Geological Storage of Carbon Dioxide
Geological Society Special Publications Society Book Editors R. J. PANKHURST (CHIEF EDITOR) P. DOYLE F. J. GREGORY J. S. GRIFFITHS A. J. HARTLEY R. E. HOLDSWORTH
J. A. HOWE P. T. LEAT A. C. MORTON N. S. ROBINS J. P. TURNER Special Publication reviewing procedures The Society makes every effort to ensure that the scientific and production quality of its books matches that of its journals. Since 1997, all book proposals have been refereed by specialist reviewers as well as by the Society's Books Editorial Committee. If the referees identify weaknesses in the proposal, these must be addressed before the proposal is accepted. Once the book is accepted, the Society has a team of Book Editors (listed above) who ensure that the volume editors follow strict guidelines on refereeing and quality control. We insist that individual papers can only be accepted after satisfactory review by two independent referees. The questions on the review forms are similar to those for Journal of the Geological Society. The referees' forms and comments must be available to the Society's Book Editors on request. Although many of the books result from meetings, the editors are expected to commission papers that were not presented at the meeting to ensure that the book provides a balanced coverage of the subject. Being accepted for presentation at the meeting does not guarantee inclusion in the book. Geological Society Special Publications are included in the ISI Index of Scientific Book Contents, but they do not have an impact factor, the latter being applicable only to journals. More information about submitting a proposal and producing a Special Publication can be found on the Society's web site: www.geolsoc.org.uk.
It is recommended that reference to all or part of this book should be made in one of the following ways: BAINES, S. J. & WORDEN, R. H. (eds) 2004. Geological Storage of Carbon Dioxide. Geological Society, London, Special Publications, 233. GUNTER, W. D., BACHU, S. & BENSON, S. 2004. The role of hydrogeological and geochemical trapping in sedimentary basins for secure storage of carbon dioxide. In: BAINES, S. J. & WORDEN, R. H. (eds) 2004. Geological Storage of Carbon Dioxide. Geological Society, London, Special Publications, 233,129-145.
GEOLOGICAL SOCIETY SPECIAL PUBLICATION NO. 233
Geological Storage of Carbon Dioxide
EDITED BY SHELAGHJ.BAINES BP Exploration and Production Company, Sunbury, UK and
RICHARD H. WORDEN Department of Earth & Ocean Sciences, University of Liverpool, Liverpool, UK
2004
Published by The Geological Society London
THE GEOLOGICAL SOCIETY
The Geological Society of London (GSL) was founded in 1807. It is the oldest national geological society in the world and the largest in Europe. It was incorporated under Royal Charter in 1825 and is Registered Charity 210161. The Society is the UK national learned and professional society for geology with a worldwide Fellowship (FGS) of 9000. The Society has the power to confer Chartered status on suitably qualified Fellows, and about 2000 of the Fellowship carry the title (CGeol). Chartered Geologists may also obtain the equivalent European title, European Geologist (EurGeol). One fifth of the Society's fellowship resides outside the UK. To find out more about the Society, log on to www.geolsoc.org.uk. The Geological Society Publishing House (Bath, UK) produces the Society's international journals and books, and acts as European distributor for selected publications of the American Association of Petroleum Geologists (AAPG), the American Geological Institute (AGI), the Indonesian Petroleum Association (IPA), the Geological Society of America (GSA), the Society for Sedimentary Geology (SEPM) and the Geologists' Association (GA). Joint marketing agreements ensure that GSL Fellows may purchase these societies' publications at a discount. The Society's online bookshop (accessible from www.geolsoc.org.uk) offers secure book purchasing with your credit or debit card. To find out about joining the Society and benefiting from substantial discounts on publications of GSL and other societies worldwide, consult www.geolsoc.org.uk, or contact the Fellowship Department at: The Geological Society, Burlington House, Piccadilly, London W1J OBG: Tel. +44 (0)20 7434 9944; Fax +44 (0)20 7439 8975; E-mail:
[email protected]. For information about the Society's meetings, consult Events on www.geolsoc.org.uk. To find out more about the Society's Corporate Affiliates Scheme, write to
[email protected]. Published by The Geological Society from: The Geological Society Publishing House Unit 7, Brassmill Enterprise Centre Brassmill Lane BathBA13JN,UK (Orders: Tel. +44 (0)1225 445046 Fax+ 44 (0)1225 442836) Online bookshop: http://bookshop.geolsoc.org.uk The publishers make no representation, express or implied, with regard to the accuracy of the information contained in this book and cannot accept any legal responsibility for any errors or omissions that may be made. © The Geological Society of London 2004. All rights reserved. No reproduction, copy or transmission of this publication may be made without written permission. No paragraph of this publication may be reproduced, copied or transmitted save with the provisions of the Copyright Licensing Agency, 90 Tottenham Court Road, London W1P 9HE. Users registered with the Copyright Clearance Center, 27 Congress Street, Salem, MA 01970, USA: the item-fee code for this publication is 0305-8719/047$ 15.00. British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library. ISBN 1-86239-163-7 Typeset by Servis Filmsetting Limited, Manchester, UK Printed by Cromwell Press, Trowbridge, UK
Distributors USA AAPG Bookstore PO Box 979 Tulsa OK 74101-0979 USA Orders: Tel. + 1 918 584-2555 Fax+ 1918 560-2652 E-mail
[email protected] India Affiliated East-West Press PVT Ltd G-l/16 Ansari Road, Darya Ganj, New Delhi 110002 India Orders: Tel. +91 11 2327-9113, 2326-4180 Fax+ 91 112326-0538 E-mail
[email protected] Japan Kanda Book Trading Company Cityhouse Tama 204 Tsurumaki 1-3-10 Tama-shi, Tokyo 206-0034 Japan Orders: Tel. + 81 (0)423 57-7650 Fax+ 81 (0)423 57-7651 E-mail
[email protected]
Contents BAINES, S.J. & WORDEN, R.H. Geological storage of carbon dioxide GALE, J. Why do we need to consider geological storage of CO2? CHADWICK, R.A., HOLLOWAY, S., BROOK, M.S. & KIRBY, G.A. The case for underground CO2 sequestration in northern Europe PEARCE, J., CZERNICHOWSKI-LAURIOL, I., LOMBARDI, S., BRUNE, S., NADOR, A., BAKER, J., PAUWELS, H., HATZIYANNIS, G., BEAUBIEN, S. & FABER, E. A review of natural CO2 accumulations in Europe as analogues for geological sequestration SHIPTON, Z.K., EVANS, J.P., KIRSCHNER, D., KOLESAR, P.T., WILLIAMS, A.P. & HEATH, J. Analysis of CO2 leakage through 'low-permeability' faults from natural reservoirs in the Colorado Plateau, east-central Utah BAINES, S.J. & WORDEN, R.H. The long-term fate of CO2 in the subsurface: natural analogues for CO2 storage ROCHELLE, C, CZERNICHOWSKI-LAURIOL, I. & MILODOWSKI, A.E. The impact of chemical reactions on CO2 storage in geological formations: a brief review JOHNSON, J.W, NITAO, J.J. & KNAUSS, K.G. Reactive transport modelling of CO2 storage in saline aquifers to elucidate fundamental processes, trapping mechanisms and sequestration partitioning GUNTER W.D., BACHU, S. & BENSON, S. The role of hydrogeological and geochemical trapping in sedimentary basins for secure geological storage of carbon dioxide HOVORKA, S.D., DOUGHTY, C., BENSON, S.M., PRUESS, K. & KNOX, PR. The impact of geological heterogeneity on CO2 storage in brine formations: a case study from the Texas Gulf Coast ZWEIGEL, P., ARTS, R., LOTHE, A.E. & LINDEBERG, E.B.G. Reservoir geology of the Utsira Formation at the first industrial-scale underground CO2 storage site (Sleipner area, North Sea) ARTS, R., EIKEN, O., CHADWICK, A., ZWEIGEL, P., VAN DER MEER, B. & KIRBY, G. Seismic monitoring at the Sleipner underground CO2 storage site (North Sea) LAENEN, B., VAN TONGEREN, P., DREESEN, R. & DUSAR, M. Carbon dioxide sequestration in the Campine Basin and the adjacent Roer Valley Graben (North Belgium): an inventory WORDEN, R.H. & SMITH, L.K. Geological sequestration of CO2 in the subsurface: lessons from CO CO2 injection enhanced oil recovery projects in oil fields BACHU, S. & GUNTER, W.D. Acid-gas injection in the Alberta Basin, Canada: a CO2-storage experience STENHOUSE, M.J. & SAVAGE, D. Monitoring experience associated with nuclear waste disposal and its application to CO2 sequestration projects Index
1 7 17
29
43 59 87 107 129 147 165 181 193
211 225 235 249
This page intentionally left blank
Geological storage of carbon dioxide SHELAGH J. BAINES1 & RICHARD H. WORDEN2 1
BP Exploration and Production Company, Chertsey Road, Sunbury-on-Thames, TW16 7LN, UK (e-mail: bainess @bp. com) 2 Department of Earth and Ocean Sciences, University of Liverpool, 4 Brownlow Street, Liverpool L69 3GP, UK (e-mail:
[email protected])
Abstract: Carbon dioxide is the main compound identified as affecting the stability of the Earth's climate. A significant reduction in the volume of greenhouse gas emissions to the atmosphere is a key mechanism for mitigating against climate change. Geological storage of CO2, or the injection and stabilization of large volumes of CO2 in the subsurface in saline aquifers, existing hydrocarbon reservoirs or unmineable coal-seams, is one of the more technologically advanced options available. A number of studies have been carried out aimed at understanding the behaviour and long term fate of CO2 when stored in geological formations.
Global warming, and through it, climate change has been generally accepted as a problem; the consensus proffered by the International Panel on Climate Change (IPCC) in 2001 was that climate change is real and that its causes and effects should be mitigated. Recent studies suggest that the earth's surface has warmed by 0.6° over the past 100 years (Hulme & Jenkins 1998). A significant reduction in the volume of greenhouse gas emissions to the atmosphere was identified as a key mechanism for achieving stability. Carbon dioxide (CO2) is the main compound identified as affecting the stability of the Earth's climate, representing 62.5% of all greenhouse gases generated globally (IPCC 1996, 2001; review by Gale). Geological storage of CO2, or the injection and stabilization of large volumes of CO2 in the subsurface, is one option being employed against rising CO2 emissions. Most of the additional CO2 added to the atmosphere has been generated by human activity; mainly through fossil fuel combustion (e.g. power plants, refineries, motor vehicles). Roughly one third of CO2 emissions in the USA come from power plants and other large point sources (US DOE 2004). In the UK alone, the volume of CO2 emitted has increased by one third since the industrial revolution (IPCC 1996). Predictions of CO2 emissions suggest the increase is, and will remain, exponential under the current energy-use pattern.
Geological storage options There are a number of ways by which CO2 emissions can be reduced. The technology exists to capture the CO2 at a power plant, separating it from the flue
gases and producing a stream of concentrated gas. In addition, a number of oil and gas reservoirs contain significant volumes of CO2 (e.g. Baines & Worden) which must be stripped from the hydrocarbon before processing or sale. Historically, this CO2 would have been vented to the atmosphere. Once the stream of CO2 is produced, it needs to be removed from the earth's atmosphere for a period of time sufficient to allow climate stabilization. The actual time period required is not well understood, however, the general consensus is that CO2 would need to be prevented from reaching the atmosphere for hundreds, if not thousands of years (Gunter et al.). The three main options for subsurface storage of CO2 are saline aquifers, existing oil and gas fields, and unmineable coal seams. These can be split into two economic end-members. First are purely storage options, whereby the CO2 is injected without any positive benefit other than to the atmosphere (e.g. saline aquifers), although carbon tax credits may improve the financial viability of this option. The second are utilization options where the CO2 injection process has additional, economic benefits through enhanced oil recovery or coal-bed methane production (e.g. existing oil and gas fields and coal seams). Alternative options for captured CO2 exist (Chadwick et al; Gale) but these are at an early stage of technology development. They include deep ocean sequestration (e.g. Omerod et al. 1999), carbonate production, and injection into deep carbonate sediments. Research into geological storage of CO2 is still at a relatively early stage, although the technology has matured over the past decade from feasibility and theoretical studies (e.g. Cox etal 1996; CzernichowskiLauriol et al. 1996; Holloway 1997; 2001; Gupta et
From: BAINES, S. J. & WORDEN, R. H. (eds) 2004. Geological Storage of Carbon Dioxide. Geological Society, London, Special Publications, 233,1-6.0305-8719/047$ 15.00 © The Geological Society of London 2004.
2
S. J. RAINES & R. H. WORDEN
al. 2002) to application in the hydrocarbon industry in storage (Sleipner field, North Sea) and utilization (Weyburn field, Canada) projects.
Saline aquifers Saline aquifers can be defined as units which do not have any potential to act as sources of potable water. The volumetric potential for saline aquifer storage is huge (Holloway 1997; Gunter et al. 1998). Studies suggest that, on a global scale, saline aquifers could account for between 20% and 500% of the projected total CO2 emissions to 2050 (Davidson et al. 2001). It has been estimated that deep saline formations in the United States alone could potentially store up to 500 billion tonnes of CO2 (US DOE 2004). Chadwick et al. suggest that within the UK, the Bunter sandstone can provide a storage volume equivalent to around 350 years of UK derived emissions. A similar study of the Campine Basin in Northen Belgium (Laenen et al.} has identified a CO2 storage capacity of several hundred million tonnes in Triassic and Carboniferous formations. Economically, saline aquifer storage is less attractive than other utilization options. However, carbon tax credits, or carbon trading systems may improve the economics. Gas from the Sleipner field in the Norwegian sector of the North Sea contains 4.0-9.5% CO2; the sales gas agreement allows a maximum CO2 content of 2.5% (Baklid et al. 1996). Venting to atmosphere of the approximate 1 Mt/a of stripped CO2 would increase the total Norwegian emissions by 3% but also cost the operator in carbon taxes under Norway's tax scheme. The solution was to strip the CO2 from the produced gas at the offshore top-side facility and then re-inject it, as a supercritical phase, into a saline aquifer 800m below the sea-bed. Although not 100% cost-effective (DOE EIA 2002), the project was the world's first geological storage programme, providing a demonstration of the technology required to store CO2. An accompanying research project studied the first five years of CO2 injection (Torp & Gale 2003; Zweigel et al. 2002; Arts et al.\ Zweigel et al.}. Further storage projects are already being planned, for example, the Gorgon field on the North West Shelf, Australia where the produced gas contains up to 14% CO2. CO2 stripped from the gas will be reinjected 2000m below the producing reservoir into a saline formation (IEA 2003).
Existing oil and gas fields The alternative to storage of CO2 is to use the gas to improve the economics of the capture process. The oil industry uses CO2 to improve the recovery effi-
ciency of heavy oils, a process known as enhanced oil recovery (EOR). Around 32 million tons of CO2 per year are injected into oil reservoirs for EOR in the US (NETL web-site). The two main EOR processes are discussed by Holm & Josendal (1974), Monger et al. (1991), and briefly by Worden & Smith. Historically, the long term storage of CO2 was not considered by EOR operators; CO2 which reached the surface during production was either vented or captured for recycling. However, technology to increase the volume of CO2 that remains in the subsurface does exist. An additional economic benefit of CO2 injection into oil reservoirs is the extension of field life by improving rate of recovery. CO2 injection into gas fields for improved pressure support is also being considered as an option. Oil and gas fields also have potential as storage reservoirs after they cease to be economic. The geological understanding of old petroleum fields is typically much greater than that for saline aquifers and potential exists for recycling of production and enhanced oil recovery (injection) infrastructure already in place. A major research project, combining EOR with CO2 storage was started in 2000 at the Weyburn field, Canada. CO2 purchased from a synthetic fuel plant in North Dakota, USA was piped 320km to the field in Saskatchewan province. The EOR project is expected to produce 130 million barrels of incremental oil through CO2 displacement, extending field life by 25 years (Moberg 2001). Studies on the phase behaviour of the CO2-reservoir fluids, the chemical reactions between the CO2 and reservoir rocks, the impact of the CO2-rock interaction on fluid flow, formation stability, seal integrity and storage capacity were undertaken (Moberg 2001). Detailed time-lapse monitoring programmes, using various remote and direct techniques, were planned to study the physical distribution of the CO2 during the injection period, providing data for history matching reservoir models. The overall aim was to provide a comprehensive risk and economic assessment of CO2 storage in the area. Acid gas, a variable mixture of hydrogen sulphide (H2S) and CO2 derived from the 'sweetening' of sour gas, is also a candidate for geological storage. In Alberta, Canada, instead of flaring the acid gas or stripping the sulphur (desulphurization) and storing it in solid form, the gas mixture has been injected into several geological formations, including both saline aquifers, old oil and gas reservoirs and the water leg of producing reservoirs (Bachu & Gunter).
Unmineable coal-seams Coal seams typically contain economically large amounts of methane-rich gas that is adsorbed onto
GEOLOGICAL STORAGE OF CARBON DIOXIDE
the surface of the coal. Extraction of the gas requires depressurization, usually by pumping water out of the reservoir. Roughly twice as much CO2 can be adsorbed on coal as methane such that injected CO2 has the potential to displace methane and remain sequestered in the coal seam (Reeves 2003). A pilot project in San Juan, New Mexico suggests that methane production from the extensive coal beds there could be increased 75% by injecting CO2 (Reeves etal. 2001). For any CO2 storage or utilization process to be successful, it must be economically competitive compared to alternative options, and effective in significantly reducing emissions. It must be a long-term option that can prevent emissions from reaching the atmosphere for sufficient periods of time and be safe both to the population and to the environment. In addition, the technology must be flexible to allow storage in a range of geological regimes, ideally, close to the point source of CO2. A number of studies have been carried out aimed at understanding the behaviour of CO2 when stored in geological formations and to help create guidelines for the selection and development of CO2 storage facilities.
Natural analogues for CO2 storage CO2 occurs naturally in the subsurface, often in large volumes (Pearce et al. 1996; Baines & Worden; Pearce et al; Ship ton et al). There are a number of possible sources (Baines & Worden, and references therein); CO2 is often associated with igneous processes, with high temperature metamorphism of carbonate-bearing rocks and de-volatilization of CO2-bearing fluids. Alteration of organic matter can also produce abundant CO2. The retention time of these CO2 accumulations can vary from extremely short term, active vents (Pearce et al.; Shipton et al), to fields that have undergone CO2-EOR with a CO2 residence time of months to several years (Wolcott et al 1989; Worden & Smith), to CO2 in subsurface traps that have been in place for many thousands or even millions of years (Baines & Worden). The study of these accumulations provides ideal natural analogues for studying the longterm consequences, both physical and chemical, of introducing CO2 into the subsurface in a range of geological environments.
Geochemical fate of CO2 CO2 is a reactive gas and, unlike methane, interacts with both the rock and formation fluid in the injected reservoirs. A number of studies have attempted to predict the behaviour of CO2 under a range of physi-
3
cal and chemical conditions (Czernichowski-Lauriol et al 1996; Gunter et al 1998; Gunter et al, Rochelle et al; Worden & Smith). CO2 solubility is dependent on the temperature, pressure and composition of the formation fluids. The proportion of an injected volume which dissolves is controlled by the surface area of CO2-water interaction and the degree of mixing. Injection rate also plays a role. Dissolution of CO2 creates chemical disequilibria in the storage formation, resulting in reactions involving mineral dissolution and precipitation (Bachu et al. 1994; Baines & Worden; Rochelle et al., Worden & Smith). Precipitation of carbon-bearing minerals results in much more stable, longer term trapping, or sequestration, of the CO2; however, studies from natural analogues suggest that mineral precipitation is relatively minor, slow, and controlled by the concentration of divalent cations available for reaction (Baines & Worden). The extent of CO2-water-rock interaction during migration of the injected CO2 is the main control on the ultimate fate of the CO2. Reactive transport modelling of a Sleipner-like storage reservoir suggested that only 1% of the CO2 precipitated as carbonate minerals whereas 15-20% was still dissolved in the formation fluids after 20 years. The remainder stayed as an immiscible phase (Johnson etal.). Migration and trapping of CO2 The pressure, hydrodynamic and geothermal regimes in a basin have a significant impact on its potential for CO2 storage (Gunter et al). At the basin scale, the tectonic regime affects the rate at which fluids migrate up-dip, or along fault zones. The physical and chemical controls on CO2 state post-injection affect the migration of the injected volume. CO2 that remains in an immiscible phase is driven by buoyancy forces, displacing the pore fluids, up-dip to the surface unless constrained by stratigraphical or structural traps in the same manner as oil and gas accumulations. This upward migration may be extremely slow taking more than a million years (Gunter et al), thus forming transient 'hydrodynamic traps' (Bachu et al. 1994). CO2 dissolved in formation fluid migrates at an even slower pace, at the rate of the in situ formation fluids of the basin, which may have subsurface residence times of many millions of years. A key point is that fluids and gases do not remain stationary in the subsurface but migrate at rates related to their composition, state and the 'plumbing' of the basin. Oil and gas accumulations provide a good analogy for the long-term security of injected CO2; hydrocarbons form continually migrating fronts; however, their progress is slowed by subsurface traps allowing significant accumulations to form.
4
S. J. BAINES & R. H. WORDEN
The geological fill of a basin and the internal heterogeneities at all scales, including faults and fractures, depositional patterns, and porositypermeability trends, affect the capacity and effectiveness of CO2 storage (Hovorka et al.; Johnson et al.). The buoyancy-driven migration of immiscible CO2 is controlled by barriers and baffles to flow, increasing the CO2-water-rock interaction. This internal heterogeneity can also have a negative impact on storage effectiveness. Barriers to flow, particularly close to the injection point can create pressure gradients resulting in barrier or seal failure and continued migration of the CO2 along the path of least resistance. It is necessary to understand the storage volume at a greater scale than merely the primary trapping mechanism; secondary or even tertiary traps may occur over time. Studies carried out at the Sleipner site in Norway provide some understanding of how CO2 migrates immediately post-injection (Arts et al. 2000; Chadwick et al 2000; Lindeberg et al 2000; Zweigel et al 2000; Arts et al., Zweigel et al.}. Reservoir models built for the injected reservoir, the Utsira Formation, were used to model CO2 migration. These were history-matched using data from time-lapse 3D seismic over the injected volume. The results demonstrated the impact of the internal heterogeneity, present as shale bands of variable thickness and shallow structural doming, of the Utsira Formation on CO2 migration.
