediction of Blue Nile Soil Proﬁle

1.2. Models Validation [2] The purpose of the model validation phase is to ensure that the model has the ability to generalize within the limits set by the training data. The coefficient of multiple determinations (R2) and the coefficient of determination (r2) are the main criteria that are used to evaluate the performance of the ANN models developed in this work. Generally in this study our estimation for models performance depends on R2. A perfect fit would result in an R2 value of “1”, a very good fit near “1”, and a very poor fit less than “0”. R2 is not the ultimate measure of whether or not network is producing good results, especially for classification nets. We might decide the network is OK without obtain a high value for R2 and the judgment be by the number of correct classifications.

2. Results and discussion The procedure that used after gets the output value by each model is to find the percentage of success achieved for each case study this done after compare prediction output values with the actual values for all cases. This percentage of success calculated then tabulated and summarized in Table 2. General notes in Table 2. a) “100%” means prediction profile is duplicated with actual profile. b) “0%” means prediction profile is not matched with actual profile in any depth. c) Remark “- “means the layer for the model in the actual profiles was not found. 2.1. Global classification model”1” All data passing through this model to be separated into: course-grained soil “sand” or fine-grained soil “clay/silt”. The result for this model summarized in column “1” of Table 2, the average degree of success as percentage achieved by this model for all case study reach 75.5%. These results with good R2 value indicate that back-propagation neural networks have the ability to predict the global soil classification with an acceptable degree of accuracy. 2.2. Fine \grade sand model”2” The course-grained soil “sand” obtained by global classification can be separated by this model into fine (SM & SC) or grade (SW & SP). The degree of success as percentage achieved by this model summarized in column “2” of Table 2. The average degree of success as percentage achieved for all cases reaches 75.6%. The results for fine/grade sand indicate that ANNs have a number of significant benefits that make them a powerful and practical tool to predict the type of soil layers. 2.3. Fine sand model”3” The fine sand obtained by fine/grade sand model can be separated by this model into SM or SC as fine sandy layers. The degree of success as percentage achieved by this

H. Elarabi and M. Mohamed / The Potential of Using ANNs for Prediction of Blue Nile Soil Proﬁle 583

model summarized in column “3” of Table 2. The average degree of success as percentage achieved for all case study reach 79.9%. Table 2. Percentage of success between actual and ANN predicted for soil classification Column Suleim BLDG Osman BLDG M. Energy Dental Hosp Elfatih Tower F.Elfatih Tower M. of Industry Lebanon Rest Taj Elsir Tower Alawgaf Tower G.ManshiBDG Health Insur Abode BLDG Kamal BLDG M. Nimir BDG Islamic Bank Medical City E.P.C .Kilo 10 Drwish BLDG Ani/Reso.BLD M.Of Defence Soba Housing St. Cathedral Hagar Com. Plot 18 -.Mansh Wadaa Hall School of Math BLDG - 72/Z Civil Aviation Chinese Emb Na. Rec.Office Custom institut M.Energy 61St Football Asso.

1

2

3

4

5

6

84.5 76.6 94.7 64.3 94.5 68.5 60 88.9 88.9 89.4 76.4 88.5 71.4 67.4 90 40.2 67.5 73.4 76.6 66.7 82.1 75.2 100 47.4 72.3 93 88.5 50 65 75.1 80 65.5 78.2 65.2

89.7 96.3 54.9 100 76.7 85.7 100 37.8 56.4 69.2 91.5 100 45 24 59.7 91 100 51.9 100 61.5 86 100 100 100 100 100 81.1 97.7 89.7 71.9 100 89.5

63 79.8 40 92.3 48 58.3 100 82.9 84.5 90.5 100 100 92.5 100 94.3 65.9 70.6 100 47.8 83.9 91 100 100 86.4 100 100 81 7 93.1 100 100 84.2

29.4 100 9.7 0 100 100 100 100 71.9 96.3 100 0 89.5 92.9 100 70.7 69 100 100 57.7 -

61.9 81.8 100 74.5 68.9 67 87.5 62.5 100 16 84.1 100 86.1 49 100 83.3 100 100 80 88.5 33.3 90.9 32 100 100 82.5 35.1 20 100 94.7 52.2 57.4 78.7 59.2

13.3 100 36.8 76.6 40.2 56.3 100 100 16.7 100 100 71.4 100 75 91.7 100 100 65.3 65.3 91.7 100 100 0 23.1 44 93.6 100 0 100 46.2 33.3 63.6 100 28.6

9

10

11

12

13

100 0 100 - 100 68.8 29.2 100 63.3 0 51.7 62.5 93.9 10 4.1% 91.3 60 75 100 75 80 100 100 - 100 60 100 66.7 100 93.8 100 100 100 - 100 100 100 95 100 93.9 63.9 - 100 43.3 100 100 94 100 98.2 93.3 100 98.2 93.3 - 91.7 91.7 40 17 100 100 58.8 100 - 100 100 100 23.1 100 100 64 100 65.9 87.2 - 55% 80 100 - 100 - 100 100 100 100 36.4 100 - 100 100 64.3 67.9 82.4 - 100 100 50 71.4

7

8

100 100 100 49 38.8 100 86.7 74.1 57.7 51.7 62.1 53 0 100 100 100 100 78.4 100 100 92 54.5 56 100 100

100 100 96 63.3 0 50 63 75.9 96.1 100 100 57.1 100 70 24 100 52.2 100 100 100

100 100 100 100 100 100 100 100 62. 100 100 87 50 100 100 63 60 100

100 100 100 76.5 71.4 100 73.3 100 96.2 95 100 100 50 90 100 73.9 100 100 100 100 92 100 52 100 84.6

2.4. Grade of sand model”4” The grade of sand obtained by fine/grade sand model can be separated by this model into SW or SP as grade sandy layers. The degree of success as percentage achieved by this model summarized in column “4” of Table 2. The average degree of success as percentage achieved for all case study reach 70.4%.The worst percentage achieved is 0% (one case - Elfatih Tower), due to: a) The percentage of the grade sand in the actual profile for this site represents small value 3% so can be neglected.

584 H. Elarabi and M. Mohamed / The Potential of Using ANNs for Prediction of Blue Nile Soil Proﬁle

ďͿ The actual number of boreholes conducted for this site is 6 and the grade of sand actually found in one borehole in depth range from 17 – 21 m. For this reason we can neglect the presence of grade of sand in this model͘ 2.5. Clay\silt model”5”: The clay/silt layers obtained by global classification can be separated by model “5” into clay or silt and the degree of success as percentage achieved by this model is summarized in Column “5” of Table 2. The average degree of success as percentage achieved for all case study reach 75.1%. The worst percentage achieved by grade of sand model is 16% (one case - Alawgaf Tower), due to: a) The actual number of boreholes conducted for this site is 3 and the presence of clay/silt actually found in one borehole at depth of 0 – 0.5 m clay and 0.5 – 3m silt, and is a small portion of the sample. 2.6. Fine\coarse clayey model”6” The clayey layers can be separated by this model into fine clayey layers or coarse clayey layers, and the degree of success as percentage achieved by this model is summarized in Column “6” of Table 2. The average degree of success as percentage for all case study reach 64.4%. 2.7. Fine clayey model”7” The fine clayey layers obtained by fine/coarse clayey model can be separated by this model into silty clay or clay, and the degree of success as percentage achieved by this model is summarized in Column “7” of Table 2.The average degree of success as percentage achieved for all case study reach 87%. The worst percentage achieved by grade of sand model is 0% (one case - Dental Hospital), due to: a) The percentage of the fine clayey layer in the actual profile for this site represents small value 9%. b) The actual number of boreholes conducted for this site is 4 and the clayey layer actually found in one borehole in depth range from 7 – 8.5 m, this fine clay also combined with clayey silt “double symbol in the actual profile” and the predicted profile by ANN gives clayey silt. 2.8. Coarse clayey model”8” The coarse clayey layers obtained by fine/coarse clayey model can be separated by this model into sandy clay or sandy silty clay, and the degree of success as percentage achieved by this model is summarized in Column “8” of Table 2.The worst percentage achieved is 0% (one case - Sulieman F.), this to: a) The percentage of the coarse clayey layer in the actual profile for this site represents small value 4%. b) The actual number of boreholes conducted is 3 and the coarse clayey layer actually found in one borehole at depth of 0.5 – 1 m, and is a small portion of the sample.

H. Elarabi and M. Mohamed / The Potential of Using ANNs for Prediction of Blue Nile Soil Proﬁle 585

2.9. Clay plasticity model”9” The clayey layers obtained by clay/silt model, can be separated by this model into clay of low plasticity or clay of high plasticity, and the degree of success as percentage achieved by this model is summarized in Column “9” of Table 2. The average degree of success as percentage achieved for all case study reach 76%. 2.10. Fine\coarse silty model”10” The silty layers can be separated by this model into fine silty layers or coarse silty layers, and the degree of success as percentage achieved by this model is summarized in Column “10” of Tables 2. The average degree of success as percentage achieved by for all case study reach 72%. The worst percentage achieved is 0% (one case E.P.C.Kilo 10), due to: a) The percentage of all silty layers in the actual profile represents small value 3%. b) The actual number of boreholes conducted is 5 and the coarse clayey layer actually found in one borehole in depth range from 4 – 5 m, and is a small portion of the sample. 2.11. Fine silty model”11” The fine silty layers obtained by fine/coarse silty model can be separated by this model into clayey silt or silt, and the degree of success as percentage achieved by this model is summarized in Column “11” of Tables 2. The average degree of success as percentage achieved for all case study reach 90%. 2.12. Coarse silty model”12” The coarse silty layers obtained by fine/coarse silty model can be separated by this model into sandy silt or sandy clayey silt, and the degree of success as percentage achieved by this model is summarized in Column “12” of Tables 2. The average degree of success as percentage achieved for all case study reach 84%. 2.13. Silt plasticity model”13” The silty layers obtained by clay/silt model, can be separated by this model into silt of low plasticity (ML) or clay of high plasticity (MH). The degree of success as percentage achieved by this model is summarized in Column “9” of Table 2. The average degree of success as percentage achieved for all reaches 76%.

3. Conclusions Based on the results obtained by ANNs, the following conclusions have been found: 1. The advantages of using ANNs to predict soil profiles is that neural networks are able to automatically create an internal distributed model of the problem during training process, which make them a powerful and practical tool for soil classification prediction.

586 H. Elarabi and M. Mohamed / The Potential of Using ANNs for Prediction of Blue Nile Soil Proﬁle

2. 3. 4. 5.

Results obtained from the ANNs Atterberg limits show that ANNs with proper training is a good tool in prediction especially for liquid limit. ANNs are effective tools when used as pattern classifier more than used for parameters prediction. Increasing training process time leads to reliable results. ANNs may be used as a good decision support and source of information for soils profiles.

References [1] [2]

RALPH B. PECK, Foundation Engineering ,professor of Foundation Engineering, University of Illinois at Urbana – Champaign Mohamed A. Shahin1; Holger R. Maier2; and Mark B. Jaksa3, Predicting Settlement of Shallow Foundations using Neural Networks, (2002),

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 587 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-587

Using a Modified Plate Load Test to Eliminate the Effect of Bedding Errors Hennie BARNARDa and Gerhard HEYMANN b a Aurecon, Pretoria, South Africa b University of Pretoria, Pretoria, South Africa Abstract. Plate load tests have been used extensively in the past to determine the bearing capacity and the stiffness of soil. Two of the main advantages of plate load tests are the cost-effectiveness of the test and the relative straight forward test procedure. The plate test can either be performed vertically or horizontally. The test consists of a plate that typically varies in diameter between 150 mm and 600 mm, which are loaded using a hydraulic pump and jack. The displacement of the plate is typically measured with two or more calibrated displacement measuring devices attached to the plate. This paper reports the results of a vertical plate load test designed to eliminate the effect of bedding errors that occurs during plate tests by using telescopic probes to measure the displacement below the centre of the plate. A series of plate tests were performed to determine the effectiveness of installing telescopic probes to eliminate the bedding errors. The measured vertical displacement of the plate was compared with the relative displacement of the telescopic probes and the stiffnesses were compared. The test apparatus, methods and results are discussed in this paper. Keywords. Plate load test, bedding errors, telescopic probes.

Introduction Geotechnical engineers are continuously searching for more accurate and cost effective tests to determine the stiffness of soil. Two of the main advantages of plate load tests are the cost-effectiveness and the relative straight forward test procedure. This paper reports on research that was done on vertical plate load tests using a modified apparatus. Bedding errors affect the soil stiffness measurements during plate load tests and therefore it needs to be eliminated or kept to a minimum. Telescopic probes were used to measure the displacement below the center of the plate in order to eliminate effects of bedding errors. The experiment equipment is discussed as well as the test procedure and data interpretation. The results of stiffness measurements for both the external instruments and telescopic probes are compared. Final conclusions are summarised and the way forward for plate load testing is proposed. The conclusions will help practical engineers realize the importance of bedding errors that occurs during plate load tests and argue that engineers should be more critical when interpreting plate load data.

588

H. Barnard and G. Heymann / Using a Modiﬁed Plate Load Test

1. Experiment Equipment The conventional plate load test consists of a selection of plates, a hydraulic pump and jack, some means of applying a reaction force and two or three calibrated displacement measuring devices. In the experiments of this project, a steel beam with a weight of 1,3 ton was use for the reaction force, together with grouted anchors. A hydraulic jack and pump system were used to provide the required contact pressure by jacking against the anchored steel beam. The applied load was directly measured with a 10 ton load cell, placed on top of circular plates, and logged with an automatic logging unit. 300 mm and 450 mm steel plates were stacked on top of each other and placed on the prepared surface. The vertical displacement of the bottom plate was measured with three calibrated DCDT’s (Direct-Current Displacement Transducers) placed 120˚ apart and approximately 25 mm from the plate perimeter to accommodate for any tilt that might occur during testing.

Figure 1. Schematic illustration of the modified plate load test set-up

The three external measuring devices were supported by two, 3 m long wooden, reference beams which were placed on both sides of the test area without interfering with any test equipment. Figure 1 illustrates the modified plate load test set-up and the instrument details. The telescopic probe consisted of a solid inner aluminum rod (8 mm), designed to slide freely inside a 13 mm aluminum tube. Three bended spring steel strips were welded onto a bolt and screwed unto the threaded tube. A calibrated LVDT (Linear Variable Differential Transformer) were attached to the probes to measure the relative displacement at two positions below the centre of the plate. Figure 2 shows the telescopic probe set-up.

H. Barnard and G. Heymann / Using a Modiﬁed Plate Load Test

589

Figure 2. Telescopic probes set-up

2. Test Procedure The plate load tests were performed on a uniform residual andesite (silty clayey sand) at the experimental farm of the University of Pretoria. A steel beam and four anchors, grouted to a depth of 1,5 m, were used to provide the reaction force needed to generate the required contact pressure between the plate and the soil. Tests were conducted in 2 m x 2 m holes and approximately 300 mm deep. The surface was leveled as smooth as possible and cleaned before the test commenced. A 25 mm hole was drilled by means of a hand bore, in the center of the 2 m x 2 m hole, up to a depth of one plate diameter (450 mm) below ground surface. Once the plates, hollow load cell and hollow hydraulic jack were stacked, the external DCDTs were installed on the perimeter of the 450 mm plate. The final step before the test could commence was to install the telescopic probes. The inner aluminum rod was grouted at the bottom of the 25 mm hole using ROCSET® grout. The spring steel unit together with the aluminum tube was release at a depth of 225 mm below the plate using a 25 mm steel release pipe. The LVDT with a range of 5 mm was installed at the top of the telescopic probes to record the relative displacement between the two probe points, 225 mm and 450 mm below the plate, respectively. A load sequence was applied which comprised three cycles (8 kN, 24 kN and 100 kN). The applied loads was recorded during the load and unload cycles and logged at 10 reading per second throughout the tests together with the four displacement transducers measurements. A number of tests were performed but only one test is discussed due to limited space.

3. Data Interpretation The average vertical displacement of the plate (ρ ) obtained from the three DCDT’s was used in Eq. (1), together with the plate diameter (D) ; and Poisson’s ratio (v) to determine the external stiffness (Eext ) in MPa [2]. The contact stress (q ) was taken as the applied load divided by the plate area, therefore assuming a uniform pressure distribution across the plate.

590

H. Barnard and G. Heymann / Using a Modiﬁed Plate Load Test

Eext

(

π ⋅ q ⋅ D ⋅ 1 − v2 = 4⋅ ρ

)

(1)

The strain level below a loaded plate varies with depth. In order to allow comparison of stiffnesses from Eq. (1) and those calculated from the telescopic probes, the average strain of the soil was taken as the plate settlement divided by 1.5 times the plate diameter. This was taken from the influence depths for circular foundations based on Boussinesq’s theory where less than 20% of the applied stress occurs below 1,5 times the plate diameter. The measured relative displacement of the telescopic probes (∂L ) was used in Eq. (2) together with the distance between the two probe points (L ) , to determine the strain levels for the local stiffness (ELocal ) . The vertical stress at depth z below the centre of the circular plate with diameter D=2R, carrying a uniform pressure (q ) , is calculated using Eq. (3). Values of the influence factor (I c ) are always between zero and unity. The local stiffness (ELocal ) was calculated for every load step using Eq. (4) with the average vertical stress (Δσ z ) between the two fixed points, 225 mm apart; and the corresponding strain level calculated with Eq. (2).

Δε = ∂L

L

32 ⎡ ⎧ ⎫ ⎤ 1 ⎥ = qI c σ z = q ⎢1 − ⎨ 2⎬ ⎢⎣ ⎩1 + (R z ) ⎭ ⎥⎦

ELocal =

Δσ z Δε

(2)

(3)

(4)

4. Results The main study objective was to investigate the bedding errors that may occur below the plate during plate load testing. The external and local stiffness at different strain levels were plotted and compared. Figure 3 shows the comparison between the measured external and local stiffness for the full strain range.

591

H. Barnard and G. Heymann / Using a Modiﬁed Plate Load Test

CSW Stiffness

Local Stiffness External Stiffness

Figure 3. Comparison between external stiffness and local stiffness (Cycle 3)

It is clear from Figure 3 that the stiffness determined with the telescopic probes showed significantly higher values than that from the conventional external measurements up to 0.05% strain. Table 1 summarises the local and external stiffness that was determined in all three cycles. It is interesting to notice the higher stiffness measured during the third cycle compared to the first two cycles. Table 1. Comparison between external and local instruments Stiffness E (MPa) Strain %

Cycle 1

Cycle 2

Cycle 3

LS*

ES**

LS*

ES**

LS*

ES**

0.001

137

65

139

50

249

106

0.002

104

54

145

40

225

101

0.01

67

28

152

25

162

89

0.02

64

23

140

19

163

80

0.05

n.a

n.a

114

15.3

153

76

Note: * Telescopic probe stiffness

**

External stiffness

592

H. Barnard and G. Heymann / Using a Modiﬁed Plate Load Test

Table 2 shows the ratio between the external and local stiffness. In most cased the external stiffness was less than 50% of the local stiffness. Continuous surface wave tests (CSW) on the same site showed small strain stiffness (E0) values of 250 MPa. The local stiffness in cycle 3 at 0.001% to 0.002% compare very well with the CSW value. Table 2. Percentage of external stiffness compared with telescopic probe stiffness Strain %

Cycle 1 External vs. Local (%)

Cycle 2 External vs. Local (%)

Cycle 3 External vs. Local (%)

0.001

47%

36%

43%

0.002

52%

28%

45%

0.01

42%

16%

55%

0.02

36%

14%

49%

0.05

-

13%

50%

5. Conclusions and Way Forward Plate load tests have an important role in the future of geotechnical engineering and the determination of soil stiffness. The results shown in this paper clearly demonstrates that bedding errors can have a significant effect on stiffnesses determined with the traditional method of measuring plate settlement. This modified plate load test opens new opportunities to be explored and developed. The results of these plate load tests also showed the importance and effectiveness of using telescopic probes to determine the soil stiffness. The authors believe that bedding errors were eliminated by means of telescopic probes and therefore recommend this method for the future use.

Acknowledgements The authors would like to thank the following institutes: • University of Pretoria for the technical support and facilities; • Aurecon SA, for the financial support during the study period.

References [1] G. Heymann, The stiffness of soil and weak rocks at very small strains, University of surrey, 1998. [2] B. Wrench, Plate load test, University of Pretoria, 1994. [3] B. Wrench, Plate test for the measurements of modulus and bearing capacity of gravels,1984.

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 593 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-593

Geotechnical characterization and design considerations in the Moatize Coalfields, Mozambique Gary N. DAVISa, T. E. B. VORSTERa and Célia BRAGA b a Aurecon (Pty) Ltd, Pretoria, South Africa b Odebrecht, Lisboa, Portugal

Abstract. The Moatize coalfields in the Tete Province of Mozambique are currently seeing an explosion of development. These new developments have been preceded by intensive geotechnical investigations in order to characterize founding conditions, and prepare geotechnical designs for new infrastructure in a relatively undeveloped area. This paper describes the findings of design-level geotechnical investigations and characterization for Vale’s Moatize Coal Project primarily for the plant infrastructure. Specific geotechnical design considerations for the coal stockyard and primary crusher, which provided unique geotechnical challenges, are described Keywords. Moatize, Coal, Site Investigation, Stockyard, Crusher

Introduction The coal deposit near the village of Moatize, has been explored since the beginning of the twentieth century. At the time the coal mining was opencast and small-scale. In 1940 work began underground, with an annual coal output of about 10 000 t. In early 1950’s the annual production was 250 000 t, reaching a peak of 575 000 t in 1975. In 1977 the State coal company, Carbomoc, took over the exploitation of the mineral deposit and launched detailed studies of the complex Moatize coal basin. Between 1978 and 1982, the basin was the exploration target of the Soviets and the Germans, and intermittently from 1990 onwards a number of companies including CVRD (Companhia Vale do Rio Doce) carried out further exploration studies, culminating in the awarding of the concession to CVRD in 2004. In 2004 Vale, as Companhia Vale do Rio Doce (CVRD), was awarded a concession to explore coal deposits in the Moatize area, signaling the beginning of an exponential increase in coal production. Extensive geotechnical investigations were conducted for a Bankable Feasibility Study (BFS) by Golder Associates Africa [2]. A number of other geotechnical investigations were then conducted by Ingérop. Africon, now part of Aurecon (Pty) Ltd, was subsequently appointed; initially for the conceptual layout of the mine infrastructure, but then also for the detailed design [3]. This paper only deals with the findings of the conceptual layout and detailed design investigations.

594

G.N. Davis et al. / Geotechnical Characterization and Design Considerations

1. Geological setting and weathering The oldest rocks in the area of interest belong to the Tete (Gabbro-Anorthosite) Suite, which is a layered intrusive body [4]. Structurally-controlled, block-faulted basins within the rocks of the Tete Suite are filled with Karoo sedimentary strata comprising sandstone, siltstone, tillite, conglomerate and the economically-important coal. The main mine infrastructure has been optimized to be primarily located on the gabbro-anorthosite rocks of the Tete Suite, rather than on the coal-bearing Karoo strata. The Tete Suite is predominantly composed of gabbro, with subordinate leucogabbro, norite and anorthosite and minor but widespread ultramafic rock types, mostly pyroxenite. The rock fabric is generally massive un-metamorphosed with medium to very coarse or even pegmatitic grain sizes. The geological succession has been intruded by younger dolerite dykes. The orientation of these dykes is variable but they are commonly associated with faulted contacts between the Tete Suite rocks and the Karoo strata. Tertiary and Quaternary deposits comprising alluvial and colluvial (hillwash and limited talus) materials cover these older rocks. In terms of the general structural geology, these rocks have been subjected to metamorphism and highly complex folding. Faulted contacts between the Proterozoic Tete Suite and the Karoo strata are common, with regional faulting aligned NW – SE and secondary faulting aligned NE – SW. The Zambezi River valley is a known seismically active area. A detailed study of the seismological setting and seismic hazard of the area in the vicinity of Tete was conducted [5] which determined an upper-bound magnitude m max = 6,28 ± 0,26. The lineament studies and the likely presence of active faults in the Tete area imply that probably much larger earthquakes are possible. It was noted, however, that no earthquakes with a magnitude greater than 4 have been recorded in the area, but that the earthquake catalogue is relatively short. For design a Peak Ground Acceleration (PGA) of 0.08 g was used, corresponding to a 10% probability of exceedence in 50 years. The climate of the region is sub-tropical and semi-arid. Temperatures and evaporation are generally high and rainfall low. Tete has a Weinert N-value of 2.1 [6] which indicates that chemical weathering predominates. The parent rocks, at least in certain areas, comprises basic igneous rocks (i.e. gabbro), and the N-value of 2 further indicates that montmorillonite clays might develop.

2. Outline of the development The key elements of the mine infrastructure for which geotechnical information was required comprised a coal stockyard, coal processing plant, utilities yard including LDV platform, a 3 Ml water reservoir and rising main, primary crusher, various internal and haul roads, the railway loop, a tailings storage facility and solid and hazardous waste site. Investigations also needed to identify and prove sources of construction materials, specifically gravel sources for the infrastructure, railway and access roads.

G.N. Davis et al. / Geotechnical Characterization and Design Considerations

595

3. Investigation methodology Field investigations primarily comprised excavation of test pits, either using a light TLB (tractor-loader backhoe) or a tracked excavator, and drilling of rotary core boreholes. A comprehensive programme of laboratory testing was also followed. In addition to determination of Foundation Indicators and compaction characteristics (Modified AASHTO density and CBR’s), shear box and oedometer testing as well as chemical tests were conducted in order to assess the potential soil aggressiveness or corrosivity. Geophysical resistivity surveys were carried out at proposed positions of various electrical infrastructure items, such as electrical sub-stations, in order to characterise the ground resistance for earthing design. Some 230 test pits were excavated during the detailed design investigations and more than 60 boreholes were drilled.

4. Geotechnical Design: Coal Stockyard and Primary Crusher One of the most challenging components of geotechnical design was the design of the coal stockyard and specifically the trapezoidal embankments (‘bunds’), which would support the settlement sensitive stacker and reclaimer equipment. The stockyard is a spatially extensive development, covering a length of more than 1 km and comprising a series bunds between which coal is stacked according to quality grading before being ‘reclaimed’ for export. Stacker and reclaimer equipment operate on rails placed onto the bunds. Two rail systems were considered, namely (a) a rail on ballast system, which requires regular maintenance, and (b) a less maintenance-prone, but settlementsensitive inverted T-beam foundation system. The benefit of a rail and ballast system is that grade adjustments can be done easily as part of a routine maintenance. The disadvantage is that it affects the operation of the stockyard whilst maintenance is underway. In the case of the inverted T-beam solution the rail is permanently fixed to the T-beam. This means that maintenance is difficult and disruptive if required, which places significant onus on designers to provide a robust and settlement resistant solution to avoid future maintenance interference and reconstruction. The stockyard site is underlain by clayey and silty sand hillwash (up to 0.5 m deep), underlain by medium dense to dense, silty gravelly sand residual gabbro to depths of 2.2 m, generally. Gabbro bedrock underlies the soil and occasionally extends to more than 4 m below natural ground surface. These local variations are of interest from a differential settlement perspective. For analysis of the rail and ballast system the following core assumptions were made: (a) The system applies near static loading; (b) The ballast will be tamped regularly (as such the ballast was not modelled as part of the foundation); (c) Wheel load is 30 tons; (d) A stacker / reclaimer will have two rails on a bund (the bunds are 2 m and 3 m high with 10 m and 14 m top width respectively); (e) Rail sleepers are 0.265 m wide x 0.22 m thick x 1.7 m long and rigid (for the purpose of analysis); (f) Rails are separated at 6 m and 10 m spacing respectively for the stacker and reclaimer; (g) Sleepers are spaced at 0.85 m centres and there are approximately 12 sleepers that are loaded at one time below a bogey of 10 m length; (h) Total applied load per railway sleeper is 666 kPa load on a 0.265 m sleeper (it is assumed that there would be 300 mm of ballast and that the load induced to the first layer under the ballast is a distributed load of 183 kPa - 3600 kN over an area of 1.9 m x 10.4 m to account for load spreading

596

G.N. Davis et al. / Geotechnical Characterization and Design Considerations

through the ballast); (i) Allowable total long-term movement is 300mm, while allowable differential settlement is 25mm across the rails; (j) The operation of the stacker / reclaimer tracks will be more or less balanced, with negligible biased load to either track (as such differential settlement only relates to changes in subsoil condition). The system was analysed using Plaxis 2D finite element (FE) software in a plain strain condition using a Mohr-Coulomb failure criterion to be able to model the effect of differential subsoil conditions. Five load cases were analysed to represent different configurations in fill and cut and possible changing subsoil and bedrock conditions. Figure 1 shows a typical configuration in fill and Figure 2 shows a typical differential settlement field. It was found that, to remain within acceptable limits, the ballast needed to be founded on 1.5 m of C4 stabilised soil-gravel. This was needed to limit differential settlement and increase bearing capacity of the rail system.

Figure 1. Load case in fill to account for differential subsoil, conditions

Figure 2. Differential settlement due to varying subsoil conditions

For the inverted T-beam option a maximum of 5 mm of differential settlement between rails was allowed with a maximum of 15 mm of total settlement. For similar configurations in cut and fill described earlier, a 1.3 m wide T-beam founded at a depth

G.N. Davis et al. / Geotechnical Characterization and Design Considerations

597

of 0.4 m onto 1 m of C4 stabilised soil provided a viable solution, with significant savings in cement stabilization and benefits in quality control and material availability. The primary crusher provided a further component of interesting geotechnical design. The crusher designer provided a package solution requiring that the crusher be founded on an 800 mm thick reinforced concrete (RC) raft foundation. To limit the impact on the redesign of the crusher foundation, the geotechnical team needed to develop a solution that would address the prevailing subsoil condition, whilst providing the crusher designer with a means of checking the RC raft design. Piling was not considered a viable solution due to availability of plant in Tete and the impact it would have on redesign of the foundation. Subsoil conditions at the primary crusher site comprise 3.5 m thick sandy clay colluvium, overlying sandy clay residual coal and residual shale to depths ranging between 6.2 m and 8.3 m below ground level. The clayey subsoil is potentially expansive when dry and highly compressible when wet. To meet movement tolerances of the crusher, extensive ground replacement was required to minimize the risk of non-uniform founding conditions. A system comprising of the RC raft overlying 2 m of C4 stabilised soil-gravel founded on a maximum of 5.2 m of G6 quality soil-gravel was proposed. To validate the RC raft design, the crusher designer required spring stiffness for design. To facilitate this, a three-dimensional FE model was developed using Plaxis 3D Foundation. For analysis it was assumed that the ground replacement system would provide a uniform founding condition from a stiffness point of view. Differential movement across the RC raft would therefore be attributed primarily to individual column loads rather than variable subsoil condition. Individual columns were modeled as point loads in the FE model. The RC raft was modeled as an elastic floor element with zero weight, while the individual C4, G6 and bedrock layers were analysed using a Mohr-Coulomb failure criterion. Figure 3 shows the FE half model and corresponding movement distribution field occurring across the RC raft due to the different column loads. The analysis was not intended to be a structural model, but rather to validate spring reactions used in the crusher designer’s model and to show the effect of soil-structure interaction between an 800 mm thick RC raft founded at shallow depth and the underlying ground improvement measures. The feedback to the crusher designer was a table listing each column with a corresponding column load (kN) and vertical movement (mm), which was converted to springs in units kN/mm for each column. This enabled the crusher designer to validate his model and optimize reinforcement across the 800 mm thick reinforced concrete slab.

5. Summary The Moatize coalfields in the Tete Province of Mozambique are currently seeing an explosion of development. One of these new developments as part of Vale’s Moatize Coal Project have been preceded by intensive geotechnical investigations in order to characterize founding conditions, and prepare geotechnical designs for new infrastructure in a relatively undeveloped area. This was done employing an extensive program of test pitting, rotary core drilling and laboratory testing to characterize the subsoil and quantify geotechnical design parameters. Due to variable and generally poor founding conditions, as well as severe limitations in relation to access to sophisticated laboratory testing in Mozambique, the coal stockyard and primary crusher facilities underwent intense geotechnical design to devise suitably robust founding

598

G.N. Davis et al. / Geotechnical Characterization and Design Considerations

solutions, which are both safe and practical from a constructability perspective. These included combinations of local soil improvement, cement stabilization and ground replacement.

Figure 3. Plan view showing individual column loads and associated movement field

References [1] T. Lehto and R. Gonçalves, Mineral Resources Potential in Mozambique. Geol. Surv. of Finland, Special Paper 48, 307 – 321. 2008. [2] Golder Associates Africa (Pty) Ltd. Report on Moatize Coal Mine Project Bankable Feasibility Study Infrastructure Geotechnical Volume 1- Report. Submitted to Rio Doce Mozambique Lda. Report No 7522/8311/27/E Revision 2. August 2006. [3] Africon (Pty) Ltd. Moatize Coal Project: Geotechnical Report for Detail Design Purposes. Report to Vale, No. 07-RL-9000-X-0002. 2008 [4] A.B.P. Westerhof, A. Tahon, T. Koistinen, T. Lehto and C. Akerman. Igneous and Tectonic Setting of the Allocthonous Tete Gabbro-Anorthosite Suite. Geol. Surv. of Finland, Special Paper 48, 191 – 210. 2008. [5] Council for Geoscience. A Seismic Hazard Analysis for Tete, Mozambique. Report for Murray and Roberts, number 2007-0012. 2007. [6] Weinert, H.H. The Natural Road Construction Materials of Southern Africa. Academica, Pretoria, Cape Town. 1980.

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 599 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-599

Estimating the Heave of Clays A. D. W. SPARKS University of Cape Town, Rondebosch, South Africa

Abstract. Atterberg Limit values can be used to predict the heave of clays due to unloading, or due to the wetting of a clay. A comparison is made between a new formula for swell and the van der Merwe method. Keywords. Expansive Clays, Swell, Double Oedometer Test.

Introduction The value of the pore water suction in a clay is an important parameter relating to the initial state of a clay, e.g. prior to the wetting of the clay. This initial pore water suction (-u) is a function of the initial water content prior to wetting, and it also depends on the size of the pores in the clay. If the pore sizes are large, then the water suction (-u) will be closer to a zero suction value, and the expansion caused by soaking the soil sample will also be reduced. 1. Important Soil Properties 1.1 The Liquid Limit (%), the Plastic Limit (%), and Linear Shrinkage (%) In this paper, the Atterberg Limit tests are determined by using the whole sample. Soil samples must not be oven-dried prior to the Atterberg tests, because oven-drying can cause a large reduction in the measured Atterberg values. De-aired water must be used for the Liquid Limit test which is used as the starting point of the drying path towards the Shrinkage Limit. Large suction values develop as the sample dries out towards the Shrinkage Limit. The author uses a shrinkage trough which is open on both sides but closed on the top and bottom. This results in a straight shape for the dried sample. De-aired water can be produced by boiling water in a kettle, and then storing the water in a full screw-top bottle until it cools prior to immediate use. The Shrinkage Limit can be determined by using equation (1). Shrinkage Limit % = Liquid Limit % - 3 x Linear Shrinkage % .............................(1)

600

A.D.W. Sparks / Estimating the Heave of Clays

1.2 Type of Soil Sample Laboratory technicians should state the type of samples. Several types are listed below. a.) Undisturbed in-situ soil. b.) “Undisturbed sample” recovered from the site. c.) A re-constituted sample. ( Originally mixed at high water content ). d.) A remoulded sample without adding water. e.) A compacted sample ( e.g. as part of a CBR test ) f.) A crushed oven-dried sample ( i.e. a dry powder ). One can expect the pore diameters in re-constituted clays to be uniform. Pore sizes in compacted clays can be large. Cementation can exist between grains of in-situ clays. Cementation can cause the expansion of in-situ clays to be less than the expansion of re-constituted or remoulded clays. 1.3 Type of Clay deposit If possible, the nature of the type of clay deposit should be stated e.g. :a.) In-situ weathering b.) In-situ weathering and leaching by acid pore water c.) Originally deposited in salt water d.) Originally deposited in fresh water. 1.4 The stress path which is followed Some research workers cause soil samples to follow certain stress-paths during the wetting of a sample in a consolidometer. These stress-paths must be clearly described. 1.5 Matric water suctions versus Total water suctions Any device which relies on the condensation or movement of water vapour is measuring the total water suction. Matric water suction (-u) is the water pressure which is used in the effective stress calculations. Matric stresses are measured by devices ( e.g. filter paper, manometers) which are in direct contact with the pore water. At high suctions the total water suction is a few percentage points larger than the matric suction. 1.6 Beware of the phrase “Swell Pressure”. Swell Pressure can be defined as the vertical effective pressure required to prevent the soil from swelling when the soil is wetted. At least four different methods are used to measure the Swell Pressure. It is necessary for an author to define the method which is being used. In one method the vertical load on the consolidometer sample is incrementally increased in order to prevent the sample form expanding, while the sample is being wetted. The final load is called the “Swell Pressure”. In another method the consolidometer sample is permitted to expand under a small loading, and it is then loaded in increments until it reaches its original void ratio. The vertical pressure at this stage is called the “Swell Pressure”. The Swell Pressure from this latter method can be approximately 3 times the Swell Pressure from the first method.