Monitoring CO2 storage sites Monitoring provides confidence that the distribution of the injected CO2 can be proven both immediately post-injection and in the medium term after the injection period is completed. If mitigation is required at any point, i.e. through unexpected migration direction, an accurate picture of CO2 distribution would be necessary. Verification of the injected volume may also become a requirement. The monitoring process is an integral part of the storage development plan. A number of monitoring options have potential, including both surface and downhole (cross-well and vertical seismic profiling) methods. Any monitoring option should be cheap, repeatable, easy to acquire, and have a short interpretation turnaround time The use of 3D seismic has been demonstrated at two CO2 injection sites. At Sleipner, a pre-injection 3D seismic survey was shot over the area (Arts et al 2000) and, although not optimized for the Utsira Formation, formed the baseline survey. Subsequent surveys indicated that time-lapse seismic allows the CO2 distribution to be mapped effectively (Arts et al.) in this offshore field. For the onshore Weyburn CO2-EOR project, a comprehensive multicompo-
nent 3D seismic survey was shot over the injection area prior to any CO2 injection. This was followed by a second survey after 14 months of CO2 injection. Strong seismic anomalies around the CO2 injector wells in the second survey suggest good imaging of the CO2 migration fronts (Jazwari 2002).
Regulation of CO2 storage Regulatory frameworks exist for the disposal of hazardous wastes; radioactive and chemical. Monitoring is a key component of the development and operation of nuclear waste repositories (Stenhouse & Savage), requiring continuing geotechnical, groundwater and environmental checks to ensure the continued security of the stored material. Injection or long-term storage in geological formations has been taking place for a number of years. Although there are distinct physical and chemical differences between CO2 and liquid or radioactive waste, the experience gained from these industries may be valuable when designing guidelines for CO2 storage. The technology for CO2 storage is developing rapidly. If successful, a large number of wells, both onshore and offshore could be required for injecting the volumes of CO2 available for storage. At present, no dedicated regulatory framework is in place in any country. The two areas should be developed together, to ensure technology meets the needs of the regulatory controls, i.e. monitoring guidelines and operational constraints.
The future of geological storage of CO2 Geological storage of CO2 remains a controversial concept amongst the public and environmental groups. Even if all technical obstacles to geological storage are met and if sufficient capacity at an economic rate is available to mitigate much of the present CO2 emissions, the safety and acceptability of storage has to be demonstrated. However, national governments are beginning to realize that today's policies regarding carbon emissions and the environment cannot be sustained and will not meet the targets set for future emissions. Between 2002 and 2004 both the US and UK governments set targets for reducing greenhouse gas emissions. In 2002, the UK government issued a White Paper (UK DTI 2002) stating their energy goals which included plans to reduce CO2 emissions by 60% by 2050, with the aim of making significant progress by 2020. Similarly, in 2004, the USA government Global Climate Change Initiative (US DOE 2004) stated an ambition to reduce greenhouse gas intensity by 18% by 2012. Both countries highlighted geological storage as the technologically most advanced option for achieving
GEOLOGICAL STORAGE OF CARBON DIOXIDE these aims, with the US indicating it would have commercially-ready storage or sequestration technologies available for assessment by 2012. The focus of future carbon management is on energy efficiency and alternative energy supplies such as lower carbon fuels (e.g. switching from coal to natural gas-fired power stations) and renewable energy mechanisms. However, the current global energy demand is such that these options cannot supply sufficient energy at present and the timeframe for development means that fossil fuels will remain the main fuel source for the short to medium term. However, by capturing the carbon at source and removing it from the atmosphere and surface environment of the planet, fossil fuels can become a low-carbon option, thus reducing greenhouse gas emissions. CO2 storage therefore becomes a stopgap mechanism, buying time whilst development of efficient, low- or no-carbon energy sources continues. This means that the technology required to carry out effective, safe, CO2 storage is needed now. Many of the papers in this volume arose from a technical session held at BUG 2001, entitled 'Greenhouse Gas Disposal' chaired by Shelagh Baines and John Gale. The aim of the session was to provide an introduction to the science of geological storage of carbon dioxide to a wider technical audience. The editors would like to thank the following for undertaking the technical reviews of the papers in this volume: Etienne Brosse, John Bunney, Charles Byrer, Bob Chaplow, Max Coleman, Daniel Garcia, Hal Gluskotter, Neeraj Gupta, Adrian Hartley, Stuart Haszeldine, Howard Herzog, Susan Hovorka, Ian Hutcheon, Yousef Kharaka, Rob Lander, Steve Laubach, Craig Lewis, Eric Lindeburgh, Ross McCartney, Tony Milodowski, Larry Meyers, Paul Nadeau, Tim Needham, Euan Nisbet, Sue Raikes, Beverly Saylor, Bruce Sass, Craig Smalley, Scott Stevens. Angharad Hills and Sally Oberst of the Geological Society dealt with the organization and editing of the volume. John Gale is thanked for helping with the early stages of the volume.
References ARTS, R., BREVIK, L, EIKEN, O., SOLLIE, R., CAUSSE, E. & VAN DER MEER, B. 2000. Geophysical methods for monitoring marine aquifer CO2 storage - Sleipner experiences. In: WILLIAMS, D. J., DURIE, B., MCMULLAN, P., PAULSON, C. & SMITH, A. (eds) Greenhouse Gas Control Technologies, CSIRO Publishing, Collingwood, Australia, 366-371. BACHU, S., GUNTER, W.D. & PERKINS, E. H. 1994. Aquifer disposal of CO2: hydrodynamic and mineral trapping. Energy Conversion and Management, 35, 269-279. BAKLID, A., KORB0L, R. & OWREN, G. 1996. Sleipner Vest CO2 disposal, CO2 injection into a shallow underground aquifer. Paper presented on the 1996 SPE Annual technical Conference and Exhibition, Denver, Colorado, USA, SPE paper 36600,1-9. CHADWICK, R. A., HOLLOWAY, S., KIRBY, G. A., GREGERSEN, U. & JOHANNESSEN, P. N. 2000. The Utsira Sand,
5
Central North Sea - an assessment of its potential for regional CO2disposal. In: WILLIAMS, D. J., DURIE, B., MCMULLAN, P., PAULSON, C. & SMITH, A. (eds) Greenhouse Gas Control Technologies, CSIRO Publishing, Collingwood, Australia, 349-354. Cox, H., HEEDRICK, J. P., VAN DER MEER, B., VAN DER STRAATEN, R., HOLLOWAY, S., METCALFE, R., FABRIOL, H. & SUMMERFIELD, I. 1996. Safety and stability of underground storage. In: The Underground Disposal of Carbon Dioxide. Joule II project no: CT92-0031, Final report. CZERNICHOWSKI-LAURIOL, I., SANJUAN, B., ROCHELLE, C., BATEMAN, K., PEARCE, J. & BLACKWELL, P. 1996. Analysis of the geochemical aspects of the underground disposal of CO2. In: APPS J.A. & TSANG C.F. (eds) Deep Injection Disposal of Hazardous and Industrial Waste. Academic Press Inc, pp. 565-583. DAVIDSON, J., FREUND, P. & SMITH, A. (2001). Putting Carbon Back in the Ground. IEA Greenhouse Gas R&D Programme, February. DOE EIA. 2002. Norway: Environmental Issues, http ://w w w. eia.doe. gov/emeu/cabs/norenv.html GUNTER, W. D., WONG, S., CHEEL, D. B. & SJOSTROM, G. 1998. Large CO2 sinks: Their role in the mitigation of greenhouse gases from an international, national (Canadian) and provincial (Alberta) perspective. Applied Energy, 61, 209-227. GUPTA, N., SASS, B. M., SMINCHAK, J. R. & HICKS, J. E. 2002. Feasability of long term carbon dioxide storage in deep saline formations, http://www.netl.doe.gov/ publications/proceedings/98/98ps/ps4-7.pdf HOLM, L. W. & JOSENDAL, V. A. 1974. Mechanisms of oil displacement by carbon dioxide. Journal of Petroleum Technology, December 1974. HOLLOWAY, S. 1997. An overview of the Joule II project: The underground disposal of carbon dioxide. Energy Conversion and Management, 38S, S193-S198. HOLLOWAY, S. 2001. Storage of fossil fuel-derived carbon dioxide beneath the surface of the earth. Annual reviews of Energy and the Environment, 26, 145-166. HULME, M. & JENKINS, G. J. 1998. Climate Change Scenarios for the U.K.: Scientific Report. TJKCIP Technical Report No:l, Climate Research Unit, Norwich. IEA. 2003. Australia's Gorgon gas development will reinject reservoir CO2. Greenhouse Issues, 66. http://www.ieagreen.org.uk/ IPCC. 1996. Climate Change 1995: The Science of Climate Change, Summary for Policymakers and Technical Summary of the Working Group I Report. Cambridge University Press, Cambridge, UK, 56pp. IPCC. 2001. Climate Change 2001: The Scientific Basis, Summary for Policymakers and Technical Summary of the Working Group I Report. JAZWARI, W. 2002. Monitoring CO2 injection at Weyburn. Greenhouse Issues, 61, 5-6. IEA Greenhouse Gas Programme, U.K. LlNDEBERG, E., ZWEIGEL, P., BERGMO, P., GHADERI, A. &
LOTHE, A. 2000. Prediction of CO2 dispersal pattern improved by geology and reservoir simulation and verified by time lapse seismic. In: WILLIAMS, D. J., DURIE, B., MCMULLAN, P., PAULSON, C. & SMITH, A. (eds) Greenhouse Gas Control Technologies, CSIRO Publishing, Collingwood, Australia, 372-377.
6
S. J. BAINES & R. H. WORDEN
MOBERG, R. 2001. The Weyburn CO2 monitoring and storage project. Greenhouse Issues, 57, 2-3. MONGER, T. G., RAMOS, J. C. & THOMAS, J. 1991. Light oil recovery from cyclic CO2 injection: influence of low pressures, impure CO2 and reservoir gas. SPE Reservoir Engineering, 6, 25-32. OMEROD, W. G., FREUND, P. & SMITH, A. 1999. Ocean Storage of CO2. IEA Greenhouse Gas R&D Programme Report. NETL. http://www.netl.doe.gov PEARCE, J. M., HOLLOWAY, S., WACKER, H., NELIS, M. K., ROCHELLE, C. & BATEMAN, K. 1996. Natural occurrences as analogues for the geological disposal of carbon dioxide. Energy Conversion and Management, 37,1123-1128. REEVES, S. 2003. Enhanced CBM recovery: Coal-bed CO2 sequestration assessed. Oil and Gas Journal, July 14th. REEVES, S., PEKOT, L. & CLARKSON, C. 2001. Geologic sequestration of CO2 in deep, unmineable coal-beds: An integrated research and commercial scale field demonstration project. SPE 71749. Proceedings of the SOE Annual Technical Conference and Exhibition, New Orleans, September 30-October 3 2001. TORP, T. & GALE, J. (2003), Demonstrating Storage of CO2
in Geological Reservoirs: The Sleipner and SACS Projects. In: GALE, J. & KAYA, Y. (eds) Proceedings of the 6th International Conference on Greenhouse Gas Control Technologies, Kyoto, Japan, Elsevier Science. UK DTI. 2002. Our Energy Future - Creating a Low Carbon Economy, http://www.dti.gov.uk/energy/ whitepaper/index. shtml US DOE. 2004. http://www.fe.doe.gov/programs/ sequestration WOLCOTT J. M., MONGER T. G., SASSEN, R. & CHINN, E. W. 1989. The effects of CO2 flooding on reservoir mineral properties. 1989 SPE International Symposium, Houston, Texas. SPE Paper 18467, 101-109. ZWEIGEL, P., HAMBORG, M., ARTS, R., LOTHE A. & T0MMERAs, A. 2000. Prediction of migration of CO2 injected into an underground depository: Reservoir geology and migration modelling in the Sleipner case (North Sea). In: WILLIAMS, D. J., DURIE, B., MCMULLAN, P., PAULSON, C. & SMITH, A. (eds) Greenhouse Gas Control Technologies, CSIRO Publishing, Collingwood, Australia, pp. 360-365.
Why do we need to consider geological storage of CO2? JOHN GALE IE A Greenhouse Gas R&D Programme, Stoke Orchard, Cheltenham, Glos., GL52 4RZ, UK (e-mail:
[email protected]) Abstract: To meet targets for greenhouse gas emission reduction set by the Kyoto Protocol, many countries are considering a range of near-term options such as, fuel switching, energy efficiency improvements and use of renewable sources of energy, to reduce their greenhouse gas emissions. However, to meet the goal of the UN Framework Convention on Climate Change, namely stabilization of greenhouse gas concentrations in the atmosphere, it is likely that deeper reductions in emissions will be needed. This will require additional measures such as the geological storage of CO2. Geological storage of CO7 would be used to sequester CO2 captured from large anthropogenic sources, such as power and large industrial plants. There are a number of reservoirs suitable for geological storage of CO2 including depleted oil and gas fields and deep saline aquifers. Many of these reservoirs have stored hydrocarbons and fluids for million of years, which gives confidence that CO2 can be stored for similar durations, but research is needed to confirm this.
The topic of geological storage of CO2 forms the main focus of this volume. This opening paper sets the scene by debating why this particular mitigation option should be considered in the context of reducing greenhouse gas emissions globally. The paper reviews the international arena and reviews what actions are underway to reduce greenhouse gas emissions globally. Currently CO2 capture and storage is not included in the list of internationally agreed mitigation measures. However, the paper aims to discuss the issue of why this mitigation option should be included in the future as a complementary action to those already in place. The paper is not a technical article that is consistent with the normal scope of this journal and, therefore, contains references to different sorts of evidence in presenting its conclusions. Three sorts of references are included; scientific peer-reviewed articles, conference proceedings which have not been peerreviewed and reports by a number of organizations that may not be considered to have the same rigour of quality of evidence of the other references. The paper is based on work undertaken by the IEA Greenhouse Gas R&D Programme which is an international collaborative programme focusing on the assessment of technologies for reducing greenhouse gas emissions. The Programme is supported by sixteen countries and the European Commission and aims to provide expert reports for its members on measures for the mitigation of the full range of greenhouse gases. The Programme's work is used by its members to support national and international policy development.
International actions to reduce greenhouse
gas emissions In the late 1980s, increased public awareness of international environmental issues moved the climate change debate from the scientific to the political arena. Concerns about the possibility of global warming due to anthropogenic emissions of greenhouse gases prompted governments to form the International Panel on Climate Change (IPCC) in 1988 (Grant 1999). The key aims of the IPCC were to assess the available scientific information on climate change; to examine the potential environmental and social impacts of climate change and to formulate national and international response options. The IPCCs third assessment report was published in 2001, in the report the scientific committee stated that there was 'newer and stronger evidence' (from modelling of the global climate, reconstructions of past records and studies of the temperature records) that 'most of the observed warming over the last 50 years is likely to have been due to the increase of greenhouse gas concentrations' (IPCC 2001). The findings of the IPCC provide evidence that global climate change is a real effect and the potential consequences of its impact are now well acknowledged by governments and the public alike. International actions to reduce global emissions of greenhouse gases are undertaken through the United Nations Framework Convention on Climate Change (UNFCC). The UNFCC was signed by 155 countries at the so called 'Earth Summit' held in Rio de Janeiro in 1992 and came into force in 1994 after ratification by 50 countries (Grant 1999). A national government becomes a party to the convention by ratifying it. The ultimate objective of the Framework
From: BAINES, S. J. & WORDEN, R. H. (eds) 2004. Geological Storage of Carbon Dioxide. Geological Society, London, Special Publications, 233, 7-15. 0305-8719/047$ 15.00 © The Geological Society of London 2004.
8
J. GALE
Convention was 'to achieve stabilization of greenhouse gas concentrations in the atmosphere at a level that would prevent dangerous anthropogenic interference with the climate system'. The implementation of the convention is shaped by the Conference of the Parties (COP) which convenes at regular intervals (Grant 1999). The third Conference of the Parties (COP-3) was held in Kyoto, Japan in December 1997 and was where the parties debated and adopted the Kyoto Protocol. The main features of the Kyoto Protocol were that it called on the developed countries to reduce their greenhouse gas emissions by an average of 5.2% below 1990 levels by the end of the first five-year commitment period, 2008 to 2012. In recognition of their different circumstances, countries agreed different reduction targets. For example the European Union agreed an 8% reduction, whereas Norway and Australia were actually allowed to increase their emissions by 1 and 8% respectively, relative to their 1990 levels (Weyant & Hill 1999). The ratification of the Kyoto Protocol has been more protracted than was initially conceived. Several of the parties, led by the European Commission have now ratified the treaty but many such as the USA and Australia remain opposed to ratification.
Greenhouse gas abatement under the Kyoto Protocol Under the Kyoto Protocol, the greenhouse gas reduction commitments apply to six gases or groups of gases namely; carbon dioxide (CO2), methane (CH4), nitrous oxide (N2O), hydrofluorocarbons (HFCs), perfluorocarbons (PFCs) and sulphur hexafluoride (SF6). Those substances that contribute to ozone depletion, namely chlorofluorocarbons (CFCs) and halons are covered by the Montreal Protocol, a separate international agreement (Weyant & Hill 1999). The contributions of the different gases are weighted according to their Global Warming Potentials (GWP) (Weyant & Hill 1999). GWP is defined by IPCC as the time-integrated commitment to climate forcing from the instantaneous release of 1kg of a trace gas expressed relative to that of a reference gas (CO2). The time horizon used for the GWP index is typically 100 years (IPCC 2001). The contribution of CO2 to climate change is the most significant of all the basket of gases covered by the Kyoto Protocol. The contributions to global warming from anthropogenic sources from pre-industrial times to date are shown in Figure 1. Therefore, to make significant long-term reductions in global wanning, significant reductions in global anthropogenic CO2 emissions will be needed, as well as cuts in the other gases (Weyant & Hill 1999). The abatement measures proposed under the Kyoto Protocol to reduce emissions were:
Fig. 1. Contribution to global warming by greenhouse gas emissions from anthropogenic sources: pre-industrial times to date.
improved energy efficiency both in end-use and in the supply and conversion sectors; fuel switching to reduce the carbon intensity of fossil fuel use, such as substituting natural gas for coal; use of renewable energy; and use of nuclear power. The nuclear power option was promoted strongly by a number of parties at the outset of the process but technical doubts remain, primarily relating to safety which, along with attendant political issues, mean that nuclear power is not universally accepted as a mitigation measure. Many countries are focusing their greenhouse gas reduction targets on the first commitment period for the Kyoto Protocol (2008-2012) and will concentrate on the low cost, easily achieved options. These options can include: fuel switching (coal to natural gas), abatement of N2O emissions at adipic acid plants and methane emission reduction from natural gas pipelines and from coal mining. However, the low cost easy to achieve options will soon be used up and other more expensive abatement options will then be required, for later commitment periods. There is a growing recognition that if the UNFCC target of stabilization of atmospheric greenhouse gas emissions is going to be achieved then deep reductions in greenhouse gas emissions and in particular CO2 emissions will be required. To achieve deep reductions, wide-scale changes in the world's energy system would be needed, for example wide-scale use of renewable energy, substantial improvements in energy efficiency and fuel switching. Such changes are potentially attainable but the complexity of the change needed to the world's energy system is substantial and is unlikely to be practically achievable for at least another decade (Edmonds et al 2000; Lewis & Shinn 2001). The alternative would be to introduce CO2 capture at existing power and industrial plants combined with CO2 sequestration, which would have three advantages:
THE NEED FOR CO2 STORAGE
It would significantly reduce the complexity of any change in the world's energy system. It would allow the development of alternative sustainable technologies to take place at a reasonable technical pace (Edmonds et al. 2000). It could act as a stepping stone to a renewable hydrogen-based economy by helping to establish the necessary infrastructure to support such a low-carbon economy (Simbeck 2002). One criticism often levelled at CO2 capture and sequestration is that it is a high-cost abatement option (Edmonds et al. 2000). At its current technical status the costs are significant (of the order US $50 per tonne of CO2 abated); however, extensive research work is underway to reduce these costs and it has been estimated that by 2020 the costs of CO2 capture and sequestration will be cost-competitive with other alternative power generation technologies (Edmonds etal 2000).
CO2 sequestration options The options for sequestering CO2 fall into two broad categories. The first category includes the enhancement of natural sinks, such as forests and soils, and the second is capture and storage of CO2.
Enhancement of natural sinks The use of forests and soils to capture and store CO2 is considered by many to be an environmentally attractive method of reducing global CO2 emissions. However, in practical terms there are a number of disadvantages to both forestry and soils as CO2 sequestration options. Large land masses are required to store relatively modest amounts of CO2 (Chadwick et al 2000). IPCC has estimated that some 700 million hectares of forestry might be available globally, principally through reforestation schemes in the tropics, which could sequester between 220-320 Gt CO2 (IPCC 2001). This is about 11-16% of the atmospheric CO2 that would need to be absorbed to stabilize global emissions (IPCC 2001). Increasing use of land for forestry could also compete with agricultural use which would not be acceptable, particularly in areas of the world where population growth is increasing the demand for food. Agricultural practices can be modified to promote the uptake of carbon in soils but the process cannot be considered as either permanent or irreversible (IPCC 2001). The IPCC has estimated that another 84-161 Gt CO2 could be sequestered in agricultural soils, that is 4-8% of the total CO2 that needs to be absorbed (IPCC 2001). In both cases there is the issue of permanence
9
(Trexler 1999; Lewis & Shinn 2001). The carbon (CO2) stored in trees could be held for 50-100 years or more provided that the area remains forested. Problems with permanence will arise if deforestation occurs due to logging or clearance activities or if the increased temperature due to climate change makes the areas unsuitable for forestry. Ideally, the CO2 should be stored well past the end of the fossil fuel economy which could conceivably last for 200 years or more with the development of new exploration technology and development of non-conventional fossil fuel sources like natural gas hydrates. Another issue, particularly relevant to international carbon trading, is that of verification of sequestered CO2 It is extremely difficult to monitor the amount of CO2 that is stored physically within forests with a reasonable degree of accuracy; hence it is extremely difficult to verify how much CO2 will be stored in a forestry sequestration scheme (Trexler 1999; Lewis & Shinn 2001). Recently there have been concerns that some tropical forests, in particular the Amazon rainforest may not be a net absorber of CO2 despite previous assertions (Di Paola 2002). Also, afforestation may take place for reasons other than greenhouse gas reduction and it will be difficult to distinguish between those measures that would have occurred anyway and those that are deliberate sequestration schemes. Sequestration of CO2 in soils is another option that has been advocated. Most of the issues relating to forestry are also common to those for soils. In particular, there are concerns over the permanence of CO2 in soils. In certain regions of the world agricultural practices, such as no-till, have been introduced to reduce soil erosion and such measures also indirectly improve CO2 sequestration. However, the very nature of farming means that agricultural practices can change quickly, even year by year, due to factors other than the need for greenhouse gas reduction and such changes mean that the transition from CO2 sequestration to a net release of CO2 could occur within several years. Concerns have also been raised that areas of the world that are currently net absorbers of CO2 could release CO2 if the global temperature increases much further (White 1998; Di Paola 2001). Overall, neither of these methods for enhancing natural sinks of CO2 are, on their own or in combination, going to make deep reductions in CO2 emissions achievable.
CO2 capture and storage CO2 can be captured from a variety of anthropogenic sources such as power and large industrial plants and either stored in the oceans or in geological reservoirs. The oceans already act as a large natural sink for CO2; therefore, there is some merit in attempting
10
J. GALE
to increase their storage potential (Ormerod et al 1999). There are two potential routes for ocean storage of CO2: direct injection of captured CO2 into the deep ocean; and fertilization of the oceans with nutrients to increase the draw down of CO2 from the atmosphere. Although very different in concept, both these methods have the potential to increase the amount of CO2 storage in the oceans significantly (Ormerod et al. 1999). Of the two options, there exist greater uncertainties regarding the concept of ocean fertilization. The principal concerns include: the impact on the marine community and ecological balance (Ormerod et al 1999). Modelling work has indicated that the increased amount of CO2 sequestered could be offset by biological production (Ormerod et al 1999). The science associated with the injection of CO2 into the deep ocean, although imperfect, is somewhat better understood. However, there are considerable uncertainties over the environmental impact of deep ocean disposal (Omerod et al 1999). There is also a political dimension to ocean storage in that the disposal of CO2 in the ocean is currently not allowed under international law. In addition, there is considerable opposition from the NGO community to ocean storage that makes it a difficult option for governments to consider. Because of the uncertainties over the science and the attendant political issues, there is considerable reluctance internationally to consider ocean storage as a mitigation option at present. The alternative option is the use of geological reservoirs for the injection and storage of CO2 This option is discussed in detail in the following sections of the paper.
The geological options for storing CO2 Several potential geological reservoirs can be used to store captured CO2 from power and large industrial plants. These include: depleted and disused oil and gas fields, deep saline aquifers and deep unminable coal seams (Lewis & Shinn 2001). The global storage capacity for these geological reservoirs has been estimated and is compared with the projected total emissions between 2000 and 2050 according to IPCCs 'business as usual' scenario in Table 1 (Davison etal 2001). The capacity estimates for these reservoirs show that geological storage of CO2 can make a substantial impact on CO2 emissions reduction. The estimates for storage in aquifers were made in the early 1990s; research work underway should assist in firming up these numbers. A study completed for NW Europe suggested some 800 Gt CO2 could be stored in that region alone, mostly in deep saline aquifers (Gale etal 2001).
Table 1. Global storage capacity of potential geological reservoirs for carbon dioxide.
Global capacity Storage option
GtCO 2
% of emissions to 2050
Depleted oil and gas fields 920 45 Deep saline aquifers 400-10 000 20-500 Unminable coal seams 20 <2
The geological reservoirs under consideration have already held petroleum, natural gas and saline water for many millions of years. In addition, there are many examples of gas fields that contain high CO2 concentrations (up to 80%) such as the Natuna field offshore Indonesia, the Blue Whale prospect offshore Vietnam and some fields in the Otway Basin offshore Australia (Baines & Worden 2000). The CO2 in these fields has been contained for geological timescales. In the case of the Blue Whale field, CO2 entered the field some 50 million years ago (Baines & Worden 2000). In addition, many natural geological CO2 reservoirs throughout the world, have held CO2 for many thousands of years. Natural CO2 reservoirs are known to exist in the USA (a typical case is Bravo Dome in New Mexico) and throughout Europe (Stevens et al 200la). Research has indicated that the CO2 in fields such as Bravo Dome began entering 50000 years ago (Baines & Worden 2000). Taken together these points should give confidence that by careful selection of the reservoirs the injected CO2 can be stored for similar timeframes. Thus the injected CO2 should be stored well past the end of the fossil fuel economy. Many geological reservoirs have been converted successfully to store natural gas. In Canada and the USA alone some 400 reservoirs, mostly oil and gas fields, now store 300 Gm3 of natural gas (EIA 2000). The duty imposed on these reservoirs is more severe than storage of CO2 because the natural gas is injected and withdrawn several times per year at regular intervals to meet fluctuations in consumer demand. Such cyclical operations have the potential to place additional geomechanical stresses on the reservoir compared to CO2 injection and storage. Again, we can draw from the natural gas storage experience which indicates that oil and gas reservoirs and aquifers can be adapted successfully to store gases.
Current geological storage operations Storage of CO2 in geological formations is not a new technology. Most of the equipment required can be
THE NEED FOR CO, STORAGE
readily adapted from existing oil and gas operations. CO2 injection into oil fields has been practised since the mid 1980s as part of enhanced oil recovery (EOR) operations. In the USA, there are some 74 CO2 enhanced oil recovery (CO2-EOR) operations underway that in total inject some 33Mt of CO2 annually, most of which comes from natural CO2 accumulations. However, some 3Mt of CO2 is used from anthropogenic sources, namely gas processing and fertilizer plants (Stevens et al 2001b). CO2EOR projects are also currently underway in Trinidad, Turkey and Canada. In Canada recently, a CO2-EOR project has been established by EnCana at the Weyburn oil field in Southern Saskatchewan. This project will inject 19Mt CO2 and extend the life of the oil field by 25 years (Brown etal 2001). To date, most CO2-EOR operations have focused on the use of CO2 to enhance oil production and the storage of the CO2 in the oil field has been or still is a secondary issue. This situation would not change unless there was a fiscal inducement to encourage CO2 storage. However, when the CO2 is injected into the oil field typically up to 50% of the injected volume remains in the reservoir either in the immobile oil or dissolved in the formation water. The remaining CO2 is extracted from the produced oil and recycled for re-injection. Monitoring activities concentrate on injection volumes for commercial reasons. Although smaller in scale than the CO2-EOR operations, in Canada (mostly in Alberta) there are 31 acid-gas injection projects currently operating. Many of these injection projects have been underway since the early 1990s (Bachu & Gunter 2004). Acid-gas streams, consisting primarily of hydrogen sulphide (H2S) and carbon dioxide (CO2), are commonly generated as by-products of the gas sweetening process used to bring produced gases up to safety and quality standards and to meet pipeline specifications for transport and sale. In the current economic climate and attendant low sulphur prices, storage of the acid gases in geological reservoirs offers an economic and environmentally acceptable way of disposing of the acid gas. In 2001, some 1 Mt CO2 was injected into geological reservoirs; these reservoirs included both disused oil and gas fields and deep saline reservoirs (Bachu & Gunter 2004). There is already one commercial CO2 injection project at the Sleipner gas field in the North Sea operated by Statoil. Here CO2 has been injected into a deep saline aquifer, known as the Utsira formation, since 1996 (Torp 2001). The CO2 comes from the Sleipner West gas field where it is stripped from the produced gas and reinjected into the aquifer that overlies the gas field. The aquifer lies at a depth of 800-1500m below the sea floor. So far over 5MT CO2 have been reinjected into the Utsira formation at a rate of nominally 1 Mi/a. Interest is also develop-
11
ing in the use of CO2-EOR operations in the North Sea sector. The only pilot-scale project involving the injection of CO2 into coal seams is in the San Juan Basin in the USA (Gale & Freund 2001). However, further projects are under development in Canada and Poland (Gale 2003). The technology for CO2 storage in deep unmineable coal seams is at an earlier stage of development than for injection in oil/gas fields and aquifers.
Addressing the outstanding issues The technical and economic issues related to geological storage have not yet been fully resolved and the outstanding issues must be addressed before the technology can be accepted by the policy makers and public for wide scale implementation (Stevens et al 2000). The key barriers to wide scale implementation of the technology have been considered in a study undertaken by the IEA Greenhouse Gas R & D Programme. The study indicated that the main barriers were: the high costs of capturing, processing and transporting anthropogenic CO2; incomplete understanding of reservoir processes and storage methods; monitoring, verification and environmental safety of CO2 storage; lack of functioning emission trading system and storage regulations; and commercial/organizational conflicts between CO2 storage and production in EOR or natural gas recovery. In the first two cases, extensive R & D activity is needed to understand the mechanisms involved in the storage process and to develop evolutionary improvements to existing methods of CO2 capture, as well as breakthrough development of technologies that can dramatically lower capture costs. Several substantial research programmes are underway throughout the world (e.g. Canada, USA, Europe, Japan and Australia), supported by both governments and industry, should begin to address these issues within the next few years (Gale 2003). The issue of the effectiveness of storage of the CO2 is related to the understanding of reservoir processes. It will only be overcome by demonstration of the technology to prove that the injected CO2 can be monitored and the integrity of the reservoir will not be compromised either directly (i.e. through reservoir over pressurization) or indirectly (via reaction of CO2 with the caprock) by storing CO2. Two major monitoring programmes now underway, along with others that are planned, will also begin to address the
12
J. GALE
Fig. 2. Monitoring of injected CO2 using seismic surveying at Sleipner (Torp 2001).
issues raised. These monitoring projects are the Saline Aquifer CO2 Storage (SACS) project, which is monitoring the injection of CO2 into the Utsira formation in the North Sea and the Weyburn Monitoring Project that is monitoring CO2 injection into the Weyburn oil field in Canada (Moberg el al 2003; Torp & Gale 2003). To date, the SACS project has provided information to help address the issues of environmental safety of CO2 storage. The project has
shown that the injected CO2 can be monitored within the Utsira formation using time-lapse seismic surveying (Fig. 2; Arts et al. 2001). Results have shown that the injected CO2 migrates to the top of the reservoir below the caprock and then migrates laterally under the caprock as the injected volume increases. Discrete shale layers within the reservoir provide barriers to this upward migration, causing the CO2 to disperse more widely in the reservoir and increasing
THE NEED FOR CO2 STORAGE
the contact area between the CO2 and the formation water. This should assist dissolution of the CO2 in the brine solution within the reservoir (Chadwick et al. 2003). The caprock layer contains many local domal and anticline structures, which act as traps and or channels for CO2 migration within the reservoir. Reservoir simulation tools have been developed to monitor the migration of the CO2 in the reservoir. The simulation packages have been calibrated against repeat seismic surveys and have demonstrated that they are capable of replicating the position of the CO2 in the reservoir (Van der Meer 2000; Zweigel et al 2000). Good agreement has also been achieved by comparing the volume of CO2 injected with results derived from the reservoir simulation software (Arts et al 2003). The CO2 can, therefore, not only be monitored within the reservoir but verification of injected CO2 can be achieved. Geochemical studies are underway to study the reaction of the CO2 with caprock and the results of this work were available in early 2003 (Pearce et al 2001). The cumulative experiences of the SACS project has been embodied in a best practice manual that has been published by the project to assist other organizations planning CO2 injection projects to take advantage of the learning process from this project and to facilitate new CO2 storage projects (IEA 2003). A number of research projects, for example Weyburn, are studying the environmental and safety issues concerning geological storage (Moberg et al 2003). Several projects are undertaking detailed risk assessment studies on representative geological storage reservoirs (Gale 2003). These risk assessment studies will address the key areas such as the main events that could cause migration of the injected CO2 out of the reservoir (both natural and man made), the seepage rates that might be expected and their timescales. Knowledge of the seepage rates from reservoirs will allow the environmental impact of CO2 migration from a reservoir to the sea floor or the ground surface to be determined. Much of this work was not due to be completed until late 2003/early 2004 and hence a scientifically substantiated assessment of the environmental impact and safety of geological storage is now in progress (Gale 2003). Emissions trading regimes are now being formulated in countries such as Canada, UK and Denmark and by the European Commission and many other countries are considering trading schemes to counter the commercial barriers. Although these schemes are at an early stage of development the potential for some form of incentive (e.g. taxation or credits) for CO2 sequestration would help to overcome these barriers and to build a commercial basis for CO2 capture and storage to operate from. The one issue that has not been fully considered concerns the regulatory system that will be required,
13
in particular codes of practice, standards and regulations to ensure safe and effective capture, transmission and storage of CO2. In North America, there are existing standards relating to pipeline transportation of CO2 and industry guidelines for the use of CO2 in EOR operations in the USA (Gale & Davison 2003). In other areas of the world, where CO2 capture and storage is likely to be deployed, there are currently no major CO2 pipelines and no major industry sector using CO2 in large volumes for production or sequestration purposes. Consequently, standards for CO2 capture and storage have not been developed. However, there are analogous gas storage installations in these countries, which might provide a basis for comparison - natural gas storage is a regulated activity that is practised throughout the world with limited public concern. It is expected that as the technology begins to gain impetus a regulatory regime will quickly follow.
Conclusions If significant reductions in anthropogenic greenhouse gas emissions are needed to stabilize atmospheric concentrations of greenhouse gases then it is unlikely that the mitigation measures proposed under the Kyoto Protocol will be sufficient to achieve this objective. Therefore the proposed technologies (fuel switching, energy efficiency and renewable energy) need to be supplemented with additional options to achieve the necessary deep reductions. Several CO2 sequestration options have been proposed that include enhancing natural sinks such as forests and soils. Both these options suffer a number of disadvantages in that the amount of CO2 that can be stored is small in comparison to geological storage and there are issues relating to the permanence of the stored CO2. In both cases, these options can contribute to a broad portfolio of mitigation options but neither cumulatively, or alone, will they contribute significantly towards deep reductions in CO2 emissions. Capturing the CO2 from power plants and large industrial plants, however, does have the potential to achieve deep reductions in emissions. The ocean or geological reservoirs can be used to store the CO2. Currently there is uncertainty concerning the impact of storing CO2 in the ocean and there are also questions regarding the legality of ocean storage of CO2 under international conventions. It is therefore unlikely that ocean storage will be accepted internationally as a mitigation option in the foreseeable future. As far as geological storage is concerned, there is sufficient storage capacity worldwide to enable deep reductions to be achieved if the technology is globally implemented. The geological reservoirs concerned are; depleted oil and gas fields, deep saline aquifers and deep unmineable coal
14
J. GALE
seams. Such reservoirs, have already stored petroleum, natural gas and saline water for millions of years and have been adapted for use as natural gas stores which is a more arduous duty than CO2 storage alone. This gives confidence that CO2 storage in geological reservoirs can be stored well past the end of the fossil fuel economy (200 years or more from now) and storage should be both safe and environmentally acceptable. In addition, by capturing the CO2 from power plants, the existing energy supply infrastructure can be maintained whilst achieving significant reductions in CO2 emissions. Geological storage of CO2 is, therefore, the key mitigation technology to be used in conjunction with the other abatement measures such as energy efficiency, fuel switching and renewable energy to achieve deep reductions in atmospheric greenhouse gas concentrations.
sequestration offshore, a provable greenhouse mitigation strategy, OffshoreMagazine,34-135, November. CHADWICK, A., ZWEIGEL., GREGERSEN, U., KIRBY, K. A. & JOHANNESSEN, P. N. 2003. Geological characterization of CO2 storage sites: Lessons from Sleipner, Northern North Sea. In: GALE, J. & KAYA, Y. (eds) Proceedings of the 6th International Conference on Greenhouse Gas Control Technologies. Elsevier Science, Kyoto, Japan. DAVISON, J., FREUND, P. & SMITH, A. 2001. Putting carbon back in the ground, IEA Greenhouse Gas R&D Programme, February. Di PAOLA, M. 2001. Global look at carbon sinks. Global Environmental Change Report, XII (22), 23 November, Aspen Publishers. Di PAOLA, M. 2002. Mighty Amazon, sink or source. Global Environmental Change Report, XIV (8), 26 April, Aspen Publishers. EDMONDS, J., FREUND, P. & DOOLEY, J. J. 2000. The role of carbon management technologies in addressing The work from which this paper has been drawn has been atmospheric stabilisation of greenhouse gases. In: carried out by the IE A Greenhouse Gas R&D Programme. DURIE, R., PAULSON, C. & SMITH, A. (eds) Any opinions expressed are those of the author and do not Proceedings of the 5th International Conference on represent the views of the Operating Agent or the members Greenhouse Gas Control Technologies, pp. 46-51. CSIRO, Cairns, Australia. of the Programme, nor of the International Energy Agency or its secretariat. EIA 2000. Natural gas storage, energy information administration publication available at http://www.eia.doe. gov/fuelnatgas. html. GALE, J. 2003. Geological storage of CO2: what's known, References where are the gaps and what more needs to be done. In: GALE, J. & KAYA, Y. (eds) Proceedings of the 6th ARTS, R., BREVIK, I., EIKEN, O., SOLLIE, R., CAUSSE, E. & VAN DER MEER, L. 2001. Geophysical methods for International Conference on Greenhouse Gas Control monitoring marine aquifer CO2 storage - Sleipner Technologies. Elsevier Science, Kyoto, Japan, pp. 207-212. experiences. In: DURIE, R., PAULSON, C., & SMITH, A. th GALE, J. & FREUND, P. 2001. Coal-bed methane enhance(eds) Proceedings of the 5 International Conference ment with CO2 sequestration: worldwide potential. on Greenhouse Gas Control Technologies, pp. Environmental Geosciences, 8, 210-217. 366-371. CSIRO, Cairns, Australia. ARTS, R., EIKEN, O. CHADWICK, A., ZWEIGEL, P., VAN DER GALE, J. & DAVISON, J. 2003. Transmission of CO2: safety and economic issues. In: GALE, J. & KAYA, Y. (eds) MEER, L. & ZINSZNER B. 2003. Monitoring of CO2 Proceedings of the 6th International Conference on injected at Sleipner using time lapse seismic data. In: Greenhouse Gas Control Technologies. Elsevier GALE, J. & KAYA, Y. (eds) Proceedings of the 6th Science, Kyoto, Japan, pp. 207-522. International Conference on Greenhouse Gas Control GALE, J., CHRISTENSEN, N.P.C., CUTLER, A. & TORP T Technologies, pp. 347-351. Kyoto, Japan, Elsevier 2001. Demonstrating the potential for geological Science. storage of CO2: The Sleipner and GESTCO Projects. BACHU, S. & GUNTER, W. 2004. Acid gas re-injection: an Environmental Geosciences, 8,160-165. innovative storage opportunity. In: BAINES, S. J. & GRANT, J. 1999. Buenos Aries and beyond: a guide to WORDEN, R. H. (eds) Geological Storage of Carbon climate change negotiations. IPIECA. Words and Dioxide. Geological Society, London, Special Publications, 233,225-234. Publication, Oxford, UK. IEA 2003. Saline Aquifer BAINES, S. J. & WORDEN, R. H. 2000. Geological CO2 disCO2 Storage Project (SACS): Best Practice Manual. Report no PH4/21, June. IEA Greenhouse Gas R&D posal: understanding the long term fate of CO2 in natProgramme, Cheltenham. urally occurring accumulations. In: DURIE, R., PAULSON, C. & SMITH, A. (eds) Proceedings of the 5th IPCC, Climate Change 2001 2001. The scientific basis. Summary for Policy Makers and Technical Summary International Conference on Greenhouse Gas Control of the Working Group I Report. Cambridge University Technologies, pp. 311-316. CSIRO, Cairns, Australia. Press, Cambridge. BROWN, K., JAZRAWI, W., WILSON, M. & MOBERG, R. 2001. LEWIS, C. & SHINN, J. 2001. Global warming - an oil and The role of enhanced oil recovery in carbon sequestration. The Weyburn monitoring project, a case study. gas company perspective: prospects for geologic sequestration? Environmental Geosciences, 8, In: Proceedings of the 1st National Conference on 177-186. Carbon Sequestration, Session 2A.1. Washington, USA. NETL Conference Services: www.netl.doe.gov/ MOBERG, R., STEWART, B. & STACHNIAK, D. 2003. The publications/proceedings/01/carbon_seq/2al.pdf. Weyburn CO2 monitoring and storage project. In: CHADWICK, A., HOLLOWAY, S. & RILEY, N. 2000. Deep CO2 GALE, J. & KAYA, Y. (eds) Proceedings of the 6th
THE NEED FOR CO9 STORAGE International Conference on Greenhouse Gas Control Technologies, Elsevier Science, Kyoto, Japan, pp. 219-226. ORMEROD, W. G. FREUND, P. & SMITH, A. 1999. Ocean storage of CO2. IEA Greenhouse Gas R&D Programme, February. PEARCE, J. M, CZERNICHOWSKI-LAURIOL, I., ROCHELLE, C. A., SPRINGER, N., BROSSE, E., SANJUN, B., BATEMAN, K. & LANINI B. 2001. How will reservoir and caprock react with injected CO2 at Sleipner? Preliminary evidence from experimental investigations. In: DURIE, R., PAULSON, C. & SMITH, A. (eds) Proceedings of the 5th International Conference on Greenhouse Gas Control Technologies, pp. 355-359. CSIRO, Cairns, Australia. SIMBECK, D. R. 2002. CO2 capture and storage - the essential bridge to the hydrogen economy. In: GALE, J. & KAYA, Y. (eds) Proceedings of the 6th International Conference on Greenhouse Gas Control Technologies, Elsevier Science, Kyoto, Japan, pp. 25-30. STEVENS, S., KUUSKRAA, V. A. & GALE J. 2000. Sequestration of CO2 in depleted oil and gas fields: global capacity, costs and barriers. In: DURIE, R., PAULSON, C. & SMITH, A. (eds) Proceedings of the 5th International Conference on Greenhouse Gas Control Technologies, pp. 278-283. CSIRO, Cairns, Australia. STEVENS, S., PEARCE, J. M. & RIGG, A. A. 2001a. Natural analogues for geologic storage of CO2: an integrated global research programme. In: Proceedings of the 1st National Conference on Carbon Sequestration. Session 6A.1 Washington, USA. NETL Conference Services: www.netl.doe.gov/publications/proceedings/01/carbon_seq/6al .pdf. STEVENS, S., KUUSKRAA, V. A., GALE, J. & BEECY, D. 2001&. CO2 injection and sequestration in depleted oil and gas fields and deep coal seams: worldwide potential and costs. Environmental Geosciences, 8, 200-209.
15
TORP, T. 2001. The Sleipner CO2 injection and the SACS. In: Proceedings of the 1st Annual European Energy and Transport Conference, Session 2C Assuring Cleaner and Safer Means of Using Fossil Fuels, European Commission, CG Energy and Transport, Barcelona, Spain. TORP, T. & GALE, J. 2003. Demonstrating storage of CO2 in geological reservoirs: the Sleipner and SACS Projects. In: GALE, J. & KAYA, Y. (eds) Proceedings of the 6th International Conference on Greenhouse Gas Control Technologies, Elsevier Science, Kyoto, Japan, pp. 311-316. TREXLER, M. 1999. Forestry. Report of an IPIECA Workshop, Technology Assessment and Climate Change Mitigation, IPIECA. VAN DER MEER, L. G. H. 2000. Prediction of migration of CO2 after injected in a saline aquifer; reservoir history matching of a 4D seismic image with a compositional gas/water model. In: DURIE, R., PAULSON, C. & SMITH, A. (eds) Proceedings of the 5th International Conference on Greenhouse Gas Control Technologies, pp. 378-384. CSIRO, Cairns, Australia. WEYANT, J. P. & HILL, J. N. 1999. The costs of the Kyoto Protocol - a multi gas evaluation, introduction and overview. In: WEYART, J. P. (ed.) A Special Edition of The Energy Journal, Elsevier Science. WHITE, A. 1998. Impacts of climate change on natural vegetation in climate change and its impacts. The Meteorological Office and Department of Environment and Transport, UK. ZWEIGEL, P., HAMBOURG, M., ARTS, R., LOTHE, A., SYLTA, 0. & T0mmeras, A. 2000. Prediction of migration of CO2 injected into an underground repository: reservoir geology and migration modelling in the Sleipner case. In: DURIE, R., PAULSON, C. & SMITH, A. (eds) Proceedings of the 5th International Conference on Greenhouse Gas Control Technologies, pp. 360-365. CSIRO, Cairns, Australia.