601

A.D.W. Sparks / Estimating the Heave of Clays

1.7 Chemicals which alter their state when wetted The term “anhydride” means “without water”. Particles of anhydrite can absorb water and change to a Gypsum (calcium sulphate) . Glauber’s salt (sodium sulphate) can also expand by absorbing water. Equations used for modelling normal clays will not apply to these materials. 2. Average Effective Stress The author combines the water suction value (-u) with the external loadings (e.g. due to overburden) by using the following equation in a saturated soil :Average effective stress p* = (p'v + p’h +p’h )+ (absolute value of u)

(2a)

= p'v ( 1 + 2 K h ) + ( absolute value of u )

(2b)

An alternative definition of “swell pressure” would be, that the swell pressure is equal to the “average effective pressure” which is required to prevent the clay from swelling when it is wetted. This value of “average swell pressure” would be different from the value defined in section 1.6 above ; which is the “vertical pressure” which will prevent swelling. Average pressures can be simulated in a triaxial cell. 3. The swelling of in-situ undisturbed clay One cannot expect agreement between swell formulae for different in-situ clays. a.) For an in-situ oil, the present author suggests that the Expansion Index Cs is Cs =

{Constant M.(Clay% - 6). G s .(PI% - 10 ) n } Log R

(3)

where M= 0,0000066 (depends on degree of cementation, and type of soil structure; M can be as low as 0,000003 ) and n= 1,65 (approx), and LogR = Gs (PI%) / {0,91.(LL% - 5.7)}

(4)

Portion of equation (3), namely the expression (Clay% - 6).(PI% - 10)1.65 has been plotted in Figure 1. For comparison purposes, the van der Merwe chart has been shown in Figure 2. It will be noticed that in both Figure 1 and in Figure 2, the most expansive conditions are furthest from the origin of each diagram. Figure 1 incorporates portion of equation (3) and was derived independently of the van der Merwe method. van der Merwe uses a certain swell% for each of the zones in Figure 2. The potential expansiveness PE can be defined as the expansiveness of the clay before a correction is made for the vertical overburden pressure These potential expansiveness for Figure 2 values are as follows :Very High High Medium Low

= = = =

potential expansiveness PE = 1 inch per foot depth potential expansiveness PE = 0,5 inch per foot depth potential expansiveness PE = 0,25 inch per foot depth potential expansiveness PE <0,25 inch per foot depth

=8% =4% =2% < 2%

602

A.D.W. Sparks / Estimating the Heave of Clays

100 Evaluation of term (Clay%-6)(PI%-10)1,65 Clay %

80 Clay %

20x103

50x103

Low VeryHigh

(< 2 μm)

High Medium Low

5x103 0

1x103

0 0

50

100 150 200 PI% of finer portion

Figure 1. Portion of equation (3)

0 PI% x (fraction 70 < 0,425mm sieve) i.e. PI% of whole sample

Figure 2. van der Merwe-type Method

In the van der Merwe method the swell in a certain soil layer is equal to value of the Potential Expansiveness PE multiplied by the modification factor F which has a value of unity on the soil surface, and a value of approximately 0,2 at a depth D of 4 metres. In fact Log F = - D(metres)/6,45). The minus sign was omitted by van der Merwe(1). In other words the Potential Expansiveness PE is equal to the swell of a soil layer located at the soil surface ( where F=1 ). The value of C s as found in equation (3) is used in equation (5) which provides the Swell % of the clay at this particular depth. Swell% = (100.C s ).{Log(p *original )- Log(p * final )} / (1+ eo )

(5)

Note that the original pre-soaking water content wco% affects the Swell% in two ways. A high water content causes p*original to be low, and it increases eo. This means that in a wetter climate zone the clay will swell less. Unfortunately the method by van der Merwe does not take into account the pre-soaking water content.The van der Merwe method was devised for clays on the Highveld. Consider a clay sample (PI% of whole sample = 40%) at a depth of 1 metre below the soil surface (F = 0,68) which has a pre-soaking water content of wco% = 18 %. Figure 3 shows a comparison between the above method (equation 3) and the van der Merwe Method. It will be noticed that the van der Merwe values seem to coincide with the values from equation (3) if the value of Log (p*original/ p*final) is approximately 2, and the value of M is approximately 0,000005. 4.

A replacement for the van der Merwe Method.

The method by van der Merwe does not take into account the original pre-soaking water content of the clay. His method of calculation is based on soils located in the dry Highveld zone ( Transvaal and Orange Free State ). Section 5 and Figures 4 and 5 describe a method which can be used to estimate the swell of natural deposits.

603

A.D.W. Sparks / Estimating the Heave of Clays

10%

M=0,000005

For PI% = 40 % Log( p*o/p*f ) = 2 wco% = 18%

Swell %

M= 0,000003

Shaded region shows stepped values of van der Merwe’s method 0 0

20

40

60 %

80

Clay % Figure 3. Comparing the smooth curve with the stepped values of van der Merwe

5. A method which includes the original water content and the locality

%Swell of clay if it is on soil surface

a.) Use this chart or equation (6) to find swell of this layer if this layer were at the soil surface.

20 %

PI% 40% PI% 30% PI% 20%

10 % 0

5%

(6)

b.) Decide on depth H of influence e.g Highveld H=5,5m; Cape Town H=1,4m Durban H = 1 m; Sydney H=1,4m

40 % 30 %

⎛Swell% at ⎞ ( 10+ 0, 47 PI% − wo% ) ⎜ ⎟= ⎝Soil Surface ⎠ ( 1/ Gs) + (wo% /100 )

c.) Calculate Df for this layer 0 1,0 0

45% PI% 50% 35%

D/H

25%

10%

20%

30%

Pre-wetting Water Content wo % Figure 4. Assuming the clay layer is on the soil surface

Depth Factor D f = ( Log H - Log D) / (1.32+ Log H )

1,0

Df

D = depth to centre of this clay layer Df is defined by equation(7).

Df is the Depth Factor Figure 5. The Depth Factor Df

[ distances in metres]

(7)

The Percentage Swell in Fig. 4 is multiplied by the Depth Factor Df . An example of this method is as follows : Consider a profile in the Transvaal (H = 5.5m). A clay layer (PI% = 30%) is located at depth D=2 metre. Original water content =17%. Fig.4

604

A.D.W. Sparks / Estimating the Heave of Clays

or eqn. (6) gives Swell% = 13,1 % if the sample were on the soil surface. For 2 m depth, Df = O,213. Swell of this clay at depth of 2m is Df x (%Swell from Fig.4) = 0,213 x 13.1% = 2,79 % . Note that these values of Df are lower than the values of multiplier F used by van der Merwe [1}. 6. Measurements by Aksoy and Kaya (2) Experiments relating to the swelling of soils were performed by Aksoy and Kaya [2]. Values for the swelling of “undisturbed natural samples” are plotted in Figure 6. The samples chosen by these authors are mainly samples which have a high in-situ water content. Their values of measured Cs are compared with values from equation 3.

Calculated values of Cs from eqn.(3)

U4 U7 U10

U5 U3

U1

0,1 U12 U14 U13 U8 U15 U2 0

0

Value of M used for calculated values is 0,0000066

0,1 Values of Cs from Measurements by Aksoy and Kaya

Figure 6. Approximate correlation for “undisturbed” samples.

Conclusion The author has attempted to warn against certain practices such as the oven-drying of clays prior to performing Atterberg Limit tests. Measured swell values for undisturbed partly-cemented clays will be less than those for remoulded clays. The method used in equation (3) might have merit. But it is likely that the method described in Figs 4 and 5 will be the simplest method for practising engineers. The author wishes to thank the authors Aksoy and Kaya for supplying extra information to the author. References [1] D.H. van der Merwe, “The prediction of Heave from the Plasticity Index and Percentage Clay Fraction of Soils”, (June 1964),The Civil Engineer in South Africa, Journal of S.A.I.C.E.; p103-p105. [2] Y.Aksoy and A.Kaya, (Aug 2011) , “Predicting soil swelling behaviour from specific surface area”,Geotechnical Engineering, Inst. Civil Engrs, London.p229 - p238. [3] A.D.W.Sparks and T.Pidgeon ( April 2011 ) “Simplifying Expansion of Clays”, Civil Engineering, Publication of the South African Inst.of Civil .Engineering , p 28 - p34.

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 605 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-605

Instrumentation and Monitoring during Construction of the Ingula Power Caverns G. J. KEYTER, M. KELLAWAY and D. TAYLOR Braamhoek Consultants Joint Venture, South Africa

a

Abstract. An extensive programme of instrumentation and monitoring was carried out during construction of the Ingula hydro power caverns given the size of these excavations in Karoo mudrocks, for validation of assumptions made during design. In this paper, some of the monitoring results are compared to predicted convergence and an incident is reviewed where excessive overbreak in one of the side headings of the machine hall top heading resulted, locally, in undue roof deformation which necessitated remedial action to stabilize this part of the crown. Keywords. Cavern, hydro, instrumentation, monitoring

Introduction The Ingula pumped storage scheme is being constructed in the Drakensberg escarpment between the Free State and KwaZulu Natal provinces, South Africa. The powerhouse complex comprises two main caverns as well as other galleries, tunnels and shafts. This paper provides details of instrumentation and monitoring work carried out in the power caverns during construction and monitoring results are compared with those predicted during design. Excessive overbreak in a side heading when excavating the machine hall crown resulted locally, in undue roof deformation. Remedial actions taken to stabilize the excavation in this area are discussed.

1. Powerhouse Location and Description The high pressure waterways and underground powerhouse lie under a prominent mountain ridge off the escarpment. The main power caverns are located at a depth of almost 400 m below ground level, about halfway between the two reservoirs. The machine hall is approximately 184 m long with a span of 26 m. The crown, 23.5 m above operating floor level, has a double curvature profile with a relatively low span : height ratio of 2.5. The cavern width reduces to 24.6 m below crane beam level. The turbine pits extend another 27 m below operating floor level. The adjacent transformer hall has a 19 m span, a height of 21 m and a length of approximately 176 m, with a cable and pipe gallery on one side running the length of the cavern and extending another 6 m below operating floor level.

606 G.J. Keyter et al. / Instrumentation and Monitoring During Construction of Ingula Power Caverns

2. Geology The powerhouse is being constructed in horizontally bedded siltstones, mudstones and carbonaceous mudstones of the Volksrust Formation of the Ecca Group, Karoo Supergroup. The power caverns are located some 25 m below a 40 m thick dolerite sill. Intact rock properties derived from field and laboratory testing are presented schematically in Figure 1 in relation to the powerhouse. A decreasing trend in intact rock strength and stiffness with depth below the dolerite sill is evident in Figure 1. This can be attributed to induration effects given the proximity of the sill above as well as changes in mudrock composition with depth. The durability of mudstones at and above cavern roof level was classified [1] as ‘good’ to ‘excellent’ and at lower elevations, further away from the influence of the dolerite sill, as ‘poor’ to ‘fair’.

Figure 1. Rock material properties in relation to the power caverns

Faults in the project area generally trend E-W and ESE-WNW with a further two sets of small displacement faults striking NW-SE and NE-SW. A sub-vertical, sheared and faulted dolerite dyke with strike orientation NNW-SSE intersects the power caverns and main access tunnel at an oblique angle. A normal fault zone comprising slickensided, striated joints, infilled with calcite and mylonitic material, was intersected in access tunnels near the powerhouse with a few of these fault planes intersecting the far eastern end of the transformer hall. A number of bedding parallel shears have also been identified in boreholes and underground excavations in the powerhouse area and surrounds, three of which are located in and above the machine hall crown. However, there are no known seismically active faults in the immediate project area. The Tugela Fault which follows the boundary between the Kaapvaal Craton and the Namaqua Province in this region, is located some 50 km to the south.

G.J. Keyter et al. / Instrumentation and Monitoring During Construction of Ingula Power Caverns 607

In situ ground stress was measured in hydrofracture tests in boreholes and in a small number of overcoring tests. The minimum horizontal stress is orientated NNW-SSE. The major horizontal stress is greater, and the minor horizontal stress slightly lower, than the estimated vertical overburden stress. Hydrofracture tests at cavern level gave a horizontal / vertical stress ratio (K-ratio) of between 0.5 and 0.9 while overcoring tests indicated a K-ratio of approximately 1.0 in the powerhouse area. Time dependent deformation of the rock mass was noted during construction of the access tunnels to the powerhouse [2]. Taking into account scale effects given the size of the main power caverns compared to that of the tunnels constructed earlier, most of this time dependent deformation in the power caverns are expected to occur within 6 months to a year following excavation down to operating floor level and for about a year in the turbine pits after turbine floor level has been reached. The rock mass at powerhouse level is characterized by closed joints with a resultant low rock mass permeability. Virtually no groundwater has been encountered at cavern level during excavations carried out to date.

3. Cavern Support Design Initial estimates of the support required in the Ingula power caverns were based on precedent experience. The final support design was based on a detailed evaluation of all available geological and geotechnical information followed by numerical modelling using UDEC, Phase2 version 7 and FLAC3D [3, 4]. The convergence of different points located on the crown and sidewalls of the main power caverns was estimated on the basis of the results of this modelling work, for the anticipated excavation sequence as well as specified timing of support installation.

4. Instrumentation and Monitoring Design Instrumentation monitoring arrays were designed to monitor ground displacements as well as ground anchor loads during construction of the power caverns and adjacent excavations. This was critical for validation of design assumptions and analyses and to timely detect unanticipated cavern convergence and loads in ground anchorages. Instrument arrays typically comprised multiple point borehole extensometers (MPBX), rock anchor load cells and optical convergence targets, with array layouts optimized on site during installation to suit actual conditions. Figure 2 shows one of the main MPBX arrays installed at one of the turbine pits. Most of this instrumentation will become redundant after construction with only a limited number of instruments to be maintained during scheme operation to monitor longer term effects. Limits were set on cavern convergence and anchor loads in terms of so-called trigger levels as follows, to allow a rapid but appropriate response to actual monitoring data: A baseline level on which the support design is based, with observed excavation convergence and anchor loads falling within expected limits in line with that predicted during design; a warning level, which is approached when convergence and anchor loads exceed the baseline level; and an alarm level, which is approached when convergence and loads exceed the warning level.

608 G.J. Keyter et al. / Instrumentation and Monitoring During Construction of Ingula Power Caverns 50m

40m

50m

50m

25m

25m

25m

25m

25m

13m

45°

13m

13m

6.5m

6.5m

6.5m

20m

20m 10m

45° 6.5m

6.5m

13m

5m

5m

5m 20m

25m

5m

5m

6.5m

6.5m

13m

20m

10m

25m

10m

10° 5m

Transformer Hall

Machine Hall 13m

5m

20m

6.5m

6.5m

20m

25m

10m 10m

5m

13m

10m 13m

10m

10m 25m

6.5m

10°

25m

20m

20m

13m 6.5m

25m 13m

20m

13m

10m

5m

10m

20m

5m

10m

20m

5m

0 1m 2

5

10

15

Figure 2. Multiple point borehole extensometer array in power caverns

Baseline convergence is satisfied where measured increases in working load in an anchor is projected not to exceed 50% of the yield capacity of the anchor on cavern completion [5]. The warning level is reached where the anchor load is projected to exceed 50% of the yield capacity of the anchor and the alarm level when the anchor load is projected to surpass 62.5% of the anchor’s yield capacity on cavern completion. Given the number of instruments installed in the power caverns and adjacent excavations, a system was adopted whereby two flags were assigned to each instrument installation, with the flag status indicated on an instrumentation plan of the powerhouse complex as follows: (1) A first (or left) flag looking at ‘total convergence’ with a green flag designating convergence within the baseline; an orange flag convergence approaching the warning level; and a red flag indicating convergence approaching the alarm level; and (2) a second (or right) flag looking at ‘rate of convergence’ with a green flag indicating the excavation is stabilizing; an orange flag that creep deformation is noted; a magenta flag indicating blasting nearby; and a red flag accelerating convergence. An extract from this summary with instrumentation flags is shown in Figure 3 for that part of the machine hall where the dolerite dyke was intersected. This system of flags allowed for a quick and easy, high level overview of cavern convergence and support performance in different parts of the cavern. For example, in Figure 3, all the instruments in the area of the dyke indicate total convergence within the baseline level (i.e. all the ‘left flags’ are green). However, note that all the instruments on the western side of the dyke are showing time dependent creep deformation (all the ‘right flags’ on this side of the dyke are orange) whereas instruments on the eastern side of the dyke are showing an excavation that is stabilizing (i.e. all the ‘right flags’ on this side of the dyke are green).

G.J. Keyter et al. / Instrumentation and Monitoring During Construction of Ingula Power Caverns 609

Instrument position

‘Right flag’

‘Left flag’

N

FLAG KEY

Total Convergence Within expected

Rate of Convergence Stable

Within expected

Creep

Within expected

Blasting nearby / Accelerating

Approaching warning level

Stable

Approaching warning level

Creep

Approaching warning level

Blasting nearby / Accelerating

Approaching alarm level

Stable

Approaching alarm level

Creep

Approaching alarm level

Blasting nearby / Accelerating Decreasing convergence

Dolerite dyke Figure 3. Instrumentation flags in part of the machine hall cavern

5. Actual Cavern Convergence and Support Performance Less than expected convergence has been noted to date on cavern centre line, possibly due to permanent ground anchorages in the roof being installed closer to the face than originally designed. Convergence measured in the cavern side headings are nearer to that predicted during design as shown in Figure 4. However, in October 2009, overbreak occurred along a bedding shear plane in one of the machine hall side headings, see Figure 5. The overbreak developed over 3 days in as many blast rounds in this heading and increased to more than 2 m beyond the theoretical excavation line by the third blast. Increases in anchor loads were noted in load cells and a jump in convergence was noted in MPBX readings in the cavern crown. By then, cracking of shotcrete developed in the ‘brow’ in the roof as shown in Figure 4. With primary support rock bolts already installed up to the face of the side heading at that point in time, all blasting work in this heading was stopped and the following remedial action taken in the affected area: All rock bolts already installed were full column grouted; all cable bolts were installed up to the face and grouted up; a layer of steel fibre reinforced shotcrete was applied over the area of cracked shotcrete followed by a layer of weldmesh reinforced shotcrete; permanent cable anchors were installed closer to the side heading face; and voids behind the cracked shotcrete were grouted to ensure good contact between the shotcrete and the rock substrate. The monitoring frequency was also increased to check the adequacy or otherwise of the above remedial measures; these measures successfully stabilized the roof in this area.

6. Acknowledgements This paper could not have been published without permission from the client, ESKOM, and the support of the Braamhoek Consultants Joint Venture (BCJV). The monitoring results presented in this paper are based on regular readings of instruments and presentation of data as prepared by the main contractor for construction of the underground works, CMI JV, as submitted to the BCJV site supervision team.

610 G.J. Keyter et al. / Instrumentation and Monitoring During Construction of Ingula Power Caverns 1220

WARNING LEVEL ALARM LEVEL MPBX E4-N1 MPBX E4-S1 MPBX E4-N2 MPBX E4-S2 MPBX E4-N3 MPBX E4-S3 Excavated invert level (masl)

120 105 90

1214 1208 1202 1196

75

1190

60

1184

45

1178

30

1172 Expected Convergence

1166

Long term creep deformation

Turbine pits excavated

Sixth bench excavated

Fifth bench excavated

Fourth bench excavated

Third bench excavated

Second bench excavated

Drainage gallery excavated

Central heading excavated

0

First bench excavated

15

Excavated Cavern Invert Level (masls))

135

Side headings excavated

Machine Hall Side Heading Convergence (mm)

150

1160

Figure 4. Measured convergence in machine hall side headings

As-built profile Bedding shear

Extent of cracked shotcrete

Theoretical excavation line

Figure 5. Overbreak on bedding shear in machine hall side heading

References [1] H. J. Olivier, A new engineering-geological rock durability classification, Engineering Geology, Vol. 14 (1979), 255–279. [2] G. J. Keyter, M. Ridgway and P. M. Varley, Rock engineering aspects of the Ingula powerhouse caverns, 6th International Symposium on Ground Support in Mining and Civil Engineering Construction, SAIMM (2008), 409-445. [3] G. J. Keyter and P. M. Varley, Design of the Ingula powerhouse caverns: General design considerations, SANCOT Seminar, SAIMM, Ladysmith, South Africa, 2008. [4] G. J. Keyter and P. M. Varley, Ingula pumped storage scheme: Excavation and support of the main powerhouse caverns, Braamhoek Consultants Joint Venture, Construction design brief – Main underground works, 2010. [5] Civil Engineering and Building Structures Standards Policy Committee, British Standard Code of Practice for Ground Anchorages, British Standards Institution, London, 1989.

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 611 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-611

Piezocone Investigation of Paleo River Channels at Changane River, Mozambique, for a Railway Embankment H.A.C. MEINTJES and G.A. JONES SRK Consulting, Johannesburg, South Africa

Abstract. Development of a heavy minerals mine at Chibuto, Mozambique, required the construction of a railway embankment across the Changane River, a tributary of the Limpopo. The embankment will be about 5.5 kms long and between 4m and 6m high. The investigation, at the bankable feasibility stage, comprised boreholes at three bridge crossings and piezocone testing across the flood plain. These revealed four distinct paleo channels with depth of up to 30m infilled with recent alluvial deposits. Analysis of the piezocone data, from which compressibility and consolidation parameters were derived, showed that an average embankment settlement of about 1.5m will take place with a maximum of 3m. Time for consolidation will vary from a couple of years up to many decades. The proposed solution is a combination of preloading, surcharging and in the worst cases and at bridges, the addition of vertical sand drains. These together will reduce the post-construction settlements to less than 250mm over a decade; differential settlements over short distances will be negligible hence the situation will be manageable despite the extremely poor subsoil conditions. Keywords. Paleo channels, Alluvial Deposits, Piezocone, Embankment Settlement

Introduction A proposed heavy minerals mine at Chibuto, Mozambique, about 200 kms north of Maputo and 50 kms inland from Xai-Xai, necessitated a 90 km extension of an existing railway line from Barragem to Chibuto, a distance of about 90 kms. Close to Chibuto the railway crosses the flood plain of the Changane River a tributary of the Limpopo. The confluence is about 5 kms downstream of the railway crossing and the flood plain extends for a length of 5.5 km. The embankment will be up to 6m high to minimise flooding problems, hence settlement of the embankment over deep recent alluvial deposits is a significant potential problem. The investigation comprised boreholes and piezometer cone testing which graphically revealed old buried river channels. Geotechnical solutions were developed to minimise problems arising from the embankment stability and settlement and its influence on bridge foundations. The study was at the bankable feasibility level at which stage detailed design was not required but reliable cost estimates were necessary. Foundation solutions are therefore discussed in relatively general terms rather than in detail .

612

H.A.C. Meintjes and G.A. Jones / Piezocone Investigation of Paleo River Channels

1. Geology The geological history which lead to the deposition of the heavy minerals has been described by Meintjes and Jones, [1]. The flood plains of the Limpopo and Changane are however the result of much more recent geological processes. These river systems, like those of the east coast of South Africa, have developed towards the end of the Flandrian transgression after the most recent ice age about 18 000 years ago when the sea levels rose about 80m drowning the river valleys. The alluvial deposits are generally up to 50m deep for the largest rivers, about 20 to 30m thick for most of the rivers and somewhat less, 10 to 15m for the rivers rising closer to the coast. A feature of many of the rivers is that their courses have frequently changed hence old infilled paleo channels are common, particularly in the more extensive flood plains towards the northern parts of Natal e.g. Richards Bay. In central and southern Natal the topography is more hilly close to the coast and hence the river courses more confined, except immediately adjacent to the shore where in some cases lagoons have formed and channel movements have taken place. The alluvial deposits are extremely heterogeneous, ranging from sands for the largest rivers (Umgeni and Tugela in Natal) to organic silty clays (Sea Cow Lake Durban). Generally, however, any one site has a full range of materials reflecting the local geological and flood history. In some cases, and this is so at the Changane flood plain, the more clayey softer materials occur at depth and these are covered by more sandy materials at the surface. Although this has an advantage in that access becomes straightforward and minor works present no difficulties, it has the disadvantage that there is little if any surface expression of the underlying soft materials and buried river channels. The latter do, however, often show up clearly on air photographs.

2. Project Description The proposed railway line will be about 5.5 km long across the Changane River flood plain. The mine will rely heavily on the railway both for receiving various materials and primarily for the exporting of the processed heavy minerals ores of titanium, ilmenite, rutile and zircon. The area is notorious for cyclone induced extreme flooding events, and those in 2000 and from the Demoina cyclone in 1984 are still vivid in memory from extensive media coverage. The height of the proposed railway embankment is therefore necessarily a compromise; it must be sufficient for “normal” floods, but cannot be economically designed to be above exceptional flood levels. In the latter case, however, the design should take account of the probability of occasional overtopping. When the investigation was undertaken the proposed implementation and construction programme was such that at least one year could be available for preloading after the building of the railway embankment and before the line needed to be operational.

H.A.C. Meintjes and G.A. Jones / Piezocone Investigation of Paleo River Channels

613

3. Site Investigation The Changane flood plain extends from Ch 71.4 to Ch 76.9 and includes three bridge sites. The investigation comprised six boreholes at the bridge sites, and thirty CPTU’s. 3.1. Boreholes Table 1. Bridge 1; Ch 74.7 BH No. RBH1

RBH2

Depth m 0 – 12 12 -20 20 – 26 0 – 7.5 7.5 – 25 9.5 – 12.5 12.5 – 18.5 18.5 – 25 25 – 26

Description Dark brown silt to firm clay Light brown silty firm sand Brown very stiff clay Brown firm to stiff clay Dark brown silty sand Black soft clay Light brown silty f.m.c. sand Slight brown clayey sand Light brown speckled black clay

SPT No. 2–6 10 – 40 29 – 43 6 – 20 7 – 12 4–5 20 – 30 10 – 20 42

Description Mottled dark brown silty clay Light brown silty clayey sand Brown and grey mottled clay Grey and light brown silty clay Brown clayey gravely f.m.c. sand Dark brown silty clay

SPT No. 4–9 19 -33 3 - 11 4 – 14 50 – 70 48 - 65

Description Dark brown silty clay As above very stiff

SPT No. 3 – 11 15 - 63

Table 2. Bridge 2; Ch 72.4 BH No. 3 and 4

Depth m 0 – 15 15 – 18 18 – 23.5 23.5 – 34 34 – 36.5 36.5 – 46.5

Table 3. Bridge 3; Ch 71.4 BH No. 5 and 6

Depth m 0 – 24 24 – 25

3.2. Piezocone Testing - CPTU Thirty piezometer cone tests with dissipation tests were carried out along the embankment line with some of these in close proximity to the bridge site boreholes to enable correlations to be made between them. The CPTU’s are variable but two typical examples are given in Figures 1 and 2. The first showing an upper sandy layer and the second, in a paleo channel, a thin upper sandy layer and very deep soft clay. 3.3. Laboratory Testing A feature of the investigation was that laboratory testing of the soils across flood plain was minimised since reliance could be placed on the results of piezometer cone testing and visual examination of the borehole samples. The project was at the bankable feasibility stage and at this level reliance on geotechnical parameters derived from the piezometer cone testing is justified

the the the by

614

H.A.C. Meintjes and G.A. Jones / Piezocone Investigation of Paleo River Channels

international experience and more particularly by extensive experience of geologically similar deposits along the Natal, South Africa east coast described by Jones [2].

Figure 1: Typical CPTU –Sand/Clay/Sand

Figure 2: Typical CPTU Sand/Clay

4. Interpretation of Investigation Data 4.1. Stratigraphy The boreholes and penetration tests allow a reliable picture of the stratigraphy to be made and this is illustrated in Figure 3. The description of the material was derived from the piezocone soils identification chart by Jones and Rust [3] and direct comparison with the borehole samples. The embankment height averages about 4m with a higher 6m section at about Ch 71 – 72. The upper erratic mixed horizon comprises sands, silts and clays and varies from about 5m to 10m elevation except at about Ch 72.2, Ch 72.8 and Ch 73.7 where the clay almost daylights. Underlying this uppermost mixed layer there is the dominant soft clay stratum which continues down to an extremely variable elevation of from about -10m to -30m except from Ch 74.5 where it thins out towards the end of the flood plain. This clay stratum, contains a sandy layer generally about 2m to 3m thick which follows the upper profile of the clay layer and is about 3m or so below it.

H.A.C. Meintjes and G.A. Jones / Piezocone Investigation of Paleo River Channels

615

The most striking aspect of the stratigraphy, however, is the definition of four distinct sub-channels within the overall flood plain which are old infilled courses of the meandering river. 4.2. Geotechnical Parameters The relevant geotechnical parameters are derived from the piezocone data using conventional correlations. cu = q c N c

(1)

When ću is undrained shear strength; qc is cone pressure and Nc is a factor taken as 14 for normally consolidated clays. mv =

1 α m qc

(2)

Where mv is the coefficient of compressibility and αm is the constrained modulus coefficient taken at 2.75 ± 0.55 for recent alluvial deposits, Jones [2]. cv = 150/t50

(3)

Where cv is the coefficient of consolidation in m²/yr and t50 is the time for half consolidation in minutes for recent alluvial deposits, Jones and Rust [4] From the above equations and the piezocone results in Figures 1 and 2 it can be seen that in the clay layers typical cone pressures of about 1.5 MPa lead to undrained shear strengths of 100 kPa, to mv of 0.2 m²/MN, and that measured t50 times in the dissipation tests of 30 to 50 minutes give cv of 5 to 3 m²/yr.

Figure 3: Changane Floodplain Stratigraphy

616

5.

H.A.C. Meintjes and G.A. Jones / Piezocone Investigation of Paleo River Channels

Geotechnical Problems

The potential problems for the railway embankment crossing of the flood plain are the stability and settlement of the embankment and the founding of the three bridges. 5.1. Stability At the feasibility level a conservative view was taken of the embankment side slopes which were selected as 1 (v) : 4 (h). At extreme flooding events the embankment may be overtopped hence flat slopes will also give erosion protection, miminise rapid draw down problems and allow, if necessary, future increase in embankment height to compensate for excessive settlement without the undesirable increasing of embankment side slopes. Since over most of the embankment length the uppermost stratum was mixed clays and sands generally of firm or better consistency which forms in effect a sandier raft, stability of a 4m to 6m high embankment will not be a problem with such flat slopes. At three locations the overlying layer is absent and the soft clay close to the surface. It is envisaged that these areas of potential instability can be dealt with by including stabilising berms. Typical red dune sands are available for embankment and berm construction from borrow areas at the end of the flood plain. 5.2. Settlement The major problem with embankment settlements is not so much the amount of settlement but the differential settlements and the time for full settlement to take place. At major road embankments along the Natal coast, over similar recent deposits, settlements of over 1.5m are common, and up to 3m have occurred without causing problems. In most areas it was possible to construct the embankments well before final pavement construction. At bridges, however, the interaction of embankment settlement with the abutment structures creates considerable difficulties and may result in extreme measures such as preloading, sand drains and piles designed for downdrag and lateral loading. At the Changane site the large differential settlements along the embankment caused by the extremely variable thicknesses of soft subsoils causes additional problems. Estimates of settlements and times for settlement were required and these were obtained from the results of the piezometer cone testing using conventional methods. Settlements were calculated of up to 3m and consolidation times in the range of decades for the thickest clay layer. Figure 4 shows the initial settlements, the clay settlements and the total settlements along the embankment. Figure 5 shows the thicknesses of the clay layer and the times for 90% consolidation using cv’s of both 10m²/yr and 3m²/yr. These figures show very large settlements should be expected and they may take decades before completion. The proposed problem solutions are discussed in the following section.

H.A.C. Meintjes and G.A. Jones / Piezocone Investigation of Paleo River Channels

617

Figure 4: Changane Floodplain settlement estimates

Figure 5: Changane Floodplain estimates of consolidation time

6.

Discussion

Analysis of the data from the piezocone testing shows that the average settlement along the 5.5 km railway embankment will be about 1m with a range of from 0.2m at the edges of the flood plain up to about 3m at Ch 72 where the embankment is at its highest and the soft subsoil at its thickest. Settlement times vary from a couple of years

618

H.A.C. Meintjes and G.A. Jones / Piezocone Investigation of Paleo River Channels

to a couple of decades and at the deepest infilled channels of up to 80 years, using the most conservative cv. These are daunting predictions and clearly unmodified or normal construction would result in totally unacceptable post construction settlements with heavy ongoing maintenance and re-leveling of the railway line. Fortunately, however, the planning resulted in at least one year being available after initial construction and before track laying and this time allows the problem to become manageable without resorting to extreme measures. Preloading for one to two years is feasible since conventional consolidation theory indicates (T50 = 0.196 and T90 = 0.848), that half the settlement will occur in a quarter of the 90% consolidation time. The preload will include a 50% surcharge load then, for example, for an estimated 10 year consolidation time, two years of preload and 50% surcharge load will result in 75% of the estimated final settlement. Along about half the embankment length this will be sufficient leaving acceptable post construction settlements of about 0.25m. The longitudinal changes in thicknesses of the subsoil are in reality fairly gradual with the exaggerated vertical scale of the figures giving a false appearance of canyon like channels. Nevertheless it is these channels which create the major problem since the thickness of subsoil results in much longer drainage path lengths. Preloading for two years (at 50% surcharge) will only result in about a quarter of the final settlement when that is expected after say 50 years, so a further two metres of post construction settlement would then occur. This is unacceptable and the solution is the use of sand drains. In the deep channels the present drainage path is estimated to be about 14m at Ch 74.3 i.e. half the clay thickness, hence the 90% consolidation time of 70 years. Sand drains with a conventional spacing can readily reduce the drainage path lengths to 3m hence reduce the 75% consolidation time to about 1.5 years so that after installation of the drains and preloading at 50% overload, most of the final settlement will have occurred. By judicious increasing of the spacing and depth of the drains a gradual transition from drained to undrained zones can be made with the intention of having uniform post construction settlements and minimum differential settlements along the embankment. The three bridges create a potentially more severe problem because of very large differential settlements between the necessarily piled structures and the contiguous embankments. Acceleration of the rate of settlement of the embankments at the bridges will be necessary and this will be achieved by a combination of preloading, surcharging and sand drains. Even with these precautions it is unlikely that downdrag forces on the piles will be eliminated and therefore allowance for this will be made.

7.

Summary

The construction of a 5.5 km long railway line over the Changane river flood plain as part of the infrastructure of a proposed heavy minerals mine presents some difficult problems, the first of which was to assess the stratigraphy and geotechnical parameters of the subsoil. The investigation comprising boreholes and piezometer cone testing, with the emphasis on the latter, revealed up to a 30m depth of recent alluvial deposits with four

H.A.C. Meintjes and G.A. Jones / Piezocone Investigation of Paleo River Channels

619

distinct paleo river channels. This was not unexpected in the general Limpopo flood plain of which the Changane is a part. Over most of the flood plain there is a mixed sandy and clayey uppermost stratum which overlies the soft clays and fortuitously forms a partial raft which lessens stability problems for the 4 to 6m high embankment. Nevertheless analysis show large settlements of up to 3m with an average of about 1.5m, will take place and that the times for 90% consolidation may be decades. These problems can, however, be overcome by a combination of preloading and 50% surcharging for a period of at least a year at the most severe places, i.e. at the three bridges and above the deepest infilled valleys, and the addition of vertical sand drains to reduce post construction settlements to manageable amounts. It is reiterated that the study was conducted at the bankable feasibility level hence detailed solutions were not required. Nevertheless the piezometer cone investigation allowed sufficient information to be obtained so that reliable cost estimates and construction programmes could be developed.

References [1] H.A.C. Meintjes and G.A. Jones, Geotechnical characteristics of the red sands at Chibuto, Mozambique, 15 Afr. Reg. Conf. SMGE, Maputo, Mozambique 2011 [2] G. A. Jones, Development of sounding equipment for the assessment of the time-settlement characteristics of recent alluvial deposits when subjected embankment loads, Ph D Thesis, University of Natal, South Africa, 1992. [3] G.A. Jones and E. Rust, Piezometer penetration testing CUPT, 2nd Europ. Symp. Penetration Testing, ESOPT 11 Amsterdam 1982 [4] G.A. Jones and E. Rust, Piezocone settlement prediction parameters for embankments on alluvioum, Int. Symp. On Cone Penetration Testing, Linkoping, 1995

620 Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-620

Site Selection of the Mathemele Landfill a

a

and Ivan MINDOa TÉCNICA Engenheiros Consultores, Maputo, Mozambique Carlos QUADROS,

Abstract. Both Maputo and Matola Cities do not have a properly designed landfill, however there are several waste disposal sites in use that are environmentally unacceptable. The proposed solution is to develop the first municipal landfill at Mathemele, located at the border between Maputo and Matola. The Site is underlain by deposits of the Congolote formation. The upper 18 m of this formation is made of fine to medium sand, and the ground water level varies between 13 to 15.5 m, conditions that are favorable to the construction of a landfill. The presentation is based on field work, physical modeling of the ground and the waste management principles adopted and aims at providing inputs for the design of the Mathemele landfill. The only landfill to date in Mozambique that could be considered as environmental friendly is the Mavoco landfill for hazardous waste. Keywords. Landfill, environment, waste management, soil characteristics, soil investigation

Introduction The Municipalities of Maputo and Matola face significant problems in providing environmentally and publicly acceptable sustainable solid waste management services, and one of the reasons is the lack of a well designed landfill for waste disposal . The only landfill to date in Mozambique that could be considered as environmentally friendly is the Mavoco landfill for hazardous waste. A feasibility study was carried out for the development of a landfill in the Mathemele area, located within the Matola municipal boundaries. This landfill should serve the city of Maputo and the greater Maputo, which includes the industrial area of Matola. The estimated annual waste production for both cities is nearly 705 000 T at the horizon 2033, planned as starting year for the landfill operation. The selection of the proposed landfill area was based on geotechnical features and characteristics of the site as well as geohydrological considerations and surface hydrology. This paper reports the work done over a period of three months covering the site selection process from the geotechnical point of view and the design considerations. For the selection process a site investigation was carried out that was decided on the base of the limited available budget. Besides the topographic survey, geophysical survey, drilling and laboratory testing were undertaken. For the design of the structures the bearing capacity of the soils was assessed and the potential for soil liquefaction and settlement was analyzed.

C. Quadros and I. Mindo / Site Selection of the Mathemele Landﬁll

621

1. Site Investigations and Surveys The conditions of Mathemele, the site proposed for the waste landfill for Maputo and Matola municipalities were investigated considering the topography, geophysics, soils and water quality. 1.1 Site Location Analysis The proposed site of the new landfill facility at Mathemele is located about 20 km to the northwest of Maputo and 17 km to the north-northeast of Matola City, just west of the Infulene suburb and east-northeast of Matola Railway Station (Fig. 1).

LANDFILL SITE

MAPUTO

MATOLA Figure 1. Mathemele site location

1.2 Topography The morphology of the region is characterized by flat terrain with very little variation. The study area itself forms a gentle slope of about 0.25 %, falling towards the Infulene Valley. 1.3 Site Geology According to the Geological Map of Maputo at scale 1:50 000, the study area is geologically located in the Congolote Formation (QCo) (Upper Pleistocene age – 800 000 years) (Fig. 2). This Formation is a sand plain constituted by yellowish, orange and white occasionally slightly consolidated sand. On the eastern side close to Infulene River this Formation occasionally shows sand ridges. The soil does not vary much with depth. In the first 18 m, thick sands form the upper layer of the horizon with deep to

622

C. Quadros and I. Mindo / Site Selection of the Mathemele Landﬁll

very deep (5-15 m) stratified fine-grained silty sands. These silty sands are sometimes inter-layered with clay lenses, which in places are red to dark red in colour. The clay contents vary greatly, with moderately high contents within the upper layers associated with the valleys (colluvial), with inter-bedded lenses of fine silt and clay in places at depth and silty sands with very little clay occupying the dune ridges and wind-blown deposits. Grain-size composition is dominated by fine to medium well sorted sand of aeolian origin with about 5% of silty clay. Chemical analyses have shown high silica content, with about 91% of SiO2. Calcium Carbonate analyses revealed low content of CaCO3, varying between 1.1% to 1.8%. The sand samples recovered from the three boreholes drilled at the site confirmed the presence of fine grained silty sands, however it was not possible to identify clay lenses from the disturbed samples.