This page intentionally left blank
The case for underground CO2 sequestration in northern Europe R. A. CHADWICK, S. HOLLOWAY, M. S. BROOK & G. A. KIRBY British Geological Survey, Kingsley Dunham Centre, Key worth, Nottingham NG12 5GG, UK (e-mail: rach @ bgs.ac. uk) Abstract: In northern Europe numerous industrial point sources of CO2 surround the North Sea Basin, which contains a number of viable underground sequestration opportunities. These include injection into depleted oil and gas fields and into major regional aquifers; the latter probably offering the greatest ultimate storage potential. At the Sleipner gas field, CO2 is being injected into the Utsira Sand, a large saline aquifer. More than 6Mt of CO2 have currently been injected, with a projected final target of about 20 Mt. Time-lapse seismic reflection data are being used to monitor the operation and have provided clear images of the CO2 plume and its development with time. Moreover, CO2 volumetrics derived from the seismic data are consistent with the well injection figures. In conjunction with reservoir simulation studies, time-lapse seismic monitoring seems, therefore, to offer an effective means of predicting the future growth, migration and dispersion of the CO2 plume. Another important aquifer, the Bunter Sandstone, stretches from Britain to Poland. In the UK sector alone, the pore volume in structural closures is equivalent to about 350 years' worth of current CO2 emissions from UK power generation. Industrial CO2 sources in northern Europe are well placed to exploit these major subsurface reservoirs and European countries are technically very well equipped to use and develop this emerging technology.
Atmospheric levels of CO2 have risen from about 280 ppm immediately prior to the industrial revolution to about 370ppm today, due largely to fossil fuel combustion. Current global anthropogenic emissions amount to some 23 Gigatonnes (Gt) of CO2 per year. Radical dislocations to contemporary life would be required to reduce consumption of fossil fuels and provide alternative non-fossil energy sources in sufficient quantity to stabilize CO2 levels in the atmosphere. For example, stabilization at 450ppm would require anthropogenic CO2 emissions to drop below 1990 levels by about 2030 (sharply reversing the current upward trend), and continue to decrease steadily thereafter. Ultimately, anthropogenic CO2 emissions would have to decline to the level of persistent natural land and ocean sinks, probably less than 0.73 Gt CO2 per year (Prentice et al. 2001), requiring a 97% cut in current emissions. Bearing in mind that power generation sensu stricto accounts for only about 30% of global emissions, an effective emissions reduction strategy must also address other anthropogenic sources of C02. A potentially vital tool in achieving these goals is CO2 sequestration, which can be defined as 'the storage of anthropogenic CO2 in a domain of the planet, other than the atmosphere, until any greenhouse crisis has passed'. Thus, sequestered CO2 would have to remain effectively isolated from the atmosphere until atmospheric CO2 levels were well into decline. This implies the need for secure storage until well past the end of the fossil fuel era - some hundreds of years into the future. An important
benefit of sequestration is its potential for lessening environmental damage to the atmosphere without requiring profound changes to current lifestyles. The aim of this paper is to argue the case for largescale underground sequestration of CO2 in northern Europe. To help set the context, we will briefly review two other commonly proposed CO2 mitigation methods, reforestation and deep ocean disposal.
Direct uptake of atmospheric carbon by terrestrial ecosystems Abstracting CO2 directly from the atmosphere by photosynthesis has the immediate advantage of potentially sequestering all anthropogenic emissions to the atmosphere irrespective of source. It would also avoid any (potentially costly) modifications to industrial plant. Terrestrial ecosystems take up carbon if there is an excess of primary production over respiration and other oxidative processes (decomposition or combustion of organic material). Since the 1980s the total amount of carbon stored in terrestrial ecosystems has actually increased, due to enhanced plant growth caused by higher levels of atmospheric CO2, fertilization by atmospheric nitrogen deposition, and changes in land management practices (Prentice et al. 2001). This is despite the fact that other land-use changes, particularly deforestation, have greatly reduced the amount of terrestrial carbon stored in many parts of the world. Reforestation is considered a primary method by
From: BAINES, S. J. & WORDEN, R. H. (eds) 2004. Geological Storage of Carbon Dioxide. Geological Society, London, Special Publications, 233,17-28. 0305-8719/047$ 15.00 © The Geological Society of London 2004.
18
R. A. CHADWICK ETAL
densely populated nations of northern Europe. To illustrate, the Drax coal-fired power-station in northern England (Fig. la) generates nearly 4000 MW of electricity, emitting some 20 Mt of CO2 per year. Two million hectares of forest would need to be planted to sequester seventy years' worth of its emissions; an area that would cover much of central England (Fig. Ib). Furthermore, the mechanisms controlling overall carbon budgets of terrestrial ecosystems are complex and not fully understood. Cao & Woodward (1998) have suggested that if global warming continues, by the middle of this century, forests and other plant ecosystems may no longer act as net carbon sinks due to increased respiration rates releasing stored carbon from soils. In addition, a worrying increase in the frequency of damaging wind-storms casts further doubt on tree planting as an effective CO2 mitigation option for Europe (Chadwick et al 20010). In October 1987 a relatively small storm destroyed 17 million trees in SE England. This event was dwarfed by the severe storm which swept across Europe around Christmas 1999, destroying over 270 million trees in France alone. In a European context therefore, although tree planting may well be desirable for a host of other environmental reasons, its value as a carbon sequestration strategy is severely limited.
Carbon capture from industrial plant
Fig. 1. (a) The Drax B coal-fired power-station, northern England [photograph courtesy of Vincent Lowe] (b) New forest (rectangle) required to absorb 70 years' worth of CO? emissions from Drax B.
which an increase in the total amount of carbon stored in the terrestrial biosphere could be achieved. Approximate upper limits for the impact of reforestation on atmospheric CO2 levels over a century timescale indicate that if all land-use change were completely reversed over the twenty-first century, reductions of between 293 Gt CO2 (about 40ppm) and 513Gt CO2 (about 70ppm) could be achieved (Prentice et al. 2001). However, such a radical reversal of land-use is not a practical option for the
Given the practical limitations of creating new terrestrial ecosystems to act as carbon sinks, there is significant scope for a technology-based approach to carbon sequestration. This involves the capture and separation of CO2 from flue gases and other waste gas streams (Holloway 2001), followed by longterm storage in a location that is isolated from the atmosphere. Though it is realistically applicable only to large industrial point sources of CO2, capture and sequestration can make a significant impact on emissions. For example, in 1998, total UK emissions of CO2 amounted to 0.572 Gt. Of this, more than 20% was produced by the 25 largest industrial point sources (power-stations or integrated steel manufacturing plants). If, say, 85% of this CO2 could be captured and sequestered, it would provide a 17% reduction in total UK emissions.
Costs Much of the debate about costs of CO2 capture is based around power plants, which form by far the largest category of industrial point sources. Costs of CO2 capture (including compression for transport by pipeline) from power stations depend mainly on the type of plant, and are likely to be in the range
CO2 SEQUESTRATION IN NORTHERN EUROPE
19
Fig. 2. Major industrial point sources of CO2 in the area surrounding the North Sea and the approximate extents of the Utsira Sand and Bunter Sandstone.
US$18-72 per tonne of CO2 avoided, which equates to about US$66-264 per tonne of carbon avoided. Even allowing for some additional costs for transport and injection, prices compare favourably with the cost of the UK renewables obligation, which is estimated at £310 ($430) per tonne of carbon avoided (Department of Trade and Industry 2000). There is great potential for technological improvements to significantly lower costs and also for the development of new types of power plant and power cycles (Herzog 1999). Furthermore, if CO2 were to be sequestered in depleted oil or gas fields (see below), part of the cost could be offset by enhanced oil or gas recovery (EOR).
Transport Captured CO2 would most likely be transported to a sequestration site by pipeline, similar to those already used in the USA, connecting sources of CO2 with EOR projects in the Permian Basin of Texas (Doctor et al 2001). In northern Europe a large number of point sources of CO2 in several countries surround the North Sea (Fig. 2). These include power stations, integrated steel manufacturing
plants, cement factories, oil refineries and petrochemical complexes. The mature oil and gas production infrastructure already in place in the North Sea could perhaps be used to sequester emissions from these facilities. Transport by ship to offshore sequestration sites could also be an option.
Sequestration options Ocean sequestration The oceans provide the largest natural sink of atmospheric CO2, having absorbed roughly a third of anthropogenic CO2 production since the start of the industrial revolution (Siegenthaler & Joos 1992). However, the rate at which the oceans currently absorb CO2 is not keeping pace with recent large inputs from fossil-fuel burning and deforestation. By injecting man-made CO2 directly into the deep ocean it is possible to 'short-circuit' the relatively slow exchange of carbon across the major natural interfaces (atmosphere - ocean and shallow ocean deep ocean) and thereby effect a partial solution to the greenhouse problem. The sequestration efficacy of deep ocean injection (the ability of the ocean to
20
R. A. CHADWICK ETAL.
retain CO2 over time) depends on the injection location and ocean circulation patterns. Computer models of CO2 disposal into the deep ocean (e.g. Orr et al 2001; Orr 2002) predict that about 30 to 60% of CO2 injected at depths of 1500m, and about 50 to 90% of CO2 injected at 3000m, could be retained after 500 years. Still higher proportions could possibly be retained for longer periods if the CO2 were injected in liquid form at depths beneath 3000m. This might result in the accumulation of a 'lake' of liquid CO2 or the formation of CO2 hydrate on the sea bed (IEA Greenhouse Gas R & D Programme 2000). In less favourable circumstances however, dissolution and equilibration within the oceanatmosphere system would take place on a shorter timescale. The main effect of ocean sequestration therefore would be to act as a buffer, working on a scale of hundreds of years, which would prevent a very high early 'peak' of atmospheric concentration arising from CO2 added directly to the atmosphere. In practical terms, northern Europe is not well placed to use ocean sequestration, the nearest suitable deep water being over 1000km to the west. More generally, the possible effects of ocean sequestration on the marine environment are not well constrained. Although considerable research effort is underway to understand and minimize these issues, under current legislation, sequestration of CO2 into the oceans could be illegal, for example under the terms of the London Convention that prohibits the dumping of waste at sea. Thus, although the ocean potentially represents an enormous carbon sink, environmental issues may prevent its widespread
Fig. 3. (a) Density of CO2 over a range of geothermal gradients and depths, (b) Phase diagram for CO2.
use
Subsurface sequestration CO2 can be stored underground by direct injection into porous reservoir rocks such as are found in depleted oil and gas fields and in regional saline aquifers. A specific advantage that underground storage offers over ocean sequestration or terrestrial sinks is that it aims to isolate the CO2 from the biosphere for very long periods of time (tens to hundreds of thousands of years). Under most reservoir conditions CO2 is buoyant and will tend to migrate towards the top of the reservoir until it reaches the caprock. The fact that many oil and gas fields, and a number of natural CO2 accumulations, have existed for millions of years, amply demonstrates that buoyant fluids can be retained in the subsurface for geological timescales. Subsurface sequestration potentially offers the very large storage capacities necessary to make a serious impact on global emissions. Storing CO2 underground is volumetrically very efficient. As CO2 is injected into the subsurface it
undergoes a sharp increase in density associated with a phase change from gas to liquid or supercritical fluid (Fig. 3). The depth at which this takes place is generally between about 500 and 1000m, and is principally dependent on the geothermal gradient (Fig. 3a). At typical reservoir conditions, one tonne of CO2 occupies less than 6m3 of rock (assuming 30% porosity and displacement of 80% of in situ pore-water). Ideal conditions for subsurface storage therefore are at depths of about 1000m, where CO2 is in the supercritical phase, pore-waters are saline (non-potable) and porosities, in many sedimentary basins, are likely to be high. As CO2 is injected into the pore spaces of the reservoir rock, it displaces much of the pore fluid. If the permeability of the rock is low, or there are barriers to fluid flow, such as faults, injection will cause a progressive increase in the fluid pressure centred on the injection point. This will limit the rate at which CO2 can be injected, and may ultimately limit the amount of CO2 that can be practically stored. Structurally complex and compartmentalized reservoirs are therefore likely to be less suited to CO2 storage than large unfaulted ones.
CO2 SEQUESTRATION IN NORTHERN EUROPE
Oil and gas fields. The simplest sequestration option is to inject CO2 into a closed, sealed structural feature. Exhausted or depleted oil and gas fields offer the a priori benefit of having caprocks of proven sealing efficacy. Furthermore, they commonly have lowered formation pressures, which renders them particularly suitable for subsequent reinjection. In addition, CO2 acts as a solvent for hydrocarbons making it an effective agent for enhanced oil recovery (EOR), whose cost benefits can usefully offset the costs of capture and injection. A potential drawback of storage in old hydrocarbon fields includes the possibility of leakage through abandoned production and exploration wells. There is also the possibility that reduction of formation pressures during the production stage, followed by re-pressuring during CO2 injection, may damage reservoir caprock integrity. Regional saline aquifers. Regional saline aquifers provide an important alternative option for subsurface CO2 sequestration, commonly having good porosity and permeability characteristics and offering very large potential storage volumes. In the North Sea Basin some aquifers such as the Utsira Sand (Fig. 2), are relatively flat-lying, whereas others such as the Bunter Sandstone (Fig. 2) have major structural closures. Providing the aquifer is sufficiently large, it is not necessary to inject CO2 into a specific closed structure to ensure stable long-term containment. CO2 injected into a relatively flat-lying subsurface reservoir will rise to the reservoir top and accumulate in any small domes or other closed structures. As each structure becomes full, the CO2 will spill from it and migrate laterally to the next such structure along the migration path and so on (Zweigel et al 2001). Thus, as the CO2 migrates within the reservoir, it may become distributed over many pools within many small closures. With time, depletion of these accumulations is likely to take place as a consequence of dissolution, an important process capable of sequestering CO2 more or less permanently. The solubility of CO2 under typical reservoir conditions at a salinity of 3% is about 49kgm~ 3 , corresponding to a volume of free CO2 of about 7% of the pore volume (Lindeberg 1996). As formation water becomes saturated with dissolved CO2 its density increases and it tends to sink in the reservoir, a process which effectively sequesters the CO2. The rate of dissolution depends on the amount of mixing of CO2 and formation water. Diffusion of CO2 into the water is assisted by thin but widespread accumulations, with a high surface area to volume ratio. Similarly, the presence of internal permeability barriers such as intra-reservoir shales will make the migration path of the CO2 through the reservoir more tortuous and encourage
21
mixing and dissolution. Nevertheless, for many accumulations, dissolution could be slow, in the order of a few thousand years for some injection scenarios (Ennis-King & Paterson 2001), unless there is some form of active mixing induced by fluid flow or convection within the reservoir (Lindeberg & Wessell-Berg 1997). Even so, if a relatively small (a few tens of Mt) amount of CO2 is injected into a very large reservoir, the combination of topographic 'roughness' at the reservoir top and gradual dissolution of the CO2 into the formation water means it may never reach the edge of the reservoir, even in the absence of a major structural closure (Lindeberg 1997). Another process leading to long-term sequestration is chemical 'fixing' by reaction of the injected CO2 with either the formation water or the reservoir rock. The amount of chemical fixing and the reaction timescales depend on pore-water chemistry, rock mineralogy and the length of the migration path (Czernichowski-Lauriol et al. 1996). In some circumstances, hydrodynamic trapping of the CO2 may also have a part to play (Bachu et al. 1996), though this should perhaps not be relied upon as a primary sequestration mechanism, not least because prior identification of reliable hydrodynamic traps is exceedingly difficult. To summarize, the interaction of three main processes will determine the fate of the CO2 in the reservoir. These are: immobilization in structural traps, dissolution into the surrounding formation water and geochemical reaction with the formation water or minerals making up the rock framework. All three of these processes acting either alone or in concert can produce long-term, effectively permanent subsurface storage.
Current and potential CO2 subsurface sequestration options Reservoir rocks suitable for CO2 storage are abundant beneath the North Sea. Many of these reservoirs are sealed by mudstones and shales whose seal efficiency is proven by the presence of hundreds of hydrocarbon fields. Published calculations suggest that the oil and gas fields of the North Sea may theoretically be able to store about 30Gt of CO2 (Van der Straaten 1996). Storage figures for gas fields are based on the assumption that the underground volume of CO2 that could be injected into a depleted gas field is the same as the underground volume of the produced gas, i.e. there has been no water drive during or after depletion, or irreversible compression of the reservoir rock. For the oil fields it is based on the principle that a volume of pore space equivalent to that occupied by the proven reserves of the field could be filled by
22
R. A. CHADWICK ETAL
Fig. 4. Regional seismic line across the Utsira Sand near Sleipner. (Seismic data courtesy of Schlumberger Ltd.)
CO2, injected in conjunction with EOR. These figures therefore probably represent likely upper storage limits. More detailed discussion of the sequestration potential offered by oil and gas fields, and related EOR issues, is beyond the scope of this paper. Instead we focus on regional saline aquifers, which ultimately may offer greater storage possibilities. Two examples are discussed below; the Utsira Sand is currently utilized for CO? storage, whereas the Bunter Sandstone has large, but so far untested, storage potential.
The Utsira Sand CO2 separated from natural gas produced at the Sleipner field (Baklid et al. 1996) is currently being injected into the Utsira Sand, a major saline aquifer lying within the thick Cenozoic post-rift succession of the North Sea (Figs 2 & 4). Injection started in 1996 and is planned to continue for about twenty years, at a rate of about one million tonnes per year. The CO2 injection point is at a depth of 1012m below sea level with an estimated formation temperature of about 36°C. Between this and the top of the reservoir at about 800m, with an estimated formation temperature of 29 °C, the CO2 forms a supercritical fluid with a roughly constant density of around 700kgirr3(Fig.3a). The regional extent of the Utsira Sand renders it potentially suitable for large-scale, multi-site disposal of CO2, or perhaps additionally as a temporary storage hub for CO2 EOR operations.
Reservoir properties. The Utsira Sand comprises a basinally-restricted deposit of Mio-Pliocene age forming a clearly defined seismic unit, pinching out to east and west, and seismically distinct from overlying and underlying strata (Fig. 4). The reservoir is highly elongated, extending for more than 400km from north to south and between 50 and 100km from east to west, with an area of some 26100km2 (Fig. 5). Its eastern and .western limits are defined by stratigraphical lap-out. To the SW it passes laterally into shaly sediments, and to the north it occupies a narrow, deepening channel. Locally, particularly in the north, depositional patterns are quite complex with some isolated depocentres, and lesser areas of non-deposition within the main depocentre. The top Utsira Sand surface (Fig. 5a) generally varies relatively smoothly, mainly in the range 550 to 1500m, but mostly from 700 to 1000m. Isopachs of the reservoir sand show two main depocentres (Fig. 5b). One is in the south, around Sleipner, where thicknesses locally exceed 300m. The second depocentre lies some 200km to the north of Sleipner. Here, the Utsira Sand is locally 200m thick with an underlying sandy unit adding further to the total reservoir thickness. Macroscopic and microscopic analysis of core and cuttings samples of the Utsira Sand show a largely uncemented fine-grained sand, with medium and occasional coarse grains (Fig. 6a). The grains are predominantly angular to sub-angular and consist primarily of quartz with some feldspar and shell fragments. Sheet silicates are present in small amounts (a few percent). The unit is interpreted as
C02 SEQUESTRATION IN NORTHERN EUROPE
23
Fig. 5. (a) Depth map to the top of the Utsira Sand, (b) Isopach map of the Utsira Sand.
being deposited by mass flows in a marine environment in water depths of 100m or more. The porosity of the Utsira Sand core ranges generally from 27-31 %, but reaches values as high as 42% (Zweigel et al 2001). Regional log porosities are uniform, in the range 35-40% over much of the reservoir. The proportion of clean sand in the reservoir unit varies generally from about 0.7 to 1.0. Finegrained lithologies are present mostly as a number of thin, intra-reservoir shales or mudstones, typically about 1 m thick. The total estimated pore volume of the reservoir is about 6 x 10nm3, similar to that of the Bunter Sandstone (see below). Caprock properties. The caprock succession overlying the Utsira reservoir is variable, and can be divided into three main units (Fig. 4). The Lower Seal forms a shaly basin-restricted unit, some 50-100m thick. The Middle Seal mostly comprises prograding sediment wedges of Pliocene age, dominantly shaly in the basin centre, but coarsening into a sandier facies both upwards and towards the basin margins. The Upper Seal comprises Quaternary strata, mostly glacio-marine clays and glacial tills. The Lower Seal extends well beyond the area currently occupied by the CO2 injected at Sleipner (Chadwick et al. 200Ib) and seems to be providing an effective seal at the present time. Cuttings samples comprise dominantly grey clay silts or silty clays. Most are massive (Fig. 6b) although some show a weak sedimentary lamination. XRD analysis typically reveals quartz (30%), undifferentiated mica (30%), kaolinite (14%), K-feldspar (5%), calcite (4%), smectite (4%), albite (2%), chlorite (1%), pyrite (1%) and gypsum (1%) together with
traces of drilling mud contamination. The clay fraction is generally dominated by illite with minor kaolinite and traces of chlorite and smectite. An accurate assessment of caprock sealing capacity is not yet available, but analysis of caprock core material, recently acquired at Sleipner, is currently ongoing. The cuttings samples are classified as nonorganic mudshales and mudstones according to the Krushin (1997) classification. Although the presence of small quantities of smectite may invalidate its predictions, typical quartz contents determined by XRD are consistent with displacement pore throat diameters in the range 14-^-0 nm. Such displacement pore-throat diameters predict capillary entry pressures of between about 2-5.5MPa (Krushin 1997), capable of trapping a CO2 column several hundred metres high. In addition, the predominant clay fabric with limited grain support resembles the type 'A' or type 'B' seals illustrated in Sneider et al. (1997), and stated to be capable of supporting a column of 35° API oil greater than 150m in height. Empirically, therefore, the caprock cutting samples suggest the presence of an effective seal at Sleipner, with capillary leakage of CO2 unlikely to occur. Geophysical monitoring of the CO2 injection plume. The CO2 injection plume at Sleipner is currently being monitored by time-lapse seismic methods. Baseline 3D seismic data were acquired in 1994, prior to injection, with repeat surveys acquired in October 1999 and October 2001, when some 2.35 and 4.26 million tonnes respectively of CO2 had been injected. The seismic data have proved notably successful
24
R. A. CHADWICK ETAL
uncertainty remains in a number of reservoir, fluid and seismic parameters, and a solution which verifies the injected volume has not yet been obtained (Arts et al 2004; Chadwick et al. 2004). Uncertainty notwithstanding, the time-lapse seismic data do illustrate that geophyisical monitoring of a CO2 sequestration operation is technically feasible, and is capable of providing the type of information necessary to demonstrate a long-term safety-case. The Utsira reservoir is particularly well suited to effective seismic monitoring, due to its favourable acoustic properties. This is not necessarily the case elsewhere, where more complex monitoring techniques such as multi-component seismic surveying may be required. Other geophysical methods such as passive seismic monitoring, microgravimetry, and electromagnetic techniques may also be usefully deployed in cost-effective long-term monitoring strategies.