Rio Infulene

Legend QC0

- Congolote Formation

QMc

- Machava Formation

TPv

- Ponta Vermelha Formation

TIn

- Inharrime Formation

TSa

- Santiago Formation

Qa

- Alluvial Deposits

Figure 2. Geological Cross-Section of an area close to project site

1.4 Geophysical Investigations and Subsurface Water Surveys A geophysical investigation of the area was carried out. Nine VES (Vertical Electrical Soundings) were done and aimed at a depth of about 40 meters. The interpretation of the geophysical data shows basically descending type curves with a high resistivity layer on the top (sand dunes) and an intermediate layer over a low resistivity layer (aquifer). The knowledge obtained from the field data can be summarized as follows: •

The subsurface geology of the area is uniform and there are no major changes in lithology to be expected. This is indicated by the similar behaviour of the apparent resistivity curves over the studied area. The surface defined by the water level

623

C. Quadros and I. Mindo / Site Selection of the Mathemele Landﬁll

surface follows the morphology of the terrain. The direction of the groundwater flow was determined. •

The groundwater level indicated by geophysics is shallow and varies from 10 m to about 20 m with higher depths being observed in the eastern side. The measurement of Electrical Conductivity (EC) which is an indication of the salinity of water, is sensitive to the depth of the well/borehole. Deeper levels show higher values (1500 to 2000 μS/cm of EC compared to shallower levels from 500 to 800 μS/cm) – see Table 1.

Table 1: Groundwater quality test results

Parameter

Units

Borehol e 01

Borehol e 02

Borehol e 03

Admissible limits (Decree nº15/2004 of 15 June) minimum

pH

maximum

-

7.89

7.51

7.29

6.5

8.5

Electrical Conductivity

μS/cm

521

868

466

50

2000

Turbidity

NTU

50

0.6

13

Colour

-

Coloured* * colour due to its recent opening.

Colourless

Colourless

TDS* (deposit)

-

Absent

Present

Present

-

5

-

Nitrate (NO3)

mg/l

13.64

9.68

28.02

Nitrite (NO2)

mg/l

0.08

<0.03

0.03

Colourless

-

Absent

-

50

-

3

Chloride (Cl) Ammonia (NH4) Total hardness (Ca2CO3) Faecal coliform

mg/l

120.53

159.52

88.62

250

-

1.11

0.12

0.11

-

1.5

mg/l

38

130

56

-

500

number/ 100ml

<1

<1

2

-

10

* Total Dissolved Solids To confirm the findings of the geophysical study, three boreholes were drilled to a depth of between 25m and 33m. In all three boreholes, piezometers were installed for future monitoring of ground water level variations and water quality. The borehole data analysis shows the following:

624

C. Quadros and I. Mindo / Site Selection of the Mathemele Landﬁll

•

After a thin layer of topsoil, the soil profile is made up of medium to very fine sands with the fine sands being the largest component. Ground water level is found at depths from 13m to 15.5 m.

•

Permeability of the fine sands is estimated in the range of 10 -4 to 10-5 cm/s.

According to the sampling testing results and also the Hydrogeological Map of Mozambique, the project is located in an area of local intergranular aquifers with limited productivity or without significant groundwater, and with low to very low permeability (generally with Q< 5m3/h). The aquifer is developed in very fine to medium clayey sand of aeolian or marine origin, and consists of brackish to saline groundwater resources. The ground water quality results are shown in Table 1. Groundwater aquifers are recharged directly by precipitation or by groundwater seepage or flow from water bodies.

2. Soil Bearing Capacity and Soil Stability The existing soils are non-plastic sands. Their shear parameters were determined as being c’= 0 and Ø ‘= 30 º. The bearing capacity of these soils is estimated to be in the order of 150 – 180 kPa at the depth of 1m for a standard footing of 1 x 1 m 2 . For static loads the expected settlements are very small. The ground water level is relatively deep such that no influence for the bearing capacity and stability of shallow foundations is anticipated.

3. Seismicity The proposed site is located in an area of potential seismic activity. According to the Seismic Intensity Chart prepared by the Meteorological Service of Mozambique, the area investigated is classified as having a local seismic intensity of between VI and VII on the Modified Mercalli scale (MMS) with a 90% probability of not being exceeded during a 100-year recurrence period. The location of potentially active faults in the southern region of Mozambique was investigated. The distance of these faults to the site is at least 200 km. For the quasistatic structural design of the buildings and other structures it is adequate to adopt a local seismic coefficient of k=0.07. Due to the position of the groundwater level, soil profile, density of the sands in the natural state, no liquefaction effects under earthquake loading is likely.

Conclusions Based on the site investigation, physical modeling of the ground and the waste management principles adopted to select the location for the new urban landfill, Mathemele is a suitable location.

C. Quadros and I. Mindo / Site Selection of the Mathemele Landﬁll

625

Apart from various local ponds that may develop during the rainy season, there are no surface water bodies near the site. The nearest surface water body is the Infulene River some 7 km away. There is a high risk of ground water contamination and therefore appropriate measures have to be taken into consideration during the development and operation at the waste disposal site. It is recommended that an air and water quality buffer zone of at least 200 meters be established around the site. No natural cementation of the sands was found. The load bearing capacity of the sands can increase or decrease with the void ratio. However, for the constructions that are planned in connection with the landfill, no special measures are suggested. The buildings are light-weight and can sustain a certain degree of settlement without detrimental consequences. Infiltration of rainwater and drainage water near the building foundations should be avoided to minimize the effects of sand structural collapse. Due to the position of the groundwater level and soil profile, earthquake induced liquefaction is unlikely.

References [1] Momade F., Ferrara M. e Oliveira, J.T., 1995. Carta Geológica de Maputo na esclala de 1: 50 000, 1ª Edição. [2] Momade F., Ferrara M. e Oliveira, J.T., 1996. Notícia Explicativa da Carta Geológica de Maputo na escala 1: 50 000. [3] DNA, 1987. Carta Hidrogeológica de Moçambique na escala 1:1 000 000, 1ª Edição. [4] Costa Jr., L. and Sénvano, A., 1988. Água Subterrânea (Região de Maputo). UEM, Maputo.

626 Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-626

Hazard Assessment on Shallow Dolomite a

Tony A’BEARa,1 and Lindi RICHER a Bear GeoConsultants, Johannesburg, South Africa

Abstract. Research comparing a number of sites where dolomite bedrock occurs close to surface indicates that not all such sites should automatically be considered highly hazardous. There is good cause to believe that shallow dolomite sites should be examined in detail, possibly involving additional techniques, and that they should then be classified on merit. The factors normally used to evaluate dolomitic sites should be used to derive the degree of hazard. It also appears that there may be merit in using a factor related to the variation in depth to bedrock within such a site to help characterise the degree of hazard. Shallow dolomite sites require a modified approach to the investigation and this requires a change in approach to the gravity survey, the positioning of boreholes and includes using trenching on a large scale. Keywords. Shallow dolomite, hazard assessment, sinkhole, subsidence

Introduction Traditionally, shallow dolomite is considered to have a high hazard rating in terms of the potential for sinkholes to develop, usually resulting in an automatic classification of the site with an Inherent Hazard Class rating of 5 [2,3] indicating that it has a high potential for small sinkholes to develop. The origin of the reasoning behind this is not certain but probably relates primarily to bad experiences in certain low cost housing projects where numerous sinkholes developed in shallow dolomite terrain. Recent experience indicates that shallow dolomite should by no means be assumed to be highly hazardous in terms of the potential for sinkholes to develop. Recent work on the development of South African National Standards (SANS) relating to development of dolomitic land [1] has changed the use of the word risk to hazard when referring to the potential for sinkholes to develop in dolomite areas. This was done to bring the assessment of risk, or hazard, into line with the approach used overseas and with the terminology used in assessing other geotechnical hazards such as seismicity and slope stability.

1 Corresponding Author: Tony A’Bear, Bear GeoConsultants, PO Box 28334, Kensington, 2101, South Africa; Email: [email protected]

T. A’Bear and L. Richer / Hazard Assessment on Shallow Dolomite

627

1. Definition of Shallow Dolomite The current SANS [2] definition of shallow dolomite reads “dolomite land, where the average bedrock head is less than 8 m”. The definition also indicates that such areas are typically associated with gravity plateaus. In our experience, as shown later as well, areas in which dolomite pinnacles outcrop can have an average depth to bedrock well in excess of 8 m if deep grykes, solution features usually associated with weathering along joints, are present. The average depth to bedrock in boreholes may not truly reflect the average depth to bedrock across a site as boreholes are often sited to avoid outcrop or pinnacle heads. It is perhaps more appropriate to understand that the investigation of sites on shallow dolomite requires a different approach. Shallow dolomite areas could perhaps be better defined as being those areas in which bedrock, or pinnacle bedrock is commonly found within reach of a large track mounted excavator. This would indicate that a significant portion of the site is underlain by rock at depths of 4 m or less.

2. Study Areas Several study areas have been selected to examine dolomite sites which are considered to be shallow dolomite or closely approximating shallow dolomite. A number of these are known to be highly hazardous and have several sinkholes close to or within the site. A number of sites have also been selected which are considered to have a low hazard associated with the potential for sinkholes to develop or which have not had sinkholes develop despite the area having been developed for many years. The locations of the sites chosen for this study are shown in Figure 1. There are some limitations with respect to the methodology used in this assessment as the authors are not familiar with all of the sites and the hazard assessment has not taken into account factors such as uniformity of bedrock topography as would normally be indicated by a gravity survey. In addition, the sinkhole data base is not complete and may not truly represent the level of hazard associated with the site. Nevertheless, the sites selected should give a reasonable variation between low and high hazard sites underlain by shallow dolomite.

628

T. A’Bear and L. Richer / Hazard Assessment on Shallow Dolomite

Figure 1. Site localities in relationship to dolomite areas.

3. Characterisation of hazard The hazard assessment of sites underlain by dolomite usually employs the following investigation techniques: • A geophysical survey, usually employing a gravity survey on a 30 m grid, is carried out to form a picture of the variation in bedrock topography. This usually forms the basis for zoning the site. • Rotary percussion boreholes are drilled to establish and evaluate the strata overlying dolomite bedrock. The following factors are typically used to evaluate the degree of hazard associated with sinkhole and subsidence (doline) development in areas underlain by dolomite [3]: • • • •

Nature of the overburden, including its potential to erode into underlying cavities and its potential to absorb, or reduce the velocity of, water flowing vertically through it. The depth to the present groundwater level and its position relative to bedrock and overburden. The presence and depth of cavities as well as air and sample loss when drilling in boreholes. Bedrock morphology. For sites underlain by shallow dolomite the variation in depth to bedrock plays an important role as this reflects the degree of gryke and pinnacle development.

629

T. A’Bear and L. Richer / Hazard Assessment on Shallow Dolomite

The sites selected for the study have been analysed primarily using depth to bedrock and the nature of the overburden. An assessment of the level of hazard associated with the site is based on the occurrence of sinkholes in the immediate vicinity and within similar conditions. The results have been summarised in Table 1. As a measure of how much the depth to dolomite bedrock varied on a site the statistical measure of dispersion of data, variance, was used. The variance is defined as the square of the standard deviation. The variance showed a more pronounced difference between the low hazard and high hazard sites as opposed to the standard deviation. The low hazard sites, K, L, and M especially, show a marked difference from the others in that the variance in depth to bedrock is much lower, well below 10 m 2, and in addition show a definite lack of air loss or the presence of cavities. Sites with a low to moderate hazard rating tend to have a fairly low percentage of boreholes showing air loss or cavities. In these instances the variance in bedrock level would appear not be a good indicator of the level of hazard to be expected. However, further research involving a more detailed analysis of the data may find that there is some significance in the variations. Table 1. Summary of boreholes and Hazard Level for various sites on dolomite Minimum depth to bedrock in boreholes (m)

Maximum depth to bedrock in boreholes (m)

Average depth to bedrock in boreholes (m)

Variance in depth to bedrock in boreholes (m2)

Boreholes with cavities or airloss (%)

Site A

1.0

18.0

8.7

13.2

28.21%

Site B Site C Site D Site E Site F

5.0 3.0 8.0 2.0 6.0

21.0 20.0 17.0 35.0 21.0

12.1 10.3 14.0 19.6 11.8

31.6 53.7 27.0 173.7 29.4

0.00% 53.85% 100.00% 5.13% 83.33%

Site G

0.1

28.0

10.0

93.7

0.00%

Site H

0.0

25.0

10.4

49.7

11.54%

Site I Site J Site K Site L Site M

2.0 2.5 0.0 1.9 1.5

17.8 20.0 4.7 12.0 6.0

7.2 14.1 1.2 4.9 3.5

22.3 45.3 1.0 6.3 1.8

20.00% 16.67% 0.00% 0.00% 0.00%

Site

Site Hazard rating

Medium/ High Medium High High Medium High Medium/ High Medium/ High High High Low Low Low

Sinkholes

None Present Present Present Present Present None None Present Present None None None

4. Conclusions There is clear evidence that the current approach defining all shallow dolomite sites as being highly hazardous is not appropriate. Shallow dolomite sites should be investigated using a similar approach to that used where bedrock is deep although variations in the nature of the investigation are warranted.

630

T. A’Bear and L. Richer / Hazard Assessment on Shallow Dolomite

Similar to deeper dolomite, it is obvious that the presence of cavities or the occurrence of air loss during drilling is a good indicator of highly hazardous conditions. The variance in depth to bedrock appears to be a useful indicator of low hazard conditions but does not always appear be relevant to medium and high hazard sites. It is probable that in areas where few or no sinkholes have occurred, Site H for example, but where the variance is still high, that other factors, such as the nature of the material in the grykes, play an important role.

5. Proposed approach for analysing shallow dolomite Gravity surveys on shallow dolomite terrain tend to show little variation in depth to bedrock and are unable to pick up the narrow grykes which are characteristic of such terrain. This is largely due to the grid spacing being too wide. Narrowing the grid spacing to, say, 5 m tends to improve the detail recovered dramatically but this is usually only cost effective on small sites. On larger sites, in excess of one or two hectares in area, the increase in cost is usually too large for the project to bear. It is nevertheless recommended that all sites be covered by a gravity survey, unless too small, and that this should be done on as small a grid as possible. Should this not be feasible then it is recommended that a number of traverses with closely spaced stations be added into the conventional survey. These should be at least four traverses and these should be split between two orthogonal directions. This should allow for smaller features to be picked up and some estimate made of direction and spacing. Trenching using large excavators to expose the bedrock over long distances is highly recommended and has been used successfully on a number of sites by the authors. This allows for pinnacle spacing and gryke widths to be determined as well the nature of the gryke infill to be examined. The geotechnical parameters of the gryke infill can sometimes be determined by taking undisturbed samples or at least be estimated from disturbed samples. Drilling to establish the depth of the grykes and the nature of them is still required. However, it is strongly recommended that clusters of boreholes be used to characterise the area with spacing between boreholes in each cluster being set at between 5 and 10 m. As shallow dolomite sites usually result in a substantial reduction in the total depth drilled it becomes cost effective to drill more holes but in a different pattern from the norm. In this sense it is recommended that the minimum number of boreholes recommended per hectare by the authorities be increased by at least 15% in shallow dolomite sites. As a change in approach is required when investigating shallow dolomite sites it is recommended that test pitting, albeit limited in extent, proceed ahead of the normal stability survey in order to determine whether the site is a shallow dolomite site or a deep dolomite site if this is not known.

6. Further work The Council for Geosciences in South Africa is currently creating a database of boreholes and sinkholes which will, in the not too distant future, allow for far more

T. A’Bear and L. Richer / Hazard Assessment on Shallow Dolomite

631

useful research to be carried out in terms of better defining the characteristics of low, medium and high hazard shallow dolomite areas. There appears to be some validity in looking at a parameter using the variance in depth to bedrock within a shallow dolomite site and it is recommended that this be further pursued. It would also appear to be critical that a better understanding be obtained of the nature of wad rich soils in shallow dolomite areas and the ease with which they are mobilised as it would appear that some shallow dolomite areas, located primarily in Formations which are deficient in chert, are less susceptible to sinkhole and doline development. A set of tests, possibly involving large scale field trials, to determine the erodability and permeability of these soils is required.

Acknowledgments The authors wish to thank members of the Council for Geosciences, in particular Lindy Heath, Samantha Richardson and Therina Oosthuizen who kindly gave of their time, provided information and discussed their research with us.

References [1] SABS, SANS 1936-1:2010 South African National Standard: Development of dolomite land Part 1: General principles and requirements, SABS Standards Division, Pretoria, Committee Draft. [2] SABS, SANS 1936-2:2009 South African National Standard: Development of dolomite land Part 2: Geotechnical investigations and determinations, SABS Standards Division, Pretoria, Committee Draft. [3] D.B. Buttrick, A. van Schalkwyk, R.J. Kleywegt and R. Watermeyer, Proposed method for dolomite land hazard and risk assessment in South Africa. Journal of the South African Institution of Civil Engineering 43(2) 2001, pp 27-36.

632 Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-632

Correlations of DCPT and SPT for Analysis and Design of Foundations Dalmas L.NYAOROa,1 and Mwajuma IBRAHIM b University of Dar es Salaam , Dar es Salaam, Tanzania b Mbeya Institute of Science and Technology, Mbeya, Tanzania a

Abstract. Both Standard Penetration Test (SPT) and Dynamic Cone Penetration Test (DCPT) have been carried out in Tanzania for many years whereby data obtained from these tests have been used for analysis and design of foundations. Data from various geotechnical investigation projects where the SPT and DCPT were carried out alongside one another were analysed and used to formulate correlations between the two tests data for different soil types while considering testing depth for each soil type. From the analysis, it was observed that a strong correlation between DCPT values and SPT values exist. These correlations can be useful in soil investigation works whereby the number of SPT testing involving expensive boreholes drilling can be reduced while supplementing them with DCPT thus enjoy time and cost saving ability of the DCPT at the same time obtaining more detailed information of the ground within the site being investigated. Keywords. DCPT, geotechnical investigation, penetrometers data correlations, penetration tests, SPT

Introduction A thorough and comprehensive soil investigation is essential prior to design and construction of any civil engineering project of significance. The extent and scope of the investigation will depend partly on the nature of the site and partly on the type of the structure to be erected. The depth of exploration is related mainly to the types of material present and their susceptibility to compression under load. Reasonably good estimates of properties for cohesive soils can be made by laboratory tests on undisturbed samples but it is nearly impossible to obtain undisturbed samples of cohesionless material for strength testing. For this reason, relative density, strength and compressibility estimates are usually obtained from penetration tests or other in-situ methods. Penetration tests have long been used to evaluate soil consistency and relative density. There are numerous penetrometers of standardized design; the Standard Penetration Test (SPT) and the Dynamic Cone Penetration Test (DCPT) are among them. SPT is currently the most popular means of obtaining subsurface information. For example 85% – 90% of conventional foundation design in America is done using the test [1] and it is also popular in Great Britain [2]. Experience in using SPT has enabled a considerable knowledge of the behaviour of foundations in sands and gravels to be accumulated [3]. As a result, relationships have been developed between SPT “N” 1

Corresponding Author.

D.L. Nyaoro and M. Ibrahim / Correlations of DCPT and SPT

633

values and different soil parameters. Despite the usefulness of SPT, the test results are sensitive to test details and so are the correlations derived from them. Continuous penetration soundings such as DCPT serve in general to investigate soils with irregular profile, to indicate possible zones with lack of resistance that could not be determined from the investigation executed by discrete soundings such as SPT. Given that soils of irregular stratigraphy constitute a greater part, continuous sounding is therefore advantageous. Besides the advantages, the dynamic penetration test with conical point is not universally standardized [1] and also there is little information published regarding DCPT correlations to justify using DCPT in preference to SPT [3]. Where correlations exist, they are usually specific to a locality.

1. Background Both SPT and DCPT have been carried out in Tanzania for many years. However, consulting firms still design foundations using SPT correlations that are not adjusted to suit the local soil conditions, equipment and practice. This is because there are no such correlations or guidelines that can be used to properly interpret the test results. The cost of soil investigation depends on the method of investigation employed as well as the extent of investigation. When compared to DCPT, SPT allows sample recovery for further laboratory testing and analysis but it is more expensive and more time consuming. On the other hand DCPT does not allow recovery of soil samples. Reliable correlations between the two penetrometers test data can be used to relate them and hence facilitate their best use. The main objective of this study is to establish correlations between SPT and DCPT values based on local Tanzanian conditions. This will promote the use of the best alternative among the methods or a combination of the two methods depending on the problem in hand.

2. Field and Laboratory Data The fact that both DCPT and SPT are common tests for soil investigation in Tanzania has enabled the use of some of the existing data obtained from previous investigation projects. A total of 37 SPT boreholes done side by side with corresponding 37 DCPT obtained from five project sites in Dar es Salaam and Coastal region were used to establish correlations. The type of equipment used for data collection in all the project sites were: For SPT, automatic trip hammer of 65Kg with a falling height of 0.76m as specified in BS 1377:1990 and for DCPT, a 10Kg hammer falling at 50cm and 25.2mm cone diameter as specified in DIN 4094: 1974. The borehole logs indicate the presence of clayey SAND and silty SAND as the major dominating sub soils for the sites while the water table was below 10 meters for all the sites and since the DCP testing was possible to penetrate the ground to a depth not more than 10 meters then the DCPT and SPT values used in formulating the correlations were not affected by water table.

Figure 1. DCPT instrument to DIN 4094: 1974

634

D.L. Nyaoro and M. Ibrahim / Correlations of DCPT and SPT

3. SPT – DCPT Correlations In formulating the correlations, two major factors were considered in grouping the data; the soil type and the test depth. All these have a significant effect in data values for both SPT and DCPT. In this way, soils of the same type were grouped together and the correlations were formed by first considering smaller intervals of test depth and then a wider range of test depth to observe the effect of depth in the correlations. 3.1. Correlations in Clayey SAND The relationship between SPT N values and DCPT n values for clayey SAND tested at 2 – 4 m depth is as shown in Figure 1 where the best fit equation is N = 0.514n + 2.122 with R2 value of 0.827. As it can be seen from the figure, the zone of relationship is bounded between two nearly parallel lines with upper boundary represented by the equation N = 0.514n + 8.122 and the lower boundary equation is N = 0.514n – 2.420. By considering test depth of 4 – 7m range in the same clayey SAND, the relationship between N and n values is as shown in Figure 2 where the best fit line is N = 0.597n – 1.162 which is different from the one presented in Figure 1. The difference suggests the importance of considering test depth while formulating and using the correlations. A combination of data from Figures 1 and 2 represents the correlation for a depth range of 2 – 7m as shown in Figure 3. The correlation equation becomes N = 0.548n + 0.971 with R2 value of 0.777. This combination suggests an alternative way of formulating the correlations for the same soil by considering a single correlation to save for any depth range which can be accompanied by some correction factors to cater for specific depth range and therefore minimizing the number of necessary correlations.

Figure 2. Correlation between N and n values in Clayey SAND at 2 – 4 m depth

D.L. Nyaoro and M. Ibrahim / Correlations of DCPT and SPT

Figure 3. Correlation between N and n values in clayey SAND at 4 – 7m depth

Figure 4. Correlation between N and n values in clayey SAND at 2 – 7 m depth

635

636

D.L. Nyaoro and M. Ibrahim / Correlations of DCPT and SPT

3.2. Correlations in Silty SAND Correlation formed by combining data for well graded and that for poorly graded silty SAND have very low R2 value but they become better if each soil is considered separately. Figures 4 and 5 indicate the correlations formed by considering well graded silty SAND and poorly graded silty SAND respectively. Basically the correlations are in form of zones bounded between two straight parallel lines.

Figure 5. Correlation between N and n values in well graded silty SAND at 2 to 6 m depth

Figure 6. Correlation between N and n values in poorly graded silty SAND at 2 – 7 m depth

D.L. Nyaoro and M. Ibrahim / Correlations of DCPT and SPT

637

3.3. Correlations in Silty and Clayey SAND A correlation formed by considering the three types of soils together is as seen in Figure 6 and like when each of the soil was considered separately, it takes the form of a zone bounded between two parallel lines. The value of R2 is large enough to comment that the penetration values relate strongly. This suggests the possibility of formulating a universal correlation to save for many types of soils and providing correction factors for specific soil types.

Figure 7. Correlation between N and n values in clayey and silty SAND at 2 – 9m depth

4. Conclusion This study has enabled the formulation of correlations between SPT N values and DCPT n values with consideration of test equipments that are common and useful in Tanzania. The effect of soil type and test depth has also been considered. It is observed that the relationships of the penetration values are expressed in form of zones bounded between two parallel lines with the best fit line passing nearly through the centre of each of the zones. These correlations can be useful in soil investigation works whereby the number of SPT testing involving expensive boreholes drilling can be reduced while supplementing them with DCPT thus enjoy time and cost saving ability of the DCPT at the same time obtaining more detailed information of the ground within the site being investigated.

References [1] J.E. Bowles, Foundation Analysis and Design, McGraw-Hill, Singapore, 1996. [2] R. Whitlow, Basic Soil Mechanics, Pearson Education Ltd, Essex, 2001. [3] M.J. Tomlinson, Foundation Design and Construction, Pearson Education Ltd, Essex, 2001

638 Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-638

The Effective Porosity Paradigm and the Implications on Empirical Permeability Estimations a

Matthys A. DIPPENAARa,1 and J. Louis VAN ROOY Department of Geology, University of Pretoria, Pretoria, South Africa

Abstract. Porosity can generally be subdivided into soil material or texture and soil mass or structure, ranging over macro-scale and influencing the complete profile to submicro-scale. A number of empirical grading-based approaches used widely relate porosity to the uniformity coefficient and subsequently the grain size representing the finest 60% and 10% of the sample respectively. Methods such as these, however, clearly do not acknowledge two important aspects: (1) effective porosity is required for flow and is influenced by soil structure; an undisturbed sample will therefore not suffice in adequately describing the interconnected pore spaces, and (2) one sample is typically confined to a fairly small amount of material from one soil horizon and the scale of investigation is therefore limited and does not include larger structures or erosion features which may serve as pathways of more rapid flow, thereby increasing the seepage parameters based on the estimated porosity. This paper aims to address some of the problems pertaining to estimate porosity η, effective porosity ηe and the importance of better evaluation techniques. Hydraulic conductivity and intrinsic permeability are often calculated as a function of some constant, a porosity function and an effective grain size de where de is often preselected (usually as equal to d10 or the median grain size d50). However, during this ongoing work it was found that this value should vary for soils of different texture, but that the effective grain sizes corresponding to these d-values remain fairly constant and is almost independent of soil type. Calculation of conductivity and permeability are, therefore, influenced significantly by assumptions in porosity estimates and need to be addressed. Keywords. porosity, permeability, hydraulic conductivity, grading curve, effective grain size

Introduction Assessment of unsaturated seepage parameters – notably porosity (η) or void ratio (e) and effective porosity (ηe); hydraulic conductivity (K) or intrinsic permeability (k); and the influence of changing moisture content (θ) with gradual saturation (S) – is often based on single near-surface field tests, laboratory experiments mimicking saturated conditions in disturbed samples, or grading-based empirical relations. These, however, pose a direct problem due to intrinsic assumptions and issues pertaining to representativeness of the sample material or test site. It is not always possible to

1

Corresponding Author.

M.A. Dippenaar and J.L. van Rooy / The Effective Porosity Paradigm

639

conduct field tests through the lateral, vertical and temporal heterogeneity and subsequently values are estimated based on a non-representative elementary volume. On a more empirical scale, methods are mostly based on “Hazen materials”, i.e. sands with a uniformity coefficient below 4. This rarely exists in nature and makes the methods essentially inapplicable to more non-uniform and/ or finer-grained materials.

1. Empirical Relationships The mean grain size diameter – assumed representing the effective (or flowcontrolling) particle size de– can be calculated as the harmonic mean, weighted with the fraction fi between the upper and lower sieve sizes, di(max) and di(min) respectively (Eq. 1). Based on this, comparison can be made between the assumptions of de = d5, de = d10, de = d20 and even de = d50. 1 = de

n

(

)

⎡ d i (max) + d i (min) ⎤ ⎥ 2 fi ⎦ i =1

∑ ⎢⎣

(1)

The methods applied are those according to Hazen ([1], Eq. 2), USBR ([2], Eq. 3); Kozeny ([3], Eq. 4) and Zamarin ([4], Eq. 5). Where required, groundwater temperature was assumed constant at 15Σ͘ 2 K S = c ⋅ T1 ⋅ d10

T1 = 0.70 + 0.03 ⋅ T and c = 4.6 ×10 −3 + 4.6 × 10 −2 (η − 0.26)

(2)

2.30 K S = 0.36 ⋅ d 20

(3)

⎡ η 3 ⎤ 2.30 ⋅ d 20 K S = 5400 ⋅ ⎢ 2⎥ ⎢⎣ (1 − η ) ⎥⎦

(4)

⎡ 1 ⎤ 2 K S = 8.07 ⋅ ⎢ ⋅ (0.926) ⋅ (1.275 − 1.5η ) 2 ⋅ d10 (at 15o C : d e = d10 ) 2⎥ ⎢⎣ (1 − η ) ⎥⎦

(5)

Calculating porosity from grading results is a function of the uniformity coefficient CU only according to the method of Istomina [5] and is shown in Eq. 6. In order to be applicable to fine-textured materials, determination of a d10value becomes problematic as more than 10% of the sample passes 0.002mm and no further refinements are supplied. For this reason, an estimate of d10 is required and can range through the complete spectrum of approximately 1 x 10-10 – 2 x 10-3mm, allowing for seven orders of magnitude difference in CU despite most methods assuming CU< 21 or even CU< 4. For the sake of this dataset, d10 was assumed between 10-5 and 10-3, depending on the shape of the grading curve.

640

M.A. Dippenaar and J.L. van Rooy / The Effective Porosity Paradigm

(

)

n = 0.255 1 + 0.83CU where CU =

d 60 d10

(6)

2. Data and Results The dataset comprise 32 residual to highly weathered granite to granite-gneiss samples (of the Goudplaats-Hout River suite and Bushveld Igneous Complex) and 29 residual to highly weathered gabbro-norite samples of the Bushveld Igneous Complex. The aim was to assess the grading curves as a function of soil origin as noted above (with granite and gneiss differing significantly from gabbro-norite in terms of mineralogy and nature of weathering). The maximum, minimum and average grading curves, as well as the calculated range of effective grain size diameters, were calculated for the granite-gneiss soils (Fig. 1) and the gabbro-norite soils (Fig.2). The empirical results were also correlated with 17 percolation tests on granitegneiss and four on gabbro-norite. The tests were conducted in 150 mm diameter, 400 mm deep trial holes excavated from surface. The percolation rate was measured and is shown in Fig. 3. Note the overall poor correlation with the empirical methods as well as the large range of K-values for the respective empirical approaches.

Figure 1.Granite-gneiss grading curves and calculated effective grain size diameter (arrow indicates highly weathered becoming residual; shaded portion indicates range of effective grain sizes and the corresponding d-values).

641

M.A. Dippenaar and J.L. van Rooy / The Effective Porosity Paradigm

Percolation Rate/ Hydraulic Conductivity (m/s)

Figure 2. Gabbro-norite grading curves and calculated effective grain size diameter (arrow indicates highly weathered becoming residual; shaded portion indicates range of effective grain sizes and the corresponding d-values).

1.00E-01 1.00E-03 1.00E-05 1.00E-07 1.00E-09 1.00E-11 1.00E-13 1.00E-15 P (m/s)

GRANITE max K (m/s)

-

ave

-

Hazen

min

| USBR

max

-

ave Kozeny

-

min NORITE Zamarin

Figure 3. Percolation rates (P) versus hydraulic conductivities according to Hazen, USBR, Kozeny, Zamarin and average empirical conductivity (K).

642

M.A. Dippenaar and J.L. van Rooy / The Effective Porosity Paradigm

3. Discussion Based on the results, effective grain diameters to be used in empirical calculations of porosity and hydraulic conductivity therefore differ for the different textures with essentially the following generalizations being evident: •

For granite-gneiss, de ranges between d6 (0.0035 mm) for highly weathered bedrock to d15 (0.0004 mm) for residuum.

•

For gabbro-norite, de ranges between d3 (0.0045 mm) for highly weathered bedrock to d40 (0.0004 mm) for residuum.

•

The assumption of an effective grain size diameter relating to the size relating to the finest fraction of the sample appears to apply to granite-gneiss as the coarse materials yield de ≈ d10 as suggested by Hazen.

•

For fine-grained material, de >> d10 although de ≈ d10 for the coarser less weathered products with less clay minerals and with lower activity.

•

For granite-gneiss the average de ≈ d10 corresponding to 0.004 mm and for gabbro-norite de ≈ d20 corresponding to 0.002 mm, indicating overall minor deviations for highly to completely weathered materials.

It is probably much more useful to relate the effective grain size diameter directly to the particle size as opposed to the percentage represented by a certain fraction. This becomes independent of mineralogy or rock type and appears to be solely a function of the state of weathering, i.e. approximately 0.0004 mm for residuum to approximately 0.0035 – 0.0045 mm for highly weathered rock over the range of extreme intrusive igneous rock compositions. However, mineralogy does play a role in the hydraulic properties of a material and effective grain size diameters cannot solely be responsible for flow through a medium. This, therefore, also puts a question mark over the validity of using arbitrary d-values to force hydrological parameters on soil types. As for porosity, calculations according to Istomina yield 0.255 ≤ η ≤ 0.280 (average 0.256) which, once again, indicate much less variation than anticipated. This can most probably be ascribed to the ineffectiveness of basing porosity estimates solely on the uniformity coefficient and the excluded influence of sphericity and packing of grains and secondary porosity. The generally poor correlation between field percolation tests and empirical approaches can potentially also be ascribed to the issues pertaining to porosity and effective grain size estimates as well as changing moisture contents and soil mass in the field.

Acknowledgements The authors wish to thank the Water Research Commission for continuous funding of this project.

M.A. Dippenaar and J.L. van Rooy / The Effective Porosity Paradigm

643

References [1] Hazen, A. Water supply: American civil engineers handbook. New York, Wiley, 1930. [2] Kozeny, J. Soil permeability. SitzungberOesterrAkad. Abt 136:271 (1927). [3] Van Schalkwyk, A. and Vermaak, J. J. G. The Relationship between the Geotechnical and Hydrogeological Properties of Residual Soils and Rocks in the Vadose Zone. WRC Report No. 701/1/00, Water Research Commission, Pretoria, 2000. [4] Vukovic, M. and Soro, A. Determination of hydraulic conductivity of porous media from grain-size composition. Littleton Colorado, Water Resources Publications, 1992. [5] Istomina, V. C. Seepage stability of the soil. 1957 Translated from Russian. Moscow. In: Van Schalkwyk and Vermaak, 2000.

644 Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-644

Numerical Modelling of Wave Propagation in Ground Using Non-Reflecting Boundaries S.J. MBAWALA, G. HEYMANN, C.P. ROTH and P.S. HEYNS University of Pretoria, Pretoria, South Africa Abstract. A critical aspect when modelling wave propagation in the ground with the finite element technique is using non-reflecting boundaries. The choice of element size and domain size to fit non-reflecting boundaries becomes difficult when a forcing function with a large range of frequencies is used. This paper presents the results from a finite element study designed to gain insight into wave propagation in soils due to an applied vertical harmonic load with varying frequency content ranging from 10 Hz to 95 Hz using Abaqus/Explicit. Comparison is made between numerical results and measured field data. It is concluded that Abaqus/Explicit can reliably model the wave propagation problem. Particularly, the phase velocity determined by finite element modelling compared favourably to values obtained theoretically, as well as in the field. In addition, good correlation was found between the dispersion data from the model and field measurements. Keywords. Phase velocity, Finite element technique, Non-reflecting boundaries, Forcing function and Wave propagation

Introduction Modelling wave propagation with finite element techniques for soil–structure interaction problems has received increased interest in recent years. This includes problems of analysis and design of machine foundations, ground borne vibration induced by railway traffic, Rayleigh wave dispersion techniques used to determine shear wave velocity and wave propagation techniques used to detect near surface discontinuities. Different finite element software codes have been developed to analyse the propagation problems. The disadvantage of the finite element method (FEM) is that it needs large amounts of memory to solve the matrixes. This is more critical for wave propagation problems as very fine meshes are needed. The current advancement and availability of powerful computers certainly reduces the problem related to memory. However, the cost of computing time remains an issue for analysing large problems. Different techniques have been introduced to reduce computing time. Amongst others is the introduction of non-reflecting elements at the boundaries to avoid spurious reflections. Liu and Jerry [1] introduced a decaying function in the finite element method to completely damp out Lamb wave propagating along a plate. The strip element method (SEM) was developed for bounded bodies. The method was extended to deal with infinite bodies by introducing infinite elements and non-reflecting boundaries condition [2]. Abaqus uses a user element for seismic

S.J. Mbawala et al. / Numerical Modelling of Wave Propagation

645

analysis to absorb waves due to free-field motion formulated by combining a viscous boundary and free–field soil column [3]. The free–field soil column accounts for seismic motion along the edge of the model while viscous boundary absorbs radiating waves. Other important aspects to be followed during modelling include the estimation of mesh size and time increment. The estimation should consider the frequency range, wavelengths of interest and wave propagation characteristics of individual finite elements. The objective of this study is to establish the reliability of Abaqus/Explicit in numerical modelling of wave propagation problems. Abaqus/Explicit is a specialpurpose analysis under Abaqus products that uses an explicit dynamic finite formulation [4]. The wave propagation in ground is investigated by simulating the continuous surface wave (CSW) tests. The CSW is a seismic technique that can be used to determine ground stiffness by measuring the velocity of a Rayleigh wave propagating along the ground surface [5]. The finite element study was designed to gain insight into the physical wave propagation in the ground due to an applied vertical harmonic load with varying frequency content ranging from 10 Hz to 95 Hz using Abaqus/Explicit. The size of the model was 100 m x 100 m in plan and 50 m deep. The analysis considered only one quarter of the model in plan taking advantage of symmetry. The model was partitioned in two sections, the section with finite elements (25 m x 25 m) and infinite elements with length of 25 m placed at the left bottom and right top of the model as illustrated in Figure 1.

1.

Boundary Conditions

Every finite element model must be terminated at some finite boundary. The use of the finite element method in analysing problems involving wave propagation encounters the problem of wave reflection at the boundaries. The simplest solution to this problem is to define a domain large enough so that waves reflected from the boundary do not have time to return to the region of interest. However, due to the relatively high wave speeds of soil this is not often a practical option. Therefore, it is important to introduce boundaries that do not reflect waves. Numerous solutions have been proposed on the subject of non-reflecting boundaries [1-3]. Lysmer and Kulhmeyer [6] suggested the first local non-reflection boundary. This solution is only efficient for waves propagating normal to the boundary (P-waves and S-waves). Therefore, it is important to verify the efficiency of nonreflecting boundaries to absorb Rayleigh waves. The solution that is most popular to absorb incident waves is the viscous boundary traction (dashpots) [3, 6]. Abaqus implements the principles of this theory by defining a non-reflecting boundary condition using infinite elements [7].

646

S.J. Mbawala et al. / Numerical Modelling of Wave Propagation

25 m

25 m

A

25 m

4m

Infinite elements

50 m

Finite elements

Symmetry line

Figure 1. Finite/Infinite Element Model of Vertical Harmonic Load

2.