The Bunter Sandstone
Fig. 6. (a) SEM image of Utsira Sand core material showing moderately well-sorted, subrounded to subangular fine to medium-grained sand, (b) SEM image of caprock cuttings material showing massive mudrock with several well-rounded fine-grained quartz grains (arrowed).
in imaging the CO2 plume in the subsurface (Fig. 7). The CO2 is evident as a series of high amplitude subhorizontal reflections above the injection point. These are interpreted as arising from individual thin layers of CO2 trapped beneath the thin intra-reservoir shales, with the incremental growth of the plume from 1999-2001 clearly evident. Reflectivity variations within the plume can be mapped (Fig. 7b) and the evolution of each individual CO2 layer can also be tracked in detail (Fig. 7c). The CO2 accumulation also produces a pronounced velocity pushdown of reflections beneath the injection point (Arts et al 2004). Detailed quantitative treatment of the seismic data is beyond the scope of this paper. Suffice to say that analysis of the reflectivity, whereby changes in reflection amplitude can be related directly to thickness variations in the CO2 layers gives results broadly consistent with the known injected volumes of CO2. Similarly, synthetic pushdown values calculated from CO2 saturation models are roughly consistent with observations. However considerable
The Bunter Sandstone is a major Triassic reservoir rock in the Southern North Sea Basin (Fig. 2), with proven natural gas accumulations. In the UK sector the Bunter Sandstone lies at depths between 200 and 3000m, and much of it is in the zone where CO2 would exist in a dense (liquid or supercritical) phase (Fig. 3a). Offshore of the UK, the Bunter Sandstone is characterized by many large dome-shaped structures (Fig. 8) formed by the movement of the underlying lower Zechstein Salt. The salt has formed domes, pillows and diapirs, folding the overlying Bunter Sandstone into a number of large traps suitable for CO2 storage (Fig. 9). Many of these traps have never been filled with hydrocarbons and can therefore be categorized as saline aquifers. The porosity of the Bunter Sandstone ranges between 8% and 46%, with an average value of about 23% (Cameron et al. 1992), creating large volumes potentially available for CO2 storage. In a typical structure (Fig. 8), the Bunter Sandstone has a pore volume of around 5.5 X 109m3. If just 10% of this pore volume were used, sufficient storage would be available to sequester nearly 20 years' worth of emissions from the Drax power-station (Fig. 1). In the UK offshore sector (Fig. 9) the Bunter Sandstone is estimated to have a total pore volume of 1.10 X 101] m3 within closed structures. If all of this volume were to be utilised for CO2 storage, approximately 7 X10 10 tonnes could be sequestered. Assuming an annual CO2 output of 200 Mt from electricity generation, this would represent 350 years of UK power-generation emissions. More realistically, perhaps 10% of the aquifer pore-space could be used, providing storage for 35 years' worth of UK emissions. If, as discussed above, the whole
CO2 SEQUESTRATION IN NORTHERN EUROPE
Fig. 7. Time-lapse seismic data from around the Sleipner injection point (a) Inline showing the change in reflectivity from 1994 (pre-injection) through 1999 (2.35Mt CO2 in situ) to 2001 (4.36Mt CO2 in situ), (b) Difference data showing map of total plume reflectivity in 2001 (c) growth and change in reflectivity (thickness) of the largest CO2 layer (arrowed in a). IP, CO2 injection point.
Fig. 8. Seismic profile through a typical domal structure in the southern North Sea (Seismic data courtesy of WesternGeco).
25
26
R. A. CHADWICK ETAL
Fig. 9. Major structural closure in the Bunter Sandstone of the southern North Sea, UK sector.
aquifer were targeted for storage, not just the structural closures, the whole pore volume of the Bunter Sandstone beneath about 800m depth could potentially be exploited. This would increase total storage volumes by up to an order of magnitude. Issues over the practical suitability of the Bunter Sandstone for CO2 injection have still to be resolved. A key parameter is CO2 injectivity. This is influenced by porosity, permeability and reservoir connectivity and has to be investigated further. Each potential storage site will require site-specific evaluation to resolve issues such reservoir quality, trap volume, integrity of caprock, pore-water chemistries and suitability of already existing infrastructure.
Discussion The key issues for underground sequestration are cost, available storage capacity, safety, stability and sustainability of storage, and public acceptance. Costs, as discussed above, are significant but not overwhelming. In relative terms, fossil-fuel based power generation coupled with sequestration is likely to prove significantly less expensive than alternatives such as biofuels, hydro, wind and solar power (Stromberg 2001). Storage capacity in northern Europe does not appear to be a limitation. This leaves safety and environmental issues as the main points of debate. One way of considering safety and environmental issues is by reference to the permitted storage of
other liquids and gases in reservoir rocks underground. For example the disposal of hazardous and non-hazardous liquid waste and acid gas in deep sedimentary formations is a well-established practice. More than 35 Mt of waste (most of it hazardous) has been injected underground in Ohio alone (Gupta et al. 2002). There are numerous examples of natural gas storage in aquifers and depleted gas fields at various localities around the world. Also, large quantities of CO2 are currently being injected into depleted oil fields, particularly in the Permian Basin of Texas, to enhance oil recovery (Bonder 1992). The presence of oil and gas fields and, indeed, large natural fields of carbon dioxide in the subsurface (Studlick et al 1990), demonstrate that fluids can be retained in the subsurface for millions of years under favourable circumstances. There are also areas of the world where CO2 is emitted naturally at the surface, even to the extent of being used for medicinal or therapeutic purposes (e.g. Pearce et al. 1996). The US Environmental Protection Agency gives seven specific criteria to be used in the evaluation of injection proposals (Walker & Cox 1976): 1. 2. 3.
All reasonable alternatives have been explored and found less satisfactory in terms of environmental protection. Adequate pre-injection tests have been made for predicting the fate of the materials injected. There is conclusive technical evidence to demonstrate that such injection will not inter-
CO2 SEQUESTRATION IN NORTHERN EUROPE
4.
5.
6.
7.
fere with present and potential use of water resources or result in other environmental hazards. The subsurface injection system has been designed and constructed to provide maximum environmental protection. Provisions have been made for monitoring both the injection operation and the resulting effects on the environment. Contingency plans that will obviate any environmental degradation have been prepared to cope with all well shut-ins or any well failures. Provisions have been made for plugging injection wells when abandoned and for monitoring plugs to ensure their adequacy in providing continuous environmental protection.
Provided therefore that site selection is suitably rigorous, there is every reason to believe that large amounts of CO2 can be stored underground, in situations wholly isolated from the atmosphere, for very long periods of time - certainly well beyond any atmospheric CO2 maximum caused by anthropogenic emissions. The ongoing Sleipner operation demonstrates that the underground behaviour of the injected CO2 can be monitored, modelled and, within current limits of uncertainty, in situ amounts can be verified. In conclusion, northern Europe is particularly well provided for in terms of suitable underground CO2 sequestration sites. Issues surrounding storage capacity and site security are technically surmountable and, if cost issues can be overcome, underground sequestration may be able to play a key role in reducing CO2 emissions in the years to come. We thank our colleagues in the SACS and GESTCO projects for their valuable discussions and opinions. Reviewers H. Herzog and C. Lewis provided helpful comments and suggestions. SACS and GESTCO are funded by the EU, national governmental organizations such as the UK DTI, and industrial partners, BP, Compagnie Fran9ais de Geothermale, Danish National Oil and Gas Company, ExxonMobil, Greek Public Power Company, Norsk Hydro, Shell Global Solutions, Statoil, TotalFinaElf and Vattenfall AB. We acknowledge the consent of Schlumberger Ltd and WesternGeco to publish examples of 2D seismic data. This paper is published with permission of the Executive Director of the British Geological Survey (NERC).
References ARTS, R., EIKEN, O., CHADWICK, R. A., ZWEIGEL, P., VAN DER MEER, L. & KIRBY, G.A. 2004. Seismic monitoring at the Sleipner underground CO2 storage site (North Sea). In: BAINES, S. J. & WORDEN, R. H. (eds) Geological Storage of Carbon Dioxide. Geological Society, London, Special Publications, 233,181-191.
27
BACHU, S., GUNTER, W. D. & PERKINS, E. H. 1996. Carbon dioxide disposal. In: HITCHON, B. (ed.). Aquifer Disposal of Carbon Dioxide: Hydrodynamic and mineral trapping - proof of concept, Geoscience Publishing Ltd, Sherwood Park, Alberta, Canada, 11-22. BAKLID, A., KORB0L, R. & OWREN, G. 1996. Sleipner Vest CO2 disposal, CO2 injection into a shallow underground aquifer. Paper presented on the 1996 Society of Petroleum Engineers Annual technical Conference and Exhibition, Denver, Colorado, USA, SPE paper 36600,1-9. BONDOR, P. L.I992. Applications of carbon dioxide in enhanced oil recovery. Energy Conversion and Management. 33, 579-586. CAMERON, T. D. J., CROSBY, A., BALSON, P. S., JEFFERY, D. H., LOTT, G. K., BULAT, J. & HARRISON, D. J. 1992. The Geology of the Southern North Sea. United Kingdom Offshore Regional Report. British Geological Survey, Nottingham. CAO, M. & WOODWARD, F. I. 1998. Dynamic responses of terrestrial ecosystem carbon cycling to global climate change. Nature, 393, 249-252. CHADWICK, R. A., HOLLOWAY, S. & RILEY, N. 20010. Deep subsurface CO2 sequestration - a viable greenhouse mitigation strategy. Geoscientist, 11,4-5. CHADWICK, R. A., HOLLOWAY, S., KIRBY, G. A., GREGERSEN, U. & JOHANNESSEN, P. N. 2001&. The Utsira Sand, central North Sea - an assessment of its potential for regional CO2 disposal. In: WILLIAMS, D., DURIE, B., MCMULLAN, P., PAULSON, C. & SMITH, A. (eds) GHGT-5 Greenhouse Gas Control Technologies, CSIRO, Australia, 349-354. CHADWICK, R. A., ARTS, R., EIKEN, O., KIRBY, G. A., LINDEBERG, E. & ZWEIGEL, P. 2004. 4D seismic imaging of a CO2 bubble at the Sleipner Field, central North Sea. In: DAVIES, R., CARTWRIGHT, J., STEWART, S., UNDERHILL, R. & LAPPIN, M. (eds) 3D Seismic Technology: Application to the Exploration of Sedimentary Basins. Geological Society, London, Memoirs 29, 311-320. CZERNICHOWSKI-LAURIOL, L, BANJUM, B., ROCHELLE, C., BATEMAN, K., PEARCE, J. & BLACKWELL, P. 1996. Analysis of the geochemical aspects of the underground disposal of CO2. In: APPS, J. A. & TSANG, C. (eds) Deep Injection Disposal of Hazardous and Industrial Waste. Academic Press, 565-683. Department of Trade and Industry. 2000. New and Renewable Energy - Prospects for the 21st Century. London, DTI. DOCTOR, R. H., MOLBURG, J. C. & THIMMAPURAM, P. 2001. Transporting carbon dioxide recovered from fossilenergy cycles. In: WILLIAMS, D. DURIE, B., MCMULLAN, P., PAULSON, C. & SMITH, A. (eds) GHGT-5 Greenhouse Gas Control Technologies, CSIRO, Australia. 567-571. ENNIS-KING, J. & PATERSON, L. 2001. Reservoir engineering issues in the geological disposal of carbon dioxide. In: WILLIAMS, D., DURIE, B., MCMULLAN, P., PAULSON, C. & SMITH, A. (eds) GHGT-5 Greenhouse Gas Control Technologies. Collingwood, Australia, CSIRO, 290-295. GUPTA, N., SASS, B. M., SMINCHAK, J. R. & HICKS, J. E. 2002. Feasibility of long term carbon dioxide storage
28
R. A. CHADWICK ETAL
in deep saline formations, http://www.netl.doe.gov/ publications/proceedings/98/98ps/ps4-7.pdf HERZOG, H. J. 1999. The economics of CO2 capture. In: ELIASSON, B., RIEMER, P. W. F. & WOKAUM, A. (eds) Greenhouse Gas Control Technologies. Elsevier Science, Oxford, 101-106. HOLLOWAY, S. 2001. Storage of fossil fuel-derived carbon dioxide beneath the surface of the Earth. Annual Reviews of Energy and the Environment, 26,145-166. IEA Greenhouse Gas R & D Programme. 2000. Carbon dioxide capture and storage. IEA/OECD, Paris. KRUSHIN, J. T. 1997. Seal capacity of nonsmectite shale. In: SURDAM, R. C. (ed.) Seals, Traps, and the Petroleum System. American Association of Petroleum Geologists Memoir, 67, 31-47. LINDEBERG, E. 1996. Phase properties of CO2/water systems. In: HOLLOWAY, S. (ed.) Final report of the Joule II project No. CT92-0031: The Underground Disposal of Carbon Dioxide, British Geological Survey, Nottingham, 165. LINDEBERG, E. 1997. Escape of CO2 from aquifers. Energy Conversion and Management, 38 (supplement), S235-S240. LINDEBERG, E. & WESSEL-BERG, D. 1997. Vertical convection in an aquifer column under a gas cap of CO2. Energy Conversion and Management, 38 (supplement), S229-S234. ORR, J.C. 2002. Global Ocean Storage of Anthropogenic Carbon, Final Report, EC Environmental and Climate Programme, 129pp. Available at http://www.ipsl. jussieu.fr/OCMIP/reports/GOSAC_finalreport_hires. pdf. ORR, J.C., AUMONT, A., YOOL, K. & 8 OTHERS. 2001. Ocean CO2 sequestration efficiency from 3-D ocean model comparison. In: WILLIAMS, D., DURIE, B., MCMULLAN, P., PAULSON, C. & SMITH, A. (eds) GHGT-5 Greenhouse Gas Control Technologies, CSIRO, Australia, 469-474. PEARCE, J.M., HOLLOWAY, S., WACKER, H., NELIS, M.K., ROCHELLE, C. & BATEMAN, K. 1996. Natural occurrences as analogues for carbon dioxide disposal. Energy Conversion and Management, 37,1123-1128. PRENTICE, I. C., FARQUHAR, G. D., FASHAM, M. J. R. & 7 OTHERS. 2001. The carbon cycle and atmospheric carbon dioxide. In: HOUGHTON, J. T, DING, Y.,
GRIGGS, D. J., NOGUER, M., VAN DER LINDEN, P. J., DAI, X., MASKELL, K. & JOHNSON, C. A. (eds) Climate Change 2001: The Scientific Basis. Contribution of Working Group 1 to the Third Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA, 184-237. SIEGENTHALER, U. & Joos, F. 1992. Use of a simple model for studying ocean tracer distributions and the global carbon cycle. Tellus,44B, 186-207. SNEIDER, R. M., SNEIDER, J. S., BOLGER, G. W. & NEASHAM, J. W. 1997. Comparison of seal capacity determinations: conventional cores vs. cuttings. In: SURDAM, R C. (ed.) Seals, Traps, and the Petroleum System. American Association of Petroleum Geologists Memoir, 67,1-12. STROMBERG, L. 2001. Discussion on the potential and cost for different CO2 emission control options in Europe. VGB Power Technology, 10/2001, 81, 92-97. ISSN 1435-3199 STUDLICK, J. R. J., SHEW, R. D., BASYE, G. L. & RAY, J. R. 1990. A giant carbon dioxide accumulation in the Norphlet formation, Pisgah Anticline, Mississippi. In: BARWIS, J., MCPHERSON, J. G. & STUDLICK, J. R. J. (eds) Sandstone Petroleum Reservoirs. New York, Springer-Verlag, 181-203. VAN DER STRAATEN, R. 1996. Inventory of the CO2 Storage capacity of the European Union and Norway. In: HOLLOWAY, S. (ed.) Final report of the Joule II project No. CT92-0031: The Underground Disposal of Carbon Dioxide, British Geological Survey, Nottingham, 16-115. WALKER, W. R. & Cox, W. E. 1976. Deep Well Injection of Industrial Waste: Government Controls and Legal Constraints. Virginia Water Resources Center, Blacksburg, Virginia, USA. ZWEIGEL, P., HAMBORG, M., ARTS, R., LOTHE, A. E., SYLTA, O. & TOMMERAS, A. 2001. Prediction of migration of CO2 injected into an underground depository: reservoir geology and migration modelling in the Sleipner case (North Sea). In: WILLIAMS, D., DURIE, B., MCMULLAN, P., PAULSON, C. & SMITH, A. (eds) GHGT-5 Greenhouse Gas Control Technologies. CSIRO, Australia, 360-365.
A review of natural CO2 accumulations in Europe as analogues for geological sequestration J. PEARCE1,1. CZERNICHOWSKI-LAURIOL2, S. LOMBARDI3, S. BRUNE4, A. NADOR5, J. BAKER5, H. PAUWELS2, G. HATZIYANNIS6, S. BEAUBIEN3 & E. FABER4 1
British Geological Survey, Keyworth, Nottingham, NG12 5GG, UK (e-mail:
[email protected]) 2
Bureau des Recherches Geologiques et Mineraux, 3 avenue Claude Guillemin, BP 6009, 45060, Cedex 2, Orleans, France
3
Dipartimenta di Science della Terra, Universita La Sapienza, P. A. Mow 5, 00187 Roma, Italy 4
Bundenanstalt fur Geowissenschaften und Rohstoffe,
Stilleweg 2, D-30655, Hannover,
Germany 5 6
Geological Institute of Hungary (MAPI), H-1143, Budapest, Hungary
Institute of Geology and Mineral Exploration, 70 Mesogion Street, Athens, 115 27, Greece Abstract: Natural geological accumulations of carbon dioxide occur widely throughout Europe, often close to population centres. Some of these CO2 deposits leak, whereas others are sealed. Understanding these deposits is critical for selecting and designing underground storage sites for anthropogenic CO2. To provide confidence that the potential risks of geological CO2 storage are understood, geologists are required to predict how CO2 may behave once stored underground. Natural CO2 accumulations provide a unique opportunity to study long-term geochemical and geomechanical processes that may occur following geological storage of anthropogenic CO2. In addition, natural CO2 springs and gas vents can provide information on the mechanisms of gas migration and the potential effects of CO2 leakage to the surface. This paper provides a description of some natural, European CO2 occurrences. CO2 accumulations occur in many basins across Europe. In addition, volcanic areas and seismically active areas allow CO2-rich fluids to migrate to the near surface. Many of these occur in areas that have been populated for hundreds and thousands of years. Stratigraphic traps have allowed CO2 to accumulate below evaporite, limestone and mudstone caprocks. Comparisons between reservoir sandstone and equivalent nearby sandstones that contain no CO2 indicate that reservoir sandstones may experience increased secondary porosity development through feldspar dissolution. Where fracture reactivation allows CO2-rich fluids to migrate, limited self-sealing may take place through calcite precipitation. Gas migration experiments indicate that, due to geochemical interactions, fine-grained seals would be able to trap smaller volumes of CO2 compared to, for example CH4. In natural systems most leakage from depth occurs along fractures and is typically extremely localized on a metre-scale.
An understanding of natural geological accumulations of carbon dioxide is critical for identifying underground storage sites for anthropogenic CO2. Natural CO2 accumulations are being studied throughout the world to understand the processes that may occur during geological CO2 injection and afterwards during long-term storage. By examining natural occurrences where CO2 has been trapped for geological timescales, we can constrain our criteria for storage site selection. Where CO2 is migrating towards the near surface, there are opportunities to examine migration pathways (mainly faults and fractures), the speed of migration and likelyfluxes, and the effects these emissions may have on local populations. Predictive modelling can therefore be further validated with data from natural accumulations. This paper reviews some of the processes
being addressed in the Natural Analogues for the Storage of CO2 in the Geological Environment (NASCENT) project, from evidence obtained at various European sites. Natural CO2 occurrences are common across Europe and, although they occur in a variety of geological settings, their distribution is principally controlled by the Cenozoic rift system (Fig. 1) and associated Tertiary volcanism. These occurrences can be classified into: (1)
CO2-rich waters both at depth and in springs. These are often exploited for mineral waters: for example from the Betic Cordilleras in the west (Ceron & Pulido-Bosch 1999; Ceron et al 1999, 2000), through the Massif Central (Pauwels et al 1997) to the western (Lesniak
From: BAINES, S. J. & WORDEN, R. H. (eds) 2004. Geological Storage of Carbon Dioxide. Geological Society, London, Special Publications, 233,29-41.0305-8719/047$ 15.00 © The Geological Society of London 2004.
Fig. 1. The distribution of CO2 springs and gas pools in relation to the Alpine Fold belt and volcanicity (base map after Ziegler 1994). 1, Betic Cordillera (Ceron et al 2000); 2, Massif Central (Pauwels et al. 1997); 3, Eiffel (May 2001); 4, Bekes sub-basin (Clayton et al 1990); 5, West Pannonian Basin (this study and (Koncz 1983); 6, Derecske sub-basin (Lollar et al. 1994); 7, Latera (this study and (Gianelli and Scandiffio 1989); 8, Karlony Vary (Vrba 1996; Moller et al 1998); 9, Pannonian basin (Cornides 1993); 10, West Carpathians (Lesniak 1985); 11, Mura sub-basin (Kralj 2000); 12, SE Carpathians (Vaselli et al 2002); 13, Eger rift (Weinlich et al 1999); 14, Eger rift (Weise et al. 2001); 15, Vorderrhon (this study).
NATURAL CO, ACCUMULATIONS IN EUROPE
(2)
(3)
1985) and southeastern Carpathians (Vaselli et al 2002) in the east. These waters occur in a variety of geological settings with the sources of CO2 ranging from primary mantle degassing, volcanic activity to thermal metamorphism of limestones. Dry CO2 gas vents ('moffettes'). These are associated with Cenozoic rifts such as the Eger rift system and associated adjacent areas (Vrba 1996;Moller*f0/. 1998; Weinlichef a/. 1999; Weise et al 2001) and the Tyrrhenian rift system (Chiodini et al 1999). They are also associated with hydrothermal fields and Quaternary to Recent volcanic activity such as in the Eiffel volcanic complex within the Rhenish Massif (May et al 1996; May 2001); the Larderello geothermal field (Gianelli & Scandiffio 1989; Gianelli et al. 1998; Gianelli & Ruggieri 2000) and, of course, currently active volcanism. CO2 gas accumulations. These occur in Cenozoic extensional basins such as within the sub-basins of the back-arc Pannonian Basin (Koncz 1983; Cornides et al 1986; Clayton et al 1990; Cornides 1993; Koncz and Etler 1994) and the Florina-Ptolemais-Aminteo graben system (see below). In addition, older basins subject to Cenozoic tectonism such as the Triassic to Jurassic Southeast Basin of France (see below) and sub-basins within the Southern Permian Basin in Saxo-Thuringia, Germany (see below) can also host CO2 accumulations.
The NASCENT project is examining both leaking and 'non-leaking' systems. The study sites include trapped supercritical CO2 at Mihalyi-Repcelak in western Hungary, the Vorderrhon region of central Germany and Montmiral in the French Southeast Basin (Fig. 1). These accumulations are, or have recently been, exploited for industrial CO2 gas supplies. Systems studied where CO2 leaks directly to the surface include Fiorina in northern Greece (also currently exploited), the Latera geothermal field in Tuscany and Matraderecske in northern Hungary (Figsl&4).
CO2 gas pools CO2 gas pools occur in small traps of the Vorderrhon in the Thuringia area of the Southern Permian Basin, Germany, the Southeast Basin of France and extensively across the Miocene back-arc Pannonian Basin in Hungary. Host lithologies vary from Triassic sandstones in the Southeast Basin, fractured Rotliegend siltstones in the Vorderrhon and Pannonian (Miocene) conglomerates and sandstones, plus minor
31
fractured Silurian and Devonian basement metapelites and pelites at Mihalyi-Repcelak.
The Vorderrhon deposits, a fractured reservoir The Vorderrhon CO2 deposits, located in the western part of Thuringia, have been known since the end of the nineteenth century. Despite industrial demand for pure CO2, systematic exploration only started in 1960, when CO2 was found by chance during potash exploration drilling. Four small deposits are known in the area: Bernardshall, Schorngraben-Wolferbutt, GehausHohenwart and Oechsen. All occur along north-s outh trends. CO2 was produced from 1958 to 1994 and transported to a production plant at Leimbach. Total production amounted to about 528 million kg CO2. The Vorderrhon area is located within the South German Block, a major unit of the Central European fault-block mosaic formed during Saxonian tectonics (Fig. 2). The study area lies within the SaxoThuringian outcrop at the NW margin of the Central German Crystalline Basement Zone and the weakly metamorphosed Northern Phyllite Zone. Rotliegend and Zechstein sediments were deposited in a series of sub-basins on the margins of the Southern Permian basin. The Rotliegend consists of a coarse sandstone and conglomerate overlain by red, usually finegrained siltstones deposited in a low-energy environment and forms the main reservoirs. The overlying Kupferschiefer, Zechstein Limestone, and Lower Werra Anhydrite are also reservoir rocks. A high degree of cementation in these reservoirs limits their storage capacity to fractures (Fig. 3). The Werra Rock Salt and the intercalated Thiiringen and Hessen potash seams form the seal for the CO2 deposits. The potash has been extensively mined in the Werra potash mine. Porosities in the Rotliegend, the main CO2 reservoir, vary up to 8% with values above 6% being attributable to microfractures (Ziegenhardt 1978). A mean permeability of less than O.OlmD was determined with a few values up to 0.1 mD which were attributed to microfractures. Capillary pressure measurements, determined from two samples from the Rotliegend siltstone sequence, indicated that up to 70% of the pores were less than 75 nm in diameter. The spatial distribution of the CO2 accumulations is controlled by the north-south aligned fracture network and not by variations in reservoir quality. A network of intersecting Hercynian (NW-SE), Rhenian (N-S) and Eggian (NNW-SSE) structural trends come together in the Vorderrhon area. Late Tertiary basalts form caps to hills, exploiting the Rhenian and Eggian fault zones and typically located at the intersections of these with Hercynian-trending
32
J.PEARCE£rAL.
Fig. 2. Location of the Vorderrhon study area in relation to major structural elements of central Western Europe. Base map after (Ziegler 1982).
fault zones that were repeatedly used by successive eruptions. A direct relationship between NorthSouth-trending faults, volcanic rocks and CO2 is demonstrated by the exposures in the Werra potash mines (Miiller 1958; Hoppe 1960; Koch & Vogel 1980). Production histories suggest that fractures contain both CO2 and formation water. The less dense CO2 accumulated in the upper parts of the reservoir rock and the water collected in the deeper parts. In the most favourable cases, a 'gas cap' could form, from which 'dry' gas was produced at first. Differences of up to 100m were observed in the depth of the gas cap, relative to the base of the Werra Rock Salt, indicating hydrodynamic separation of the different occurrences. Of particular importance for estimates of reserves is the phase of the CO2 which is controlled by reservoir pressures and temperatures and the critical point of CO2 (pc = 73.8bar, tc = 31.04°C). The base of the Werra Rock Salt lies at depths between 465 and 995m (145-555m below m.s.l.) around the CO2
traps. The formation pressure at these depths is 45-70 bar and the temperatures range from 2035 °C. Under these conditions, the CO2 is in both the gas (Bernhardshall) and liquid (SchorngrabenWolferbiitt and Gehaus-Hohenwart) phases, as well as in the supercritical gas phase (Oechsen). The raw gas has a mean composition of 97-99.5 vol. % CO2, c. I vol.% N2, <7 vol.% CH4, with traces of H2 and He and higher hydrocarbons. The CO2 from the Zechstein Limestone (Schorngraben deposit) sometimes also contains traces of crude oil. S13CCO value of -5.4%o for gas within the Vorderrhon accumulation and its close spatial relationship with Tertiary basalts within the Rhenian fault system would suggest that it is of volcanic origin. May (2001) established a similar relationship in the nearby Rhenish Massif. This is in contrast to o13Cco values that range from -12.1 to -21.7 %0 in overlying aquifers and soil gases which indicate the CO2 in these waters is biogenic, suggesting that CO2 does not migrate through the evaporites. The distribution of CO2 deposits also depends on
Fig. 3. Schematic geological section of the CO2 accumulations in the Vorderrhon.
34
J.PEARCECTAL.
the sealing capacity of the Zechstein salt beds and CO2 accumulations only occur from the Lower Werra Anhydrite down to the uppermost Rotliegend. The seals are a complex sequence of anhydrites and claystones, from the Lower Werra rock salt, 200-300m thick, to the Lower Letten, several hundred metres thick. However, in the southwestern and southeastern parts of the study area, leaching has reduced the thickness of the Werra rock salt, which is locally removed where Hercynian fault zones intersect other trend directions, especially the Rhenian. Correlated rock bursts occurred in the potash mines in 1975 and 1989, the latter causing damage to local buildings. These blow-outs provide graphic evidence for CO2 migration where salt is extensively leached. Studies of the mechanisms and rates of gas migration through caprocks from the Vorderrhon area (Hildenbrand 2002) indicate that CO2 flow through the matrix is orders of magnitude lower than in fractures but that, compared to inert gases such as CH4 or N2, rates are higher due to geochemical interactions between the CO2 and caprock.
Mihalyi-Repcelak, a siliciclastic matrix porosity reservoir The Mihalyi-Repcelak area comprises two small CO2 gas fields in the central part of the Little Hungarian Plain within the Pannonian Basin (Fig. 4) and is 30km long and 7-8km wide at 130-150m above sea level. These are two of several known CO2-rich gas accumulations that occur throughout the Pannonian Basin (Fig. 4; see for example Cornides et al 1986; Cornides 1993; Lollar et al 1994, 1997). The Mihalyi reservoir was discovered at the beginning of Hungarian oil exploration during the early 1930s. The first gravimetry survey in 1933 subdivided a large gravimetric anomaly into three smaller parts: Mihalyi, Repcelak and Mosonszentjanos. The first borehole was drilled through the Mihalyi anomaly in 1935, passing through Neogene sediments before hitting a CO2producing zone (94.6 volume %) in early Palaeozoic phyllite at a depth of 1602m. In the MihalyiRepcelak area 26 CO2 occurrences (>90 % CO2) and two non-burning mixed gas pools (CO2 content: 65-75%) are known. At Repcelak a basement high, at a depth of 1460m, was explored in 1945^46 and produced CO2 from two Lower Pannonian sandstone beds (Fig. 5). Industrial exploitation started in the 1950s, when a large number of wells were drilled (up to 43 boreholes by 1979). Two concessional blocks, the Mihalyi and the Repcelak fields, have been exploited since 1993. The present-day geological setting of the Pannonian Basin (Fig. 4), surrounded by the Carpathians to
the north and west and by the Dinarides to the east and south, is the final result of a complicated, multiple evolutionary history (for a summary of the formation and evolution of the Pannonian Basin see Horvath and Tari (1999) and references therein). Pannonian (Miocene-Pliocene) basal conglomerates and Lower and Upper Pannonian sandstones form the main CO2 reservoirs, and unconformably overlie the Miocene formations or rest directly on the metamorphic basement (Fig. 5). Intergranular porosity and permeability of these sedimentary rocks varies according to lithofacies. The Lower Pannonian sediments comprise basal pelites, followed by fine-grained sandstones and claystones (100-350 m), and the uppermost part of the section comprises sandy calcareous claystone with finegrained sandstone lenses. The Upper Pannonian sediments (1150-1450m thickness) consist of rapidly alternating fine-grained sandstones and claystones. Gas migration is assumed to be from the west, since sandstones that pinch out on the eastern flanks of the structures are barren. Although the origin of the Mihalyi-Repcelak CO2 has yet to be established, in the Mihalyi area there is no difference in the gas composition between areas close to the metamorphic basement and Upper Pannonian reservoirs, suggesting that it migrates upwards along fractures. 813CCO values range from -3.1 to -5.5 and suggest a mantle-derived origin. Elsewhere, the Pannonian Basin contains significant CO2-gas pools, associated with petroleum and natural gas reserves. Cornides (1993) and Cornides et al (1986) have determined 813CC02 values of -5 ±2.5%o and 3He/4He ratios of -6.2 to — 5.2%o, suggesting that the CO2 is largely mantle derived. Helium and neon isotopic analyses throughout the region also indicate a mantle source for these gases which are often associated with CO2rich gas pools (Lollar et al. 1994, 1997). However, Clayton et al (1990) determined isotopically heavier gases and attributed their source to thermal decomposition of limestones.
Montmiral, a matrix porosity carbonate reservoir Several CO2-rich gas pools occur throughout the Ardeche palaeomargin of the Southeast Basin of France (Fig. 6). The Southeast Basin is located to the SE of the Massif Central and is bounded on the east and south by Tertiary thrust belts of the western Alps and Pyrenees-Provence. From the Late Trias to the end of the mid-Jurassic, this area underwent major extension as part of the Tethyan rifting that preceded the opening of the Ligurian ocean (Lemoine et al 1986; Roure et al 1992; Razin et al 1996). During the Late Jurassic, opening of the Ligurian ocean caused regional subsidence, followed in the Late
NATURAL CO, ACCUMULATIONS IN EUROPE
35
Fig. 4. Locations of CO2 occurrences in the Pannonian Basin with respect to oil and gas fields and Neogene volcanics. Base map after (Horvath and Tari 1999). 1 - (Clayton 1995), 2 - (Cornides 1993), 3 - (Lollar et al. 1994).
Fig. 5. Cross-section of the Mihalyi-Repcelak CO2 gas accumulations, Western Hungary.
36
J.PEARCECTAL.
Fig. 6. Distribution of CO2 springs and gas pools in the Massif Central and Southeast Basin. Basemap after Roure et al. (1992).
NATURAL CO, ACCUMULATIONS IN EUROPE
Cretaceous by basin inversion in a north-south compressive regime associated with the Pyrenean collision with Europe (Phillippe et al 1998 and references therein). Since the Late Miocene an E-W compressional regime has been established corresponding to the main Alpine orogen, with development of predominantly vertical fracturing. In the Neogene, uplift of the Massif Central was accompanied by three magmatic events of increasing scale. Oil exploration in the Southeast Basin has often been disappointing due to the discovery of CO2 rather than hydrocarbons (Mascle et al 1996), eight accumulations having been discovered in total. This CO2 is now extensively exploited by the sparkling mineral water industries centred around Vichy, Badoit and Perrier. As with other significant CO2 occurrences throughout Europe, 813CCO values obtained in this study of —2.7%c would suggest the CO2 to be of mantle or deep crustal origin. This is close to one 813C measurement of — 3.2%c determined by Blavoux and Dazy (1990). Chemical and isotopic determinations on the many carbonated springs found throughout the Massif Central and Southeast Basin area indicate that the CO2 and associated noble gases are predominantly (though not entirely) of mantle or deep-crustal origin (Matthews et al. 1987; Pauwels et al. 1997). Reservoirs are in Lower Jurassic and Triassic limestones, dolomites and sandstones at depths of between 2000 and 5000m. Among them, only the Montmiral site is currently exploited (Fig. 6), with production having started in 1990. Supercritical CO2 fluid occurs below 2400m depth at Montmiral in both the Early Jurassic Hettangian marine and pelagic deposits, and Late Triassic Rhaetian and earlier, terrestrial or continental shelf evaporitic sediments. Late Triassic anhydrite, clays and dolomite separate the two CO2 occurrences. The main CO2 reservoir is within the lower Triassic sandstone and caprocks are partly siliceous Sinemurian clayey limestone. Mean reservoir sandstone porosity and permeability is variable. Between 2455-2462 m, moderate reservoir properties of 6% porosity and 0.5mD permeability occur. However, a network of wellconnected open fractures maintains reservoir productivity. The CO2 content is estimated at about 70%. Deeper, between 2462-2411 m, the matrix porosity varies between 8-12% with corresponding permeabilities of 0.3-120mD. In this zone, CO2 saturation reaches 30-60%. Over the ten years of exploitation, average fluid composition has been 1.33% water, 0.12% oil and 98.55% CO2. Comparisons between reservoir sandstones and equivalent sandstones in a nearby well that do not contain CO2 suggest that secondary porosity has been enhanced in the presence of CO2 through increased dissolution of K-feldspar and, to a lesser
37
extent, increased dissolution of carbonate, mainly dolomite, cements. This is currently the focus of further study. Fluid inclusion studies indicate that CO2-rich fluids have migrated out of the reservoir along fractures. These open fractures have been partially lined by calcite precipitated from these CO2rich fluids. The timing of this migration is not clear. Although CO2 is locally trapped below Triassic sequences of claystones, limestones and evaporites (mainly anhydrite but also halite), the many CO2rich springs and mineral waters provide evidence for CO2 migration to the surface, through these caprock sequences. It is assumed that migration occurs along fractures and faults. Leakage and migration In addition to the Vorderrhon, Mihalyi-Repcelak and Montmiral gas fields, several sites are being studied where CO2 is actively leaking to the near surface. These sites are the small village of Matraderecske in Northern Hungary, the Fiorina CO2 production facility in northern Greece, and the Latera geothermal area of Italy.
Matraderecske Matraderecske is a small village in the northeastern foreland of the Matra Mountains in northeastern Hungary, which form part of a Middle Miocene andesite volcanic complex, close to a major NNESSW fault zone (Darno Zone). An Eocene andesite and andesitic tuff basement is overlain by Eocene and Oligocene clays and sands. The basement is faulted into stepped blocks that rise to the shallow subsurface. CO2 and methane seepages and carbonated springs have been known for a long time in this area. The CO2 area is characterized by hills 300-400m high, dissected by elongated valleys at an elevation of 180-220m with permanent streams. In this area the basement comprises a 20km long, NNE-SSW striking subsurface ridge, composed of Triassic shale, quartzite and limestone series. The basement rocks have a low transmissivity, but locally karstified parts may be more permeable. On the southern flank of this ridge (surroundings of Parad, Recsk and Matraderecske) the basement is overlain by a 200-800 m thick, strongly weathered Eocene andesite which has good transmissivity and contains the uppermost shallow groundwaters. The basement groundwaters have a high CO2 content: the gas-water ratio is about 10-15m3m~3 by volume, and gases within the subvolcanic groundwaters comprise 50% CO2 and 50% CH4. Hydrocarbons are often present in both areas. Petroleum has been produced from the Oligocene
38
J.PEARCECTAL.
layers in Biikkszek since 1940, whereas hydrocarbon traces are known from the covering andesite and the Oligocene strata in the southern area. The gases migrate upwards along faults and fractures from a deep-seated karst water reservoir at a depth of about 1000m and disperse in the overlying weathered andesite. Where there is a seal, the gases migrate laterally and can escape along other faults opening to the surface. The average gas flux is about 5-101 hour" 'm~ 2 , but along faults it can reach 4001 hour" l m~ 2 . Gas seepages can be observed as bubbling in wells, local streams and in strongly carbonated springs. CO2-rich gas (95% CO2, 4-5% CH4, with less than 1% SO2) leaks into residential basements in Matraderecske village and has lead to asphyxiation problems in the past. Successful mitigation options have included alterations to house and basement ventilation systems and, in some cases, forced ventilation in basements (Pearce et al. 2002). In addition to the CO2, the high radon levels (222Rn = 125 kBqm~ 3 ) also cause potential health problems. Recently the upwelling CO2, locally called 'mofetta', has been used for medical purposes. In Matraderecske, and elsewhere in Hungary and Romania, CO2 is used in the treatment of blood pressure and other circulatory problems. A medical facility has been built above a dry CO2 gas vent enabling the CO2 to collect in the base of a room with a sunken floor (CO2 is denser than air). Patients sit or stand within the CO2, which is absorbed by the skin. As the CO2 concentration in the blood stream increases, blood vessels dilate resulting in improved circulation.
Epicentres are located mainly in pre-Neogene rocks to the SE of the basin, coinciding with several NE-SW striking faults which created a horst between the Fiorina and Amynteon-Ptolemais basins. Relationships between seismic activity and CO2 leakage are currently being assessed. Metamorphic rocks intruded by granites flank the west margin of the Fiorina basin. The fault zone which created this basin, runs along this western margin where rapid filling with coarse clastic sediments took place. Mesozoic crystalline limestones, schists and gneisses occur in the east where a slower rate of sedimentation deposited finer elastics on the basement limestones. The Neogene sediments, overlying the Mesozoic crystalline limestones, comprise fluvial, coarse to fine siliciclastics with clayey tops to the sequence. Lignite seams are developed mainly from the eastern margin, starting at the contact with basement, westwards to the basin centre. CO2 occurs at various levels in the Mesozoic limestones and in overlying Neogene sandstones. Carbonated springs are also common throughout the area. Local cap rocks comprise clayey sediments. Surface gas seeps occur in areas covered by limestones free of the Tertiary clays and silt cover. At Fiorina, the productive life of a well is about one year because of scaling problems. Only one well has produced pure CO2 continuously for about 10 years. New wells are drilled between 50 and 100m from the old wells. The mean depth of producing wells is about 300m with CO2 being produced between 180 and 260m. Gas pressure is 8 atm and water pressure is higher at c. 10 atm (in the well). Approximately 1000 tonnes of dry ice are produced annually.
Fiorina Latera The Fiorina CO2 field lies in the northern part of a NNW-SSE oriented graben, which extends northwards to the town of Monastiri in the Former Yugoslav Republic of Macedonia, with a total length of about 150km. The middle Tertiary graben formed in metamorphic rocks of the Pelagonian tectonic zone of the internal Hellenides during extensional stresses following the Alpine orogenesis. Along this narrow, long graben a number of sedimentary basins were subsequently filled with fluviatile and lacustrine Miocene and younger sediments which exceed 1000m in thickness. The Fiorina CO2 field is the only commercial, naturally sourced CO2 producer in Greece. It was discovered by chance in the 1960s during lignite exploration in the northeastern part of the Fiorina basin. Production started in 1980 with annual production currently around 30000 tonnes of liquid CO2. This is a seismically active area with maximum magnitudes of around 5 on the Richter scale.
Latera is a commercially-exploited geothermal field with common CO2-rich springs and small local accumulations throughout the area. The geology of the west-central Italian peninsula is the product of two main tectonic events, a compressive phase during the Eocene to Late Miocene and a subsequent tensional phase (Carmignani & Kligfield 1990). The regional basement consists of upper metamorphic phyllites and micaschists, above a PalaeozoicPrecambrian gneissic complex. A series of 'tectonic units' were placed above this basement during the initial compressive period, including (from base to top): a tectonic accretionary wedge complex; the Tuscan' limestone nappe units that host the geothermal reservoir and CO2, and the Cretaceous to Eocene 'Ligurian' allochthonous flysch and ophiolite sequence (Baldi^ al 1995;Chiodini^a/. 1995; Minissale et al 2000). Travertine deposits, located SW of Latera, are associated with CO2-rich springs.
NATURAL CO9 ACCUMULATIONS IN EUROPE
The Tuscan limestone subsurface structural high is the eastern remnant of the first Latera caldera in the Mt Vulsini volcanism, now located in the centre of the present caldera (Barberi et al. 1984). The subvertical faults associated with these calderas have either remained open for fluid flow (resulting in springs or gas vents on surface) or have become sealed by fault gouge or secondary mineral precipitation (thereby isolating entire blocks and creating heat/water convection cells and CO2 reservoirs). The Latera geothermal field is water dominated, and as such most gas is dissolved and exsolves at the lower surface pressures. Typically reservoir waters contain 3-6% 'non condensable gases' (NCG) in the separated steam; the NCG is primarily composed of CO2 having a partial pressure of about 90bar (Bertrami et al 1984). A representative composition is 98.35% C02, 0.05% CH4, 1.22% H2S, 0.4% N2 with trace levels of H2 (Bertrami et al. 1984). Analysis of a separate gas from a dry well consisted of 98% CO2 and, in decreasing quantity, H2S, H2 and CH4 (Lombardi et al 1993). The observed CO2 is believed to be the result of decarbonization of carbonate minerals (Duchi et al \992a). The carbon isotopes vary from +3.2 to -0.7 813C% (PDB) (Bertrami et al 1984), with the more negative values corresponding to the springs occurring outside the caldera. Results for other isotopic studies suggest the CO2 is derived from thermal metamorphic processes in the underlying Mesozoic formations (Panichi & Tongiorgi 1976; Marini 1994; Minissale etal 1997). Gas vents, which can be either dry or associated with springs, are relatively common along a NW-SE structural lineament inside the eastern boundary of the Latera caldera as well as generally throughout the geothermal region. The composition of these gases are more spatially variable than those encountered in the deep wells. Although compositions are dominated by CO2 (above 90%), with 5-10% of other constituents such as N2, O2, CH4 and H2S, some vents have almost equal amounts of CO2 and N2 (see table 2 Duchi et al (1992/?)). In a more regional study Minissale et al (1997) determined that high CO2 contents were usually associated with cold springs or dry vents. These compositional differences are closely linked to large scale flow paths, as thermal waters with CO2 produced in the Mesozoic limestone aquifer, mix with cold meteoric recharge waters (Duchi et al 19920).