Finite Element Modelling

The most reliable solution to the wave propagation problem can only be achieved if the finite element program is well chosen. Abaqus is a general finite element software code used by researchers to investigate among other ground vibration. Abaqus/Explicit has been used to model Rayleigh waves being propagated along the surface of soil mediums, and acceptable agreement has been reported between the numerical simulation and the field measurements, with judicious choices of appropriate domain scale, mesh size and boundary conditions [8-12]. Therefore, in this study the dynamic commercial finite element program Abaqus was used in the analysis using the Abaqus/Explicit scheme without Rayleigh damping. Rayleigh damping assumes the damping matrix is a linear combination of the mass and stiffness matrixes [13]. 2.1

Infinite Element Modelling

The idealised domain size of the finite element model used in this paper is shown in Figure 1. The ground is assumed a homogeneous isotropic, elastic medium with soil properties shown in Table 1.

647

S.J. Mbawala et al. / Numerical Modelling of Wave Propagation

Table 1. Soil properties of the ground Soil bulk density (kg/m3) 2000

2.2

Young’s modulus (E) MPa 260

Poisson’s ratio 0.3

Mesh Size

The dimension of the finite elements must be selected taking into consideration the wavelength of the propagating perturbation. The finite element mesh size determines the highest frequency in the finite element analysis [14]. The mesh acts like a low–pass filter. Therefore, large element dimensions filter short wavelengths [10]. On the other hand, employing very small elements can cause numerical instability. The element size must be chosen according to the frequency content of the applied load. Differences in element sizes should be as small as possible as they may cause spurious reflections. In Eq. (1), if ( f max ) is the maximum frequency of the applied load, and Vi is the propagation velocity of a wave which can be a P-wave, S-wave, or Rayleigh wave in material, then the mesh element size should satisfy the following relationship [13].

g ≤ ςλmin = ς

Vi f max

(1)

where g is mesh size and λmin is a minimum wavelength, which is estimated using maximum frequency of the wave f max . The constant ς depends on whether the mass matrices are consistent or lumped. For consistent mass matrices the constant

ς = 0.25

and for lumped mass ς = 0.2 . In a lumped mass, the mass of the element is represented by particle masses lumped at the node matrixes while for consistent mass matrix the mass of element is linearly formulated and uses the same shape function used to formulate stiffness matrix [9, 13]. In Abaqus the first order elements have a lumped mass formulation. In this study, the element size was estimated using the material properties shown in Table 1. A Rayleigh wave travels with lower velocity than P-waves or S-waves. For the given material, the theoretical Rayleigh wave velocity is 205.7 m/s. The maximum frequency of the forcing function is 100 Hz, which gives the maximum approximate mesh size of 0.41 m. After mesh size refinement, the minimum mesh size of 0.1 m was used in this analysis. 2.3

Time Increment

The time increment is estimated from the maximum wave velocity, which in this case is the velocity of the P-wave. The time increment should be carefully chosen to maintain numerical stability and accuracy. Numerical instability may cause the solution to diverge if the time increment is too large. Contrarily, a very short time increment can cause spurious oscillation. The calculation of the time increment depends on the element dimension. To estimate time steps the following relationship has been suggested [13].

Δt =≤

g Vp

(2)

648

S.J. Mbawala et al. / Numerical Modelling of Wave Propagation

where Δt is the time step, g is the element dimension and V p is the P-wave velocity. For the material shown in Table 1, the P-wave velocity is 418 m/s and the time step is estimated to be 0.0002 s. Abaqus/Explicit uses Eq. (2) to adjust time increment size automatically. Therefore, automatic time increment was selected in all analyses to prevent any numerical instability. The main parts of the model were created in Abaqus/CAE as a visualization tool [15]. The finite sections comprise 4-node, linear, axisymmetric, solid continuum and reduced integration elements (CAX4R), while the infinite sections comprise 4-node linear, one-way axisymmetric solid continuum infinite elements (CINAX4). A total of 63000 elements were created in the domain. To include infinite elements in the model the infinite elements were defined subsequently by editing the created input file manually.

3.

Verification of the Selected Boundary Condition

The standard viscous boundary suggested by Lsymer and Kulhmeyer is efficient only for waves propagating normal to a boundary [6]. To check the performance of non–reflecting boundaries, fixed boundaries were introduced on the right hand and at the bottom of the model. Thus, the infinite elements were replaced with finite elements. In all cases, a vertical harmonic force along the axis of symmetry excited the model and the time history of vertical displacement responses was observed at point A, 4.0 m from the source of excitation as illustrated in Figure 1. Figure 2 illustrates wave displacement for non-reflecting boundaries and fixed boundaries. From Figure 2, it can be seen that the displacement response of the model with fixed boundary is higher than the wave displacement for the model with nonreflecting boundaries by about 85 %. This confirms the importance of introducing nonreflecting boundaries when dealing with wave propagation problems in finite element analysis.

4. Finite Element Modelling by Characterising Wave Propagation in the Ground The phase velocity of a wave is the rate at which the phase of the wave propagates in the medium. For a Rayleigh wave this is the speed at which the phase of any frequency component of the wave travels and is given in terms of the wavelength λ and frequency f , VR = fλ . An experimental result on normally dispersive soil where stiffness increases with depth shows that the phase velocity increases with decrease in frequency [16]. The increase in the phase velocity is due to higher stiffness of the soil with depth, therefore, for homogeneous soil profiles the phase velocity is constant (nondispersive) at all frequencies. The aim of this investigation was to determine the wavelength and velocity of a Rayleigh wave for each vibration frequency. A vertical harmonic load along the axis of symmetry excited the model, with frequencies ranging from 10 Hz to 95 Hz. The vertical responses in the time domain were observed at intervals of 1 m, starting at 1 m away from the applied load. The dominating frequency and phase angle at each

649

S.J. Mbawala et al. / Numerical Modelling of Wave Propagation

observation point was determined by calculating the Fourier transform of the displacement by means of the fast Fourier transform (FFT) algorithm. The phase angle at the dominating frequency was determined from a real (Re(z)) and an imaginary (Im(z)) component of the phase vector using Eq. (3). From the response of each point, the phase changes were determined.

⎛ lm( z ) ⎞ φ = tan −1 ⎜⎜ ⎟⎟ ⎝ Re( z ) ⎠

(3)

0.40 Fixed Boundaries Non reflecting boundaries

0.30

Displacement (mm)

0.20 0.10 0.00 -0.10 -0.20 -0.30 -0.40 0.0

0.1

0.2

0.3

0.4

0.5

0.6

Time (sec)

Figure 2. The effect of Fixed Boundaries in Wave Propagation Problems

The phase angles for frequency between 10 Hz and 95 Hz were used to estimate the phase velocity. Figure 3 shows the plots of phase angle versus distance for frequencies from 10 Hz to 95 Hz. The straight lines indicate that the phase velocity is constant within the domain. The wavelength λ for each vibrating frequency was determined using Eq. (4).

λ=

d Δφ n+ 2π

(4)

where d is the distance between two observation points, Δφ is the phase change from one observation to another and n is an integer, which depends on the number of wavelengths between the observations. The theoretical shear wave velocities of the soil profile were calculated using the soil properties shown in Table 1.The calculated theoretical shear wave velocity is 223.6 m/s. The velocity of a Rayleigh wave in an elastic half-space is slower than a shear wave and is non-dispersive. For a Poissons ratio of 0.25, the Rayleigh wave travels at a

650

S.J. Mbawala et al. / Numerical Modelling of Wave Propagation

velocity of approximately 0.9194 of the shear wave [17]. Using this relationship the theoretical Rayleigh wave velocity is 205.6 m/s. The Rayleigh wave velocity, VR = fλ from the finite element analysis ranges between 193 m/s and 205.9 m/s with an average of 203.25 m/s as shown in Figure 4. On average, the difference between phase velocities from the finite element analysis and theoretical value is about 1.1 %, which suggests that the modelling agrees with the theory. A comparison was made between field measured phase velocities using the seismic method of continuous surface waves [5] and the finite element method. The dispersion curves that were obtained using field measured data, finite element method and theoretical values are presented in Figure 4. 10 0

Phase angle (rad)

-10 -20 -30 -40 F10 F20 F30 F40 F50 F60 F70 F80 F90 F95

-50 -60 -70 -80 0

2

4

6

8

10

12

14

16

18

20

22

24

Distance (m)

Figure 3. Determination of Phase Velocities between Observation Points

Comparison between theoretical dispersion curve and dispersion curve obtained by the finite element method shows that the phase velocities from the finite element model are slightly lower compared to the theoretical model at low frequencies. The dispersion curve obtained from field measured data is somewhat scattered, but shows some dispersion as the wave velocity increases at lower frequencies. The average phase velocity measured in the field is 212.4 m/s, which is higher than the average phase velocity from the finite element analysis by about 4.3 %. This marginal difference between field results and the results from the finite element indicate that the model gives realistic results. If the ground medium is excited with very low frequency, it will produce very long wavelengths while at higher frequencies the wavelengths become shorter [16]. This phenomenon was further used to validate the finite element model by comparing with wavelengths obtained from the measured field data and theory with different frequency content. The plot of wavelengths versus frequencies for both cases is shown in Figure 5. In both cases as the frequencies increases the wavelength decreases. Therefore, the finite element model depicts the same behaviour as suggested by the theory. The finite element method predicted response that is closer to the theoretical response for nondispersion profile.

651

S.J. Mbawala et al. / Numerical Modelling of Wave Propagation

5.

Conclusion

In this paper, an elastic isotropic half-space is simulated with the finite element method to model surface wave propagation. The phase velocity travelling in the soil medium obtained numerically on average is lower by about 1.1 % compared to the values obtained theoretically. This is within an acceptable range. The average Rayleigh wave velocity from continuous surface wave test measured in the field is higher by about 4.3 % compared to the results obtained numerically, which indicate that Abaqus/Explicit can reliably model the wave propagating in ground using finite element method. 300

250

Phase velocity (m/s)

200

150

100 FEM - Abaqus/Explicit Theoretical Measured field data - CSW

50

0 0

10

20

30

40

50

60

70

80

90

100

Frequency (Hz)

Figure 4. Dispersion Curves for Measured Field Data, Theoretical and Finite Element Method

25

Wavelength (m)

20

FEM - Abaqus/Explicit Field Measured Data - CSW Theoretical

15

10

5

0

0

10

20

30

40

50

60

70

80

Frequency (Hz)

Figure 5. Variation of wavelength with frequency

90

100

652

S.J. Mbawala et al. / Numerical Modelling of Wave Propagation

Acknowledgements The authors wish to acknowledge the financial contribution made towards this project by Anglo Technical, a division of Anglo American plc as well as THRIP (Grant no. 71875).

References [1] G.R. Liu and S.S. Jerry, A non-reflecting boundary for analyzing wave propagation using the finite element method. Finite Elements in Analysis and Design 39 (2003) 403–417. [2] G.R.Liu and J.D. Achenbach, A strip element method for stress analysis of anisotropic linearly elastic solids, Journal of Applied Mechanics 61(1994) 270–277. [3] A.H.Nielsen. Absorbing boundary conditions for seismic analysis in Abaqus, (2006) ABAQUS Users` conference. [4] Abaqus/Explicit User’s Manual, Vol. I-II, Version 9.10, 2010, Hibbitt, Karlsson &Sorensen, Inc [5] G.Heymann, Ground stiffness measurement by the continuous surface wave test, Journal of the South African institute of Civil Engineering V 49 No.1( 2007) 25 -31. [6] J.Lysmer and R.L, Kuhlemeyer, Finite dynamic model for infinite media. Journal of the Engineering Mechanics Division of the ASCE, American Society of Civil Engineers 95(EM4) (1969) 859–877 [7] Abaqus Theory Manual ..Version 6.9.10, Simulia Dassault Systemes. [8] R.Motamed, K.Itoh, S.Hirose, A.Takahashi and K.Osamu, Evaluation of wave barriers on ground vibration reduction through numerical modelling in Abaqus. 2009 SIMULIA Customer Conference. [9] A.Zerwer, G. Cascante and J.Hutchinson, Parameter estimation in finite element simulations for Rayleigh waves. Journal of Geotechnical and Geoenvironmental Engineering, American Society of Civil Engineers 128 (3)(2002) 250-261. [10] A.Zerwer, M.A.Polak, J.C.Santamarina, Rayleigh wave propagation for the detection of near surface discontinuities: Finite element Modelling, Journal of Non-destructive Evaluation, Vol,22,No.2 (2003) 31 -51. [11] G.Inci, Numerical modelling of wave propagation in elastic-half space with Imperfections 21st SAGEEP, Symposium on application of Geophysics to Engineering and Environmental Problems, New partnerships, New discoveries. Marriott Philadelphia Downtown Pennsylvania, April 6 – 10 (2008) ,www.x-cdtech.com/SAGEEP08/pdfs/190.pdf . [12] L.Hall and A.Bodare, Analyses of the cross-hole method for determining shear wave velocities and damping ratios, Soil Dynamics and Earthquake Engineering 20 (2000) 167-175. [13] R.D. Cook, D.S.Malkus, M.E, Plesha and R.J. Witt, Concept and Application of Finite Element Analysis, Fourth Edition. John Wiley &sons ,USA, 2002. [14] H.J.Alheid, Seismic response of deep underground openings. Soil interaction: Numerical analysis and modeling, Edited by John W. Bull. E & FN SPON, 1994. [15] Abaqus/CAE(Complete Abaqus Environment). The Abaqus Interactive FE modeling system. Simulia, Dassault Systemes. [16] R.Jones, In-Situ measurement of the dynamic properties of soil by vibration methods. Geotechnique. Vol.8.No.1 March (1958)1-21. [17] X.Ianghai, D.M.Richard, C.B.Park and J.Ivanov, Utilization of high-frequency Rayleigh waves in nearsurface geophysics. Near-surface problems and solutions SEGs 2002 Annual meeting.

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 653 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-653

Geotechnical characteristics of the red sands of Chibuto, Mozambique a

a

HAC MEINTJES and GA JONES SRK Consulting,Johannesburg, South Africa

Abstract. Red sands occur along the east coast of southern Africa from central Tanzania, through Mozambique to southern Natal and Eastern Province, South Africa. In a number of places the sands include heavy minerals which form economically viable mining deposits. One such deposit is located close to Chibuto, Mozambique and an extensive investigation was undertaken for the mineral processing plants. The geology of the area was known to be complex and the investigation which comprised primarily about 400 penetration tests (CPT, SPT and DPSH) confirmed that the red sands are extremely variable. A statistical method was adopted to define the variability so that a rational approach to foundation design could be used. The different penetration test results were correlated and confirmed that the conventional correlation factors were valid. Plate load tests, before and after, dynamic compaction proved the efficacy of the compaction but these tests when correlated with CPTs showed that moduli derived from the latter were conservative by a factor of two. Pile tests showed that the pile loads for driven cast in situ piles estimated from CPTs were also conservative by a factor of two. The overall conclusion is that the red sands are more complex than may be assumed and that although their foundation performance may be better than the general correlations with penetration testing indicate, the extreme variability should lead to considerable caution when assessing investigation results. Keywords. Collapse potential, cone penetration testing, plate load tests, pile tests, foundation investigation and design

Introduction Heavy minerals, including ilmenite, rutile, zircon and titanium, are found in the red dune sands along the east coast of Mozambique and Natal and are often exposed as a black fine sand on the beaches. The minerals are valuable for various industrial processes and the prize is to locate a deposit where the concentration is sufficient to be economically viable, as occurs for example at Richards Bay in Natal. Such a deposit was found by Southern Mining near Chibuto in Mozambique which is about 200 km north-east of Maputo and 50 km inland from Xai-Xai. The plant and smelter will comprise conveyors and stockpiles, thickeners, crushing and screening plant, furnaces, slag processing plant and offices stores and workshops together with roads and other infrastructures. The geotechnical investigation showed that the red sands are extremely variable in the characteristics represented by penetration testing.

654

H.A.C. Meintjes and G.A. Jones / Geotechnical Characteristics of the Red Sands of Chibuto

1. Geology and geotechnical characteristics The red sands of the eastern coast of Southern Africa, generally called Berea Red in Natal, are intriguing in that far from being recent dunes they have a much more complex history. Tracing this reveals the dramatic influence of climate change on the geology of the east coast in relatively recent geological times. The sand dunes in which the ore deposit occurs form a prominent elevated area which has a relief of more than 100m above the Limpopo and Changane river flood plains. The origin of the dunes and heavy minerals lies in the ancient river system, which formerly included the Zambezi, Sashi and Kafue rivers. Continental uplift resulted in coastal rivers cutting back inland and capturing the Zambezi, causing it to flow eastwards and enter the Indian Ocean north of Beira over a 1000 km north of its former mouth at Xai-Xai, where only the Limpopo now flows into the sea. The substantial width of the present river flood plain near Chibuto (about 5km wide) and at Xai-Xai (8 km wide) is an indication of the flow requirements of the Limpopo river system. The age of the mineral sand deposits is about 5 million years (Pliocene). The ancient river system flowed over the Swazian Greenstone belts, which provided the source of the heavy minerals. The present Limpopo course would not have provided such a source. The heavy minerals were concentrated in the dunes, which formed on the coast, as a result of the dual winnowing action of longshore currents and sea winds. Although the dunes extend from the coast about 50 km inland, the heavy mineral concentration occurred at a time when the coastline and the ancient river mouth coincided in the Chibuto area. The sea has gradually receded leaving the deposit at its present inland location. The Red Sands vary from a dark red and red brown, through brown orange and light yellow to almost an off white colour. Their composition is also variable from slightly clayey silty sand to silty fine sand with the darker colours generally more clayey, whereas the lighter colours are non-plastic. The grading moduli are generally approximately 1.0 and as a result the full range of soils compact well and give high CBR’s. Typical profiles are summarized in Table 1. Table 1. Penetration test profiles Depth m 0 -5 5 – 10 10 – 15 15 – 20 20 – 25 25 +

CPT MPa 1–5 5–7 7 – 15 10 – 20 10 -25 25 +

SPT N 1 -10 10 -15 15 – 30 30 – 40 20 – 50 50

Description Very loose Loose Medium dense Medium dense to dense Medium dense to very dense Very dense

2. Site investigation The investigation had two primary objectives: • to define characteristics of upper material for earthworks and shallow foundations. • to establish depths to dense material for piling.

H.A.C. Meintjes and G.A. Jones / Geotechnical Characteristics of the Red Sands of Chibuto

655

In addition to the broad scale investigation over the site, more detailed investigation was carried out in Area A at the smelter site and Area B at the plant site. The dominant method of the investigation was penetration testing of various types and boreholes. This was supplemented by trial dynamic compaction (DC) controlled inter alia by plate load tests and pile tests for both driven cast in situ (DCI) and continuous flight auger (CFA) piles. The tests are listed in Table 2. Table 2. : Schedule of in situ testing Test

Initial Plant

BH CPT DPSH Pile Plate

14 12 9

2nd Phase Plant 16 23 138

Area A Smelter 1 84 25 3 8

Area B Smelter 1 48 11 3 8

Total 32 167 188 6 16

2.1. Cone Penetration Tests It was clear from the early variable CPT results that the site fell into a complex geology and geotechnical category and the site was therefore considered in three areas: • Overall site – CPT’s at large spacings • Test areas A and B with CPT’s at about 10m spacing For each area, the individual CPT cone resistances have been averaged for each depth increment and with maximum and minimum values. These have been plotted and in addition lines of two standard deviations above and below the average are shown. The results for all areas are given in Figures 1, 2 and 3. 0 5

Depth (m)

10 15 20 25 30 35 0

10

20

30

40

50

CPT cone resistance (MPa) average

min

max

ave-2xstd

Figure 1. Basic statistical assessment of all CPT results

ave+2xstd

656

H.A.C. Meintjes and G.A. Jones / Geotechnical Characteristics of the Red Sands of Chibuto 100% 90% 80% 70%

Frequency

60% 50%

20% strength 40% 30%

10% strength variation

20% 10% 0% 0

2.5

5

7.5

10

12.5

15

17.5

20

22.5

25

27.5

30

More

Cone resistance (MPa) 0-2.5m

2.5-5m

5-7.5m

7.5-10m

15-17.5m

17.5-20m

20-22.5m

22.5-25

10-12.5m

Figure 2. Cumulative distributions of all CPT end resistance with depth 0

5

Depth (m)

10

15

20

25

30

35 0.0

0.1

0.2

0.3

0.4

0.5

0.6

Std Div / Average Cone resistance)

Figure 3. Ratio of Std Deviation to Average cone resistance with depth

For Area A, the ratio of the standard deviation to the average cone resistance typically varies between 0.2 to 0.4, which means that the actual variation could be as much as 40% to 80% of the average. In civil engineering practice the variability is usually assumed to be 30% for cohesion and 10% for friction angle. Harr [1] so the results indicate that the variability could be double that normally expected in practice.

H.A.C. Meintjes and G.A. Jones / Geotechnical Characteristics of the Red Sands of Chibuto

657

2.2. Comparison of CPT, SPT AND DPSH Tests Table 2 shows that a total of 32 boreholes with SPT’s, 167 CPT’s and 188 DPSH’s were carried out. Many of these were in locations close to one another so direct comparisons can be made of the relative values of the different penetration tests. These correlations can then be compared with those given in the literature. Figure 4 summarizes these results and shows that for this site: • SPT, N = 2/3 DPSH, N (N in blows / 300mm) (1) (2) • CPT, qc = 0.4 SPT, N (qc in MPa) The former agrees well with the correlation generally used in South Africa and the latter is a little different from the 0.5 factor often used for sands. Frankipile [2]. This factor is material dependent and for silty sands 0.4 as measured at this site is appropriate. The striking feature of Figure 1 is that many results are outside the two standard deviation lines and that the two standard deviation lines are so far from the average line. Figure 2 shows data for the whole site plotted as cumulative distributions of CPT values for 2.5m depth ranges. This figure also shows for the 20 MPa CPT the 10% and 20% strength variation lines for that depth for a model soil. The slopes of these lines can be visually compared with those for the actual data and it is clear that the actual slopes are very much flatter and increasingly so as the depth range increases. For example at the 10 – 12.5m depth range at a cone resistance of 15 MPa, a typical founding depth and CPT value for a DCI pile, the slope of the distribution is equivalent to about a 30% strength variation. The other most obvious factor is that the data lines are not evenly spaced as they would be if there was linear increase in strength with depth. The close spacing reflects a low increase in strength with depth, whereas a wide spacing reflects a relatively large increase in strength with depth. 2.3. Dynamic Compaction Eight plate load tests were carried out, four before and four after dynamic compaction, in Areas A and B to measure the increase in stiffness that was achieved, at depths of 2m and 4m. Table 3. Table 3: Plate bearing tests before and after dynamic compaction Area SPLA SPLA SPLA SPLA SPLB SPLB SPLB SPLB

Test No. / Depth 1/2 2/2 3/4 4/4 1 /2 2/2 3/4 4/4 Average

Stiffness MPa Before DC After DC 10 125 6 30 20 110 10 90 8 120 8 50 10 130 10 130 10 100

Cone Penetration Tests were also carried out before, during, and after dynamic compaction. Table 4 shows cone pressures at the positions of the plate load tests and at the same depths. A range of cone pressures is shown for each position.

658

H.A.C. Meintjes and G.A. Jones / Geotechnical Characteristics of the Red Sands of Chibuto 0

5

10

Depth (m)

15

20

25

30

35 0

10

20

30

40

50

60

70

80

9

SPT (N) SPT Mid

SPT Min

SPT Max

DPSH min

DPSH Average

CPT min

CPT max

CPT average

DPSH Max

Figure 4: Comparison of CPT, SPT and DPSH results Table 4: Cone pressures before and after dynamic compaction Cone Pressure MPa Before DC After DC 1.7 – 3.4 7.5 – 14 1.7 – 1.7 12.2 – 12.2 1.7 – 5 7.5 – 18.8 3.4 – 5 11.9 – 13.5 3 12

CPT No. SPCA 17 and 43 SPCA 18 SPCA 18 and 41 SPCA 40 Average

The CPT derived stiffness is given by: (3) E = a m qc Where E = stiffness, am = constrained modulus coefficient and qc = cone pressure. am is dependent on the material type which for this material is estimated to be 3.0 If the after final DC CPT results are expressed as stiffness the values are as shown in Table 5, and these are compared with the equivalent plate bearing E values. Table 5: Dynamic compaction comparison of stiffnesses from CPTs and plate bearing tests CPT qc CPT E Plate E

14 42 125

MPa 12.2 38 30

Average 18.8 54 110

13.5 41 90

44 88

It can be seen that the correlation of CPT cone pressure with stiffness in this case gives projected stiffness values of only half those measured by the plate loading tests.

H.A.C. Meintjes and G.A. Jones / Geotechnical Characteristics of the Red Sands of Chibuto

659

The overall increase in stiffness for the plate bearing test is a factor of ten (Table 3) whereas the increase in cone pressure is only four times (Table 5). Even using the more conservative data, i.e. CPT, the four times improvement achieved by the dynamic compaction is very large and changes the material from being unsuitable for almost all foundations, terraces and roads, to be suitable albeit with localised near surface compaction depending on the purpose. 2.4. Pile Tests Experience has shown that for large depth sand deposits the most suitable types of pile are driven cast in situ (DCI) and continuous flight auger (CFA) hence both types were tested to determine the more suitable for the site. The pile testing included before and after wetting. In all respects the testing closely followed the UK CIRIA Pile Load Testing Procedures and SABS 1200 F. No standard exists for the wetting up procedure; four percussion holes were drilled at each pile and simply filled with water until a significant increase in moisture content was believed to have occurred. Testing was carried out to assess the suitability of both driven cast insitu (DCI) and continuous flight auger (CFA) piles. The latter showed very much higher settlements than the DCI piles despite being founded at greater depths. This is not surprising because at this site the upper 10m or so is relatively loose and negligible contribution is made to the shaft friction in this zone. This does not entirely discount the use of CFA piles at the site because the DCI piles have a construction depth limit of about 20m: it is possible for very high loads, or where group effects become significant, that CFA piles could be the optimum solution. The pile testing was intended to test the lower CPT cone resistance zones at typical piling depths and hence the pile depth was selected on the basis of the CPT carried out at that precise location. 2.4.1. Driven Cast Insitu - DCI The test results for the 520mm diameter driven cast insitu piles are given in Table 6. This pile has a nominal working load of 1200 kN. Table 6: Summary of DCI pile load tests Pile No. SPCA 1 SPCA 3 SPCB 5 SPCB 6

Depth m 11 8 12 8

1200 2.0 3.0 2.5 3.7

Dry displacements mm at load kN 1800 2400 3.9 5.9 6.0 11.0 4.4 8.9

3000 8.4 41

Conventional pile design using CPT values along the shaft and above and below the base are given in Table 7 below which shows the CPT values and the pile design assuming a factor of safety of 2.5. qc and qb are the allowable pressure and Qs and Qb the allowable loads for the 520mm diameter pile.

660

H.A.C. Meintjes and G.A. Jones / Geotechnical Characteristics of the Red Sands of Chibuto

Table 7: Summary of design loads for DCI piles from base and shaft CPTs. CPT No. SPCA 1 SPCA 3 SPCB 5 SPCB 6

qc shaft MPa 3.0 2.8 4.0 3.0

qb base MPa 8.0 6.2 9.5 5.0

Qs Shaft 0.44 0.36 0.65 0.29

Qb base

Q Total

2.04 1.58 2.42 1.27

2.48 1.94 3.07 1.56

Design kN 1000 800 1200 600

The displacements at the above design loads are measured from the load-displacement diagrams and this gives values of 1.3; 1.6; 2.4 and 1.0mm respectively. It can be seen from Table 6 that at the nominal loading of 1200 kN all the piles, although founded in selected poorer zones, give satisfactory performance with an average displacement of only about 2.5mm at the design load. At 50% overload the piles all also behaved satisfactorily. From the design Table 7, together with the displacements in Table 6, correlation can be made between the calculated design loads and the measured displacements at these loads and the latter can be seen to be very low i.e. less than 2mm. Figure 5 shows the load displacement data for the tests. Figure 6 shows the data from Table 7, i.e. the conventional design loads, and also the pile test data for 3mm and 5mm displacement. The design model line shows that the nominal capacity of the 520mm diameter DCI pile of 1200 kN can be achieved provided the cone resistance at founding level is not less than 10 MPa. The pile tests, however, also shown in Figure 6 can be read as showing that this is readily attainable and that settlements of less than 3mm will occur under full load or alternatively that the 1200 kN load can be achieved with a base cone pressure at only 5 MPa. At an acceptable pile settlement criterion of 1% of the pile diameter i.e. 5mm, the pile test derived loads are practically double the conventional design loads. 2.4.2. Wet Pile Tests The piles were wetted as described previously and tested in the same way as the dry testing. Table 8 shows the wet pile performance for the DCI piles which can be directly compared to the dry test results. Table 8: Summary of wetted up DCI Pile Tests Pile No. SPCA 1 SPCA 3 SPCB 5 SPCB 6

Depth (m) 11 8 12 8

1200 2.2 3.0 4.0 42

Wet Displacements mm at load kN 1500 1800 2.7 3.6 4.0 7.5 5.0 10.5

2200 4.2 12.0 21.0

If the dry pile and wet pile performances are compared for the DCI piles (Table 9) it will be seen that the average value of displacements is about 50% higher at the higher loads but not much different at the design loads. The design load for SPCB6 is only 600 kN so only comparison in this range is valid.

661

H.A.C. Meintjes and G.A. Jones / Geotechnical Characteristics of the Red Sands of Chibuto ϯϬϬϬ

ϮϯϬϬ ϮϭϬϬ

ϮϱϬϬ

ϭϵϬϬ ϭϳϬϬ

>ŽĂĚ;ŬEͿ

>ŽĂĚ;ŬEͿ

ϮϬϬϬ

ϭϱϬϬ

ϭϱϬϬ ϭϯϬϬ ϭϭϬϬ

ϭϬϬϬ

ϵϬϬ ϱϬϬ

ϳϬϬ ϱϬϬ

Ϭ Ϭ

Ϯ

ϰ

ϲ

ϴ

ϭϬ

ϭϮ

Ϭ

ϭϰ

Ϯ

ϰ

ϲ

ŝƐƉůĂĐĞŵĞŶƚ;ŵŵͿ ^Wϭ

^Wϯ

^Wϱ

^Wϲ

Figure 5: Load-displacements for DCI dry pile tests

ϴ

ϭϬ

ϭϮ

ϭϰ

ƋĐďĂƐĞ DŽĚĞů

ϯŵŵĚŝƐƉůĂĐĞŵĞŶƚ

ϱŵŵĚŝƐƉůĂĐĞŵĞŶƚ

>ŝŶĞĂƌ;DŽĚĞůͿ

>ŝŶĞĂƌ;ϯŵŵĚŝƐƉůĂĐĞŵĞŶƚͿ

>ŝŶĞĂƌ;ϱŵŵĚŝƐƉůĂĐĞŵĞŶƚͿ

Figure 6: Theoretical design model load based on pressure and pile test results for 3mm displacement

Table 9: Wet and Dry Pile Test Displacement Comparison Pile No. SPCA 1 SPCB 3 SPCB 5 SPCB 6

Pile Type

Ratio Wet Displacements mm at load kN 1200 1500 1800 2200 1.1 1.1 1.1 1.4 1.0 0.7 1.2 1.5 1.6 6.4 11.3

800 1.2 1.0

DCI DCI DCI DCI

Note also that if a settlement criterion of 10mm is applied then the piles are satisfactory at their design load even after wetting. The question arises of how effective and representative of real possible conditions was the artificial wetting up process? In order to evaluate this further CPT’s were carried out immediately adjacent to the wetted up piles for comparison with the CPT’s carried out before pile installation. Table 10 shows the comparison with the CPT values at nominal piling depth for the DCI piles. Table 10: Influence of Wetting up on Pile Capacity Pile No.

Type

Depth (m)

SPCA1 SPCA3 SPCB5 SPCB6

DCI DCI DCI DCI

11 8 12 8

Influence of wetting on CPT Pile Capacity None (average) None None (average) None +30% None -40% -50%

662

H.A.C. Meintjes and G.A. Jones / Geotechnical Characteristics of the Red Sands of Chibuto

It would be difficult to draw definitive conclusions from the above table. Solely on the basis of SPCB6 it would appear that wetting up occurred, i.e. the CPT values are reduced, and there is a reduction in pile capacity to half the dry value. Soil suction results show that there is only one soil sample with significant soil suction of 207kPa, BH1 at a depth of 6m to 6.5m. When these soil moisture test results are compared to those measured at Mozal by Rust et al [3]it can be seen that the soil suctions at Mozal were typically higher and consequently the wetting up of the pile tests was more significant at the Mozal site as the change in suction with wetting was greater than at the Chibuto site. 3. Summary An extensive geotechnical investigation was carried out for the proposed plant site area. It comprised CPTs, SPTs in boreholes, DPSHs, plate bearing tests, pile tests both before and after dynamic compaction and before and after wetting. These and the associated laboratory tests showed the in situ material comprised silty fine red sands typical of the east coast of Mozambique and Natal and called Berea Red at the latter. Two valuable findings can be made from the investigation and test results, namely on the variability of the material and on the stiffness measurements and hence pile design. The extensive penetration testing provided data so that the material variability could be statistically characterized. A striking feature was that many cone resistance values fell outside the two standard deviation envelopes which greatly exceed the usual evaluation of soil variability. Furthermore the standard deviation values as a ratio of the average cone penetration is higher than expected. The variability can be expressed as cumulative distributions of CPT values with depth (Figure 2) and this proved invaluable in assigning the required depth to achieve a CPT value at any probability level. In other words despite the extremely high variability a rational design approach could be adopted. The pile tests for the dry condition showed much better results than would have been expected from the conventional design equation based on Cone Penetration Tests. From this and from a comparison of plate bearing tests with CPT’s it is suggested that the interpretation of CPT’s in the Red Sands should be re-evaluated. It would appear that the stiffness of these materials is underestimated by comparison with normally consolidated materials and should in fact be approximately doubled. This is, if true, of major significance in foundation design and research should be undertaken to confirm this. References [1] M.E. Harr,

Mechanics of Particulate Media – a Probabilistic Approach. McGraw Hill, 1997. [2] Frankipile, A Guide to Practical Geotechnical Engineering in Southern Africa. Table 20.3.4 p258. 1997. [3] E. Rust, G Heymann and G A Jones, Collapse Potential of Partly Saturated Sandy Soils from the East Coast of Southern Africa. J. SAICE, Vol 47, No.1, pp 8 – 14, 2005.

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 663 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-663

Simple Expansion Model Applied to Soils from Three Sites A. Dereck W. SPARKS University of Cape Town, Rondebosch, South Africa Abstract. Practical experiments by Pidgeon ( 1987 ) involved wetting similar samples which carry different vertical loadings. A formula by Sparks predicts the “free swell when carrying 1 kPa vertical loading”. A formula by Pidgeon is then used to predict the swell for other loadings. Keywords. Expansive Clays, Swell, Swelling Pressure, Thornthwaite Index

Introduction Methods for predicting the expansion of clays due to wetting, often contain the complicated use of suction pressures. These suction pressures are converted to equivalent average effective stresses, which are deemed to be the values which determine the total volume of the clay specimen. The present author has also previously used such methods. In this paper, a different approach is used. Excellent agreement is shown in Tables 1, 2, and 3 between the predicted and the measured values of swell at each of the different sites Kilner Park (at Pretoria), Kwanagxaki (at Port Elizabeth) and Leeuhof (at Vereeniging). 1. Different Values for “Swell Pressures” Swell pressures can be defined and measured by at least five different methods. The “swell pressure” used by Pidgeon (1987) can be defined as the vertical loading which will prevent the soaked clay from swelling ( Figure 1, and Figure 3 in ref.[1] ). This is also shown as point B in Figure 1 of this paper. Note that the pore water suction u is equal to zero along the line AB in Figure 1. Some methods, such as the filter paper method provide suction values which are converted to pre-wetting swell pressures which are as high as 2500 kPa ; whereas other methods provide values which are approximately one third of these high values. Softening of the clay takes place when it is soaked. Note that the “swell pressures” used by Pidgeon are “post-wetting” values. His swell pressure values will be lower than the “pre-wetting” values found by other methods. The Pidgeon Method has a further simplification. Only the values of the vertical pressures are used (even though it is known that average effective pressures determine the total volume of the clay ).

664

A.D.W. Sparks / Simple Expansion Model Applied to Soils from Three Sites

2. The Clay Softens as it Swells A clay softens when it is wetted, even while it is swelling. The author suggests that this diminishes the value of the swell pressure from its pre-wetting value, and hence only a proportion of the original swell pressure acts as a driving mechanism causing the clay to swell. The author is impressed by the simplicity and the usefulness of the “swell pressure” as used by Pidgeon ( See Figure 1 ).

Swell % “Free Swell %” carrying 1 kPa

A

e.g. Start three samples at same eo and wo% . “Swell pressure” p’s Wetting

B

0 1 kPa

15

30 kPa

112 kPa Log p’v

Figure 1. Pidgeon’s Method ( 1987 ) for finding Swell Pressures.

3. The Swell Pressure p’s due to wetting and Maximum Effective Depth of Heave Data points observed by Pidgeon for the Port Elizabeth site are listed in the last column of Table 2. These have been plotted in Figure 2 to find the total heave of a clay deposit. Note that in this case, the observed swell pressure p’s ( point B in Figure 1) is equal to 112 kPa and this is also equal to the actual vertical pressure at the point C in Figure 2. The present author decided that the reverse should also apply. In other words, if one has other evidence which suggests a maximum depth of expansive influence, then this depth can be used to estimate the overburden pressure at C in Figure 2 and this will be equal to the swell pressure at B in Figure 1. 4. Using Formulae to predict the Percentage Swell 4.1 Step No. 1 : Using equation (1). Enter Eq.(1) via the pre-wetting water content wo% and the Plasticity Index PI%. This equation provides the so-called “Free Swell “ - i.e. while carrying 1 kPa vertical loading (or use Fig. 3 ). This value also corresponds to the point A in Fig.1.

( FreeSwell%; i.e.carrying 1 kPa) =

( 10 + 0.47PI% − wo% ) .............................................(1) ⎛ 1 wo% ⎞ ⎜ + ⎟ ⎝ Gs 100 ⎠

A.D.W. Sparks / Simple Expansion Model Applied to Soils from Three Sites

665

where Gs = 2,7 (approx), PI%=PI% of whole sample, and wo% = Pre-wetting wc. 4.2 Step No. 2 : Decide on the probable maximum depth of the expansive zone, and the value of p’s . The equivalence between the vertical pressure at point C in Figure 2 and the swell pressure at point B in Figure 1 has been discussed in Section 3 above. Engineers at the CSIR plotted diagrams (ref [3]) similar to Figure 2 based on measurements at Vereeniging and Welkom, and they found that the maximum depths of the expansive zones ( to point C in Fig.2 ) were approximately 6 metres. The swell pressures of 112 kPa (point B in Fig.1 ) for clays from Pretoria and Port Elizabeth were measured by Pidgeon ( Table 1 and Table 2 ). i.e. this value of “swell pressure” is equal to the vertical surcharge at a depth of 5.5 metres. The following values can be used for the “swell pressure” p’s :Highveld ( Transvaal and Orange Free State ) p’s = 110 kPa ; Cape Town p’s = 30 kPa, Durban p’s = 20 kPa ; Port Elizabeth p’s = 110 kPa ; Sydney (Australia) p’s = 30 kPa ( ref [4] ) Melbourne (Australia) p’s = 35 kPa to 50 kPa . The Thornthwaite Climatic Index can serve as a guide. This Index has similar values for the regions of Pretoria and Port Elizabeth.

0

0,075

0,1125

% Swell / 100

0 0,75 m 1,5 m

p’v = 15 kPa p’v = 30 kPa

Plotting three of the values measured by Pidgeon for Port Elizabeth. From second column in Table 2. Assuming constant PI%= 37% and wo% = 14.1% for whole profile.

Depth metres

5,5 m

C

p’v = 112 kPa

Area of diagram = 0,29 m = total heave of surface due to soaking.

Figure 2. How to use Pidgeon’s data (expansive strain versus depth ) to find the total heave

666

A.D.W. Sparks / Simple Expansion Model Applied to Soils from Three Sites

4.3 Step No. 3 : Converting value from Eq.1 to Swell% for any vertical pressure p’v

“Free-Swell” = Percent Swell while carrying 1 kPa

⎛ Swell% while ⎞ ⎛ Log p' s − Log p'v ⎞ ⎜ ⎟ =⎜ ⎟ Log p' s ⎝carrying any p' v ⎠ ⎝ ⎠

⎛ Free Swell % ⎞ x⎜ ⎟.................................................(2) ⎝ while carrying 1kPa⎠

⎛ Free Swell% ⎞ ( 10 +0, 47PI% −wo% ) ⎜ ⎟= ⎝carrying 1kPa ⎠ ( 1/Gs)+ (wo%/100 )

40 %

PI% 40%

30 %

PI% 30%

20 %

PI% 20% 10 %

45% PI% 50% 35% 25%

Sparks 2011 ref [2]

0

0

10% 20% Pre-wetting Water Content wo %

30%

Figure 3. “Free-Swell carrying 1 kPa” versus “Pre-wetting Water Content wo%”

5. Actual Results of Predictive Calculations

The Predictive Method is based on the coming together of two formula. The first formula is Eq.(1) by Sparks which uses the pre-wetted water content to estimate the “Percentage Free Swell if the clay is carrying only 1 kPa vertical loading”; and the second formula is Eq.(2) by Pidgeon which alters the value from Eq.(1) to provide the Percentage Swell of the clay if it carries any vertical loading p’v. Table 1 displays the values for the clay ( PI%=36%,wo%=21% ) from Kilner Park (Pretoria). The measured “Free Swell” was 9,2% while the predicted value from Eq.(1) was 10,2%. Eq.(2) was then used to predict Swell % for six different vertical loadings varying from 1 kPa to 110 kPa. The six predicted values agree well with the six observed values as shown in Table 1. Table 2 displays the values for the clay ( PI%=37%,wo%=14,1% ) from Kwamagxaki (Port Elizabeth). The measured “Free Swell” was 26,2 % while the predicted value from Eq.(1) was also 26,2 %. This is a remarkable agreement. The predicted values for Swell% for the five different vertical loadings ( 1 kPa to 110 kPa) agree extremely well with the five observed values as shown in Table 2.

A.D.W. Sparks / Simple Expansion Model Applied to Soils from Three Sites

Table 1

Calculations and Observations. Kilner Park.

Predicted

Measured

Assume influence zone to 5,5 metres Swell Press = overburden = 5,5 x 20 p’s = 110 kPa (Free Swell 1 kPa ) Eq.(1) or Fig.3 = 10,2 % Using Equation 2 Overburden p’v Swell % p’v = 1 kPa ...................10,2 % p’v = 5 kPa ................... 6,7 % p’v = 15 kPa ................... 4,3 % p’v = 30 kPa ................... 2,81 % p’v = 60 kPa ................... 1,31 % p’v = 110 kPa .................. 0 %

Table 2

Measured Swell Press p’s = 112 kPa Measured “Free Swell” = 9,2 % Measured Swell % 9,2 % 5,7 % to 6,3 % 4,0 % to 4,5 % 2,79 % 1,3 % 0 % at p’v =112 kPa

Calculations and Observations. Kwamagxaki.

Predicted Thornthwaite Climate same as Pretoria. Assume influence zone to 5,5 metres Swell Press = overburden = 5,5 x 20 p’s = 110 kPa (Free Swell 1 kPa ) Eq.(1) or Fig.3 = 26,2 % Using Equation 2 Overburden p’v Swell % p’v = 1 kPa ...................26,2 % p’v = 5 kPa ...................17,23 % p’v = 15 kPa ...................11,11 % p’v = 30 kPa ................... 7,24 % p’v = 110 kPa .................. 0 %

Measured Measured Swell Press p’s = 112 kPa Measured “Free Swell” = 26,2 % Measured Swell % 26,2 % 17,3 % 11,25 % 7,5 % 0 % at p’v =112 kPa

* Note :- Free Swell in Table 2 is approx 2,5 times the Free Swell in Table 1.

667

668

A.D.W. Sparks / Simple Expansion Model Applied to Soils from Three Sites

Table 3

Predictions for Leeuhof ( Vereeniging )

Assume influence zone to 5,5 metres. Swell Press = overburden press. = 5,5 x 20 = p’s = 110 kPa At centre of layer :Layer wo% PI% “% Free Thickness Swell”

p’v kPa

% Swell Eq,2

1,83 m

17%

20% 4,61%

18,9

1,78

3,26 cm

1m 1m 1m

17% 19% 24%

34% 29% 37%

16,62% 48,9 8,26% 69,9 5,55% 90,9

3,11 0,93 0,058

3,11 cm 0,93 cm 0,058 cm

1m

25%

31%

-0,9%

111,9 -0.01

Swell = Thickness (m) x % Swell ....(cm)

negative

Total predicted heave = 7,35 cm Measured ( approx ) = 7,9 cm 6. General Method for a Layered Profile Table 3 illustrates the general method for use when the profile consists of several different layers all with different values of PI% and initial water contents. The value of the “Swell Pressure” is regarded as a Climatic Parameter and is the same for all the layers in Table 3. The Percentage Swell for each layer is multiplied by the thickness of the layer to provide the actual swell (cm) caused by soaking the layer. These swells are added to predict the total swell caused by the soaking of the whole profile. The predicted total heave is close to the value of the measured total heave for this site. 7. Conclusion This paper provides excellent agreement between 12 predicted values and 12 observed values. It is suggested that the method in this paper applies if the Liquid Limit is less than 80 %. Further research is needed to investigate dry clays which may have measured “swell pressures” higher than 250 kPa. References [1] J.T.Pidgeon ( Sept 1987 ) “The prediction of differential heave for design of foundations in expansive soil areas”. 9th Reg. Conf, for Africa on Soil Mechanics & Foundation Eng. Lagos. [2] A.D.W.Sparks, and J.T.Pidgeon (April 2011), “Simplifying Expansion of Clays”, Civil Engineering, publication of S.A.Inst. of Civil Eng. [3] J.E.Jennings and J.E.Kerrich ( Nov 1962 ), “The heaving of buildings and the associated economic consequences, with particular reference to the Orange Free State Goldfields.” [4] M.A. Allman, M.D.Delaney, and D.W.Smith ,(Aug 1998 ) “ A field study of seasonal ground movement in Expansive Soils.”, Int. Conf. Unsaturated Soils, Beijing, Vol 1 , p 309 -314.

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 669 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-669

Correlation Studies between SPT and Pressuremeter Tests Emmanuel KENMOGNE1, Jean Remy MARTIN GEOFOR S.A, Douala, Cameroon

Abstract. This article presents correlation results between SPT, N resistance and pressuremeter parameters pl and E. The presentation is on 238 data in a sedimentary basin, selected from a study of about 600 available data. These data are from three different sites of the DOUALA sedimentary. The sites are constituted of clean sand, gravely clay with transitions of sandy gravel or clayey sand. A linear correlation is obtained from N and Pl-P0, and from E and N. Correlation coefficient varies between 0.93 and 0.94. Keywords . correlation, SPT resistance, net limit pressure, clayey sand, gravely clay, deformation modulus.

Introduction Correlations obtained from parameters of different geotechnical in situ tests give us a global behaviour of the soils under different stresses. For example the correlation between Cu and Pl; Pl and qc; Pl and qd. A summary of all these correlations of soil with respect to its nature is given by Cassan [1]. The Standard Penetration Test (SPT) by the ASTM corresponds to the dynamic penetration using core barrel of the French standard NFP 94.116. It is sometimes used for boreholes drilled prior to pressure meter tests. This gives two fundamental parameters to be simultaneously obtained that is; the N value which gives the SPT resistance of the soil as well as the pressure limit, Pl and the deformation modulus, E of the soil layers. Different authors have established several correlation types for different facies of unconsolidated soils. COVILLE-LAFEUILLADE [2], obtained correlations between SPT resistance, N and the pressuremeter parameters, Pl and E for sandy and clayey formations of Paris and Northern France. Given the large number of field data from the three study sites of the Douala sedimentary basin, Cameroon, our objective is to show the correlation between SPT resistance and the pressuremeter parameters, PL and E in the different littoral formations of Cameroon. All pressuremeter tests and SPTs in this case study were carried out in the same borehole.

1

Dr Emmanuel KENMOGNE, P.O.Box 1883, Douala. Email : [email protected]

670

E. Kenmogne and J.R. Martin / Correlation Studies Between SPT and Pressuremeter Tests

1. Principles and Equipment Used

1.1. SPT Equipment It comprises: - a 64 kg hammer with a free fall height of 75 cm, - a SPT split core barrel of 45 cm long, 51mm outer diameter and 35 mm interior diameter, - a tripod cable hoisting device, - the test is carried out based on the French standard, NFP 94. 116. The SPT resistance, N conventionally represents the number of hammer blows required for the penetration of the last 30 cm after the seating phase. The N values are registered in a continuous manner (after every 45 cm) and after 3 test runs of SPT. The core barrel is brought to the surface in order to lower the pressuremeter probe for testing.

SPT being carried out

An open SPTsplit spoon core barrel

1.2. Presuremeter Equipment It comprises: - a complete pressuremeter test device, APAGEO mark with a standard 44 mm probe, - the test is carried out based on the French standard, NFP 94: 101.2 and the original pressure / volume data is plotted on the field, - the X Pressio APAGEO soft ware makes is used to bring out all the different pressuremeter parameters of the soil under studies.

E. Kenmogne and J.R. Martin / Correlation Studies Between SPT and Pressuremeter Tests

671

Pressuremeter test using an APAGEO PRESSUREMETER

2. Original Data and Method Used The original data are results from divers drilling campaigns conducted in the Douala basin and its environs. A total of 610 pairs of SPT / Pressuremeter tests were carried out. However the data were screened based on the soil facies and correlation coefficients. Certain facies of the region (mud, soft clay) are not favourable for SPT / pressuremeter tests hence are of no interest in this correlation study. Retained data are only those with correlations greater than 0.86, however to obtain satisfactory data, we considered values ranging between 0.80 and 0.86 and a total of 238 data were retained. Pearson’s correlation coefficient is the quotient of the covariance by the product of standard deviation. Table 1: Geologic formations and selected SPT / Pressuremeter pairs Number of measurements Sites Type of formations Total Pressuremeter SPT Clean Sand (Right bank 120 60 60 of river Wouri) I

Clayey Sand

56

28

28

Sandy Gravel (Left bank of river Wouri)

72

36

36

II

Sand (Right bank of river Wouri)

96

48

48

III

Gravely Clay

30

15

15

2.1. Clean Sand (fine to average) 97 pairs of data were retained from the initial 108 pairs for N values and Pl – P0 parameters. The correlation coefficient between N and E is less than 0.86 and thus not considered in this study.

672

E. Kenmogne and J.R. Martin / Correlation Studies Between SPT and Pressuremeter Tests

2.2. Sandy Gravel or Gravely Sand 33 pairs were retained for N values and Pl – P0 parameters 30 pairs of data were retained for N and E 2.3. Clayey Sand or Sandy Clay 25 pairs were retained N values and Pl – P0 parameters 26 pairs of data were retained for N and E 2.4. Gravely Clay 15 pairs of data were retained; correlation coefficient is 0.94. Note that in this case study, all pressuremeter tests were conducted in the same SPT boreholes using naked standard probe.

3. Statistical Data Analysis Correlation studies between SPT and pressuremeter parameters have been a study of years back. Bibliographic studies on this subject showed that the N and E, Pl – P0 and N parameters are adjusted to Gauss’ normal distribution curve. The adequacy of this law can be verified in our correlation study. In order to simplify the use of correlation, they were obtained from a straight line equation passing through the origin. The correlations obtained are straight lines calculated on a number of points by the least square method considering the origin to be at zero. The main difficulty, as in all statistical studies, is to obtain a formula that will enable some measurements to be rejected though retaining a good number of pairs of (Pl – P0; N) and (E; N) for approximately straight line calculations with a relatively exact percentage of inaccuracy (about 30%). 3.1. Preliminary Phase First of all, pairs have to be formed (Pl – P0; N) and (E; N) for pressuremeter tests and SPT in boreholes of the same locality and of the same geologic formation irrespective of depth. These pairs are plotted on a graph with Pl – P0 or E on the x-coordinate and N on the y- coordinate (see figures 1,2 and 3 below).The stray points from the cluster of points are removed. The percentage of these rejected points does not exceed 10 % of the initial percentage. They are as a result of the deviated measurements or tests carried out on incomparable soils. 3.2. Phase 1 The adjusted straight line passing through N with respect to Pl – P0; and E with respect to N (and inversely) are calculated from the remaining points. The coefficients are

E. Kenmogne and J.R. Martin / Correlation Studies Between SPT and Pressuremeter Tests

673

determined by the least square method. Only points close to the range of 60% around the straight line are retained. 3.3. Phase 2 The adjusted straight lines passing through the origin are then calculated for the remaining pairs. They represent the retained correlations for the geologic formation under studies. 3.4. Final Phase The adjustment straight lines for the retained geologic formations were obtained from analysing a few of the retained pairs (N, Pl – P0) or (E, Pl – P0) per formation. For each of these formations, we attribute the correlations obtained, the simplified relationships as well as their range of use.

4. Correlation Results between SPT and Pressuremeter Test Parameters Histograms of Pl-Po, E and N were studied in order to better appreciate dispersion results for the same geologic formation. The ideal situation for these histograms is that which is nearer Gauss’ distribution curve. This implies that for a locality with the same geologic formations, there is sufficient Pl – P0, E, N data at a given time interval. This study of histograms is only carried out on the sites where a good number of data (minimum 15 data) were collected. The percentage of retained data is 89.91% (for the fine, medium and coarse Bonaberi sands) and the correlation coefficient is 0.86. Correlations between Pl – P0 and N is given as: N = 2.88 (Pl – P0).

(1)

Due to data dispersion the equation can be written as: N = [1 – 5] (Pl – P0).

(2)

This coefficient is within the range of 1 and 5. Sandy gravel 80 60

N

40 20 0 0

2

4

6

8 10 Pl-PO (daN/cm2)

12

14

16

Figure 1: Correlation between Pl – P0 and N

18

20

674

E. Kenmogne and J.R. Martin / Correlation Studies Between SPT and Pressuremeter Tests

These obtained correlations are approximate straight lines computed by the least square method on a number of points, considering the origin at zero. The percentage of retained data for gravely sand, varies from 83.3% (between E and N) and 91.67% (between Pl – P0 and N).The resulting correlation coefficient is between 0.93 and 0.94. The average correlation between Pl – P0 and N is given as: N = 2.20 (Pl – P0) and E = 4.78 N

(3)

The validity criterion to be retained for this test type is: N = [1- 6] (Pl – P0) and E = [2 - 8] N

(4)

Clean sand

50

Gravely sandy

50 40

30

30 N (bar)

40

20

N

20

10

10

0

0 0

5

10 15 Pl-PO (daN/cm2)

0

20

Figure 2: Correlation between Pl – P0 and N

5

10

15 20 Pl-P0 (bar)

Figure 3: Correlation between N and E

Clayey sand: the percentage of retained data varies by 89.9% (between N and E).Its correlation coefficient is between 0.9 and 0.92; which indicates that these soils are well adapted for this type of test. The validity criterion is defined by the following relationship: N = [1 – 6] (Pl – P0) and E = [2 – 20] N

(5)

Sandy clay

Gravely clay

50

200

40

150

N (bar)

E (daN/cm2)

30

100

20 10

50 0

0 0

5

10 Pl-P0 (bar)

15

20

Figure 4: Correlation between Pl – P0 and N

0

10

20

N

30

40

Figure 5: Correlation between N and E

50

E. Kenmogne and J.R. Martin / Correlation Studies Between SPT and Pressuremeter Tests

675

5. Conclusions There is a linear correlation between SPT resistance, N and the pressuremeter parameters net limit pressure, pl* = Pl – P0 and pressuremeter modulus, E. - Sandy formations: the straight line equation for this soil, Y = a (Pl – P0) ranges between 1 and 5 taking into consideration the dispersion from the correlation straight line. - Gravely sand: The correlation straight line between N and pl* is between 2 and 3 - Sandy clay: The correlation straight line between N and pl* is given by N = apl*, where a is between 1 and 6 with a correlation coefficient of 0.92. The correlation between E and N is given as E = bN, where b is between 2 and 8 for the gravely sand whereas for clayey sand it is between 2 and 20.

References [1] M. Cassan, Les essais in situ en Mécanique des sols, Tome 1 réalisation et Interprétation, 1988. [2] M.-P. Coville.Lafeuillade, « Corrélations entre le Standard Penetration Test et le pressiomètre » mémoire présenté en vue de l’obtention du diplôme d’Ingénieur en Géophysique et Géotechnique, le 19.10.90, Institut en Science et Technologie, Université Pierre et Marie CURIE,. 1990.

This page intentionally left blank

Section 9 Slopes

This page intentionally left blank

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 679 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-679

The Value of Slope Failure Back-analysis in Open-pit Slope Design: A Case History from the South African Coalfields Mmathapelo SELOMANE and Louis VAN ROOY a Department of Geology, University of Pretoria

Abstract: A case study is presented on the back-analysis of a slope failure at an open-pit coalmine that is believed to be a progressive failure where failure in one material triggered failure through another. Back-analysis, using the method of slices was performed to obtain material properties at failure with the aim of redesigning future slopes. The back-analysis included sensitivity and probabilistic analyses using 2D limit equilibrium slope stability analysis software. A sensitivity analysis was performed to determine the material property with the most significant influence on the stability of the slope. A probabilistic analysis was performed to determine the probability of the proposed new slope geometry to fail during future mining activity. Keywords: Back-analysis, Sensitivity analysis, Probabilistic analysis, Material properties, Factor of Safety, Probability of Failure.

Introduction Mining activity changes the stress distribution and groundwater conditions in the surrounding rock mass and may therefore change the behaviour of the rock mass. Open-pits and underground mines are prone to the effects of rock mass failure where the rock mass, when under changed stress conditions (which can be mining-induced or brought about by external factors), is subjected to changes in mechanical properties [1]. Open-pit slope stability analysis is vital and it is always required to investigate the stability of a slope to maintain its safe and functional design. Furthermore, the stability analysis allows for the assessment of the physical and geometrical parameters, which can have an influence on stability. The concepts of factor of safety (FoS) and probability of failure (PoF) are used as indicators used in the comparative estimate of slope stability, where a slope is stable if its FoS > 1 and unstable if FoS < 1. In this paper a back-analysis on a failed slope at an open-pit coalmine in the Mpumalanga Province, South Africa, is performed to determine material properties at failure. These properties are then applied to propose a stable remedial slope for future mining.

a J L van Rooy: Department of Geology, University of Pretoria, Private Bag X20 Hatfield, Pretoria, 0028; South Africa; E-mail: [email protected].

680

1.

M. Selomane and L. van Rooy / Value of Slope Failure Back-Analysis in Open-Pit Slope Design

Site description

At the coalmine, the mining areas are exploited through open-pit cut and fill operations, with cuts of varying depths and varying slope angles. The rockmass in the slope under consideration is closely or heavily jointed and dry. Rocks from the Ecca Group of the Karoo Supergroup and overlying soft materials. It failed in a circular mechanism as seen in Figure 1.

Figure 1: Failed slope (Circular mechanism)

The geotechnical domains within the mining area are based mainly on the rock type and the associated geological hazards, structures, jointing, and rock properties. The geotechnical domains identified includes the dumped waste material, soft overburden (SOB), hard overburden (HOB), coal and tillite floor rocks. Waste material can be described as soil and rock material, which was removed from its original position and relocated. It consists of a mixture of weathered overburden material (thin topsoil, weathered shale, siltstone and sandstone) and, unweathered shales and sandstone (hard overburden). The waste is stockpiled adjacent to the mining areas or backfilled into mined out areas, as may be required. The SOB is mainly made up of completely to highly weathered sandstone, siltstone and shale. It extends from surface down to the moderately or unweathered bedrock and varies in thickness from 10 to 20 m. It can be classified as a very poor quality rock mass with an RMR of less than 20 [2]. This range in the rock mass quality correlates with the degree of weathering of the rock mass. The HOB consists of moderately to unweathered sandstone, siltstone and shale. It extends from the soft overburden down to the coal beds (mining horizon) varying in thickness from 15 to 25 m and is classified as generally fair quality rock mass with an RMR ranging from 30 to 70 [2]. The coal horizon includes coal and shale bands and is classified as a generally fair rock mass with an RMR of 35 to 65 and is generally associated with less instability.

M. Selomane and L. van Rooy / Value of Slope Failure Back-Analysis in Open-Pit Slope Design

681

There is however, a slope instability hazard associated with unfavourably orientated bedding, jointing and faulting in the basin structures, due to the dip of the tillite floor. Coal adjacent to dolerite intrusions occurring in the area, which cut transgressively through the coal seams forming a basin structure, is frequently baked and thus has a lower rock mass quality. This may result in unstable mining slopes. Furthermore, the spontaneous combustion of coal in slope faces can lead to undercutting of the overburden and therefore slope failures or rock falls. The Dwyka Group diamictites form the pit floor on which the subsequent sedimentary rocks, including the coal beds were deposited. No mining activity takes place in the tillite, which is classified as a generally good quality rock mass with an RMR of 40 to 80. The tillite-coal contact can be described as weak due to slickensides and polished surfaces and forms an undulating surface.

2.

Back-Analysis

The following assumptions were made during the back-analysis: • The positions of the geological contacts are reliable, • The interpreted position of the failure surface is correct, and • The influence of pore water was negligible. 2.1 Deriving material properties at failure: Sensitivity analysis using SLIDE Sensitivity analysis aims to derive material properties at failure and to determine which property has the most significant effect on the stability of a slope [3]. As failure at the mine occurred in a progressive manner (failure in one material triggered failure through another), the analysis was performed in phases of failure in order to simulate what might have happened during failure. Figure 2 is a graphical presentation of the assumed slope failure sequence (proposed phases of failure). For each material type in the slope, an applicable failure criterion was used in deriving its strength properties. However, coal was not part of this analysis due to sufficient material information and the waste dump properties were not derived through the sensitivity analysis but selected from site experience. The initial slope failure involved the rock materials, tillite and unweathered clastic sedimentary rocks (HOB). Their properties were derived using the Generalised Hoek-Brown failure criterion [4] in conjunction with the Geological Strength Index (GSI) [5], a rock mass classification system. The following input properties were derived: • Hoek-Brown constant (mi), • Uniaxial Compressive Strength of intact rock (UCS or σci), and • Geological Strength Index (GSI). Following the initial failure the soil material, SOB, failed during the second phase. The properties were derived using a Mohr-Coulomb failure criterion where the following input parameters were derived: • Drained friction angle (φ’), and • Drained cohesion (c).

682

M. Selomane and L. van Rooy / Value of Slope Failure Back-Analysis in Open-Pit Slope Design

Figure 2: Cross-section depicting proposed phases of failure

Phase 3 of the failure was considered as the ultimate failure phase and included failure of the waste dump. Properties for the waste material were known from previous analysis and it was therefore not necessary to determine them during this phase. Phase 3 is also the phase during which the major volume of material, comprising waste dump, SOB, HOB, coal and tillite was mobilised and moved down the slope until movement stabilised. Table 1 presents the average material properties derived for the different geotechnical domains. The uniaxial compressive strength of the HOB and tillite was found to have the most significant effect on the stability of the slope and the cohesion has the most significant effect for the SOB. The material properties derived in the sensitivity analysis above were confirmed during the probabilistic analysis on the geometry of the failed slope, also using the SLIDE program. The results indicated a 100 % probability of failure and a factor of safety of 0.4. The next step was to design a remedial slope using the derived material properties in Table 1. 2.2 Designing a remedial slope using SLIDE The failed slope comprises of both rock and soil materials which required the use of an applicable failure criteria for the estimation of their strength. Therefore, in the

M. Selomane and L. van Rooy / Value of Slope Failure Back-Analysis in Open-Pit Slope Design

683

designing of the remedial slope, the materials for which the same failure criterion was used for the determination of their properties were combined into one material layer.

Material Waste SOB HOB Coal Tillite

Table 1: Derived material properties in their average values UCS Cohesion Density Failure criterion Friction 3 (MPa) (kPa) angle (o) (kN/m ) 30 0 20 Mohr-Coloumb 30 44 20 Mohr-Coulomb 45 22 Gen Hoek-Brown 21 15 Gen Hoek-Brown 45 23 Gen Hoek-Brown

GSI

mi

53 48 60

15 8 10

The steps outlined below were followed to get to the stable slope geometry. • Step 1: Treated waste dump and SOB as Material 1 and determined its slope angle at probability of failure of 10% or less (acceptable risk defined by the mine). • Step 2: Treated HOB and coal as Material 2 and determined its slope angle at probability of failure of 10% or less (acceptable risk defined by the mine). Step 3: Performed a probabilistic analysis on the overall slope (Material 1 + Material 2). Table 2 shows the results obtained in the steps above; this includes the geometries, probabilities of failure and factors of safety. The values in bold and italics in the table were chosen to be the stable geometries for respective materials and the overall slope geometry. Figure 3 shows the recommended remedial slope design for future mining. Table 2: Summary of the steps carried out in the determination of the remedial slope design Material Slope height Slope angle Probability Factor of safety (m) (º) of failure Waste dump & SOB 36 44 93.33 0.916 combined (Material 1) Waste dump & SOB 36 40 9.667 1.076 combined (Material 1) Waste dump & SOB 36 38 0.33 1.144 combined (Material 1) HOB & Coal combined 26 90 5 1.203 (Material 2) HOB & Coal combined 26 84 0 1.281 (Material 2) All materials combined 102 (overall) 44 (overall) 0 1.993 (Material 1 + Material 2)

684

M. Selomane and L. van Rooy / Value of Slope Failure Back-Analysis in Open-Pit Slope Design

Figure 3: Proposed remedial slope design for future mining with overall height of 102 m and overall slope angle of 44◦

3.

Conclusion

The process of back-analysis to determine material properties at failure is useful to obtain materials properties to use in the safe redesign of the open-pit slopes. The case study presented in this paper highlights the application of back-analysis to redesign a safe slope under similar ground conditions with the confident conclusion that future failure of the remedial slope for future mining at the coal mine is not expected. This is supported by a very low (approaching 0) probability of failure and factor of safety of 2 calculated for the redesigned slope based on the material properties obtained during the back-analysis.

References [1] T. Szwedzicki, Rock mass behaviour prior to failure, International Journal of Rock Mechanics and Mining Sciences, 40 (2003), 573-584. [2] Z.T. Bieniawski, Engineering Rock Mass Classifications, Wiley, New York, 1989. [3] http://www.rocscience.com/products/Slide/Sensitivity. Analysis.asp, Accessed: 17/03/09. [4] E. Hoek and E.T. Brown, Underground Excavations in Rock, Institution of Mining and Metallurgy, London, 1988. [5] E.T. Hoek, D Wood and S. Shah, A modified Hoek-Brown criterion for jointed rock masses, Proc Rock Characterization, Symp. Int. Soc. Rock Mech.: Eurock ’92 (ed. J.A. Hudson), 209-214, Brit. Geotech. Soc. London, 1992.

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 685 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-685

General Slope Stability using Interslice Forces and Flow Nets but avoiding ru Factors A.D.W. SPARKS University of Cape Town, Rondebosch, South Africa

Abstract. Several published methods which incorporate inter-slice forces in slope stability calculations make use of the ru factor which is the ratio of the Pore Water Pressure to the Total Vertical Pressure. The value of ru will have different values for different soil slices. It is incorrect to refer to a socalled “average r u” for a slip surface. Unfortunately several recent text books continue to use an “average ru” for the stability analysis of slopes.

Keywords. Slope Stability, ru factor, Inter-slice Forces, Factor of Safety

Introduction The pore water pressures within a soil slope will greatly affect the Factor of Safety of the slope. A seepage flow net can be drawn to provide the seepage pattern. The water pressures can be determined at any point in the slope by using the equipotential lines of the flow net. In turn these pressure values can be converted to give uplift water forces which act along the sliding surfaces under each slice or as inter-slice forces. One must ask why should one use an alternative complex method involving ru for slope stability analysis ?

1. Using Interslice Forces The method described in this paper uses interslice forces and this method also ensures that all the forces acting on any soil slice are in equilibrium in the vertical and horizontal directions. The method presented in this paper will achieve the same accuracy as the Bishop General Method, but the use of the present method seems to be simpler. This method is more accurate than the Bishop shorter method which does not include the interslice forces. In the simple Swedish method of slices, the forces acting on each slice are not in static equilibrium in the vertical and horizontal directions, and the Swedish method does not include inter-slice forces. A few authors have attempted to ensure that the forces acing on each slice are also in equilibrium when moments of these forces are taken about any point within each slice. This approach seems to be an unnecessary pursuance of perfection.

686

A.D.W. Sparks / General Slope Stabilityusing Interslice Forces and Flow Nets

2. The Interslice Method

W (total weight) includes external loads acting on upper soil surface. If there is no earthquake then Whoriz = 0 Calculate W, U UL , UR for

PL

W

δL

Whoriz

UL

δR PR

α

UR

each slice for

Fd

entry into the

U

formulae.

PN Cd ΔL

Figure 1. Typical Forces on a Slice.

3. Description of the method ( ref 4) The highest slice is placed on the right and it is labelled No.1. The writer uses approximately five or six slices. A thrust line is drawn through the slices to indicate the possible line of action of the forces PL and PR . One method is to place dots on the verticals between slices at positions located at 0,6 of the distance from the soil surface to the slip surface. A smooth curve is then drawn through these dots. The slope of this thrust line provides the angles δL and δR shown in Figure 1. For slice No. 1, the lefthand vertical edge becomes a zero distance with P L and UL equal to zero. The angle δL is also zero for slice No. 1. The equations (1) to (7) are completed in turn for each slice. Note that the values of P R and UR for one slice become the values of P L and UL for the next slice. In this method a trial factor of safety with respect to shear strength Fs (e.g. 1,5 ) is assumed for the first cycle. Equation (8) permits the trial value of F s to converge rapidly in the next trial. For the last slice ( lowest slice ) assume that δR is zero for all calculations. The correct value of F s causes the value of P R to be zero for the last slice.

A.D.W. Sparks / General Slope Stabilityusing Interslice Forces and Flow Nets

687

4. The equations which apply in turn to each slice :Start with a trial Fs. Seven sequential equations are used for each slice :Tan Φ 'd = Tan Φ ' / Fs ...........................................................................................(1) X = Whoriz + U. Sin α + U L + PL ( Cos δL ) - U R - (c. ΔL / Fs ). Cos α ................(2) Y = W + PL ( Sin δL ) -U.Cos α -(c. ΔL / Fs ).Sin α

...................................................(3)

S1 = Sin α − Cos α .Tan Φ' d .......................................................................................(4)

C 1 = Cos α + Sin α .Tan Φ'd .......................................................................................(5) PR = X + {( Y x S1 / C1 ) / (Cos δ R + Sin δR x S1 / C 1 )} ............................................(6)

PN = { Y - PR x Sin δR } / C1 .......................................................................................(7) When seeking Fs , the new Trial Factor of Safety Fs(new) for the next cycle is :Fs

(new)

= { 1 - [ PRn / ( Σ (W.Sin α ) + Σ( Whoriz .Cos α ) )] }. Fs (current) ........................(8)

where PRn = compressive value of the horizontal PR for the last ( nth ) slice. ( i.e. PRn will have zero value when the correct Fs is used. ) 5. Extra Comments This method is not restricted to circular slip surfaces and it can easily cope with ponds of water at the toe or top of the slope. Extra surface loads can be included in the weight W of each slice. It will be noticed that a trial value for the “Factor of safety with respect to shear strength Fs” is used for each cycle of calculations, but in this method the value of Fs quickly converges to the correct value within two cycles of calculations ( See equation 8 and section 6 ). The correct solution is obtained when the value of PR for the last slice is zero. If a downstream pond exists, one can include the weight of free water in the last vertical slice in the weight W of the last slice and UR can be the horizontal water force on a vertical section within the pond. Figure 2 depicts a simple example of only three slices. The third slice includes the weight of the free water above the surface YZ. The horizontal water force on the vertical surface DK is equal to the value of UR for this last slice. The dotted line in Figure 2 is the thrust line for inter-slice forces representing effective pressures which act between the slices. As shown in this Figure 2, the thrust line is assumed to pass through points at positions located at 0,6 of the distance from the soil surface to the slip surface. The slopes of this thrust line provide the angles δL and δR for each slice. Assume that one is required to find the Factor of Safety F s for possible sliding along the trial slip surface EAFJK and one is also required to find the horizontal force which acts across the vertical surface NWJ on the retaining wall.

688

A.D.W. Sparks / General Slope Stabilityusing Interslice Forces and Flow Nets

Figure 2. Slice 1 is on EF . Slice 2 is on FJ. Slice 3 is on JK

6. An Example for Analysis The dotted line in Figure 2 shows the approximate thrust line for the inter-slice effective forces. The slopes of this line provides the values of the angles δL and δR . Shear strength values for the concrete-clay interface JK are Φ’=30o and c’ = 30 kPa. The average pore pressure under surface JK is 60 kPa. The convergence of this method is excellent. For example, using a trial value of Fs = 1,3 the suggested trail value at the end of one cycle was Fs = 1,713 for the next cycle. The true final value was 1,718. Table 1 will also show that the total horizontal force on the wall (i.e. on the vertical face NJ) is ( 245 + 298,2) = 543 kN per metre into the paper.. The author modified both the Bishop Simplified Method and the Swedish Method to allow for the pond at DK. This yielded F(Swedish) = 1,3 and F(Bishop) = 1,647. Table 1. Final Values of Calculation ( Using Fs = 1,718 )

Slice Total Weight Angle No. kN/m δR

Water Force Water Force U under UR on slice vertical face

Calc. Force PR

1

50 kN/m + 252 kN/m

42o

100 kN/m

80 kN/m

149,3 kN/m

2

100 kN/m+ 630 kN/m

0o

275 kN/m

245 kN/m

298.2 kN/m

294 kN/m + 625 kN/m +150 kN/m

0

o

360 kN/m

125 kN/m

0 kN/m

3

A.D.W. Sparks / General Slope Stabilityusing Interslice Forces and Flow Nets

689

Figure 3. Bold Lines are Equipotential Lines

7. How to cope with curved equipotential lines. The inter-slice water pressure force which acts on CD in Figure 3 is greater than the value of 0,5 γ w. hs2. The value of this force is actually 0,5 γ w. hs. hc . Writers of commercial computer programs may claim to use different ru factors for each soil slice, but one must ask how do they cope with problems posed by curved equipotential lines. The inter-slice water pressure force acting on AB is also 0,5 γw. hs. hc. 8. A critical discussion concerning ru factors The method based on ru factors was originally used by Bishop ( 1955 ) especially for dams with thin clay cores, which were subjected to drawdown conditions. Subsequently the use of the ru factors was extended by Bishop and Morgenstern (1960) to more complicated slope stability problems. Average ru values for a slope continue to be used in text-books on slope stability. Bishop and Morgenstern noted that using an average value for ru could lead to over-estimating the factor of safety by 7 per cent (1960). Methods for calculating the average ru for a slope were provided by Whitlow (1990, p358) and Smith and Smith ( 1998, p 180 ). However it is obvious that these formula do not embrace all the variations. Even recent text-books continue to use an average ru for a slope. For example the book by Das ( 2006 ) presents several pages ( p 552 to 560 ) which describe the results from Bishop and Morgenstern and the methods by Spencer and the method by Michalowski. It is obvious that to use these charts it is necessary to assume an average value of ru for a slope. Bishop and Morgenstern have quoted the following formula for the factor of safety with respect to shear strength : Fs = m'−n'.ru ................................................................................................................(9)

where m’ and n’ are obtained from tables or charts ( Das, 2006 ). The type of flow net in the slope is not defined ( i.e. flow with or without a toe filter etc ). Again it is obvious that some type of average value for ru for the slope is needed for use in the published equation (9).

690

A.D.W. Sparks / General Slope Stabilityusing Interslice Forces and Flow Nets

9. Closure Critical Slip Surface No.1 Critical Slip Surface No. 2 H = 20 metres

Figure 4. Slip 1 is critical for sandy soils. Slip 2 is critical for clays

Figure 4 shows that one particular slope may have two different critical slip surfaces. Slip surface No. 1 is the critical slip surface for a sandy cohesionless soil, whereas the slip surface No.2 would be the critical surface for a cohesive clay soil. In the case of the slip surface No.1 we would assume that ru is zero. The so-called average ru for the slip surface No.2 might be about 0,2 ( if it can be calculated ). There are two different critical slip surfaces, and two different “average ru” values for this one slope. Students would be wrong if they were to assume that there exists one “average ru” for a particular slope with a seepage flow net. The values of ru are also a function of the position of the trial slip surface. Whenever the position of the trial slip surface is altered, the value of the “average ru” for this slip surface will change. The use of stability tables where the student must choose a value of ru can lead to a false understanding of the ru method.. An engineer in practice might be satisfied with a computer program which claims to find the “average ru” , but a lecturer and a student are more satisfied if they can progress through a method without assuming various functions or steps which are difficult to explain. The methods in this paper use the actual water pressure forces measured from a flow net, and the steps in the method are easier for teaching to students. References [1] D.W.Taylor, Fundamentals of Soil Mechanics, Wiley 1948 p435 [2] A.W.Bishop, Embankment Dams, Chapter 9, in Hydro-electric Engineering Practice, Editor. J. Guthrie-Brown, 1965, p349 - p406. [3] A.D.W.Sparks, Methods of General Stability Analysis, 8th Int. Symp. on Landslides, Cardiff, 26-30 June 2000, Thomas Telford, p 1373 - p1378. [4] A.D.W.Sparks, Wedge-shaped Slices for Bearing Capacity and General Slope Stability, Proc.BGA Int. Conf. on Foundations, Dundee, Scotland, IHS BRE Press, .2008, p1693 - p1705. [5] A.W.Bishop, The use of Slip Circle in the stability Analysis of Earth Slopes, Geotechnique,vol 5, No 1,(1955), p 7-p17. [6] A.W.Bishop and N.R.Morgenstern, Stability Coefficients for Earth Slopes, Geotechnique,vol 10,No 4,(1960), p129-p147. [7] R. Whitlow, Basic Soil Mechanics, Longman Scientific & Technical, 1990, p358. [8} G.N.Smith and I.G.N.Smith,Elements of Soil Mechanics,Blackwell Science,1998, p180. [9] B.M.Das, Principles of Geotechnical Engineering, Cengage Learning, 2006, p552 - p559.