Conclusions The distribution of CO2 occurrences across Europe, from the Betic Cordilleras in the west to the Carpathian mountains in the east, is controlled at a continental scale by the Cenozoic rift system and
39
associated volcanism. At a regional scale CO2 occurrences are associated with Tertiary to Recent volcanism (thermal metamorphism and/or primary magma degassing) and associated hydrothermal activity, or primary mantle degassing via deep crustal fault systems. CO2 accumulations can provide natural laboratories in which to validate and calibrate, (along with laboratory-based experiments) predictive geochemical models that help to understand how CO2 will behave during long-term storage. The interactions of CO2-rich pore-waters with the reservoir lithologies and seals can provide key data for predictive geochernical models. Montmiral, Mihalyi-Repcelak and Vorderrhon provide examples of CO2-rich gas pools in siliciclastic, carbonated and fractured reservoirs. Seals are both clays and evaporites, although only at Mihalyi-Repcelak are surface leaks thought to be absent. Montmiral is situated within a regionally extensive area of CO2-rich springs that are exploited by the mineral water industry. Although surface leaks are not found in the Vorderrhon area, CO2 gas vents are common in the nearby Rhenish Massif and CO2 blow-outs have occurred in the local potash mines. At Matraderecske, Montmiral and Vorderrhon, traces of hydrocarbons are associated with CO2-rich waters. This is unsurprising since CO2 in a wellknown solvent for hydrocarbons - hence its use in tertiary enhanced oil production. However, this may be an issue when examining the risks of CO2 storage. If CO2 were to leak, it may release traces of hydrocarbons as it migrates to the near surface and aquifer-based water supplies would need to be protected. Fiorina, Latera and Matraderecske are examples of leaking systems that are being studied to determine how CO2 migrates to the near surface. Pearce et al (1996) suggested that CO2-rich groundwaters could result in release of heavy metals, through dissolution of ferromagnesian minerals in siliciclastic lithologies. No evidence has so far been found in the present research, although analyses are ongoing. Stratigraphic traps have allowed CO2 to accumulate below evaporite, limestone and mudstone caprocks. Comparisons between reservoir sandstone and equivalent nearby sandstones that contain no CO2 indicate that reservoir sandstones may experience increased secondary porosity development through feldspar dissolution. Where fracture reactivation allows CO2-rich fluids to migrate, limited self-sealing may take place through calcite precipitation. Gas migration experiments indicate that, due to geochemical interactions, fine-grained seals would be able to trap smaller volumes of CO2 compared to, for example, CH4. In natural systems most leakage from depth occurs along fractures and is typically extremely localized on a metre-scale.
40
J.PEARCEErAL.
This project is funded through the EC 5th Framework R&D Programme Energy, Environment and Sustainable Development Programme, Contract ENK5-CT-200000303. Special thanks go to staff at UGS Mittenwalde GmbH who carried out much of the initial research at the Vorderrhon site. IMP publishes with the permission of the Director of the British Geological Survey (NERC).
References BALDI, P., BELLANI, S., CECCARELLI, A., FIORDELISI, A., ROCCHI, G., SQUARCI, P. & TAFFI, L. 1995. Geothermal anomalies and structural features of Southern Tuscany (Italy). Proceedings of world geothermal congress, Florence 18-31 May. 1287-1291. BARBERI, K, INNOCENTI, F., LANDI, P., Rossi, U., SAITTA, M., SANTACROCE, R. & VILLA, I. M. 1984. The evolution of Latera Caldera (Central Italy) in the light of subsurface data. Bulletin of Volcanology, 47, 125-140. BERTRAMI, R., CECCARELLI, A. & LOMBARDI, S. 1984. L'elio nella prospezione geotermica. Rendiconti delta Societa Italiana di Mineralogia e Petrografia, 39, 331-342. BLAVOUX, B. & DAZY, J. 1990. Caracterisation d'une province a CO7 dans le bassin du Sud-Est de la France. Hydrogeologie, 4, 241-252. CARMIGNANI, L. & KLIGFIELD, R. 1990. Crustal extension in the Northern Apennines: the transition from compression to extension in the Alpi Apuane core complex. Tectonics, 9, 340-349. CERON, J. C., MARTIN-VALLEJO, M. & GARCIA-ROSSELL, L. 2000. CO2-rich thermomineral groundwater in the Betic Cordilleras, southeastern Spain: Genesis and tectonic implications. Hydrogeology Journal, 8, 209-217. CERON, J. C. & PULIDO-BOSCH, A. 1999. Geochemistry of thermomineral waters in the overexploited Alto Guadalentin aquifer (South-East Spain). Water Research, 33, 295-300. CERON, J. C., PULIDO-BOSCH, A. & BAKALOWICZ, M. 1999. Application of principal components analysis to the study of CO2-rich thermomineral waters in the aquifer system of Alto Guadalentin (Spain). Hydrological Sciences Journal-Journal Des Sciences Hydrologiques, 44, 929-942. CHIODINI, G., FRONDINI, F, KERRICK, D. M., ROGIE, J., PARELLO, F, PERUZZI, L. & ZANZARI, A. R. 1999. Quantification of deep CO2 fluxes from Central Italy. Examples of carbon balance for regional aquifers and of soil diffuse degassing. Chemical Geology, 159, 205-222. CHIODINI, G., FRONDINI, F. & PONZIANI, F. 1995. Deep structures and carbon dioxide degassing in central Italy. Geothermics, 24, 81-94. CLAYTON, C. J. 1995. Controls on the carbon isotope ratio of carbon dioxide in oil and gas fields. In: GRIMALT, J. O. & DORRONSORO, C. (eds) Organic Geochemistry: Developments and Applications to Energy, Climate, Environment and Human History. Proceedings 17th International Meeting on Organic Geochemistry, San Sebastian, Spain, 1073-1074. CLAYTON, J. L., SPENCER, C. W., KONCZ, I. & SZALAY, A.
1990. Origin and migration of hydrocarbon gases and carbon dioxide, Bekes Basin, southeastern Hungary. Organic Geochemistry, 15, 233-247. CORNIDES, I. 1993. Magmatic carbon dioxide at the crust's surface in the Carpathian Basin. Geochemical Journal, 27, 241-249. CORNIDES, L, TAKAOKA, N., NAGAO, K. & MATSUO, S. 1986. Contribution of mantle-derived gases to subsurface gases in a tectonically quiescent area, the Carpathian Basin, Hungary revealed by noble-gas measurements. Geochemical Journal,!^, 119-125. DUCHI, V., MlNISSALE, A., PAOLIERI, M., PRATI, F. & VALORI,
A. I992a. Chemical composition of natural deep and shallow hydrothermal fluids in the Larderello geothermal field. Journal of Volcanology and Geochemical Research, 49, 313-328. DUCHI, V., MINISSALE, A., PAOLIERI, M., PRATI, F. & VALORI, A. I992b. Chemical relationship between discharging fluids procuded in the Siena-Radicofani graben and the deep fluids produced by the geothermal fields of Mt Amiata, Torre Alfina and Latera (Central Italy). Geothermics, 21-3,401^13. GIANELLI, G., MEKURIA, N., BATTAGLIA, S., CHERSICLA, A., GAROFALO, P., RUGGIERI, G., MANGANELLI, M. & GABREGZIABHER, Z. 1998. Water-rock interaction and hydrothermal mineral equilibria in the Tendaho geothermal system. Journal of Volcanology and Geothermal Research, 86, 253-276. GIANELLI, G. & RUGGIERI, G. 2000. Contact metamorphism in the Larderello geothermal system. Proceedings World Geothermal Congress, Japan. 1163-1168. GIANELLI, G. & SCANDIFFIO, G. 1989. The Latera geothermal system (Italy): chemical composition of the geothermal fluid and hypotheses on its origin. Geothermics, 18,447^63. HlLDENBRAND, A., KROOSS, B.M. & SCHLOMER, S. 2002. N2
and CO2 gas breakthrough experiments on finegrained sediments. In: (ed.) ARIAULT ET AL. Poromechanics II. Swets & Zeitlinger, Lisse. HOPPE, W. 1960. Die Kali- und Steinsalzlagerstatten des Zechsteins in der DDR, Teil 1: Das Werra-Gebiet. Ereibereitung Eorschung-H., C 97/1. HORVATH, F. & TARI, G. 1999. IBS Pannonian Project: a review of the main results and their bearings on hydrocarbon exploration. In: DURAND, B., JOLIVET, L., HORVATH, F. & SERANNE, M. (eds) The Mediterranean Basins: Tertiary Extension within the Alpine Orogen. Geological Society, London, Special Publications, 156,195-213. KOCH, K. & VOGEL, J. 1980. Zu den Beziehungen von Tektonik, Sylvinitbildung und Basaltintrusion im Werra-Kaligebiet. Ereibereitung Eorschung-H. C 347. KONCZ, I. 1983. The stable carbon isotope composition of the hydrocarbon and carbon dioxide components of Hungarian natural gases. Acta MineralogicaPetrographica, Szeged, XXXVI, 33-49. KONCZ, I. & ETLER, O. 1994. Origin of oil and gas occurrences in the Pliocene sediments of the Pannonian Basin, Hungary. Organic Geochemistry, 21, 1069-1080. KRALJ, P. 2000. Thermal and mineral waters in northeastern Slovenia. Environmental Geology, 39, 488-500.
NATURAL CO9 ACCUMULATIONS IN EUROPE LEMOINE, M., BAS, T, ARNAUD-VANNEAU, A. & 8 OTHERS. 1986. The continental margin of the mesozoic Tethys in the Western Alps. Marine and Petroleum Geology, 3,179-199. LESNIAK, P. M. 1985. Open CO2-underground watersystem in the West Carpathians (South Poland) chemical and isotopic evidence. Chemical Geology, 49,275-286. LOLLAR, B. S., O'NIONS, R. K. & BALLENTINE, C. J. 1994. Helium and neon isotope systematics in carbon dioxide rich and hydrocarbon-rich gas reservoirs. Geochimica et Cosmochimica Acta, 58, 5279-5290. LOLLAR, B. S., BALLENTINE, C. J. & O'NIONS, R. K. 1997. The fate of mantle-derived carbon in a continental sedimentary basin: Integration of C/He relationships and stable isotope signatures. Geochimica et Cosmochimica Acta, 61,2295-2307. LOMBARDI, S., PlNTI, D. L., ROSSI, U. & FlORDALISI, A.
1993. 222Rn in soil gases at Latera geothermal field:a preliminary case history. Geologica Romana, 29, 391-399. MARINI, L. C. G. 1994. The role of carbon dioxide in the carbonate-evaporite geothermal systems of Tuscany and Latium (Italy). Acta Volcanologica, 5, 95-104. MASCLE, A., VIALLY, R., DEVILLE, E., BIJUDUVAL, B. & ROY, J. P. 1996. The petroleum evaluation of a tectonically complex area: The western margin of the Southeast Basin (France). Marine and Petroleum Geology, 13, 941-961. MATTHEWS, A., FOUILLAC, C., HILL, R., O'NIONS, R. K. & OXBURGH, E. R. 1987. Mantle-derived volatiles in continental crust: the Massif central of France. Earth and Planetary Science Letters, 85, 117-128. MAY, F. 2001. CO2 flux in a dormant intraplate volcanic field: the Westeifel, Germany. In: CIDU, R. (ed.) Proceedings of the 10th International Symposium on Water-Rock Interaction, A.A. Blakema, Villasimus, Italy, 2, 883-886. MAY, F, HOERNES, S. & NEUGEBAUER, H. J. 1996. Genesis and distribution of mineral waters as a consequence of recent lithospheric dynamics: The Rhenish Massif, central Europe. Geologische Rundschau, 85,782-799. MINISSALE, A., EVANS, W. C., MACRO, G. & VASELLI, O. 1997. Multiple source components in gas manifestations from north-central Italy. Chemical Geology, 142, 175-192. MINISSALE, A., MACRO, G., MARTINELLI, G., VASELLI, O. & TASSI, G. F. 2000. Fluid geochemical transect in the Northern Apennies (central-northern Italy): fluid genesis and migration and tectonic implications. Tectonophysics, 319, 199-222. MOLLER, P., DULSKI, P., GERSTENBERGER, H., MORTEANI, G.
& FUGANTI, A. 1998. Rare earth elements, yttrium and H, O, C, Sr, Nd and Pb isotope studies in mineral waters and corresponding rocks from NW Bohemia, Czech Republic. Applied Geochemistry, 13, 975-994. MULLER, W. 1958. Uber das Auftreten von Kohlensaure im Werra-Kaligebiet. Freibereitung Forschung-H,, A 101. PANICHI, C. & TONGIORGI, E. 1976. Carbon isotopic composition of CO2 from springs, fumaroles, mofettes, and travertines of central and southern Italy: a preliminary prospection method of geothermal areas. In: Proceedings of the 2nd UN Symposium on the
41
Development and Use of Geothermal Energy, San Francisco, 20-29 May 1975, 815-825. PAUWELS, H., FOUILLAC, C., GOFF, F. & VUATAZ, F. D. 1997. The isotopic and chemical composition of CO2-rich thermal waters in the Mont-Dore region (MassifCentral, France). Applied Geochemistry, 12, 411^27. PEARCE, J. M., HOLLOWAY, S., WACKER, H., NELIS, M. K., ROCHELLE, C. & BATEMAN, K. 1996. Natural occurrences as analogues for the geological disposal of carbon dioxide. Energy Conversion and Management, 37,1123-1128. PEARCE, J. M., NADOR, A. & TOTH, E. 2002. Living with CO2: Experiences from Hungary. Greenhouse Issues, 58,5-7. PHILLIPPE, Y., DEVILLE, E. & MASCLE, A. 1998. Thinskinned inversion tectonics at oblique basin margins: example of the western Vercors and Chartreuse subalpine massifs (SE France). In: MASCLE, A., PUIGDEFABREGAS,
C., LUTERBACHER, H. P. &
FERNANDEZ, M. (eds) Cenozoic Foreland Basins of Western Europe. Geological Society, London, Special Publications, 134, 239-262. RAZIN, P., BONIJOLY, D., LESTRAT, P., COUREL, L., POLI, E., DROMART, G. & ELMI, S. 1996. Stratigraphic record of the structural evolution of the western extensional margin of the Subalpine Basin during the Triassic and Jurassic, Ardeche, France. Marine and Petroleum Geology, 13, 625-652. ROURE, E, BRUN, J. P., COLLETTA, B. & VANDENDRIESSCHE, J. 1992. Geometry and kinematics of extensional structures in the Alpine Foreland Basin of southeastern France. Journal of Structural Geology, 14, 503-519. VASELLI, O., MINISSALE, A., TASSI, F, MACRO, G., SEGHEDI, L, IOANE, D. & SZAKACS, A. 2002. A geochemical traverse across the Eastern Carpathians (Romania): constraints on the origin and evolution of the mineral water and gas discharges. Chemical Geology, 182, 637-654. VRBA, J. 1996. Thermal mineral water springs in Karlovy Vary. Environmental Geology, 27,120-125. WEINLICH, F. H., BRAUER, K., KAMPF, H., STRAUCH, G., TESAR, J. & WEISE, S. M. 1999. An active subcontinental mantle volatile system in the western Eger rift, Central Europe: Gas flux, isotopic (He, C, and N) and compositional fingerprints. Geochimica et Cosmochimica Acta, 63, 3653-3671. WEISE, S. M., BRAUER, K., KAMPF, H., STRAUCH, G. & KOCH, U. 2001. Transport of mantle volatiles through the crust traced by seismically released fluids: a natural experiment in the earthquake swarm area Vogtland/NW Bohemia, Central Europe. Tectonophysics, 336,131'-150. ZIEGENHARDT, W. 1978. Uber Fehlereinfliisse bei der petrophysikalischen Speicher- und Deckgebirgsbewertung im gekliifteten Gestein. Zentralblatt angewandte Geologie,24,6-l5. ZIEGLER, P. A. 1982. Geological Atlas of Western and Centreal Europe. Shell International Petroleum Maatshappij B.V., Amsterdam. ZIEGLER, P. A. 1994. Cenozoic rift system of western and central Europe - an overview. Geologie en Mijnbouw, 73,99-127.
This page intentionally left blank
Analysis of CO2 leakage through 'low-permeability' faults from natural reservoirs in the Colorado Plateau, east-central Utah Z. K. SHIPTON1, J. P. EVANS2, D. KIRSCHNER3, P. T. KOLESAR2, A. P. WILLIAMS2 & J. HEATH2 1
Division of Earth Sciences, Centre for Geosciences, University of Glasgow, Glasgow G12 8QQ> Scotland, UK (e-mail:
[email protected]) 2 Department of Geology, Utah State University, Logan, UT 84322, USA 3 Department of Earth and Atmospheric Sciences, Saint Louis University, St Louis, MO 63103, USA Abstract: The numerous CO2 reservoirs in the Colorado Plateau region of the United States are natural analogues for potential geological CO2 sequestration repositories. To understand better the risk of leakage from reservoirs used for long-term underground CO2 storage, we examine evidence for CO2 migration along two normal faults that cut a reservoir in east-central Utah. CO2-charged springs, geysers, and a hydrocarbon seep are localized along these faults. These include natural springs that have been active for long periods of time, and springs that were induced by recent drilling. The CO2-charged spring waters have deposited travertine mounds and carbonate veins. The faults cut siltstones, shales, and sandstones and the fault rocks are fine-grained, clay-rich gouge, generally thought to be barriers to fluid flow. The geological and geochemical data are consistent with these faults being conduits for CO2 moving to the surface. Consequently, the injection of CO2 into faulted geological reservoirs, including faults with clay gouge, must be carefully designed and monitored to avoid slow seepage or fast rupture to the biosphere.
Effective design and implementation of geological CO2 sequestration projects require that we understand the storage capacity of a candidate site, the trapping mechanisms for gas, and the hydrodynamics of the system. The CO2must be segregated effectively from the atmosphere for periods of thousands of years (Rochelle et al 1999). Natural sources of CO2 include mantle degassing, metamorphism or dissolution of carbonates, oxidation or bacterial degradation of organic matter, and thermal maturation of hydrocarbons (Selley 1998). Numerous, large naturally occurring CO2 fields provide analogues for the integrity of stored gas systems (e.g. Allis et al. 2001). In some CO2 fields, however, gas leaks into the atmosphere, primarily along faults. We can study these active leaks to understand the factors that might control the feasability and safety of future CO2 injection projects and guide the design and implementation of such projects. In this contribution we examine the hydrology, stratigraphy, structural geology, and geochemistry of a naturally degassing CO2 reservoir in the Colorado Plateau of east-central Utah. The CO2 discharges from the hydrocarbon-rich Paradox Basin along the Little Grand Wash and Salt Wash faults. These faults cut sandstones, shales, and siltstones producing zones of clay-rich gouge that should theoretically be a barrier to flow (e.g. Freeman et al 1998). CO2charged springs and geysers, travertines (both active and ancient), and carbonate-filled veins are localized
along the fault traces. The faults are presently conducting CO2-rich fluids and the fault has conducted fluids for a substantial amount of time. Abandoned hydrocarbon boreholes are also active conduits for CO2 to the surface (Doelling 1994). In order to quantify the volume of CO2 that has leaked through the Little Grand Wash and Salt Wash faults, the CO2 sources, pathways, volumes, and rates of flow must be known. We have examined the distribution of the fault-related rocks and associated outcrop-scale structures of travertines and carbonate veins to characterize the flow paths. We have used the geochemistry of the spring waters, carbonates, and travertines to identify the potential sources of the fluids in the system. We have combined these data to develop a conceptual model for the groundwater/CO2 flow system (source, pathways, reservoir and caprocks).
Geological setting The Colorado Plateau and Four Corners region of the western United States contains at least nine producing or abandoned CO2 fields with up to 28 trillion cubic feet of CO 2 gas (Allis et al 2001). Most of these fields are fault-bounded anticlines with fourway anticlinal closure or fault seal along one margin of the field. We focus on the Little Grand Wash and Salt Wash faults, which are at the northern end of the
From: BAINES, S. J. & WORDEN, R. H. (eds) 2004. Geological Storage of Carbon Dioxide. Geological Society, London, Special Publications, 233,43-58.0305-8719/047$ 15.00 © The Geological Society of London 2004.
44
Z.K.SHIPTON£rAL.
Paradox Basin. The Paradox Basin contains a number of actively producing oil and natural gas fields, as well as CO2 fields including the Lisbon field and McElmo Dome field. Immediately south of the study area is the abandoned Salt Wash oil field (Peterson 1973). To the north of the Paradox Basin are the active methane and CO2 fields of the San Rafael Swell (e.g. Drunkards Wash, Perron Dome). Many of the methane fields in the area produce significant amounts of CO2 (Cappa & Rice 1995), much of which is vented to the atmosphere. The Paradox Basin is defined by the extent of organic-rich Pennsylvanian and Permian marine limestones, shales and evaporites (Fig. 1). Hydrocarbon source rocks occur in the Ismay-Desert Creek and Cane Creek cycles of the Paradox Formation (Nuccio & Condon 1996), a mixed sequence of dolostone, black shales, anhydrite, and halite. These are overlain by Triassic and Jurassic fluvial and aeolian redbeds. The oldest lithologies that crop out in the study area are red-brown finegrained sandstones of the Middle Jurassic aeolian Entrada and Curtis Formations. The Middle Jurassic Summerville Formation forms characteristic low cliffs with thin bedding and seams of gypsum. The Upper Jurassic Morrison Formation consists of stacked fluvial channels of the Salt Wash Sandstone member, overlain by the bentonite-rich lacustrine shales of the Brushy Basin member. The Lower Cretaceous Cedar Mountain Formation and the Upper Cretaceous Dakota Sandstone are conglomeratic channel sandstones. The youngest formation exposed in the field area is the Upper Cretaceous Mancos Shale, a dark organic-rich marine shale. Approximately 2500m of Cretaceous and Tertiary rocks have been eroded from the area (Nuccio & Condon 1996). The Little Grand and Salt Wash faults (Fig. 1) affect the present-day flow of gas and water. Carbonate springs, an active CO2-charged geyser, and actively forming travertine deposits are localized along the Little Grand Wash fault zone (Baer & Rigby 1978; Campbell & Baer 1978; Doelling 1994) and numerous CO2-charged springs occur in the region of the Salt Wash faults (Doelling 1994). The faults are part of a WNW trending set of 70-80° dipping normal faults in the region. Timing of continued movement along these faults is poorly known, though we present arguments below for Early Tertiary and Quaternary slip. The faults cut the Mancos Shale, consistent with substantial fault activity having occurred at least up to the Middle Cretaceous. The faults cut a north-plunging anticline (Figs 1 & 2), which could be related to salt movement in the Paradox Formation at depth. A basinwide system of salt anticlines initiated when the salt was loaded by the Pennsylvanian/Permian elastics shed off the Uncompahgre uplift to the NE.