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 691 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-691

Pit Slope Design near Tete, Mozambique, without the Benefit of Previous Slope Performance Experience Phil CLARK Pells Sullivan Meynink Pty Ltd, Australia

Abstract. One of the more important aspects of slope design is comparing a new slope design with any previous experience or performance of slopes in similar materials and environment. However, in environments such as the Moatize Coal Basin, Tete Provence, Mozambique, where no large scale open pit mining has occurred in the past, previous experience is non-existent. This paper focuses on the design of two large open pits in the Moatize Coal Basin and techniques that were employed to counteract the lack of previous experience or performance guidelines. Greater reliance is placed on the creation of an accurate geotechnical model of the mining environment. Keywords. Open pit, slope design, rock mass strength, Tete, Mozambique

Introduction Slope performance charts and back analysis techniques provide an established mine or mining area valuable guidelines during the design process as a means to calibrate an empirically derived slope design against actual slope behaviour [1, 2]. They have been used successfully as a tool to optimize existing pits or start up new pits in similar geotechnical conditions. Typically, a slope performance chart plots slope height versus slope angle and will contain groups of data points representing rock types, failed slopes, stable slopes and so forth. Geologically similar environments in both South Africa (Karoo Basin Coal Measures) and Australia (Sydney and Bowen Basin Coal Measures) contain numerous open pit coal mines. However, slope performance charts in these environments can only be used as rough guidelines because of differences in the geological history of the coal deposits. The geological history of a deposit results in different rock mass and material properties and consequently, differing pit designs (most Australian coal open pits are less than 150m deep). This paper discusses the challenges in providing operational pit slope designs in an environment where no open pits have previously been mined and where continuous surface outcrops are very limited, for Riversdale Mining Limited (Riversdale). Riversdale has recently started development on two coal mines in the Moatize Basin, near Tete, Mozambique. These two mines are known as the Benga Mine and the Zambeze Mine. Both mines could contain pits in excess of 500m depth if required by mining economics.

692

P. Clark / Pit Slope Design Near Tete, Mozambique

1. Setting 1.1. Location Riversdale’s coal projects are located in the province of Tete in north-western Mozambique (Figure 1). The two developments are composed of Permian Karoo Coal Measures and located on undulating topography that rise from the Zambeze and Rovubwe Rivers towards the north and east, where hills composed of Precambrian basement gabbro and gneiss occur. To the south, the leases are constrained by either Tete city or the Zambeze River. The Rovubwe River also forms a boundary between the two leases. Elevations across the leases vary from about 125m above sea level, along the rivers, to about 200m, where basement rocks occur.

Figure 1. Location plan of Riversdale’s developing mines in Mozambique.

1.2. Regional Geology The two mine sites are located within the Moatize Basin, one of many isolated basins in Southern Africa containing coal bearing formations from the Karoo Supergroup. Faulting, associated with the East Africa Rift, was active during the formation of the

P. Clark / Pit Slope Design Near Tete, Mozambique

693

coal seams at Benga, resulting in the coal deposit containing a series of graben-like structures. The dominant structural orientation in the Moatize Basin appears to be northwestsoutheast trend (Figure 2) as suggested by: • • • •

the orientation of the East Africa Rift in this area, the Zambeze River, the contact between the Permian Karoo Coal Measures and the underlying Precambrian basement rocks, faults shown on the 1:250,000 sheet.

An orthogonal structural set is also suggested and appears to be southwestnortheast trending. Near the main rivers, Recent to Tertiary alluvium deposits occur.

Figure 2. Regional Geological Map [3] with fault zones interpreted from coal seam correlation studies .

1.3. Lithology and Major Structures The Karoo Coal Measures are interpreted to be from the Ecca Group of the Karoo Supergroup. The measures contain interbedded sequences of sandstone, shale and coal seams. Several major coal seams and numerous other coal seams are separated mainly by massive, high strength, fine to coarse grained sandstones and conglomerates. High strength shale typically occurs near the roof and floor of the seams.

694

P. Clark / Pit Slope Design Near Tete, Mozambique

Normal faults with throws of up to 600m have been identified from coal seam correlation studies. The faults form a series of fault bounded blocks and graben structures. Bedding typically dips between 0° and 15° within the fault bound blocks. Drill core from fault zones typically contain less than 2m of crushed and sheared shale and coal with occasional minor sheared zones and brecciated sandstone above and below the main fault zone, extending laterally for approximately 10m. An interpretation of aeromagnetic data suggests there are no major intrusive structures in the Benga Mine, but there may be a major dyke structure bisecting the Zambeze Mine. Minor intrusive structures such as discrete dykes and sills are likely to occur. 1.4. Hydrological Setting Surface hydrology on the site is controlled by the Zambeze and Rovubwe Rivers. Within the mining leases, the Zambeze River appears to be well constrained with a limited flood plain and paleochannel, whereas, the Rovubwe River, one of the major tributaries on the northern bank of the Zambeze River, has an extensive alluvium filled paleochannel and floodplain. The Karoo Supergroup is a groundwater source throughout southern Africa. Typical experience in South African coal mines within the Karoo Coal Measures indicates that dolerite intrusions and large scale geological structures define groundwater regions. On the mine sites, the depth to standing water is typically 0m to 20m below ground level, with occasional boreholes, particularly near major fault zones, having minor artesian pressures.

2. Geotechnical Model An accurate geotechnical model is fundamental to a reliable slope design and in an environment with no previous slope performance criteria and limited outcrops, becomes even more important. The geotechnical model is compiled from the following four models, each of which requires good observations, accurate data recordings and in various ways, are interlinked with each other: • • • •

geological model structural model hydrogeological model rock mass and material model

2.1. Data Sources The following data sources were used in the creation of a geotechnical model for the mine sites: • • • •

regional 1:250,000 scale geological map [3]; typical geometric models of normal fault zones [4]; numerous geological and geophysical wireline logs at 100m to 1000m borehole spacing; limited geotechnical defect logs and acoustic televiewer (ATV) wireline logs;

P. Clark / Pit Slope Design Near Tete, Mozambique

• • •

695

groundwater drilling and testing results; field mapping of limited boulder outcrops; limited laboratory testing and extensive point load strength testing.

2.2. Geological Model The geological model was largely derived by Riversdale using data from geological and geophysical wireline logs, input into 3D geological modelling software. Multiple coal seams were identified from the various logs and marker beds tagged accordingly. Interburden material between coal seams was identified as predominantly sandstone, but with occasional shale and laminite units. 2.3. Structural Model The structural model was compiled at two scales; major structures and minor structures. Major structures were identified by coal seam correlation studies and selected targeted drilling undertaken to further refine the major structure characterisation. Minor structures were identified from detailed geotechnical and ATV logging. Characteristics of the main structure or defect types are provided in Table 1. Table 1. Major and minor structures characteristics Structure Normal Faults Crushed and sheared zones Bedding plane shears Bedding Joints/shears

Category Major

Description • 1-2m of intensely sheared and crushed coal and shale fault gouge Major • 100-500mm of fractured to intensely sheared parent rock Major/Minor • Slickensided planes and zones up to 100mm thick, typically located within coal seams Minor • Typically planar, smooth to rough and dipping <30° Minor • Typically planar to undulating, smooth to rough and dipping >45°

Interpretation • Regional Fault • Fault that could control the design of overall slopes • Defects that could control design of overall slopes • Defects that could control batter and berm design • Defects that could control batter and berm design

2.4. Hydrogeological Model As stated in the preceding section, groundwater across the site is relatively high with artesian pressures near some fault zones. Detailed groundwater tests are still on going, but preliminary testing on the Benga Lease indicates that the coal measures have very low hydraulic conductivity (1x10-7m/s to 1x10-10m/s). Interburden and fault zones are expected to be at the lower end of the scale, whereas coal seams are expected to be at the higher end of the scale. 2.5. Rock Mass and Material Model The rock mass and material model is a function of the other three models, coupled with laboratory test data. Table 2 provides a summary of selected rock material properties for the main interburden materials. The rock mass was classified using both RMR and GSI methods for borehole data, which ultimately allows the distribution of rock mass

696

P. Clark / Pit Slope Design Near Tete, Mozambique

properties throughout the pit as an aid for mine planning. GSI values were also used for further rock mass strength assessments and slope design methods. Table 2. Rock types and selected properties Unit Medium to coarse sandstone/conglomerates Fine sandstone/laminite Shale

Density (t/m3) 2.45

UCS (MPa) 45

E (GPa) 12

Abrasivity Very

Typical GSI 65-75

2.55 2.50

75 60

25 25

Slight to very Slight

65-70 60-65

3. Slope Design The typical slope design process involves the creation of a geotechnical model, analysis of available factual data and comparison and calibration studies against existing slope performance data. Where there is no slope performance data, then there is more reliance on an accurate geotechnical model as well as interpretation of factual data. As more data becomes available then this is fed back into the model and interpretations refined. Given the relatively massive and high strength nature of the rock masses, until the influence of in situ stress becomes apparent at depth, it was concluded that the most likely overall slope failure mechanism would be controlled by major geological structures whereas minor geological structures were more likely to control the stability of batters. As such, the Benga and Zambeze slope design studies focused on the following feasibility level analyses: 1. 2.

3. 4.

5.

Delineation of structural domains using a combination of ATV data and surface outcrop orientation mapping data and the geological model. Kinematic and probabilistic analyses using ATV data and stereoscopic methods to principally develop batter and catch berm design parameters as well as develop guidelines for interramp slope design parameters. These analyses were performed for each pit slope aspect of each structural domain. Maximum practical depth of mining assessments based on regional stress databases from Southern Africa. Two dimensional limit equilibrium and finite element numerical modelling of a generic slope model to identify likely interramp and overall rock mass slope failure mechanisms. These analyses used Hoek and Brown rock mass failure criterion parameters derived from the GSI rock mass model [5]. Sensitivity analyses on the limit equilibrium modelling considered: • • • • •

seismic loading; infrastructure positioning near the pit crest; position of the pit wall relative to adversely orientated major structures; block slide failure mechanisms based on the intersection of adversely orientated, steeply dipping normal faults and continuous low angle bedding plane shear zones; effects of groundwater depressurisation.

P. Clark / Pit Slope Design Near Tete, Mozambique

697

Once mining commences, initial pit wall face mapping and photogrammetry analyses as well as slope surveys will be used to refine the geotechnical model and provide optimised final pit wall slope designs with a greater degree of confidence. However, the effects of opening up a potentially 500m deep pit can only be monitored via an observational methodology as there is very little data to compare performance of high slopes composed of multiple interbedded coal measures.

4. Conclusion Without existing slope performance data or back analysis of previous slope failures, the typical empirically derived slope design presents a number of uncertainties and risks that can only be addressed as mining proceeds. Additional studies, coupled with actual slope performance data can be used to calibrate the geotechnical model and provide a design with less uncertainties, which ultimately results in lesser risks to the mining process. But to reduce the risk, before mining proceeds and provide a reliable slope design, without the benefit of previous slope performance data fundamentally requires an accurate, sound geotechnical model, the basis of which is good factual data.

Acknowledgement The author would like the thank Riversdale and Pells Sullivan Meynink for the use of the data and permission to publish this paper. Gratitude’s must also be expressed to the Mozambiquan geologists who have toiled in the heat collecting much of the data used in the assessments.

References [1] A. Duran, K Douglas, Do slopes designed with empirical rock mass strength criteria stand up? Proceedings 9th International Congress on Rock Mechanics, Paris (1999) 1:87-90 [2] J. Sjoberg, A slope height versus slope angle database, Slope Stability in Surface Mining, Ed: W.A. Hustralid (2000), Chapter 5: 47-57 [3] GTK Consortium, Sheet 1633 – Tete 1:250000 Geological Map, Mozambique Ministry of Mineral Resources (2006). [4] C. Childs, T. Manzocchi, J.J. Walsh, C.G. Bonson, A. Nicol, M.P.J Schopfer, A geometric model of fault zone and fault rock thickness variations, Journal of Structural Geology 31 (2009), 117-127. [5] E. Hoek, ET. Brown, Practical estimates of rock mass strength, International Journal of Rock Mechanics and Mining Science 34(8) (1997), 1165-1186.

This page intentionally left blank

Section 10 General

This page intentionally left blank

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 701 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-701

Équations et Exemple de Calcul Hydrique dans les Sols Non Saturés Abdeldjalil ZADJAOUI1 Université Abou bekr Belkaid, Tlemcen, Algérie

Résumé. Cet article présente un modèle numérique en éléments finis construit sur la base des équations de la consolidation des milieux poreux élastoplastiques non saturés. Ce modèle a été programmé dans le code de calcul TFAP (Transient Flow Analysis Program). Les équations du modèle tiennent compte de l’interaction des déformations du squelette, des écoulements de l'eau et de l'air ainsi que des mouvements de l'air dissous dans l'eau. La résolution numérique de ces équations associe la méthode des éléments finis pour discrétiser l'espace et un schéma d'intégration implicite pour discrétiser le temps. Les déformations du squelette sont décrites selon l'approche des variables indépendantes. Une application est décrite en détail et l’auteur donne les conclusions relatives à ce travail. Mots clés. Modèle, éléments finis, non saturé, élastoplastique, variationnel.

Introduction La modélisation du comportement mécanique des sols non saturés ne fait pas l'objet d'un consensus parmi les spécialistes de la géotechnique. L'extension de la notion de contraintes effectives, telle qu'elle a été proposée par Bishop [4], pour servir de cadre à la description des sols, a fait l'objet de différentes critiques, souvent étayées par des résultats expérimentaux obtenus en laboratoire ([6], [9], [11]) ). D'autres approches, comme celles d'Alonso et al. ([2], [3])), ou de Fredlund [5] semblent apporter une réponse plus complète et mieux décrire la réalité par une modélisation physique avec deux champs de contraintes indépendants. Différents modèles théoriques et numériques basés sur les modèles physiques en contraintes effectives ou en variables indépendantes sont apparus depuis une dizaine d'années ([2], [7], [8], [10]). L'étude présentée ici a été réalisée au moyen du code de calcul en éléments finis TFAP. Cet article rappelle les équations qui représentent le comportement élémentaire des milieux non saturés, décrit les étapes de leur transformation en équations matricielles adaptées à leur résolution par la méthode des éléments finis, puis présente quelques résultats de calcul.

1

Zadjaoui Abdeldjalil, BP 230 rue Abi Ayad, 13000 Tlemcen (Algérie) [email protected]

702

A. Zadjaoui / Équations et Exemple de Calcul Hydrique dans les Sols Non Saturés

1. Hypothèses et définitions La représentation mécanique que nous avons adoptée pour la modélisation numérique des sols non saturés repose sur la superposition de trois milieux continus, qui couvrent chacun l'ensemble de l'espace occupé par le sol: le milieu global (sans distinguer de phases), le milieu continu "air" et le milieu continu "eau". 1.1. Champs de Contraintes Trois champs de contraintes sont définis respectivement dans le milieu global, l'eau et l'air: la contrainte totale σ, la pression de l'eau pw et la pression de l'air pa. Pour le calcul, les pressions d'eau et d'air sont remplacées par les charges d'eau h w et d'air ha définies, selon les habitudes de la mécanique des sols, par : p p ha = a + z , h w = w + z et γw γa Avec : γw - poids volumique de l'eau; γa - poids volumique de l'air; z - altitude par rapport à un repère fixe. Le champ de contraintes du milieu global dépend directement de la pesanteur et des conditions aux limites et, indirectement, de la loi de comportement et des autres conditions aux limites. Le champ associé à la pression de l'eau (respectivement, à la pression de l'air) dépend directement de la pesanteur et des conditions aux limites sur la charge d'eau (respectivement, la charge de l'air) et, indirectement, de la loi de comportement du matériau global et des autres conditions aux limites. 1.2. Champs de Déplacements et de Déformations Trois champs de déplacements, auxquels sont associés trois champs de déformations, sont utilisés pour décrire l'état du sol non saturé : • un champ de déplacements associé au milieu continu global, noté u; •

un champ de déplacements associé à l'eau, noté uw;

•

un champ de déplacements associé à l'air, noté ua.

Les déformations totales ε sont négatives en contraction et les flux sont positifs quand ils sortent du volume élémentaire.

2. Comportement mécanique On suppose que les déformations du milieu peuvent être induites par une variation de la contrainte totale (σ+pa) et/ou de la succion (pa-pw), considérées comme variables indépendantes. La loi de comportement est de type élastoplastique avec écrouissage. Elle est décrite dans cet article avec les équations proposées par Alonso et al. [3], mais peut être transformée aisément pour accueillir d'autres formes de lois élastoplastiques avec ou sans écrouissage. Les contraintes et les déformations doivent satisfaire simultanément les équations d’équilibre et la loi de comportement.

703

A. Zadjaoui / Équations et Exemple de Calcul Hydrique dans les Sols Non Saturés

2.1. Equations d'Équilibre σij, j + Fi = 0

avec : - σij- tenseur des contraintes totales;

- F i- forces volumiques.

2.2. Loi de Comportement du Milieu Global [3] Cette loi de comportement est écrite sous la forme incrémentale :

dσ ij = D ijepkl dε kl + ( Fe - 1) dp a δ ij + Fe dp w δ ij avec - Dep - matrice de souplesse relative aux variations de (σij +paδij) ; - dεkl - incrément du tenseur des déformations ;

- Fe = Dep Ds ;

- Ds- matrice de souplesse relative aux variations de (pa-pw) ; - δij - symbole de Kronecker ;

- dpa- incrément de pression d'air ;

- dpw- incrément de pression d'eau.

3. Écoulements de l’eau et de l’air dans le milieu poreux L'eau et l'air se déplacent dans l'espace occupé par le milieu global en respectant, d'une part, des lois de conservation de la masse et, d'autre part, des lois reliant les vitesses moyennes d'écoulement aux gradients de charge (loi de Darcy pour chaque phase). Ces vitesses moyennes désignent les vitesses relatives des fluides par rapport au solide, comme si tout l’espace (solide+pores) leur était offert. 3.1. Equation de Conservation de la Masse d'Eau

∂ (ρ w nS r ) + div(ρ w v w ) = 0 ∂t avec - ρw- masse volumique de l'eau ; - n - porosité du massif ; - v w - vitesse moyenne d'écoulement de l'eau ; - Sr - degré de saturation en eau, décrit par la relation utilisée par Matyas et Radakrishna [11] et reprise par Alonso et al.[2]:

[

(

S r = S r 0 − a s + b s σ ij + p a δ ij Avec :

)] {1 - exp [-c s (p a −

- Sr0 - degré de saturation initial,

]}

pw )

- as, bs, cs - constantes.

704

A. Zadjaoui / Équations et Exemple de Calcul Hydrique dans les Sols Non Saturés

3.2. Equation de Conservation de la Masse d'Air ∂ [ρ a n ⎛⎜⎝1−Sr +HSr ⎞⎟⎠] + div ρ a v a + Hv w = 0 ∂t

[ (

)]

avec - ρa - masse volumique de l'air ; - H - coefficient de solubilité de l'air dans l'eau ou coefficient de Henry - v a - vitesse moyenne d'écoulement de l'air. 3.3. Loi de Darcy pour l'Eau La loi de Darcy s'écrit : vw = - k

w

grad h w

avec : k w - tenseur des coefficients de perméabilité à l'eau (les coefficients de perméabilité dépendent de façon générale de l'indice des vides ou de la porosité, du degré de saturation, de la température et de la nature du fluide). La fonction adoptée pour décrire les variations des coefficients de perméabilité à l'eau est la même que celle utilisée par Alonso et al. [2], Nanda [10] et Abida [1]: 3 ⎡ S − Sru ⎤ k w = a10 αe ⎢ r ⎥ ⎣ 1 − S ru ⎦ avec - a, α- constantes ;

- e - indice des vides ;

- Sr - degré de saturation ;

- h w - charge hydraulique.

- Sru - degré de saturation résiduel ;

3.4. Loi de Darcy pour l'Air On admet que l'écoulement de l'air est également régi par la loi de Darcy : v a =− k a grad h a avec :

k a - tenseur des coefficients de perméabilités à l'air.

La fonction décrivant les variations des coefficients de perméabilité à l'air est celle utilisée par Alonso et al. [2], Nanda [14] et Abida [1]. γ c k a = b a [e (1 − S r )] μa avec - b, c - constantes adimensionnelles; - γ a - poids volumique de l'air ; - μ a - viscosité de l'air ;

- e - indice des vides.

A. Zadjaoui / Équations et Exemple de Calcul Hydrique dans les Sols Non Saturés

705

4. Conditions aux Limites et Conditions aux Limites Les conditions aux limites imposées aux inconnues (déplacements et pressions) sont des déplacements ou forces imposés au milieu global, des flux ou des charges d'eau imposés pour l'eau et des flux ou des charges d'air imposés pour l'air. Pour définir un état initial en équilibre, nous avons appliqué une procédure qui consiste à calculer l'état initial de la manière suivante : • on effectue à partir d'un état initial estimé mais pas nécessairement en équilibre un premier calcul avec pour seul chargement mécanique le poids du massif de sol, jusqu'à stabilisation des charges d'eau et d'air et des contraintes ; • on prend les résultats de ce calcul comme état initial des calculs ultérieurs en annulant les déplacements.

5. Choix d'un Principe Variationnel Le couplage entre le solide et l’eau est assuré par la présence de la pression d'eau dans la première intégrale et de la variation du volume du sol dans la deuxième intégrale. Le couplage entre le solide et l’air est assuré par la présence de la pression d'air dans la première intégrale et de la variation du volume du sol dans la dernière intégrale. Le couplage entre l’eau et l’air est assuré par la présence du coefficient de solubilité de l'air dans la dernière intégrale. L'application du principe variationnel choisi à l'analyse du comportement au cours du temps des massifs de sols élastoplastiques non saturés donne le système d'équations suivant, que l'on doit à résoudre dans le domaine fixe Ω:

6. Exemple de Calcul 6.1. Présentation du Test Un test a été réalisé sur la mise en eau d’un barrage construit sur des sols argileux compressibles. Le calcul permet d’évaluer l’évolution de la charge hydraulique à l’interface AB - Figure 1. Le Tableau 1 présente l’ensemble des données introduites dans le calcul déjà utilisé par Ould Amy et Magnan [12]. L’état initial correspond à l’état hydrostatique dans la fondation. L’origine de l’axe ascendant est constituée par le substratum. On suppose que la mise en eau est effectuée instantanément. 6.2. Analyse des Résultats On étudie l’évolution de la charge hydraulique à l’interface AB. En vue de comparer les résultats, on trace sur une même figure à différents temps représentatifs de la vie de l’ouvrage : t = 1 mois ; t = 6 mois, t = 1 ans et t = 10 ans (Figure 2). Ce test numériquement assez sévère (mise en eau instantanée) a relativement bien convergé. Lorsque le barrage est imperméable, la convergence est assurée en quatre itérations, la fondation est complètement saturée. Les conditions aux limites entraînent

706

A. Zadjaoui / Équations et Exemple de Calcul Hydrique dans les Sols Non Saturés

automatiquement l’établissement progressif d’un écoulement transitoire. On peut dire que la charge hydraulique à l’interface augmente au fur et à mesure de l’écoulement vers l’aval. Aux premiers instants, les résultats sont assez comparables. Six mois après la mise en eau, l’écart entre les charges hydrauliques augmentent sensiblement, surtout à l’amont de l’interface. La conclusion de ce test est que l’étanchéité freine l’infiltration de l’eau à l’intérieur du barrage et produit une perte de charge à l’interface. Des études complémentaires peuvent être entreprises pour analyser le couplage hydrodynamique afin d’analyser l’effet de la consolidation sur ce type de problème. ⎧dF = ∫Ω D e ε ij δε ij dΩ + ∫Ω γ a (Fe − 1)h a δε ij dΩ − ∫Ω γ w Fe h w δε ij dΩ − ⎪ − ∫Ω Fi δu i dΩ − ∫ Ti δu i dSσ − ∫Ω (Fe − 1)γ a z δε ij dΩ + ∫Ω Fe γ w z δε ij dΩ = 0 ⎪ Sσ ⎪ ∂ε ij ⎪ ∂h T δh w dΩ − ∫Ω nγ 2w (g 2 + g 1Fe ) w δh w dΩ + ⎪dFw = ∫Ω γ w ng 1 m 1 D e + S r δ ij ∂t ∂ t ⎪ ∂h a ⎪ δh w dΩ + ∫Ω k w h w , j γ w δh w ,i dΩ + + ∫Ω n γ a γ w (g 2 + g 1 Fe ) ⎪ ∂t ⎪ + ∫S γ w φ w δh w dS V = 0 ⎪ w Vw ⎪⎪ ∂ε ij ⎨ T ⎪dFa == ∫Ω γ a ρ a n (H − 1)m 1 g 1 D e + (1 − S r + HS r ) ∂t δ ij δh a dΩ − ⎪ ∂h ⎪ − ∫Ω γ a γ w [ρ a n (H − 1)g 1Fe + ρ a n (H − 1)g 2 ] w dΩ + ⎪ ∂t ⎪ ∂h ∂ h a ⎪ + ∫Ω γ a2 ρ a n (H − 1)g 1Fe δh a dΩ + ∫Ω γ a2 ρ a n (H − 1)g 2 a δh a dΩ + ⎪ ∂t ∂t ⎪ ∂ h ⎪ + ∫Ω γ a2 C a n (1 − S r + HS r ) a δh a dΩ + ∂t ⎪ ⎪ + ∫Ω γ a k a h a , j δh a ,i dΩ + ∫Ω γ a Hk w h w , j δh a ,i dΩ + ∫S γ a φ a δh a dS V = 0 a ⎪⎩ Va

(

)

[

]

et auquel il faut ajouter les conditions aux limites et les conditions initiales.

7. Conclusion Cet article a décrit de façon sommaire les équations qui nous ont servi à développer un modèle numérique pour l'analyse du comportement couplé des massifs élastoplastiques non saturés. Ce modèle a été implanté dans le code de calcul aux éléments finis TFAP. Malgré des hypothèses simplificatrices, la description des processus d’écoulement en non saturé est en général très complexe, car souvent ils donnent lieu à des variations de l’état hydrique du sol pendant l’écoulement. Ces variations impliquent des relations complexent entre la teneur en eau, la succion et la conductivité hydraulique. La non saturation du milieu est directement liée à knsat, qui lui-même dépend de la teneur en eau volumique. Par contre l’évolution dans le temps est lié au coefficient d’emmagasinement (coefficient de consolidation de la théorie de la consolidation). L’application présentée ici nous ont permis d’appréciér le degré de validité de notre programme. Finalement, l’ingénieur doit être en mesure d’estimer toutes les conséquences de l’évolution de la charge hydraulique au sein de l’infrastructure surtout lorsqu’un nouveau régime d’écoulement n’est pas encore atteint.

A. Zadjaoui / Équations et Exemple de Calcul Hydrique dans les Sols Non Saturés 26 24 22 20

B

A

18 16 14 12 10 8 6 4 2 0

0

5

10

15

20

25

30

35

40

Figure 1. Interface AB entre le sol et le barrage Tableau 1. Caractéristiques de l’argile Parameter Poids volumique de l’eau γw Porosité Compressibilité de l’eau βw Coefficient de perméabilité Teneur en eau volumique Paramètre……………...Ar Paramètre………………Br Paramètre………………Cr Paramètre………………Dr

Unite kN/m3 …….. kPa-1 (m/jour) …….. …….. …….. …….. ……..

Figure 2. Evolution de la charge hydraulique à l’interface AB : a-t = 1 mois ; b-t = 6 mois, c-t = 1 ans ; d-t = 10 ans.

Valeur 10 0,4 10-3 10-4 θsat 10-3 04 10-3 04

707

708

A. Zadjaoui / Équations et Exemple de Calcul Hydrique dans les Sols Non Saturés

Références [1] Abida H. (1992). La modélisation des sols non saturés - analyse numérique. Thèse de Doctorat, Ecole Nationale des Ponts et Chaussées, Paris, 271 pages [2] Alonso A.E., Batlle F., Gens A., Lloret .A. (1988). Consolidation analysis of partially saturated soils. Application to earthdam construction. Proceedings, 6th International Conference on Numerical Methods in Geomechanics, Innsbruck, Balkema, pp. 1303-1308. [3] Alonso A.E., Gens A., Josa A. (1990). A constitutive model for partially saturated soils. Géotechnique 40, n°3, pp. 405-430 [4] Bishop A.W., Donald I.B. (1961). The experimental study of partly saturated soils. Géotechnique, vol. 13, n°3, pp. 177-197. [5] Fredlund D.G. (1989). Discussion leader's report. Proceedings, 12th International Conference on Soil Mechanics and Foundation Engineering, Rio de Janeiro, vol. 5, pp. 2861-2878. [6] Fredlund D.G., Morgenstern N.R. (1976). Constitutive relations for volume change in unsaturated soils. Canadian Geotechnical Journal, n°13, pp. 261-276. [7] Gatmiri B., Tavakoli S., Moussavi S., Delage P. (1995). Numerical approach of elastoplastic consolidation of unsaturated soils. Proceedings of the first International Conference on Unsaturated Soils, /UNSAT'95 Paris/, France, vol 2, pp 1057 - 1064. [8] Gens A., Vaunat J., Ledesma A. (1995). Analysis of hydratation of an engineered barrier in a radioactive waste repository scheme using an elastoplastic model. Proceedings of the first International Conference on Unsaturated Soils, /UNSAT'95 Paris/, France, vol 2, pp 1057 - 1064. [9] Jennings J.E.B., Burland J.B. (1962). Limitations to the use of effective stresses in partly saturated soils. Géotechnique, 12, n°2, pp. 125-144. [10] Nanda A. (1989). Finite element analysis of unsaturated coupled flow and deformation. Rapport CERMES, Ecole Nationale des Ponts et Chaussées, 18 pages. [11] Matyas E.L, Radhakrishnan A.S. (1968). Volume change characteristics of partially saturated soils. Géotechnique, vol. 18, pp. 432-448. [12] Ould Amy M., Magnan J.P: (1991). Modélisation numérique des écoulements et des déformations dans les barrages de terre construits sur des sols mous. Collection «Etudes et recherches des LPC », série Géotechnique, n°10. 145 pages.

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 709 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-709

Processus de la consolidation des sols peu cohérents saturés Mohamed Salou DIANE & Salou DIANE Centre de Recherche Scientifique de Conakry-Rogbanè Direction Nationale du Génie Rural, Conakry Abstract. L’étude du processus de la consolidation des sols peu cohérents saturés sous l’action des charges dynamiques est dictée par le développement intense de la construction des ouvrages civils et industriels sur des sites récupérés, le plus souvent, par remblayage hydraulique. L’étude a permis de résoudre les problèmes plans de la consolidation du sol totalement liquéfié contenant des sous-couches de sols peu perméables non liquéfiés et de sols à couches multiples totalement liquéfiés.

Mots-clés : Sol peu cohérent, pression interstitielle, consolidation.

Introduction Les sols peu cohérents, en particulier les sables fins et limons, sont fréquemment utilisés pour l’acquisition de nouveaux territoires en vue du développement des villes. Dans plusieurs cas, le remblayage de ces sols se fait sous l’eau, ce qui conduit à la formation d’une structure meuble, donc peu stable, de ces sols. Pour assurer leur stabilité et diminuer les tassements susceptibles de se produire en leur sein, il s’avère nécessaire de les consolider. Pour ce faire on a souvent recours aux procédés du compactage. Pour de vastes étendues de terrains constitués de sols meubles, on peut utiliser l’explosion de charges d’explosifs relativement petites à cet effet. C’est pourquoi l’étude des sols peu cohérents saturés sous l’action de l’explosion est un sujet d’actualité. L’une des particularités de la méthode [1,2,3,4,5,6,7,8,9] est l’élargissement considérable de son domaine d’utilisation et le large diapason des types de sol utilisés : des enrochements aux argiles, limons et autres résidus industriels, loess et même aux argiles.

1. Principales prémisses physiques Sous l’action du choc et de l’explosion [10] dans des sols relativement peu perméables tels que les sables, il se produit le processus de leur densification accompagné de l’expulsion de l’eau de leurs pores. La particularité du processus de consolidation de la couche de sol sans cohésion et complètement liquéfiée est la présence de deux frontières mobiles (r) entre la partie redéposée de la couche où il existe des contacts entre les grains et le sol continuant à demeurer dans l’état de liquéfaction complète. La variation essentielle de la porosité, c’est-à-dire le compactage du sol, se produit à la

710

M.S. Diane and S. Diane / Processus de la consolidation des sols peu cohérents saturés

limite entre ces zones. Ainsi il se produit une diminution progressive du domaine de liquéfaction complète du sol. Par ailleurs dans la zone de redépôt du sol à structure nouvelle, il se produit un compactage supplémentaire sous l’action du poids propre de la couche de sol redéposée dont la hauteur augmente progressivement. A la frontière de ces zones caractéristiques, dans une couche élémentaire de sol en instance de consolidation les conditions de continuité s’écrivent :

udt = (u + Δr − u − Δr ) = [n1 ( z ) − n 2 ( z )]dr

(1)

Où u – vitesse de filtration de l’eau expulsée des pores du sol ; u +∆r , u-∆r – vitesses de filtration de l’eau au-dessus et au-dessous de la couche élémentaire de sol respectivement, n1, n2 – porosités du sol dans l’état de liquéfaction et après son compactage respectivement. Dans les limites de la zone de déplacement des particules du sol par suite du compactage, la vitesse de filtration est déterminée par la loi de Terzaghi-Guertzévanov :

u−

n dH v = −K 1− n dz

(2)

Où v – débit des particules de sol à travers l’unité de surface (analogue à la vitesse de filtration ; K – coefficient de filtration du sol. En tenant compte du fait que dans le cas d’un problème plan u = -v, la relation (2) peut s’écrire sous la forme :

u = − K (1 − n)

dH dz

(3)

Ces équations déterminent le processus du redépôt des particules du sol dans les limites de la partie de la couche se trouvant dans l’état de liquéfaction. On sait que [10], le problème de la consolidation d’une couche homogène liquéfiée a été résolu par VA Florin [11] et sous l’action d’une surcharge ou d’un drainage par P L Ivanov [12] Dans les conditions naturelles, particulièrement dans les dépôts alluvionnaires, ainsi que dans plusieurs cas de remblayage de territoires, il arrive souvent que les couches de sols sableux renferment d’autres couches de sols de composition granulométrique variée, ainsi que des couches intermédiaires d’argiles peu perméables et non liquéfiables. Lors du compactage de ces types de sols préalablement saturés, il se crée artificiellement des courants de filtration constants, ascendants ou descendants, toutes choses pouvant avoir une influence sur le processus de la consolidation après la liquéfaction du sol. Ci-dessous nous examinons quelques cas de figures.

2. Consolidation de strates de sols peu perméables totalement liquéfiés 2.1. Consolidation d’une couche de sol totalement liquéfiée contenant des souscouches de sol peu perméable Lors de la liquéfaction complète de la partie sableuse d’une couche de sol d’épaisseur h(Figure 1), l’épure initiale (t=0) de la pression interstitielle dans l’eau a presque la forme triangulaire. Dans ce cas, dans un souci de simplicité, on considère que la masse

711

M.S. Diane and S. Diane / Processus de la consolidation des sols peu cohérents saturés

volumique de toutes les couches de sol est constante sur toute la profondeur. La prise en compte de la variabilité de γbzb de couche en couche ne présente pas de difficulté. Après son passage dans l’état de liquéfaction, commence le processus de redépôt des particules du sable dans la partie inférieure avec la formation de la frontière r 1. Pour u1> un il se produit la violation de la balance d’eau et la formation, sous la sous- couche peu perméable inférieure, d’une couche d’eau. Comme un>u3,ce qui équivaut à vn
r1 = K1

γ S 1 − n1 t γ n1 − n2

(4)

Les vitesses de filtration u2, u3 et un sont déterminées par les relations :

u 2 = K 22 (1 − n 22 )( u 3 = K 21 (1 − n 21 )

u n = K n (1 − n n )(

γ bzb ΔH + ) γ h2

γ bzb γ γ bzb

γ

(5) (6)

+

ΔH ) r1

(7) La valeur de la pression ∆H (Figure 1) peut être déterminée de la condition de continuité du courant de filtration, c’est-à-dire : un = u2 (8) En remplaçant (5) et (7) par leur valeur dans (8), on obtient

γ bzb ΔH γ ΔH + ) = K n (1 − n n )( bzb − ) : d’où on tire : h2 r '1 γ γ γ γ K n (1 − n n ) bzb − K 22 (1 − n 22 ) bzb γ γ ΔH = 1 1 (9) + K n (1 − n n ) K 22 (1 − n 22 ) h2 r '1 K 22 (1 − n 22 )(

A la frontière r’1 l’équation de continuité [3] doit être satisfaite sous la forme :

(u 3 − u 2 ) dt = ( n 21 − n 22 ) ) dr '1

(10)

Ou

[K 21 (1 − n 21 ) − K 22 (1 − n 22 )] γ bzb γ

−

dr ' ΔH K 22 (1 − n 22 ) = (n 21 − n 22 ) 1 r '1 dt

(11)

712

M.S. Diane and S. Diane / Processus de la consolidation des sols peu cohérents saturés

Figure 1. Répartition des pressions dans la couche de sable liquéfiée contenant des sous-couches de sols imperméables

Posant K21(1-n21) ≈ K22(1-n22), mettant (10) dans (11) et après transformation, on obtient :Adt = (B+Cr’1)dr’1 , où :

A=

[K 22 (1 − n 22 ) − K n (1 − n n )]γ bzb hn K 22 (1 − n 22 ) n 21 − n 22

B = K 22 (1 − n 22 )hn γ

;

C = K n (1 − n n )γ

D’où, en tenant compte du fait que pour t=0, r’1=0 et en intégrant, on obtient :

t=

B C 2 r '1 + r '1 2A A

(12)

Ainsi le courant de filtration ascendant, prenant naissance dans la sous-couche imperméable, ralentit le déplacement des frontières r’1 et r’’1. Pour une perméabilité très faible de la couche d’argile, c’est-à-dire Kn=0, l’équation (12) se transforme en l’équation (4) si on tient compte du fait que K21(1-n21) ≈ K22(1-n22). Dans ce cas le processus de consolidation des particules du sable se produira dans chaque couche, indépendamment l’une de l’autre. Sous toutes les sous-couches argileuses, il se formera une couche d’eau qui, après le redépôt du sable, va filtrer pendant longtemps à travers les sous-couches d’argile. L’épure des pressions interstitielles dans l’eau à cet instant est indiquée sur la Figure 1 par la ligne brisée en pointillé (t 1). Connaissant, d’après les données de sondage, la disposition des couches d’argiles isolées et leur puissance, le temps de consolidation ou, plus exactement, le temps de liquidation des couches intermédiaires d’eau se détermine par la relation (13) :

tC =

hn ∑ h1 (n1i − n 2i ) k n (H 2 − H1 )

(13)

713

M.S. Diane and S. Diane / Processus de la consolidation des sols peu cohérents saturés

Où hi – épaisseur des couches de sables sous-jacentes liquéfiées ; H 2 et H1 – pressions de l’eau à la base et au sommet de la couche d’argile respectivement (déterminées, pour chaque couche argileuse, selon la partie en pointillée t f de l’épure des pressions sur la Figure 1). La durée maximale du tassement complet de la couche du sol à compacter sera déterminée par la couche d’argile pour laquelle la valeur tc calculée par la relation (13) sera la plus grande. Dans le cas où la couche de sol complètement liquéfiée contenant des couches de sol peu perméables non liquéfiées est disposée sur une fondation drainante (Figure 2), le dépôt des particules dans la partie inférieure de la couche s’accompagnera d’une expulsion d’eau supplémentaire du côté de la fondation drainante avec la vitesse de filtration u0 et l’équation de continuité des phases solide et liquide du sol dans la couche frontière d’épaisseur dr1 s’écrit :

u1 dt − u 0 dt = (n11 − n12 )dr1

(14)

γ dH 0 ; u1 = K 11 (1 − n11 ) bzb γ dz dH 0 H (h − r1 ) − H (0) 1 [γ bzb (h − r1 ) + hγ ] = = D’où dz r1 γr1

Où u 0 = − K 12

Remplaçant u0et u1 par leur valeur dans (14), on obtient : Adt + Br1dt = r1dr1

A=

(15)

γ bzb K 12 1 1 (γ bzb h + γh) B = [K11 (1 − n11 ) − K 12 ] n11 − n12 γ n11 − n12 γ

Après intégration de (15) on obtient la relation caractérisant le mouvement, en fonction du temps, de la zone de sol redéposé (r1), sous la forme :

Figure 2 : Distribution des pressions dans une couche de sable liquéfiée contenant des sous-couches de sol non liquéfiées et peu perméables reposant sur une fondation drainante

714

M.S. Diane and S. Diane / Processus de la consolidation des sols peu cohérents saturés

t=

1 [r1 − A ln 1 + B r1 B B A

]

(16)

La présence d’une fondation drainante accélère considérablement le rétablissement des contacts entre les particules de sol dans la couche inférieure. 2.2.Compactage d’une couche de sol totalement liquéfiée chargée par un sol non liquéfié Dans une couche de sol totalement liquéfiée chargée par un sol non liquéfiée (Figure 3) de masse volumique γH, la distribution des pressions peut s’écrire sous la forme :

H0 =

γ H h1 + γ S ( h − z ) +z γ

(17)

Par suite de la présence de la couche de sol saturée non liquéfiée, le processus du redépôt des particules du sol ou le rétablissement des contacts entre les particules peut se produire simultanément aussi bien dans la partie inférieure (r 1) que dans la partie supérieure (r2) Les vitesses de filtration le long de la couche, déterminées par la relation de Darcy-Guertsévanov, s’écrivent comme suit:

u1 = − K 11 (1 − n11 )

dH 3 dH 1 dH 2 ; u 2 = − K 12 (1 − n12 ) ; u H (1 − n H ) dz dz dz

(18)

Où K11 et K12 – coefficients de perméabilité des parties liquéfiée et redéposée de la couche de sol respectivement; KH - coefficient de filtration de la couche de charge ; n 11, n12, nH –porosités des couches de sol liquéfiée, redéposée et servant de charge respectivement. Dans les limites de la partie totalement liquéfiée, c’est-à-dire pour r 1≤z≤r2, de l’équation (17) on peut écrire:

γ γ dH 1 = − S + 1 = − bzb dz γ γ

(19)

Et dans les limites de la partie supérieure des couches redéposée et chargée :

dH 1 H (h) − H (r2 ) dH 3 H (h + h1 ) − H (h) ; = = dz h − r2 dz h1

H ( r2 ) =

γ H h1 + γ s ( h − r1 ) + r2 ; H (h + h1 ) = h + h1 γ

(20)

(21)

La valeur de la pression H(h) peut être trouvée de la condition de continuité du courant de filtration, c’est-à-dire u2 = uH . Si dans cette équation on remplace u2 et uH par leur valeur et en tenant compte de l’équation (21), on obtient :

715

M.S. Diane and S. Diane / Processus de la consolidation des sols peu cohérents saturés

Figure 3 : Distribution des pressions dans une couche de sable liquéfié supportant un sol non liquéfié

H (h) =

K 12(1 − n12 )h1 H (r1 ) + K H (1 − nH )(h − r' 2 )(H (h + h1 )

(22)

K12 (1 − n12 )h1 + K H (1 − nH )(h − r2 )

Dès lors, l’équation de continuité (10) dans une couche élémentaire dr 2 attenante à la partie supérieure du sol déjà redéposé, en tenant compte de la couche de sol servant de charge h, prend la forme : (u2-u1)dt = (n11 – n12)dr2 . Introduisant les équations (18) dans cette dernière et l’intégrant en posant t=0 et r2 = h (condition initiale), on obtient l’expression déterminant la durée du déplacement de la frontière r 2 :

t=

⎤ 1 ⎡⎛ A A ⎞ ⎜ − C ⎟ ln 1 + (h − r2 ) − (h − r2 )⎥ ⎢ B ⎣⎝ B B ⎠ ⎦

(23)

Ainsi dans la couche de sol liquéfiée, se déplacent simultanément l’une vers l’autre deux frontières (r1) et (r2) de sol redéposé après la rencontre desquelles les phénomènes de liquéfaction et de compactage prennent fin. Pour déterminer la durée du compactage tC, du sol (temps de rencontre des frontières du sol redéposé (r1) et (r2)), on peut considérer que : K12(1- n12) = K11(1- n11) = K(1-n) ; dans ce cas, en procédant comme précédemment, on obtient l’épaisseur de la couche de sol redéposée sous la forme :

r ' 2 = h − r2 = C 2 − 2 At − C

(24)

Le temps pendant lequel le sol demeure dans l’état de liquéfaction se détermine de la condition de rencontre des frontières du sol redéposé, c’est-à-dire : r1 + r ' 2 = h (25) Remplaçant les relations ( 4) et (24) par leur valeur dans (25) nous obtenons :

t1

γ bzb K (1 − n11 ) + C 2 − 2 At − C = h γ n11 − n12

(26)

716

M.S. Diane and S. Diane / Processus de la consolidation des sols peu cohérents saturés 2

Ou A' t − Bt + D = 0 où A' = (

(27)

γ bzb K11 (1 − n11 ) γ bzb 2 K 112 (1 − n11 ) 2 ; (h + h1 ); B = 2 ) γ (n11 − n12 ) γ (n11 − n12 ) 2

D = h 2 + 2h

K 11 (1 − n11 )h1 K H (1 − n H )

Résolvant l’équation (24) en tenant compte du fait que K H
tC =

γ (h + h1 )(n11 − n12 ) γ bzb K 11 (1 − n11 )

(28)

2.3. .Compactage de massifs de sol multicouches totalement liquéfiés Dans le cas où la couche supérieure est plus perméable que la couche inférieure (Figure 4), il se produit trois frontières mobiles r1), (r’1), (r’2) de sol redéposé. Dès lors, les vitesses de filtration u1, u2, u3 et u4, par analogie aux relations (5), (6) et (7), s’écrivent :

u1 = A *1

γ bzb γ

γ bzb ΔH + ) γ r'2 γ ΔH ) u 3 = A2 ( bzb − r'2 γ γ u 4 = A * 2 bzb γ u 2 = A1 (

(29) (30)

(31) (32)

Où A*1 = (1- n11)K11 A1= (1- n12)K12 A2= (1- n22)K22 ; A*1 = (1- n21)K21 et la pression DH, en tenant compte de la condition de continuité du courant de filtration, s’écrit :

ΔH =

γ bzb A2 − A1 r ' 2 h2 γ A2 r ' 2 + A1h2

(33)

Et l’équation d’équilibre de la balance d’eau en considérant A1= A*1, s’écrit :

A1

ΔH dt = ( n11 − n12 ) dr ' 2 r'2

(34) En remplaçant, dans (34), ∆H par sa valeur tirée de (33) et intégrant l’équation obtenue en considérant que pour t=0, r’2=0 et C=0, nous obtenons :

717

M.S. Diane and S. Diane / Processus de la consolidation des sols peu cohérents saturés

Figure 4 : Répartition des pressions dans un massif de sol liquéfié à deux couches dans le cas où la consolidation de la couche supérieure prend fin avant celle de la couche inférieure

t=

γ bzb (n11 − n12 )0,5A2 2 γ n11 − n12 r' 2 r' 2 + γ γ bzb A2 A1 A1 ( A2 A1 )h2

(35)

Ainsi les frontières r1 et r’2 continuent de se mouvoir l’un vers l’autre jusqu’à r1 =h1. Lors de la rencontre de ces frontières, c’est-à-dire r1+r2=h1, le processus de la consolidation de la couche de sol prend fin et la condition de la rentrée en contact des frontières prend la forme de l’équation (36)

(

γ bzb 2 A12 2 2 A1 2 2 A1 γ A ) t − 2 t − (h2 + h1 ) bzb t + h1 (h1 + 2h2 1 ) = 0 (36) 2 γ γ B1 A2 B1 B1

t1, 2

⎡ A ⎤ B1γ ⎢h2 + h1 ± (h2 + h1 ) 2 − 2h2 h1 1 ⎥ A2 ⎥⎦ ⎢⎣ = A1γ bzb

(37)

Les durées du compactage des couches h1 et h2 sont déterminées par les relations :

t h1 =

(n11 − n12 )h1 (n − n12 )h2 ; t h 2 = 11 u1 u2

(38)

Le temps de filtration de l’eau de la lame sera :

t fe =

(u1 − u 2 )t h 1

u2 γ Où u 1 = K 11 (1 − n11 ) bzb γ

(39) ;

u 2 = K '11 (1 − n'11 )

γ bzb γ

Additionnant (38) et (39), on obtient le temps nécessaire à la consolidation complète de la couche de sol donnée par l’équation (40)

718

M.S. Diane and S. Diane / Processus de la consolidation des sols peu cohérents saturés

tC =

1 [(n11 − n12 )h1 + (n'11 − n'12 )] u2

(40) Dans le cas de la disposition de la couche sur une fondation drainante, le redépôt des particules dans la partie inférieure des couches s’accompagnera d’une expulsion d’eau supplémentaire du côté de la fondation drainante et la durée du déplacement de la frontière r1 est déterminée par la formule (16) Conclusion L’étude du processus de la consolidation des sols peu cohérents saturés sous l’action de l’explosion a permis de résoudre les problèmes plans de la consolidation d’une couche de sol contenant des sous-couches de sols peu perméables et non liquéfiables et de sols multicouches totalement liquéfiées. Elle montre aussi que dans ces cas, les épures de la pression interstitielle dans l’eau peuvent se distinguer sensiblement des formes triangulaires et trapézoïdales, caractéristiques du cas des couches de sol homogènes liquéfiées. Remerciements : Ce travail a été réalisé à la chaire de Mécanique des Sols et Fondations de l’Université Technique d’Etat de Saint Petersburg, Russie. Que le chef de chaire, Feu Prof P.L. Ivanov et ses collaborateurs trouvent ici l’expression de ma profonde gratitude. Références bibliographiques [1] C. DAMITIO , La consolidation des sols sans cohésion par explosion, Paris , Construction (25) , 3, pp.100 – 108, 9, pp. 292 – 302, (2), 7/8; pp. 264 – 271, (27), 3, pp.90 – 97, 1970-72 [2] N.DEMBISKI & all, Compactage des fonds marins sableux à l’explosif, Col Int sur le compactage, Vol. 1, Paris, 1980, pp.301 – 305. [3] P.DONCHEV, Compaction of loess by saturation and explosion, Proc. Int. Conf. on compaction, Paris, Avril 1980 pp. 313 – 317 [4] A. SIMON, Comparaison de l’efficacité de trios procédés de compactage d’une couche de sol de grande épaisseur sur remblai hydraulique. Col. Int. sur le compact age, Paris Avril 1980, Vol. 1, pp. 363 – 368. [5] V.A. FLORIN, Le phénomène de la liquéfaction et les procédés de consolidation des sols sableux meubles et saturés, en russe, Annales de l’Académie des Sciences de l’URSS No 6, 1952, pp. 824 – 833 [6] P.L. IVANOV, Compaction of cohesion less soils by explosion, Proc. 6th, ICSMFE, Montreal, 1965, Vol. 3, pp. 325-329 [7] N. KISIELOWA &All, Consolidation dynamique des vases à l’explosif, Coll Int. Sur le compactage, Paris, Avril, 1980, Vol. 1, pp. 295-299 [8] N. KISIELOWA & R. Osiiecimski, A method of underwater compaction of cohesionless soils, Coll Int. Sur le compactage, Paris, Avril, 1980, Vol. 1, pp. 455-460 [9] L. MENARD, La consolidation dynamique des sols de fondations, Revue des sols et fondations, 1974, No 320, pp.79-82 [10] M.S. DIANE&S. DIANE, Liquéfaction et consolidation des sols sableux sous l’action des charges dynamiques, Proc. 12th Reg. Conf on Soil Mech. and Geot. Eng Durban 25-27 Oct 1999, pp. 151- 165 [11] V.A. FLORIN, Du problème de la liquéfaction des sables fins propres et saturés, Construction hydrotechnique, No 7, 1951, pp. 34 – 36. [12] P.L. IVANOV, Liquéfaction des sols sableux, « Gosenergoizdat »M. – L, 1962 , 260 p. [13] M.S. Diané & S. Diané, Phénomène de la liquéfaction et consolidation des sols peu cohérents sous l’action du choc et de l’explosion, Proc.of the Int. Conf on Geot. Eng. Hammammet 24 – 26 Mars 2007.

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 719 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-719

The African Regional Conferences as an Indicator of Research Trends in South Africa Philip PAIGE-GREEN CSIR Built Environment, Pretoria, South Africa

Abstract. The 2011 African Regional Conference on Soil Mechanics and Geotechnical Engineering is the 15th in the series of 4 yearly conferences. The majority of the early conferences were held in sub-Saharan Africa with significant contributions from South African researchers and practitioners. Recently, more of the conferences have been held in northern Africa with smaller South African participation. An analysis of the origin and type of papers in terms of Academic, Research, practitioner and international over the past 13 conferences has been carried out and shows interesting trends of research in relation to the available research funding in South Africa. It is clear that there is a declining trend in fundamental and basic geotechnical research in South Africa. Recommendations to overcome this problem are suggested. Keywords. Geotechnical, research, university, CSIR

Introduction This 2011 African Regional Conference on Soil Mechanics and Geotechnical Engineering (the title has changed a number of times over the past 65 years) is the 15 th in the series of 4 yearly conferences. The majority of the early conferences were held in sub-Saharan Africa with significant contributions from South African researchers and practitioners. Recently, more of the conferences have been held in northern and francophone Africa with smaller South African participation. An analysis of the origin and type of papers in terms of countries of origin and the field of activity of the authors since the first conference has been carried out and the results are briefly discussed in this paper. The role of the South African Council for Scientific and Industrial Research (CSIR) has been specifically assessed and analyzed. The trends in research in relation to the available research funding in South Africa are reviewed, in the context of the apparent declining trend in fundamental and basic geotechnical research in South Africa. Recommendations to overcome this problem are suggested. The author has been privileged to attend 6 of the past 9 ARC conferences held since 1975 and had papers included in 7 of the Proceedings.

720

P. Paige-Green / The African Regional Conferences as an Indicator of Research Trends

1. History of African Regional Conferences The first African Regional Conference on Soil Mechanics and Foundation Engineering (ARC) was held in Pretoria in 1955. Although no copy of the Proceedings of this could be located (copies of all other Proceedings were obtained), reference was made to the conference in the preface to the 5th Conference in Luanda. The 1st conference apparently had three themes and a total of 12 papers were included in the Proceedings [1]. The ARCs have been held every 4 years since the first one apart from the 1980 conference in Ghana which was held after a five year delay and then the 1991 ARC in Lesotho which was held only 3 years after the Lagos, Nigeria ARC, apparently to coincide the ARCs better with the relevant International Conferences. In the early years, it appeared to be an “un-written” rule that the ARC was hosted by the home-country of the immediate past Vice-President for Africa of the International Society for Soil Mechanics and Foundation Engineering (ISSMFE) [2]. This appears to have fallen away after about 1980. The history of the years and locations of the ARCs is summarized in Table 1. Table 1. Years and locations of ARC’s ARC number and [reference] 1 2 [3]

Year

Location

1955 1959

3 [4] 4 [5]

1963 1967

5 [1] 6 [6] 7 [2]

1971 1975 1980

Pretoria, South Africa Lourenco Marques, Mozambique Salisbury, Rhodesia Cape Town, South Africa Luanda, Angola Durban, South Africa Accra, Ghana

8 [7] 9 [8] 10 [9] 11 [10] 12 [11] 13 [12]

1984 1987 1991 1995 1999 2003

Harare, Zimbabwe Lagos, Nigeria Maseru, Lesotho Cairo, Egypt Durban, South Africa Marrakesh, Morocco

14 [13]

2007

Yaoundé, Cameroon

Theme and comments Not known No theme African soils present their own problems Soil forming processes and associated engineering problems Tropical and subtropical soils No theme African problem soils in engineering practice (South Africans barred) No theme No theme (South Africans barred) Geotechnics in the African environment No specific theme Geotechnics for developing Africa The involvement of geotechnical engineering in infrastructure development in Africa Soils of Africa

A number of the ARCs had specific overall themes and these are also summarized in Table 1. In 1980, just a few weeks before the conference held in Accra, Ghana, the South African delegation was informed that visas would not be issued to them. As a result, a separate conference for South Africans was held in Pretoria, at which the South African contributions to the Accra ARC were presented as well as 17 additional papers, published separately [14]. In 1988 South Africans were barred from attending the Conference in Lagos, Nigeria for political reasons. However, there was one South African at the conference, who was the Vice-President for Africa of ISSMFE at the time and he was given special dispensation to attend.

721

P. Paige-Green / The African Regional Conferences as an Indicator of Research Trends

2. Statistics of papers A number of the ARCs have published the papers presented in a volume that has been available at the opening of the conference as well as a second volume including the Keynote presentations, summaries of workshops held and, in some cases, papers that were submitted too late for inclusion in the pre-conference Proceedings publication. In this review only the papers included in the first volume have been analyzed. A summary of the papers presented at the various ARCs is provided in Table 2. This includes the total number of papers included in Volume 1 of the Proceedings, the number prepared by non-African authors, the number of papers prepared by South African University staff and the number of papers prepared by CSIR employees in the fields of Building Technology and Road Technology. Certain papers are more general and have been classified into one of these two groups on the basis of the CSIR unit in which the author was based. Where a CSIR author was not the main author (often included with a University colleague) the paper was taken as originating from the CSIR. The remaining papers are those presented by practitioners, researchers or academics mainly from the country in which the ARC was being held but also from other African countries. It is clear from the contents of the various proceedings that in almost every case (possibly barring Lesotho) the majority of papers, not-unexpectedly, originate from the home country. Table 2. Statistics of papers in terms of numbers (and percentages). ARC Year 1955

Total papers 12

International

South African

Other African

SA Universities ?

?

?

?

CSIR Building Road ? ?

1959

30

2 (7)

13 (43)

15 (50)

2 (7)

6 (13)

2 (7)

1963

42

5 (12)

24 (57)

13 (31)

9 (21)

8 (19)

5 (12)

1967

47

5 (11)

27 (57)

15 (32)

7 (15)

10 (21)

3 (6)

1971

37

18 (49)

13 (35)

6 (16)

2 (5)

5 (14)

2 (5)

1975

36

1 (3)

30 (83)

5 (14)

6 (17)

12 (34)

3 (8)

1980

49

9 (18)

22 (45)

18 (37)

5 (10)

5 (10)

0

1984

63

24 (38)

19 (30)

20 (32)

0

7 (11)

0

1988

74

16 (22)

3 (4)

55 (74)

1 (1)

0

0

1991

50

7 (14)

40 (80)

3 (6)

6 (12)

8 (16)

4 (8)

1995

101

51 (50)

12 (12)

38 (38)

0

1 (1)

2 (2)

1999

84

17 (20)

40 (48)

27 (32)

1 (1)

2 (2)

13 (15)

2003

99

33 (33)

15 (15)

51 (52)

0

2 (2)

0

2007

53

9 (17)

4 (8)

40 (75)

0

4 (8)

0

It should also be mentioned here that session themes at the conferences have changed over the years. The first conference had 3 themes but that has increased and changed with time. A full list is not included in this paper but trends showing the emergence of new technologies are obvious. Aspects such as the use of waste materials, geosynthetics and environmental issues have become important in recent conferences and the ongoing debate regarding education and skills development has been a major issue at recent ARCs.

722

P. Paige-Green / The African Regional Conferences as an Indicator of Research Trends

3. Analysis of paper statistics The paper statistics have been analyzed in various ways to highlight some of the trends. 3.1. Origin by region Figure 1 shows a plot of the regions of origin of the papers. It is clear that until about 1980, papers from South Africa generally dominated the conferences. Since then, unless the conference was held in South Africa, papers from other African countries have become more numerous. At the conferences in Harare (1984), Cairo (1995) and Marrakesh (2003) international papers have been dominant. Other than for Harare, this could be expected as Cairo and Marrakesh are in closer proximity to Europe and Asia/Middle East than to southern Africa. Figure 2 shows an encouraging trend in the growth of the number of papers from Africa (excluding South Africa) since 1988, indicative of the rapid development of geotechnical engineering in Africa in the past two decades or so. 60 No of papers

50 40

International

30

South African

20

Other African

10 2007

2003

1999

1995

1991

1988

1984

1980

1975

1971

1967

1963

1959

0

Year

Figure 1. Main areas of origin of papers. 60 No of papers

50 40 30 20 10 0 1959 1963 1967 1971 1975 1980 1984 1988 1991 1995 1999 2003 2007 Year

Figure 2. Number of papers from Africa (excluding South Africa).

3.2. Role of CSIR In the early years, the CSIR played a dominant role in geotechnical research in South Arica. This was primarily through the National Institutes of Road (NIRR) and Building

P. Paige-Green / The African Regional Conferences as an Indicator of Research Trends

723

Research (NIBR), with a smaller contribution from the rock mechanics section of the National Mechanical Engineering Research Institute (NMERI). Figure 3 shows the overall contribution of the CSIR over the years in terms of the numbers of papers prepared by the two major units. Although NIRR (and its successors, NITRR, Transportek and Built Environment) has generally contributed the major share over the years, the declining trend in number of papers in both research areas and the significant drop after 1991 is quite noticeable (and worrying). CSIR has long been considered the major research organization for geotechnics, but the move from being fully Government funded to carrying out more commercially funded research has had a severe negative effect on research outputs suitable for publication. As the investment in site investigations is usually seen (incorrectly) as one of the lesser important construction activities, the funding of geotechnical research seems to suffer from the same apathy. If it is assumed that the mantle of Geotechnical research has moved more to the universities, this should be reflected in an increase in the numbers of papers emanating from South African universities. Figure 4 includes the contributions from Universities. Although there were a significantly larger number of papers from universities in 1999 (Durban), the overall trend remains negative. This does not bode well for the future of innovative geotechnical engineering in South Africa, particularly when it is noted that only one South African university has full time Professors in geotechnical engineering. This is also reflected in the decrease in the number of geotechnical researchers at the CSIR, from more than 15 in 1976 to only 4 in 2011. It should also be noted that the CSIR had two state-of-the-art soil mechanics laboratories and one rock mechanics laboratory in 1975 and now only has a single downscaled soil testing laboratory and a small rock testing facility. 20 15 Road

No of papers 10

Building

5 0 1959

1967

1975

1984

1991

1999

2007

Year

Figure 3. Number of papers from CSIR.

724

P. Paige-Green / The African Regional Conferences as an Indicator of Research Trends

25 20 SA Universities

15 No of papers

Road 10

Building

5 0 1959 1967 1975 1984 1991 1999 2007 Year

Figure 4. Number of papers from CSIR and South African Universities.

4. Recommendations As is the case with most public research organizations in the modern world, funding is limited. A reduction in basic/fundamental research work often results in the application of new technologies, usually developed overseas under different environmental and technological conditions, without the necessary fundamental support for local conditions. Failures, which could have been avoided, often result leading to criticism (and often abandonment) of the technology. The need for ongoing research to support the implementation of new technologies in a developing environment cannot be overemphasized and this requires adequate funding. Although some of these technologies are proprietary and should be funded by the suppliers or agents, many are independent and in the national interest and should be funded by Government. In South Africa, it is important that Government’s funding of Research and Development (R&D) be increased from the current level of 0.92% of Gross Domestic Product (GDP) to the proposed level of 1.5% as recommended by Government in 2010 [15]. This would be more in line with the levels achieved in other developing countries such as China. It is also important that geotechnical engineering facilities and staff at South African universities be expanded and improved. This will facilitate an increase in quality of post-graduate research. Good progress has been made in the establishment of the Centre for Excellence in Foundation Engineering as a joint operation among CSIR BE, the University of Pretoria and a private company, URD, with various government and parastatal bodies also being involved. If adequately funded and utilized, the Centre will assist in the coordination and advancement of post-graduate research in the field of geotechnical engineering in sub-Saharan Africa.

5. Conclusions The past 14 African Regional Soil Mechanics and Foundation Engineering Conferences (ARC) have provided excellent platforms and opportunities for advancing geotechnical engineering in Africa. These have gone from strength to strength with greater participation form African countries as well as more interchange of ideas with the international geotechnical community. A worrying issue, however, is the steady

P. Paige-Green / The African Regional Conferences as an Indicator of Research Trends

725

decline of geotechnical research in South Africa, shown by the reduction in contributions to ARCs from the CSIR, which is not being replaced by contributions from South Africa Universities. It can thus be concluded that geotechnical research in South Africa is suffering from severe financial cut-backs, a trend that needs to be reversed in the medium to long-term national interest. In general, African governments have indicated their intentions in this regard, but the fruits of increased investment in research are still to be seen.

Acknowledgements This paper has been prepared as part of the ongoing research at CSIR Built Environment and is published with permission of the Executive Director, CSIR Built Environment.

References [1] Laboratorio de Engenharia de Angola (LEA). Proceedings of the 5th Regional Conference for Africa on Soil Mechanics and Foundation Engineering, Luanda, Angola, LEA, Luanda, 1971. [2] M.D. Gidigasu, A.A. Hammond & J.O. Gogo, (Eds). African problem soils in engineering practice, Proceedings of the 7th Regional Conference for Africa on Soil Mechanics and Foundation Engineering, Accra, Ghana, Balkema, Rotterdam, 1980. [3] Anon. Proceedings of the 2nd Regional Conference for Africa on Soil Mechanics and Foundation Engineering, Lourenco Marques, Mozambique, Empresa moderna, Lourenco Marques, 1959. [4] Rhodesian Institution of Engineers (RIE). African soils present their own problems. Proceedings of the 3rd Regional Conference for Africa on Soil Mechanics and Foundation Engineering, RIE, Salisbury, Rhodesia, RIE, Bulawayo, 1963. [5] A. Burgers, J.S. Gregg, S.M. Lloyd & A.D.W. Sparks (Eds), Soil forming processes and associated engineering problems. Proceedings of the 4th Regional Conference for Africa on Soil Mechanics and Foundation Engineering, Cape Town, South Africa, Balkema, Cape Town, 1967. [6] P.J. Pells & A.Mac G. Robertson (Eds). Proceedings of the 6th Regional Conference for Africa on Soil Mechanics and Foundation Engineering, Durban, Balkema, Rotterdam, September 1975. [7] J.R. Boyce, W.R. MacKechnie & K Schwartz (Eds). Proceedings of the 8th Regional Conference for Africa on Soil Mechanics and Foundation Engineering, Harare, Zimbabwe, Balkema, Rotterdam, 1984. [8] J.O. Akinmusuru, S.S Malomo & E.A. Mesida (Eds). Proceedings of the 9th Regional Conference for Africa on Soil Mechanics and Foundation Engineering, Lagos, Nigeria, Balkema, Rotterdam, 1987. [9] G.E. Blight, A.B. Fourie, I. Luker, D.J. Mouton & R.J. Scheurenberg. (Eds). Geotechnics in the African environment. Proceedings of the 10th Regional Conference for Africa on Soil Mechanics and Foundation Engineering, Maseru, Lesotho, Balkema, Rotterdam, 1991. [10] Egyptian Geotechnical Society (EGS). Proceedings of the 11th Regional Conference for Africa on Soil Mechanics and Foundation Engineering, Cairo, Egypt, EGS, Cairo, 1995. [11] G.R. Wardle, G.E. Blight & A.B. Fourie (Eds). Geotechnics for developing Africa. Proceedings of the 12th Regional Conference for Africa on Soil Mechanics and Foundation Engineering, Durban, South Africa, Balkema, Rotterdam, 1999. [12] M. Sahli, L Bahi & R Khalid (Eds). The involvement of geotechnical engineering in infrastructure development in Africa. Proceedings of the 13th Regional Conference for Africa on Soil Mechanics and Foundation Engineering, Marrakech, Morocco, Megamix, Casablanca, 2003. [13] M. Bouassida, S.J. Ejezie, E. Rust, P. Nouanga & E.M. Kana (Eds). Soils of Africa. Proceedings of the 14th Regional Conference for Africa on Soil Mechanics and Foundation Engineering, Yaounde, Cameroun, Comitẻ National des Gẻotechniciens du Cameroun, Yaounde, 2007. [14] A.A.B. Williams, (Ed). South African Geotechnical Conference, 1985. A supplement to the Proceedings of the 7th Regional Conference for Africa on Soil Mechanics and Foundation Engineering held in Accra in June 1980. Balkema, Rotterdam, 1985. [15] Parliamentary Monitoring Group. http://www.pmg.org.za/report/20101117-department-sciencetechnology-research-development-rd-survey-results- (Accessed 27/12/2010)

726 Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-726

Geotechnical Investigations: Overregulated or Under-investigated? Tony A’BEARa,1 and Louis VAN ROOY b Bear GeoConsultants, Johannesburg, South Africa b Department of Geology, University of Pretoria, South Africa a

Abstract. The current regulatory environment encourages a minimalistic approach to geotechnical investigations allowing inexperienced or would-be practitioners to carry out, ostensibly, sufficient investigations. The paper argues that the approaches adopted over the preceding decades should be adhered to and that terrain evaluation techniques and the construction of the geological model should be of over-riding importance when carrying out geotechnical investigations. Keywords. Geological model, terrain evaluation, geotechnical standards

Introduction Currently a process is underway, and to some extent has been completed, to create South African National Standards (SANS) which establish the approach to be adopted and the minimum requirements for geotechnical site investigations. These standards include geotechnical investigations for residential urban development, characterization of and development on dolomite land as well as the description of geological materials obtained from drilling or in situ investigations. Geotechnical site characterization should lead to the development of a geological model [1] with the eventual development of the geotechnical model in which the geological model is combined with the properties of the geological units, obtained during the laboratory or field testing phase, and the geological processes.

1. Historic overview Engineering geological and geotechnical site investigations have been carried out on a formal basis in South Africa since the first dam site investigations were conducted in the early 1920’s. Since then the approach, methodologies and procedures have developed over time resulting in a well-established set of guidelines for soil profiling, borehole core logging and the description of rock chips from percussion boreholes. In addition to these guidelines a Code of Practice on Geotechnical Site Investigations [2] is available. 1

Corresponding Author: Tony A’Bear, Bear GeoConsultants, PO Box 28334, Kensington, 2101, South Africa; Email: [email protected]

T. A’Bear and L. van Rooy / Geotechnical Investigations: Over-Regulated or Under-Investigated? 727

The early approach to site investigations was strongly influenced by the terrain classification methodology that was widely applied in the United Kingdom. This was advertised and debated within the local geotechnical fraternity at the Third Symposium on Terrain Evaluation and Data Storage [3]. Terrain evaluation techniques were adopted by the National Institute for Transport and Road Research (NITRR) for road centreline investigations [4]. A number of provinces in South Africa included this approach in their manuals for centreline investigations noting specifically that where terrain evaluation was carried out the requirements for an equally spaced testing approach could be modified. This implicitly recognises that there is a cost and time advantage to using terrain evaluation techniques over a simplistic grid evaluation approach. Partridge and others [5] started the large scale application of the land system approach and characterised the Kyalami, Paardekraal and other land systems. Dolomite surface stability investigations were mainly limited to specific structure footprints, but also included regional gravity surveys covering large parts of the Far West Rand and the Centurion area south of Pretoria. The scenario supposition technique for use on dolomitic terrain [6] was developed during the 1990s and has become a well-established method to evaluate the results of site investigations on dolomite.

2. Present Site Investigation status There are a number of guidelines on geotechnical site investigations for housing projects which have been developed and published by the National Department of Housing [7] and the National Home Builder’s Registration Council [8]. Various documents also govern development on dolomite with guidelines available from the Council for Geoscience, South African Institute for Engineering and Environmental Geologists (SAIEG), the Council for Geosciences (CGS) and the Department of Public Works. The South African Bureau of Standards is presently setting up standards to regulate the approach to geotechnical investigations. The current regulatory environment pays lip service to terrain evaluation techniques but essentially sets up a number of criteria whereby an investigation is deemed to be satisfactory or not. In the experience of the authors, it is this element of the regulations that are regularly used by the authorities to test the validity of an investigation. So, for example, the investigation may not include an air photo interpretation or any other means of establishing terrain units, but, provided that the prescribed number of test pits and laboratory tests has been carried out, the report on the investigation is considered to be acceptable in terms of procedure and passes muster. This minimalistic approach encourages a grid system to be used whereby the minimum number of test pits is set out on a grid evenly spaced across the site. It is obvious that this approach does not allow terrain units to be properly explored or boundaries to be properly established. In addition, the larger the site, the greater the grid spacing becomes, allowing for terrain units to be missed altogether in the investigation. The collection of data on a grid system to obtain the prescribed minimum number of data points and samples not only defeats the advantages of the land system approach, but may also allow incompetent persons to adhere to the minimum requirements, possibly overlooking crucial indicators to on site geological constraints.

728 T. A’Bear and L. van Rooy / Geotechnical Investigations: Over-Regulated or Under-Investigated?

3. Appraisal of the current situation The current development of standards by authorities through a number of guideline documents, but most especially the SANS standards, seems to be an attempt to over regulate the established site investigation approach by ignoring the scientific skills and experience of the geotechnical or engineering geological professional and their ability to decide on the approach and methodology to be adopted, for instance, the number of test pits, depth of drilling, number and type of soil or rock tests. The development of the geotechnical model is usually a phased process where the site geological model is initially constructed, the distribution and properties of the different geological materials on site are then determined from which the ground model is presented to the design team. The ultimate goal is to determine the ground behaviour under the proposed development with changes in stress conditions, etc. The development of standards for geotechnical site investigation in South Africa is long overdue. The present use of guidelines and existing methodologies or approaches as reference works for these standards documents is flawed for a number of reasons. The most obvious shortcoming is the unscientific basis on which many issues are being addressed in the standards. Very little research and new development has been forthcoming over the past twenty years or so in the field of engineering geological site investigation. Many reasons can be put forward for this situation but that is not the purpose of this paper. It is not clear as to how the minimum numbers of tests was arrived at for the current set of standards and it is highly debatable whether this minimalistic approach has any real validity in the light of earlier work done on establishing the veracity of terrain evaluation techniques. The complexity of the terrain in terms of its geology should be the governing factor in terms of the numbers of tests applied to a site, whether the number of test holes, in-situ tests or laboratory tests. It is a serious concern that the current standards may open the field to inexperienced and inappropriately trained geological scientists who will only have to apply the minimum number of tests per hectare to adhere to the standard. The requirements put forward in these standards will also be legally binding and may lead to unnecessary legal action if less than or more than the stipulated investigation points and tests are used to develop the geotechnical model. These standards may also prevent or limit further development and research in sound engineering geological site investigation practice and methodologies. Existing techniques were well established during the 1970s and 80s and provide an ultimately sound, rational basis for the application of engineering geology and geotechnical engineering in ground investigations.

4. Conclusions It is proposed by the authors that guidelines, codes of practices, a scientifically sound approach and, very importantly, experience should govern geotechnical site investigation rather than legally binding rules and numbers of tests. The benefit of applying good practice and experience in site investigations far outweighs the rigid standard approach. Ultimately both cost and information benefits are to be had from applying the well established techniques of terrain evaluation and geological modelling.

T. A’Bear and L. van Rooy / Geotechnical Investigations: Over-Regulated or Under-Investigated? 729

The ultimate development of the geotechnical model may, in many instances, be satisfied with the minimalistic or standards type approach. However, due to our varied geology and special geomorphological history, especially compared to most other continents, the application of the minimalistic, grid type approach as advocated in current standards is not likely to be adequate.

References [1] Knill, J. Core values: the first Hans-Cloos lecture. Bulletin of Engineering Geology and the Environment. Vol. 62. pp 1-34, 2003. [2] South African Institution of Civil Engineering – Geotechnical Division. Site Investigation Code of Practice, 1st Edition, 2010. [3] Third Symposium on Terrain Evaluation and Data Storage, Division of Soil Mechanics and Foundation Engineering, South African Institution of Civil Engineering, February 1975, Kyalami Ranch, Transvaal, South Africa. [4] National Institute for Transport and Road Research. Geotechnical and soil engineering mapping for roads and the storage of materials data. Technical Recommendations for Highways: TRH-2, Pretoria CSIR, 1978. [5] T.C. Partridge. 4th land system and land type classifications: Comparisons and applications in the Southern African context. 4th Terrain Avaluation and Data Storage Symposium. Midrand Johannesburg, August 1994. [6] D.B. Buttrick, A. van Schalkwyk, R.J. Kleywegt and R. Watermeyer, Proposed method for dolomite land hazard and risk assessment in South Africa. Journal of the South African Institution of Civil Engineering 43(2) 2001, pp 27-36. [7] National Department of Housing, Geotechnical Site Investigations for Housing Developments, Project Linked Greenfield Subsidy Project Developments, Generic Specification GFSH-2. September 2002. [8] National Home Builders Registration Council (NHBRC). Home Building Manual, Parts 1, 2 & 3, Revision 1, February 1999.

730 Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-730

Challenges to Geotechnical Engineering practice in the urbanization of the city of Accra, Ghana J. K. ODDEI Architectural and Engineering Services Limited, Accra, Ghana

Abstract/Résumé. The city of Accra is the focal point of economic activity in Ghana and it is anticipated that recent discovery of oil in the country will increase the gross domestic product (GDP), which will bring about rapid infrastructural development. Further it has experienced rapid industrialization and expansion in the manufacturing and commercial sectors since 1960; the decline of agricultural activities in the rural areas coupled with the boom in the service sector in this primate city has propelled migration. The growth of Accra’s population over a thirty year period has been phenomenal. This has led to the steady rate of overcrowding and business opportunities that has influenced demands on civil infrastructures. The implications of the urbanized status have given rise to five important engineering issues, which geotechnical engineering has key roles to play to provide the needed support for Accra’s infrastructural growth. This paper seeks to identify the issues confronting the geotechnical industry in the wake of this urbanization with regards to infrastructural development based on the available logistical and other supports. Keyword. Accra Ghana, urbanization, economic activities, Geotechnical engineering practice, civil infrastructure

Introduction Ghana is located in West Africa, which borders Côte d’Ivoire to the west, Togo to the east and Burkina Faso to the north; the south is the Gulf of Guinea. It falls between latitudes 4oN and 11.5oN; longitudes 3.11o and 1.11o. It is segmented into ten (10) administrative regions (Figure 1). These regions are subdivided to give a total of 170 assemblies – 6 metropolitans, 40 municipals and 120 districts. The metropolitan areas are considered urban. These are Accra, Tema, Kumasi, Cape Coast, Tamale and Sekondi-Takoradi. Accra city is the capital of Ghana and the most urban. Between the period of 1960 and 2000, Ghana has experienced steady and speedy urbanization with the statistics of population living in urban areas as follows: 23% in 1970 to 32% in 1984; 32% in 1984 to 44% in 2000. Accra city’s urbanized status has given rise to rapid infrastructural challenges and steady growth for which geotechnical engineering has recognizable roles to play to provide the needed support. This paper has identified five issues of engineering concern with the urbanization of Accra and the needed support expected of the geotechnical industry. The available fundamental resources and other support for the geotechnical industry coupled with population data for Accra are also presented and discussed to buttress the subject. The objective of this paper is to identify the issues

J.K. Oddei / Challenges to Geotechnical Engineering Practice in Urbanization of Accra, Ghana

731

confronting the geotechnical industry in the wake of Accra’s urbanization based on available resources and other supports with regards to infrastructural development. The methodology adopted to obtain the necessary information was by sending questionnaire and personal interview via phone and in person.

1. The Accra city Accra city lies on 5o33’N and 0o15W of Ghana with a total land size of 200km2 compared to the regional size of 3,245km2 (about 6.2%). It has become a home for local and foreign industries, manufacturers, finance, marketing, insurance, tourism and workers in general. These economic and social indicators are the primary source for the population swell in Accra. Again foreign direct investments from oil revenues are expected to contribute positively to some civil infrastructure. Accra is underlain by four geological settings namely: Dahomeyan system, Togo series, Accraian formation and recent unconsolidated deposit. These rock systems are associated with inferred and observed faults. Southern Ghana, the Accra city in particular, is noted for seismic activities [1]. Recent work [2] has evaluated that peak ground acceleration (PGA) for Accra is between 0.14g and 0.57g.

Figure 2: Map with Accra city highlighted

Figure 1: Map of Ghana

Table 1: City Population data for a sixteen year period at census years City Accra Other 5No

City Population (1984) 969,195 Metros

1,004,561

City Population (2000) 1,658,937 1,885,952

732

J.K. Oddei / Challenges to Geotechnical Engineering Practice in Urbanization of Accra, Ghana

Population data, obtained from Ghana Statistical Service and the ghanadistricts.com, on urbanized cities in Ghana for a sixteen year interval is presented at Table 1. The data shows that the population growth of Accra city increased by 71.2% with 4.4% annual growth rate. The corresponding population growth and growth rate were 87.7% and 5.5% respectively for the other five cities. Population densities for Accra during those census years were approximately 4846 No/km2 and 8295 No/km2 respectively. Again Accra represented 49.1% and 46.8% respectively, in terms of urbanized zones in Ghana. At least 25% increase in population of Accra after the final census update in 2011 is predicted. The statistics presented show that Accra continues to have a strong trend towards urbanization compared with other areas in Ghana. Infrastructural systems, in Accra, such as road, bridges, buildings and others of civil engineering concern had seen fair to moderate expansion over the years. The coping of these infrastructural facilities with the rapid population growth is becoming very challenging for the city of Accra. Five issues, for the purpose of this paper, of geotechnical concern have arisen due to Accra’s urbanization. The issues are: • • • • •

Waste management - process for selection and design of waste sites, develop more technically attainable standards and introduce new waste containment technologies. Infrastructure development and rehabilitation - geotechnical engineering will play key role in an interdisciplinary coordination in the area of site characterization and performance appraisal. Construction efficiency and novelty – the need to execute geotechnical works effectively with innovative equipment and techniques. Mitigation of natural hazards – geotechnical engineering will support better land planning to effectively reduce losses in lives and monetary cost, during natural hazards. Forensic issues – geotechnical engineers would be relied on as expert witnesses to assist in arbitrations for construction cases.

2. Geotechnical Engineering in Ghana The Ghana Geotechnical Society (GGS) is a group under the Ghana Institution of Engineers (GhIE) which was established under NRCD1 143. Available data on practicing Engineers in Ghana indicates that there are thirty-six Geotechnical Engineers in the country. About 86% of this number is involved in consultancy while the remaining 14% are mostly found in the mines. The geotechnical engineer density in the consultancy sector, within Accra is 0.16 numbers of engineers per km2. Available information on fixed capital resources – probing equipment and laboratory – for the geotechnical industry in Ghana is presented in Figures 2 and 3 respectively. In Figure 2 the total number of probing equipment is fifteen (15). Hollow stem auger represents about 13%; cable percussion and rotary rigs represent about 40% and 27% respectively; other probing tool represents 20%. The preceded statistics on 1

National Redemption Council Degree

J.K. Oddei / Challenges to Geotechnical Engineering Practice in Urbanization of Accra, Ghana

733

probing equipment represent 60% holdings for private firms; 27% and 13% for public firms and academia respectively. It could be remarked that between 80% and 85% of the probing equipment are at least ten (10) years old. In Figure 3, private and public stake in laboratory resource is about 38% each with academia representing 24%. Laboratories in the academic sector are located at least 200km from Accra. With the public laboratories, two are located in Accra and the other is about 200km away. The private ones are located in Accra. In addition, the strengths (capacities) of the institutional laboratories are presented in Figure 4. Private participation ranges between 10% and 25%; public is between 40% and 80%, and that of academia ranges between 60% and 80%. Information was gathered by responses to questionnaire sent to, and a few phone calls, practicing engineers, and from database of GGS and GhIE. Laboratory Resource & their strength

Probing Equipment 3.5 33

3

3

3

10%-25% 40%-80%

2.5 2

Number

Number

2 2 1.5 1

1

1

60%-80%

2

1

0.5 0

0

000

0 Private Hollow stem auger Rotary coring

Public

Institution

Academia

0 Private

Cable Percussion Other probing tool

Figure 3: Probing equipment resource

Laboratories

Public Institution

Academia

Figure 4: Laboratory Resource

3. Discussions The five identified elements means that a well organized urban-system infrastructure that would uphold steady and functioning environment including ensuring quality of life is needed to cope with the urbanization challenge of Accra [3]. The role of geotechnical engineering in this quest cannot be overlooked. The capacity that the geotechnical industry needs to grip the identified elements is in the area of human and capital resources including the appropriate legal framework to support engineering practice in Ghana. The human resource, measured by the geotechnical engineer density of 0.16 numbers of engineers per km2, can be said to be inadequate. Therefore the Ghana Geotechnical Society (GGS) is encouraged to promote interest drive especially in the tertiary institutions to get potential engineers on board. The existing human resource must enhance their professional capabilities to attack the identified elements. In addition, geotechnical engineers must have a coordinated approach with other earth related professionals to assist in the design and implementation of disaster management

734

J.K. Oddei / Challenges to Geotechnical Engineering Practice in Urbanization of Accra, Ghana

plan for Accra. The preparation of a seismic risk map including improving the economic cost of site investigation could be the result of this coordination. Innovative and modern probing equipment are presently needed for geotechnical investigation and monitoring. It is evident that the private sector is improving on its support for the geotechnical industry. It is also clear that academia depends on the consultancy sector for further training, especially field investigations. Current equipment status does not match demands and most of them are old and do not provide the necessary efficiency. Laboratories in the private and especially in the public sectors are inadequately resourced with up to date gadgets for accurate geotechnical testing. Those of academia and just a few in the public sector are appreciably better and in most cases they are relied on for very sensitive tests. However, the cost and risk of transportation of samples through long distances to laboratories poses issues of concern. The strength of the laboratories also relates to personnel with inadequate continual training to enhance their performance. Institutional investment for probing equipment and modern laboratory including personnel development is indispensable for the geotechnical industry. The challenge has always been sources of funding considering low levels of savings of firms and high interest rates on loans from commercial banks in Ghana, which hover between 23.5% and 41.6% [4]. An alternative approach to funding is for the geotechnical industry to present persuasive proposals to plant pool outfits, which may have the financial means to purchase probing equipment for hiring. The impact of the geotechnical industry on the country would be felt with improved legal structures for general engineering practice. The NRCD 143 legal cover was put in place in the early 70’s to deal with the establishment and management of registered engineers qualified to practice in Ghana and to advance science and engineering to the benefit of society. This was in an unconstitutional regime and therefore new legislative instrument to support engineering practice in the current constitutional era is a right step. This will allow engineers to be effective in national decision making and be responsible in their practice. Though its consideration is at the parliamentary level, for now the geotechnical industry continues to make little impact toward the urbanization challenge.

4. Conclusion Population data over a sixteen year period for Accra city, the capital of Ghana, has shown positive trend in migration that has caused overpopulation. The level of infrastructure to cope with the situation has seen little to fair expansion making Accra an urbanized area. This has given rise to five identified elements of civil engineering concern which geotechnical engineering has critical roles to play. The capacities, human and capital resources couples with the legal framework, to effectively attack these roles are challenging to the geotechnical industry to provide the needed support.

J.K. Oddei / Challenges to Geotechnical Engineering Practice in Urbanization of Accra, Ghana

735

Acknowledgement I acknowledge the peer review by Ing Dr. Charles Akayuli, Ing Michael ObengKonadu and Ing Samuel Ahiamadi. In addition, much appreciation goes to the engineers who responded to the questionnaire sent.

References [1] Ghana Geological Survey, Bulletin No.13: Geological Map of Accra Area, the Geological Survey of Ghana, Accra – Ghana. [2] P.E. Amponsah et al, Seismic ground motion and hazard assessment of the Greater Accra Metropolitan Area, the Abdus Salam International Centre for Theoretical Physics preprint 55 (2008), [3] National Academies Press, Geological and Geotechnical Engineering in the New Millennium: Opportunities for Research and Technological Innovation, the National Academies Press, Washington DC – USA. [4] I.J. Mensah, G. Mensah, Nation’s Interest Rates Highest in Africa, Public Agenda Newspaper (2010).

736 Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved. doi:10.3233/978-1-60750-778-9-736

Soil Improvement through the Utilization of Agricultural Residues from Nigeria N.L. OBASI and E.B. OJIOGU Enugu State University of Science and Technology, Nigeria.

Abstract. This work deals with soil stabilization using ash obtained from both open air and oven burning of the rice husk collected from Abakaliki Rice Mill in Ebonyi State of Nigeria. The soil samples were collected from the proposed borrow pit site at Onu-Ogba Nike near Enugu, Enugu State, Nigeria at depths of 0.5 - 1.5 m. The laboratory tests were conducted in accordance with [8, 9] at the Soil Mechanics Laboratory of Civil Engineering Department, Enugu State University of Science and Technology, Enugu, Nigeria. The soil samples were prepared for stabilization by weighing out ten equal samples from the natural soil and gradually adding to each sample progressively 0 %, 2.5 %, 5 %, 7.5 %, 10 %, 12.5 %, 15 %, and 17.5 % by weight of the rice husk ash. The results obtained indicate that 59 % of the natural soil samples passed the sieve No. 200 with average plasticity index and CBR values of 21 and 8 % respectively. Similarly, the result of the blended soil samples indicate that the plasticity index (PI) values decreased from 28 – 10.4 % while the corresponding California bearing ratio (CBR) values increased from 9.1 – 25 % after which it started decreasing to 22 %. The result of this study suggests that the stabilized soil sample with 15 % rice husk ash gave the highest CBR value of 25 % with a PI value of 10.4 % indicating its high potential for soil stabilization. The adoption of rice husk ash for soil stabilization would encourage the effective disposal and utilization of the abundant rice husk waste in most developing countries. It could also offer some employment opportunities to rural dwellers. Key words. Rice husk ash, soil stabilization, plasticity index, CBR

Introduction Soils as major part of road construction material also provide the required load bearing characteristics for other engineering structures. Some of these soils which are good for gravel roads occur in tropical countries of the world, including Nigeria [1]. However, there could be situations where some of these soils may contain substantial amount of clay minerals such that the strength and stability of such soils cannot be guaranteed under certain loads particularly in the presence of moisture. Soils can only perform such services if they possess sufficient strength and stability under some adverse loading and climatic conditions to avoid undesirable structural failures. It is desirable in highway road construction to have soil materials with sufficient strength to withstand the applied wheel loads in addition to possessing good drainage properties under severe weather conditions without exhibiting swelling and shrinking qualities. It is however sometimes difficult to obtain such good soil materials with all the above qualities due to the effect of geological and climatic factors [2].

N.L. Obasi and E.B. Ojiogu / Soil Improvement Through the Utilization of Agricultural Residues

737

Generally, soil improvement could be achieved either by modification or stabilization depending on the engineering purpose. Soil modification through the introduction of additives to the soil could enhance the improvement of the strength to withstand applied loads [1]. Some of the widely used admixtures include lime, bitumen and cement. Among the soil improvement additives currently in use, cement and lime are the most prominent but are either expensive or not widely available [3]. Agricultural residues including rice husk are composed of organic constituents such as cellulose, lignin, fibre, protein and fat [4]. In addition they contain some minerals which include silica, aluminum, iron and oxides of trace elements introduced into the soil through fertilizer applications [4]. In this study, it is the ash obtained from pyroprocessing the residues that are of special interest. [5] reported that over eight million tones of rice husk are available each year during the processing of rice while in Ebonyi State of Nigeria, about sixty four thousand tones of rice husk are produced yearly during the processing of paddy rice [6]. Generally, some of the rice husks are used as fuel while the greatest portion is dumped as waste by-product constituting environmental problems to the society. Report from [7] shows that the largest quantity of rice produced comes from the developing countries and constitute major pollution source discharging variety of gaseous, liquid and solid wastes into the environment. The most effective techniques for reducing waste are usually to minimize the processes of waste generation and recycling of the materials where applicable. Recently, concern has risen over the availability of future landfill spaces because they are becoming difficult to site due to high cost of construction and operation, more stringent regulations, diminishing land availability and public opposition due to high rate of urbanization. The disposal of rice husk from milling plants has been a problem to farmers and this can easily be appreciated by observing the endless mounts of rice husk around farming communities. The current waste disposal option adopted by most farmers is setting on fire the mounts of rice husk which will smolder for several months as the open uncontrolled burning process is slow and without considerable reduction in size even as it continues to burn with its attendant air pollution. The need to further investigate the possibility of sourcing locally available materials such as ash from rice husk as a soil stabilization agent for road construction and effective method of waste management can not be overemphasized.

1. Materials and Method of Testing The materials used for this study include soil samples, rice husk and water. The soil samples were collected from a proposed borrow pit site at Onu-Ogba Nike, Enugu State, Nigeria at depths of 0.5 - 1.5 m while the rice husks were collected from Abakaliki Rice Mill, Ebonyi state, Nigeria. The water was collected from the available tap water in the Civil Engineering Laboratory, Enugu state University of Science and Technology. The rice husks were burnt in a furnace at temperatures of 400 – 600 oC and the resulting ash was allowed enough time for cooling before mild grinding and sieving. The composition of the ash is presented in Table 1. The soil samples for stabilization were prepared after classification by weighing out eight equal samples from the natural soil collected from the field and adding to each sample of the measured rice husk ash in the order of 0 %, 2.5 %, 5 %, 7.5 %, 10

738

N.L. Obasi and E.B. Ojiogu / Soil Improvement Through the Utilization of Agricultural Residues

%, 12.5 %, 15 %, and 17.5 % by weight respectively. These samples with varying percentages of rice husk ash content by weight are the stabilized soil samples. The laboratory tests conducted on the soil samples include particle size distribution, Atterberg Limits, Specific gravity, Compaction and Californian Bearing Ratio (CBR). The engineering properties of the soil samples were determined in accordance with [8] while the stabilization tests were performed in accordance with [9]. The California Bearing Ratio (CBR) tests were conducted as specified by [10] for roads and bridge works. Table 1. Chemical Analysis of Rice Husk Ash Compound Rice Husk Ash composition at 600°C (%) SiO2 85.51 CaO 1.25 MgO 2.50 Fe2O3 0.92 Al2O3 0.75 Na2O 3.16 K2O 3.80 Loss on ignition

Rice Husk Ash at 400°C (%) 86.56 2.97 2.14 0.63 3.65 2.70 1.25 17.7

Rice Husk Ash obtained from open air burning 89.50 2.00 1.32 0.18 4.32 1.48 1.38

After [6]

2. Results and Discussion •

Composition of Rice Husk Ash The ash residues obtained from the burnt rice husk samples were analyzed chemically as presented in Table 1 from which it could be seen that rice husk ash clearly has a very high percentage of silica (SiO2) 86.56 %, and low percentage of calcium Oxide (CaO) 1.97 %, indicating some potential qualities for cementatious action with soil. These values are however different from the results obtained by [11], SiO2 - 67.3 % and CaO - 1.36 % and those obtained by [4] from six countries SiO2 : 86.9 – 94.0 % and CaO: 0.25 – 1.40 %. The variation of the silica and calcium dioxide may be attributed to the incinerating process, conditions, geographical location and varieties of the planted paddy rice.

•

Soil Classification The soil engineering properties test results of the natural soil and the stabilized samples are shown in Table 2 The results indicate that the soil is predominantly clay soil with about 63.50 % by weight passing sieve No 200; liquid limit of 60.5 %, plastic limit of 25.5 % and plasticity index of 35.9 %. This implies that the sample by [12] classification system is A – 7 – 6 soil which is likely to exhibit very poor characteristics with high potential for volume change during different seasons of the year. The soil samples investigated by [13] showed A – 7 – 6 soil with 77 % passing the Sieve No 200, Liquid limit – 49.5, plastic limit – 24.4 and plasticity index of 25.1 while those of [14] showed A – 2 – 6 soil sample with 15 % passing the Sieve No. 200, liquid limit – 35.20, plastic limit – 18.13 and plasticity index of 17.07.

N.L. Obasi and E.B. Ojiogu / Soil Improvement Through the Utilization of Agricultural Residues

739

The particle size analyses result of the stabilized samples as shown in Table 2, reveals that the percentages by weight of samples that passed the Sieve No. 200 decreased with increase in the addition of the stabilizing material which are similar to those of [13] and [14]. As could be seen from Table 2, the liquid and plastic limit values of the samples increased as the rice husk ash addition increased which is similar to the results reported by [13] and [14]. The plasticity index values of the soil samples decreased as the rice husk ash addition increased until it gets to 15 % after which it showed an increase while those of [13] and [14] were at 8 % and 12 % respectively. Table 2. Engineering properties of stabilized and natural soil samples Sample No. L.L P.L P.I C.B.R. before (%) (%) (%) soaking (%) 0 % RHA 60.5 25.5 35.0 8 2.5 % RHA 55.3 23.1 32.2 8.5 5 % RHA 43.0 25.2 27.8 9.15 7.5 % RHA 40.3 24.1 16.2 11.05 10 % RHA 35.6 23.0 12.6 15.7 12.5 % RHA 30.7 20.5 10.2 20.2 15 % RHA 49.2 36.3 12.9 20.1 17.5 % RHA 56.6 35.4 21.2 13.8

% Passing sieve 200 63.80 64.50 61.65 52.64 50.41 42.74 35.00 34.63

Compaction characteristics Fig. 1 is the graphical presentation of the relationship of the Maximum Dry Density (MDD) and the stabilized soil samples. The relationship indicates that the MDD decreases with increase in the addition of rice husk ash.

Values of MDD of stabilized soils.

•

C.B.R. after soaking (%) 0.4 1.3 3.7 3.8 4.1 4.3 5.4 3.7

30 25 20 15 10 5 0 0

2

4

6 8 10 12 Percentage of RHA added to soil sample.

14

16

18

20

Fig. 1. Plot of MDD values for stabilized soils.

The decrease in the MDD could be attributed to the replacement of soil by the rice husk ash which has lower specific gravity (2.25) as compared to that of the soil (2.69). The trend is in line with [11], [13], [14], [15], and [16]. Coating of the soil by the rice husk ash may result to large particle formation with voids and hence less density. Rice husk ash serves as filler material in the soil voids. Increase in optimum moisture content (OMC) with increase in rice

N.L. Obasi and E.B. Ojiogu / Soil Improvement Through the Utilization of Agricultural Residues

husk ash contents resulted from the fact that addition of RHA reduced the quantity of free silt and clay fraction and coarser materials with larger surface areas were formed which require additional water to compact the soil blended with rice husk ash [13]. The mechanism of strength development in soil-rice husk ash stabilization is basically that of physical attributes with some form of chemical reaction brought about by the introduction of graded amorphous grain sizes to the soil clayey content thereby improving the strength of the clay mineral structures in the soil. The process is such that the smaller grain particles would fill up the intensities in the clay structure while the bigger particles would also fill up the range of sand which in turn helps to create densified composite soil structure. •

Strength characteristics The CBR is widely used for the design of base and sub-base course materials for road pavement in addition to the evaluation of the strength of the stabilized soils. The plot of CBR values against the stabilized soil samples is shown in Fig. 2. For the unsoaked samples, the CBR values increased with the addition of rice husk ash after which the values reached to a peak at the 12.5 % rice husk ash addition. It slightly dropped at 15 % and 17.5 % RHA addition. The initial decrease in the CBR is due to the reduction in the silt and clay content of the soil sample [8] which also reduces the cohesion. The increment in the CBR after 2.5 % RHA could be attributed to the gradual formation of the cementitious compounds between the RHA and CaOH contained in the soil. The gradual decrease in the CBR values after the 12.5 % RHA could be due to the excess RHA that was not mobilized in the soil – RHA reaction which probably occupied spaces within the sample resulting to a reduction in the ionic bond existing between the RHA – soil mixtures. Considering the soaked samples, the CBR values were similar to those of the unsoaked samples but vary with the result obtained by [11]. This indicates that water assists to a large extent the formation of the cementitious compounds between the soils’ CaOH and pozzolanic RHA. CBR values of stabilized soils (%).

740

25 20 15 CBR uns o aked

10

CBR s o aked

5 0 0

5 10 15 Percentage of RHA added to soil samples. Fig. 2. Plot of CBR values for stabilized soils.

20

N.L. Obasi and E.B. Ojiogu / Soil Improvement Through the Utilization of Agricultural Residues

741

3. Conclusions The natural soil sample has been classified as A – 7 – 6 soil on [12] classification system and is also clay of high plasticity (CH) according to unified system of classification (USC). The natural soil samples showed indications of poor quality construction material with plasticity index value of 35 % that obviously would require some kind of treatment to improve the engineering properties. The various stabilized soil samples obtained from the blending process showed substantial improvement in the engineering properties to a certain limit. This strongly suggests the importance of always performing trial mix designs before recommending the appropriate mix design ratios that could be used on a particular construction site. In this study, a mix design ratio of 4:0.175 yielded the maximum CBR value of 20.02 % which is an appreciable improvement of over 2.5 times that of the original soil sample of 8 %. Depending upon the soil type and results recorded in literature, the effective rice husk ash content for improving the engineering properties of the soil varies between 8 to 20 % and may require soil stabilization mix design procedure that would reflect various soil types. The use of the agricultural residues may reduce considerably the cost of road construction in addition to reducing the environmental hazards accruing from the inefficient waste management commonly practiced in most parts of the developing countries.

References [1] [2]

K.J. Osinubi, and I. Bajeh, Bituminous stabilization of laterite. Spectrum J. 1(2): (1994), 104-12. M.A. Mustapha, Effect of bagasse ash on cement stabilized laterite. Seminar paper presented at the Department of Civil Engineering, Ahmadu Bello University, Zaria, Nigeria. 2005. [3] A.M. Neville, Properties of concrete. 4th ed. (low-price ed.). Pearson education Asia publ., England, produced by Longman Malaysia. 2000 [4] A. Muthadhi, R. Anitha, and Kothandaraman, Rice Husk Ash – Properties and its uses: A Review, IE(I) Journal CV, Vol. 88, (2007) 50 – 56. [5] D.J. Cook, Using rice husk for making cement like materials, Appropriate technology 6, (1980) 9 – 11. [6] N.L. Obasi, and M.U. Alimba, 2006. The suitability of rice husk ash cement as construction material, NJOTECH, Vol. 26, No. 2, (2006) 98-105. [7] Bureau of Agricultural Economic Situation and Outlook, Rice, Australian Government Publishing Service, Canberra, (1983) 1 – 10. [8] BSI. Methods of testing soil for civil engineering purposes. (B.S.1377). British Standards Institute, London, England, 1990a.. [9] BSI. Methods of testing for stabilized soils (B.S.1924). British Standards Institute, London, 1990b. [10] Nigerian General Specifications. Roads and Bridge Works. Federal Ministry of Works, Lagos, Nigeria, 1997. [11] A. Musa, Potentials of rice hush ash for soil stabilization, Au. J.T. 11(4), (2008) 246 – 250. [12] AASHTO. Standard Specifications for Transportation Materials and Method of Sampling and Testing. Amer. Assoc. State Highway and Transportation Officials, Washington, DC, 1986. [13] F.O. Okafor, and N.U. Okonkwo, Effects of rice husk ash on some geotechnical properties of lateritic soil, Leonardo Electronic Journal of Practices and Technologies, Issue 15, (2009) 67 - 74 [14] M.D. Gidigasu, Laterite soil engineering: Pedogenesis and Engineering Principles, Elsevier, Amsterdam, The Netherlands, 1976. [15] S.A Ola, Stabilization of Nigeria lateritic soils with cement, bitumen and lime. Proc. 6th Reg. Conf. Africa on Soil Mechanics and Foundation Engineering. Durban, South Africa, 1975. [16] K.J. Osinubi. Evaluation of admixture stabilization of Nigeria black cotton soil. Nigeria Soc. Engin. Tech. Trans. 34(3): (1999) 88-96.

This page intentionally left blank

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 743 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved.

Subject Index abandoned dumpsite soil 293 acceptance specification 10 Accra Ghana 730 Africa 83 ageing 330 agglomeration 330 air-entry value 360 alluvial deposits 611 amplitude 152 anchor 215 anchored walls 215 argiles gypseuses 524 Artificial Neural Networks 580 auto-contrôle 524 axial load 221 back-analysis 679 bagasse ash 545 barrier 89 basal 507 bearing capacity 119, 127, 175 bedding errors 587 bentonite 71 Berea sand 376 berthing and mooring facility 145 bitumen stabilized materials 349 blending 382, 415 Blue Nile 580 bond stress 255 bored piles 119 British standard light 293 bulk modulus 337 California bearing ratio 389 capacité portante 158 capping 77 case history 40 cation exchange 89 cavern 605 cavités 524 CBR 396, 736 CCLs 104 charge critique 158 chaussée non revêtue 482 civil infrastructure 730

clay 175 clayey sand 669 closed form solutions 183 coal 593 coastal sediments of Lagos 145 coating 460 cofferdam 190 cohesion 249 collapse potential 653 combined pile raft foundation 243 community 476 compacted clay linings 104 compaction 293, 389, 382, 545 compartment 71 cone penetration testing 653 consolidation 63, 366, 709 constraints 104 construction equipment 337 containment structure 293 correlation 446, 669 CPT 119 crumb test 299 crusher 593 CSIR 719 damping 183 DCPT 632 deep excavation 215, 269 deep water 139 deformation characteristics 366 deformation modulus 669 deformations 228 design methodology 83 design 77, 422, 439 developing countries 468 dewatering 313 diaphragm wall 243, 269 disorders 63 dispersion 299 dispersive soils 299 displacement 183 dolomite 165 Do-nou 476 Double Oedometer Test 599

744

double twisted wire mesh 228 drag 152 dry unit weight 293 dynamic 139 dynamic compaction 307 dynamic cone penetrometer 422, 446 Dynamic Penetration Testing 565 Dynamic Probe Super Heavy (DPSH) 559, 565, 571 dynamically loaded foundation 183 earthquake 40 economic activities 730 economy 460 Ecoroute 460 EEMM 403 effective grain size 638 elastic 320 elastoplastic 63 élastoplastique 701 electrokinetics 313 electro-osmosis 313 éléments finis 701 embankment 489, 532 embankment settlement 611 empirical equivalence 565 environment 620 environmental management 83 equivalence 104 essai de chargement 158 expansive clays 599, 663 factor of safety 249, 679, 685 failure 299 falling weight deflectometer 446 fibre optics 3 field density 389 field tests 565 finite element technique 644 finite elements 165 fire fighting facility 145 flac 182 floating structure 139 forcing function and wave propagation 644 foundation breadth 133 foundation investigation and design 653 foundation 113 foundry sand 545 frequency 183

Gautrain 165 GCL 89, 104 geogrid 77, 196, 262 géogrille 482 geological model 726 geomaterial 439 geomechanical rating 10 geosynthetic clay liners 104 geosynthetics 40, 532, 97, 228, 284, 507 geotechnical 320, 453, 719 geotechnical engineering practice 730 geotechnical engineering 275 geotechnical investigation 632 geotechnical site investigations 489, 559, 565 geotechnical standards 726 geotextile 476 Gezira Formation 551 government regulations 104 grading curve 638 granular 349 granular base 389 granular materials 396 gravel 495 gravely clay 669 ground anchors 243 ground improvement 127, 313 ground treatment 243 hazard assessment 626 hazardous waste 97 heave prediction 403 hydraulic conductivity 89, 545, 638 hydro 605 hydrostatic stress 337 hyperstectral 3 identification 299 impact load 183 inertia 152 initial density 202 injection 71, 243 instrumentation 605 intake channel and pit 145 interface 355 inter-slice forces 685 iron oxide 355 jet grouting 269 jet-grouting slab 243 jetty 190

745

joint 71 Khartoum 551 landfill 77, 620 landfill liner design 97 lateral confinement 113 lateral support 249 lateral support 275 lateritic soils 10 leachate compatibility 89 limitations 104 linear shrinkage 403 liquefaction mitigation 113 load mobilisation 221 LPG terminal 145 marécageuse 343 marginal road-building materials 468 material properties 679 mechanical polymeric connector 262 mechanically stabilised layer 196 mélange de sols 343 micropile 165 Moatize 593 model 152 model pile 202 model test 40 modèle 701 modeling 320 modélisation 482 monitoring 243, 605 Mozambique 691 MRR 439 mudstones 541 nappe phréatique 343, 524 natural aggregates 415 net limit pressure 669 Niger Delta 139 non saturé 701 non-reflecting boundaries 644 Nubian Sandstone 551 numerical modelling 215, 237, 532 open pit 691 OPMC 439 optimum moisture content 293 Otta seal 495 Over-Consolidation-Ratio 366 paleo channels 611 partially saturated 366 particle 330 past 3

pavement 422, 439, 453, 460, 507 pavement rehabilitation 468 pedology 10 penetration tests 632 penetrometers data correlations 632 permeability 638 phase velocity 644 phi-scale 431 pieu 158 piezocone 611 pile load test 119 pile tests 653 piles 190 pipe 63 plasticity 389 plasticity index 736 plate load test 587, 653 Plaxis 8.6 code 145 poisson ratio 133 porosity 638 post tensioning 215 pozzolana 439 prediction 580 preloading 127 present future 3 pression interstitielle 709 pressure plate equipment 360 prime 495 probabilistic analysis 679 probability of failure 249, 679 problematic 453 quantitative 330 raft foundation 243 RCC 460 ReCap 453 re-drive correction 559 rehabilitation methods 468 reinforced separation 507 reinforced soil 284 Reinforced Soil Retaining Wall Systems (RSRW) 262 reinforced soil wall 237 reinforcement 228, 482, 507, 532 research 719 resilient modulus 396 retaining wall 40, 243 rice husk ash 736 rigid inclusion 175 road 422, 431, 507

746

road construction 284 road materials 349 rock mass strength 691 rod friction 559 rotational displacement 133 685 ru factor runway 396 rural access road 476 sables 343 sand 175, 431, 501 saprolitic soils 10 satellite imagery 3 SCDR 439 seismic load 237 seismic refraction survey 489 sensitivity analysis 679 separation 507 settlement 63, 127, 175 settlement potential 403 settlement-reducing elements 165 shaft resistance 202 shafts 275 shallow dolomite 626 shear 320 shear modulus 376 shear strength 293, 313, 396 shear zone 355 shortcomings 299 simple index properties of soil 366 simple shear deformation 376 sinkhole 626 site investigation 593 skirts 113 slope design 691 slope stability 685 soil 275, 330 soil characteristics 620 soil characterization 403 soil covers 83 soil investigation 620 soil nail 221 soil nail pull out testing 255 soil properties 307 soil reinforced wall 228 soil stabilization 313, 736 soil treated with emulsion 501 soil treatment 307 soil water characteristic curve 360

sol peu cohérent 709 South Africa 97 SPT 632 SPT resistance 669 stabilization 196, 330, 382, 415, 507 Standard Penetration Test (SPT) 565, 571 static load 202 statistical analysis 119 steep slopes 77 stiffness 320 stockyard 593 strength 320, 541 subgrade resilient modulus 446 subgrade 489, 507 subgrade soils 337 subsidence 626 suction 360 suction caisson 139 Sudan 580 sustainability 284 swell 599, 663 swell index 89 swell potential 403 swelling pressure 663 technique 453 telescopic probes 587 terrain evaluation 726 testing 320 Tete 691 theory 453 Thornthwaite index 663 torque correction 559 trafficability 476 treated sample 382 triaxial test 349 unbound 431 unbound aggregates 415 unit weight 545 United Arab Emirates 262 university 719 urbanization 730 variationnel 701 veneer stability 77 vertical reinforcement 237 waste management 620 water testing 127 weatherability 541

Proceedings of the 15th African Regional Conference on Soil Mechanics and Geotechnical Engineering 747 C. Quadros and S.W. Jacobsz (Eds.) IOS Press, 2011 © 2011 The authors and IOS Press. All rights reserved.

Author Index A’Bear, T. Akayuli, C.F.A. Akpila, S.B. Allotey, C. Allotey, N.K. Amoyo, G. Ampadu, S.I.K. Anochie-Boateng, J. Archer, A. Assinder, P. Atuboyedia, T.J. Azzam, W.R. Ballestra, G. Barnard, H. Beales, P. Belaid, M. Bello, A.A. Belounar, L. Benmebarek, N. Benmebarek, S. Blom, E. Bouafia, A. Bouassida, M. Braga , C. Breyl, J. Brink, D. Cabral, A. Caimoto, F. Chang, H.N. Cisse, I.K. Clark, P. Cruz, D. Davies, P. Davis, G.N. Day, P. de Deus, K.M.D. Derrick, A. Diane, M.S. Diane, S. Dinis, J. Dippenaar, M.A. Doulala-Rigby, C. Dsarma, M.

571, 626, 726 489 133, 152 83 83 330 446 337, 396 183 284 127 113 284 587 559, 565 63 293 482, 532 532 482, 532 307 158 63, 175 593 249, 255 97 343 190 165 460 691 71 104 593 249 468 355 709 709 518 638 262 366

Ejezie, S.U. Elarabi, H. Elhakim, A. Elsharief, A.M. Fatahi, B. Fatokun, O. Fernandes, L. Ferreira, L. Freitas, R. Fukubayashi, Y. Gawu, S.K.Y. George, E.A.J. Gidigasu, M.D. Glendinning, S. Govender, R. Hall, C.D. Harrison, B. Henniche, A. Heymann, G. Heyns, P.S. Howell, G.C. Ibrahim, M. Jacobsz, S.W. James, G. Jenkins, K. Johns, D. Jones, G.A. Kabari, B. Kalumba, D. Kamau, D. Karoui, H. Kellaway, M. Kenmogne, E. Keyter, G.J. Khabbaz, H. Kimura, M. Klompmaker, J. Kogi, S. Koseki, J. Kotheki, S. Kulabako, R. Legg, P. Lenze, B.

133, 139, 152 580 215 415, 551 237 145 495, 501 343 343 476 360 127 10 313 403 196 307, 571 158 587, 644 644 275 632 221 507 349 97 611, 653 139 313, 559, 565 453 63 605 669 605 237 476 77 453 40 330, 439 313 89 77

748

MacRobert, C. Magnani, G. Maharaj, A. Martin, J.R. Mbawala, S.J. McLennan, M. Meintjes, H.A.C. Mindo, I. Mohamed, A.M. Mohamed, E.A. Mohamed, M. Mohammad, S.B. Moses, G. Msiza, J. Mukabi, J. Mukabi, J.N. Mulusa, W. Ndeda, M. Ndimbo, R. Ngigi, L. Njoroge, B. Nobre, D. Nortjé, R. Nyako, S.O. Nyaoro, D.L. Obasi, N.L. Oddei, J.K. Ogallo, J. Ojiogu, E.B. Oju, M. Okado, J. Okang, E.K. Okonta, F. Omer, O.G. Omindo, P. Osinubi, K. Osinubi, K.J. Paige-Green, P. Parrock, A. Phalanndwa, T.S. Pidgeon, J.T. Pina, J. Pinard, M. Pinto, A.

559, 565 145 299 669 644 89 611, 653 620 415 551 580 119 545 97 320, 330, 439 453 349 453 382, 389 330, 439 453 190, 243 97 489 632 736 730 330 736 127 330 446 355, 376 415 439 545 293 422, 431, 719 3 221 403 243, 518 431 269

Pita, X. Quadros, C. Quinta-Ferreira, M. Rebelo, V. Remadna, M.S. Richer, L. Rodrigues, V. Roldão, M. Roth, C.P. Sall, A. Sarma, D. Schippers, H. Schreiner, D. Scotto, M. Selomane, M. Shahin, M.M. Shamrock, J. Shrestha, B. Sidai, V. Silva, A. Simoes, I. Sparks, A.D.W. Steyn, W.J. Tahsin, A. Taylor, D. Tayob, H. Tesha, P. Umutoni, B. van Rooy, J.L. van Rooy, L. Vicari, M. Vilar, A.C. Vorster, T.E.B. Wardle, G. Wekesa, S. Wills, P.G. Xavier, B. Yendaw, J. Yendaw, J.A. Zadjaoui, A. Zannoni, E. Zein, A.K.M. Zelalem, T.

269 389, 620 541 343 482 626 524 524 644 460 366 284 376 228 679 202 97 237 330, 439 524 495, 501 599, 663, 685 468 215 605 501 439 313 638 679, 726 228 71 165, 593 249, 255 330 262 190, 243, 518 360 489 701 228 119 453