Reactivation of the salt-related anticlines and faults occurred during Laramide (Eocene) contraction (Chan etal 2000).
The Little Grand Wash fault The Little Grand Wash fault is a south-dipping arcuate normal fault with a surface trace length of 61 km (Fig. 1). The Little Grand Wash fault consists of two parallel strands from 3.2km east to O.lkm west of the Green River; elsewhere it has only one strand (Fig. 3). Total vertical separation on the fault near the Green River is 180-210m, most of which is accommodated by the southern fault strand. The two strands of the fault were encountered at depth in an abandoned well (Amerada Hess, Green River no. 2 drilled in 1949, total depth 1798m) at 805m and 970m. Drilling records state that the deeper of the two faults has Cutler Group sediments in the hanging wall and Hermosa Group sediments in the footwall. It is therefore unclear what the offset of the fault is at this depth, or whether the fault cuts the Paradox Formation (Fig. 2). The fault is cut by several stream channels that provide excellent cross-sectional exposures of the fault zone and associated host-rock alteration. Between the two main strands of the fault, smaller faults define structural terraces with varying dips (Fig. 4a). Slickensides on subsidiary fault surfaces indicate mostly dip-slip with some oblique leftand right-lateral movement. The fault zone contains 70cm to 3 m of foliated clay gouge with occasionally well-defined, sub-planar slip-surfaces. Smaller faults with offsets less than 1 m are decorated with a thin (millimetre-thick) foliated purpleblack, clay-rich fault gouge and occasional thin calcite veins (1-2 mm thick) with subhorizontal fibres. The normally dark purple-red Summerville Formation is bleached to a pale yellow for up to several metres into the footwall. This alteration is localized along the subsidiary faults and within certain beds.
CO2 emissions and springs Several active CO2-charged springs are localized along the two strands of the Little Grand Wash fault zone (Fig. 3). The Crystal Geyser erupts to heights of up to 25m at 4-12-hour intervals (Fig. 5a). This is not a geyser in the strict sense of the term; the water in the geyser is cool and the eruptions are powered by CO2-charged waters rather than a heat source. The geyser began erupting when the Glen Ruby #1X well was drilled in 1935. This abandoned exploration well was completed to the base of the Triassic section (TD 801m). Occasionally the geyser water
CO2 LEAKAGE THROUGH LOW-PERMEABILITY FAULTS
45
Fig. 1. (a) Regional geology of the Little Grand Wash and Salt Wash faults. Stars mark the location of known CO2charged springs. Dot marks the town of Green River, Utah. Compiled from Doelling 2001; Williams 1964; Williams & Hackman 1971. The line of section shown in Figure 2 is indicated. Inset shows location of study area and approximate boundaries of the Paradox Basin, (b) Stratigraphic column, (after Doelling 2001, thicknesses converted to metres from Hintze 1993). Stippled areas indicate likely reservoir or aquifer rocks, cross hatched areas represent likely cap rocks or seals.
46
Z.K.SHIPTONZTAL.
Fig. 2. Schematic cross-section across the Little Grand Wash and Salt Wash faults, approximately along the northplunging anticline axis. Depths are above or below sea level. Abandoned oil wells (projected onto the line of section) give control on the stratigraphy. The structure of the Salt Wash faults is not constrained at depth by any well data. The line of section is indicated on Figure 1.
Fig. 3. Local map of the Little Grand Wash fault showing the location of the travertine deposits, the gas seep, oil seep, and abandoned oil wells mentioned in the text.
has a strong sulphur smell, and/or a thin film of hydrocarbons coating the water pooled around the drill pipe. The driller's records document that the well was spudded into a travertine mound and that the travertine thickness was 21.5m before hitting bedrock (Baer & Rigby 1978). The spring system must have been active prior to the well being drilled. Anecdotal reports and scant completion records indicate that the geyser erupted more regularly (at 11 hour intervals) in the past. The hole is currently open to c. 130m depth. Much damage is reported to have been done to the borehole, including dynamiting of the hole, attempts at cementing, and dumping of railroad ties. CO2-charged emissions continue despite this damage. A 1.5 m high steel pipe was added to the top of the well casing in 2001. This pipe has had no discernable effect on the geyser's eruption pattern. Three other springs are located on the system of travertine mounds around the geyser. To the NE, a water-filled pool and a chocolate-brown, mud-filled pool erupt penecontemporaneously with the geyser. These are located on the eastern edge of the travertine mound where it has been badly damaged by vehicles.
To the north, another small water-filled pool is located on the active travertine slope. The close correlation of the timing of the geyser and activity of these springs suggests that the latter either reflects some escape of CO2-charged waters from the well bore at shallow levels, or that these pools could be the original, pre-well flow paths for the CO2-charged waters to the surface. It is common for hydrothermal travertine spring sources to switch locations when the flow paths become cemented (Chafetz & Folk 1984). Although in this cool water system, cementation may not be as rapid as in hydrothermal systems, it is still likely that the flow paths switch with time when they become sealed by calcite precipitation. Within the Green River, small gas seeps produce small streams of bubbles. These can be observed on both banks of the river and are parallel to the trace of the fault. A kilometre to the east of the Crystal Geyser, in a low-lying area in the footwall of the fault, gas seeps audibly from the ground (Fig. 3). A small amount of water flows sporadically from upstream of this location leaving salty deposits and wet sand in the base of the wash. Further to the east,
CO, LEAKAGE THROUGH LOW-PERMEABILITY FAULTS
47
Fig. 4. Photos of Little Grand Wash fault, (a) Cross-sectional view looking east along fault strike, about 30m east of Crystal Geyser. At this location the fault consists of several strands separating structural terraces. The southern fault strand accommodates most of the offset here. Note the thick calcite/aragonite veins cutting the travertine that sits in the hanging wall of the central fault strand. Js, Jurassic Summerville Formation; Kmu, Cretaceous Mancos Shale. Photo about 500m across, (b) Boxwork of veins beneath a travertine mound centred on the northern fault strand. Note that the fault does not clearly cut all the way up the travertine mound, and appears to be cut by some of the veins. The host rocks are Jurassic Morrison Formation; Jmb, Brushy Basin shale member; Jms, Salt Wash sandstone member. Three people in the centre for scale.
an oil seep is located on the southernmost fault strand (Fig. 3). A shallow pit contains fresh oil indicating that there is active flow of petroleum to the surface. The outcrop close to this seep (Salt Wash member of the Morrison Formation) contains patches of oil staining.
Mineral deposits Modern travertines at the Little Grand Wash fault consist of bedded travertine mounds that were deposited from the Crystal Geyser and surrounding springs (Fig. 5). The surface of the active geyser mound has a classic rimstone texture and lobes of travertine have built out to form sub-metre scale caves with stalactites. An ochre colour indicates a small component of iron oxide. These travertines successively bury the vegetation that surrounds the geyser. The travertine surface has regions of actively forming and inactive travertines, presumably con-
trolled by switching of the source spring location or by lateral migration of flow across the surface of the mound. To the SW of Crystal Geyser, older carbonate deposits are in the process of being covered by the present day mound. These deposits are in the form of two distinct mounds of carbonates and breccias, one at the level of the river cutting an older one to the east (Fig. 5a). From a distance, it appears as though variably dipping veins are visible within these mounds, though these are not veins in the usual structural geological sense. They consist of centimetre-thick to tens of centimetres-thick subhorizontal tabular masses of radiating acicular calcite and aragonite crystals 6-15 cm long with botryoidal or mammilated top surfaces (Fig. 5b). Fresh surfaces are bright white, with occasional pale yellow banding. These veins often have paired banding and/or mammilated surfaces, which face towards the centre of the vein. Deposits of this form (described as ray-crystal crusts by Folk et al. 1985) have been interpreted to form
48
Z.K.SHIPTONE7AL.
Fig. 5. (a) Eruption of the Crystal Geyser looking east along fault. Active and inactive travertine mounds can be seen around geyser, and ancient travertine mounds and carbonates are located along fault surface. Strands of fault are marked. Je, Entrada Formation; Jet, Curtis Formation; Js, Summerville Formation; Jms, Salt Wash Sandstone (Morrison Formation); Jmb, Brushy Basin member of Morrison Formation; Kmu, Mancos Shale, (b) Thick veins cutting the lower inactive travertine mound at the edge of the Green River. Veins dip into mound, but roll over to subvertical at front of photo. Top of this mound consists of a breccia with clasts of vein material. Modern travertine deposits are covering the ancient mound in bottom left of photo. This section is about 2m high.
underwater, with the apex of the radiating crystals pointing towards the source of fluids (i.e., in the centre of the veins). The centres of some of these veins contain stalactite-like structures suggesting that subhorizontal fissures were infilled above the water table. Above these deposits lie travertinecemented breccias that include clasts of ray-crystal calcite and sandstone clasts. The surfaces of the inactive mounds have some rimstone textures preserved, though they have been extensively damaged by vehicles. The 1867 Powell expedition documented 'satin spar' at this location (Powell 1895), which we interpret to be either the travertine terraces or the bright white ray-crystal calcite veins. Other ancient travertine deposits along the fault occur at higher elevations (up to 37m; Baer & Rigby 1978) than the one presently forming. These deposits tend to form resistant caps to small buttes. The hanging wall of the fault in Figure 4a contains a thicker ancient travertine deposit than the footwall, though it is unclear if this deposit filled in the space
left by faulting, erosion of the hanging wall, or if movement on the fault cut a pre-existing deposit. The ancient travertines consist of dense-bedded layers, 1-2 mm thick, interbedded with vuggy open carbonate 1-3 cm thick. Horizontal and vertical carbonate ray-crystal veins up to 30cm thick cross cut these deposits. In an outcrop east of the geyser, an impressive array of millimetre to centimetre-thick veins with a boxwork pattern has completely obliterated the original fabric in the fault gouge (Fig. 4b). The veins in this array are not parallel to the original fault-parallel gouge fabric. This boxwork is cross cut and offset by thicker subvertical veins with occasional stalactite textures. The variation in the locations of the inactive deposits shows that the loci of active effusion of CO2-rich waters have changed in the past. The carbonate mounds form predominantly between the two strands of the Little Grand Wash fault, or in the footwall of the fault. The spatial correlation of the ray-crystal calcite/aragonite veins and the
CO, LEAKAGE THROUGH LOW-PERMEABILITY FAULTS
49
Fig. 6. Map of Salt Wash fault outcrops mentioned in text. Locations of springs and travertine/carbonate vein deposits are marked including ones not specifically mentioned in text. Abandoned oil wells are shown with their total depth (TD) and depth to top of the Paradox salt.
travertines is consistent with the latter having been the 'plumbing system' to the travertines. The Salt Wash faults The Salt Wash faults are a set of N 70° W striking normal faults. The map-scale structure of the faults reveals two normal fault systems that form a shallow graben over 15km long (Fig. 1). The faults offset Jurassic Entrada Sandstone in their footwalls against Cretaceous and Jurassic Cedar Mountain Formation in the centre of the graben. The Salt Wash faults consist of two linked en echelon graben segments (Doelling 2001) and may be structurally linked to the Moab fault system to the SE (Fig. 1), though Quaternary deposits obscure the area where this linkage potentially occurs. The depth to which these faults extend is uncertain (Fig. 2), but they may sole into the Paradox salt sequence, and could be related to salt tectonics in the region. We have studied two areas along the northernmost Salt Wash fault in detail: the Tenmile Geyser and Torrey's spring areas (Fig. 6). Tenmile Geyser The Tenmile Geyser is centred on an abandoned well 200m into the hanging wall of the northern fault (Fig.7a). A drill pipe sits within a low mound
of flaky travertine with poorly developed rimstone textures. The Tenmile Geyser erupted infrequently in the past (Doelling 1994) and continues to erupt with infrequent 1-1.5m high eruptions. A second mineral-charged spring sits on a low mound 100m into the footwall of the fault. There is anecdotal evidence that a set of travertine terraces with rimstone textures used to exist at this locality. This mound has since been excavated into a pit about 2 by 3 m in size and 1.5m deep. The bottom of this pit does not reach the base of the travertine deposit. There is an almost constant stream of CO? from three vents in the base of the pool, but this spring has not been documented to have geyser-style eruptions. There are extensive inactive travertines up to 4m thick at elevations up to 30m above the level of the present-day spring, some with well-developed rimstone textures. These travertines are presently being quarried. A 2-1 Om thick zone of fractures, intense alteration, and veining obliterates the primary sedimentary structures beneath portions of the travertine deposits. Bedding parallel carbonate veins, 2-5 cm-thick, extend up to 50m away from the fault zone. In some places mammilated ray-crystal veins change from vertical to horizontal orientation within the outcrop. These locally contain open vuggy deposits with rhombohedral calcite crystals, interpreted as forming in subaerial or spelean pools. Fractures up to two metres deep that cut the mounds have been filled with bedded travertine. The
50
Z.K.SHIPTONETAL.
gies separated by clay-rich foliated gouge (Fig. 7b). In the footwall north of the Tenmile Geyser, the Entrada Sandstone (usually red) has been extensively bleached to a light tan to pale yellow. In other places along the fault zone, the Entrada Formation is bleached in zones close to the fault and along fractures (Fig. 7c). Close to the fault, poikilotopic aragonite cements occur preferentially in certain horizons of the host rock.
Torrey 's spring
Fig. 7. Photos of Salt Wash fault, (a) Aerial view of northern Salt Wash fault looking north, showing distribution of springs and travertines in vicinity of Tenmile Geyser, (b) Cross-sectional view looking east along fault strike, about 800m west of Tenmile Geyser. At least three fault strands, consisting of thin gouge zones, can be identified based on differing lithologies. (c) Alteration focused along fractures and bedding planes adjacent to fault zone. At top of outcrop the base of an ancient travertine mound can be seen. Je, Entrada Formation; Js, Summerville Formation; Jms, Salt Wash Sandstone (Morrison Formation); Jmb, Brushy Basin member of the Morrison Formation; Kcm, Cedar Mountain Formation.
travertines tend to form resistant caps to a line of small buttes along the fault trace. All the travertines along the Salt Wash faults are localized either on the northernmost fault trace or in the footwall of this fault. The only activity seen within the graben is the Tenmile Geyser. The fault gouge is locally well-exposed, and consists of a zone up to 5 m thick of slices of host litholo-
This spring occurs in the Entrada Sandstone in the footwall to the northern fault (Fig. 6). The spring is located in the centre of a fresh-looking travertine mound about 15 m in diameter, which has slope of c. 8°. The saline spring bubbles almost constantly and occasionally smells of sulphur. To the authors' knowledge, it has never erupted in geyser-style eruptions. This spring appears to be close to the location of the Delaney Petroleum Corporation #1 drill hole drilled in 1949, which had a total depth of 299m. There was no oil show in this well. The spring was first visited by the authors in June 2000, and again in June 2001. During this interval, the travertine had advanced significantly over a large area and had created a small travertine 'frozen waterfall' into a dry river bed to the west. This is evidence of the rapid growth rate of these travertine mounds and the high CaCO2 content of the effusing waters. Growth rates of 1 mm to 10cm per year have been documented in other travertine deposits, though these were areas of warm or hot water deposition (Folk et al 1985). The main Salt Wash fault outcrops to the south of Torrey's spring. A line of small buttes (10-20 m high) capped with travertines parallels the location of the fault. Commonly, the steeply dipping veins are larger than, and cross cut, the subhorizontal veins. Fissures have been infilled with travertines, and the infill breccias contain fragments of the vein material. These outcrops also contain gypsum, though it is unclear if they are associated with the fault or if they are from the original host rock (Entrada Sandstone). Alteration of the host rock to pale yellow 'bleached' sandstone is focused along fractures close to the fault (Fig.7c). Bedding-parallel calcite veins 2-5 cm thick extend up to 350m from the fault in the footwall.
Water chemistry We collected and analysed the chemistry of waters from the Crystal Geyser on the Little Grand Wash fault, and from two CO2-charged springs along the Salt Wash fault (Torrey's spring and the excavated
51
CO, LEAKAGE THROUGH LOW-PERMEABILITY FAULTS
Table 1. Chemical data from field and laboratory testing and associated cation-anion balance errors. Samples from the Crystal Geyser are labelled CG, from Torrey 's spring T, and from the excavated spring near Tenmile Geyser SW. Cations (mmol/1) Sample number
Temp (°C)
pH
TDS (mg/1)
Ca
Mg
Na
K
Fe
Mn
Sr
CG92.1 CG92.2 CG92.3 T1/T2 SW1/SW2
16.0 16.0 16.0 19.4 23.0
6.96 6.7 6.46 6.37 6.26
13920 13685 13555 21188 20020
2.37 9.71 10.78 20.83 20.76
8.72 8.80 9.30 7.86 8.27
174.52 171.74 167.82 218.63 236.38
7.62 7.67 7.65 10.44 9.39
0.00 0.00 0.00 0.14 0.08
0.00 0.02 0.02 0.02 0.00
0.01 0.05 0.05 0.14 0.16
Anions(mmol/l) Sample number
HCO3
C03
Cl
S04
% Balance error
CG92.1 CG92.2 CG92.3 T1/T2 SW1/SW2
61.00 50.00 46.40 75.70 58.90
0.00 0.00 0.00 0.00 0.00
91.15 95.43 98.23 204.52 186.38
24.33 24.64 25.07 29.47 30.81
0.9 5.3 5.1 -8.3 -0.4
spring at Tenmile Geyser). Water samples were collected in polyethylene bottles and samples for cation analyses were acidified with reagent-grade nitric acid to pH 2 or less. Field analyses included pH and temperature. Alkalinity was not determined in the field, but relatively low balance errors for the analyses show that this did not cause significant error. Samples were kept refrigerated until analysed for major element composition. Table 1 presents the chemical data. Dissolution of calcite cannot account for all of the CO2 in the water because there is substantially less calcium in the water than bicarbonate. There must be an alternative or additional source for the CO2. Similarly, the excess sodium in the waters (Cl~/Na+ ratios range from 0.6-0.8 for Crystal Geyser system waters) could not be solely from the dissolution of halite and therefore must be derived from other minerals such as Na-bearing montmorillonite. The water we collected from the main Crystal Geyser is supersaturated with respect to calcite, aragonite and dolomite (Table 2). Modelling of our data and the water chemistry of Mayo (1991) with the programme Wateq (Truesdel & Jones 1974) indicates that the Crystal Geyser water is supersaturated with respect to aragonite, calcite, dolomite, fluorite and gypsum (Table 2). This suggests that the fluid source for the veins could have had a similar composition to present-day fluid. Results of modelling with Salt Norm (Bodine & Jones 1986) are consistent with the chemistry of the geyser water being a mixture of meteoric water and brine from redissolved marine evaporites. For a given water analysis this program calculates the salt
composition that would result if that water were evaporated to dryness. The water composition will reflect the materials with which the water came into contact; therefore the calculated salt composition can provide clues about the composition of those materials. Bodine & Jones (1986, p.37, Fig. 4) list criteria that define water types based on the calculated salt composition. Applying those criteria to our water samples suggests that the water in the Crystal Geyser system primarily is a mixture of redissolved marine evaporites and meteoric water (Table 3). Evaporites occur in the Paradox salt, and also in shallower units such as the Carmel Limestone and Summerville Formation. Water chemistry was monitored during two separate Crystal Geyser eruptions to see what chemical changes occur during an eruption cycle. There were larger changes in water chemistry for the longer eruption (30 minutes as opposed to less than 10 minutes) consistent with an influx of relatively low salinity groundwater recharging the well bore during an eruption (Fig. 8).
Isotope data Samples of carbonate veins and travertine were analysed in the stable isotope laboratory at Saint Louis University using both continuous-flow and conventional techniques. For continuous flow analysis, submilligram aliquots of sample powder were digested in orthophosphoric acid at 90 °C for several hours in an automated extraction device. Liberated gas was entrained in a helium stream, passed through
Z.K.SHIPTON£TAL.
52
Table 2. Saturation indices of water samples from the Crystal Geyser, Torrey's Spring and the Salt Wash excavated spring compared with samples from Mayo (1991). Samples labelled as in Table 1.
Aragonite Calcite Dolomite Fluorite Gypsum
Mayo 1991
CG92.1
CG92.2
CG92.3
T1/T2
SW1/SW2
1.140 1.289 2.134 3.283 0.153
0.167 0.317 1.259 NA -0.977
0.442 0.592 1.201 NA -0.372
0.217 0.367 0.730 NA -0.320
0.595 0.743 1.169 NA -0.108
0.434 0.579 0.912 NA -0.098
Table 3. Comparison of Little Grand Wash fault and Salt Wash fault spring waters with Salt Norm criteria for identifying water type. Samples labelled as in Table 1.
Re-solution of marine evaporites Meteoric source Meteoric source Meteoric source
Normative salts
CG92.1
CG92.2
CG92.3
T1/T2
SW1/SW2
>78% halite presence of thenardite presence of magnesite presence of dolomite
76.9%
80.2% 4.2%
77.3% 10.3%
80.3% 8.3%
79.3% 10.4%
6.7%
6.6%
3.7%
3.5%
a GC column for isolation of the CO2 and then transferred on-line to a gas-source, isotope-ratio mass spectrometer. Duplicates and some triplicate analyses were made for most samples. Most of the samples were also analysed by conventional (manual) techniques following the laboratory procedure described in Kirschner et al (2000). The average values of these analyses are depicted in Figure 8a. Carbonate standards (two in-house standards) were analysed with the samples. Analytical precision of 813C was below 0.10%0 (N=23- la); o18O was below 0.26%0 (N= 20 excluding one set of 3 standards; Icr). Several trends can be observed in the data (Fig. 9a). Travertine data from individual localities lie on sub vertical trends. In contrast, most ray crystal calcite/aragonite veins have lower 818O and similar or lower 813C values relative to the travertines from the same localities (dashed trends in Fig. 9a). The 818O values are consistent with precipitation of the carbonate at low temperature (c. 15-35 °C) from water with a 818O value less than -?%«?. This is similar to partly evolved (isotopically altered) meteoric water that is an important component of the water in the faults and geysers (cf. IAEA 2001). The variation in 818O values among sites could be due to variable mixing of meteoric water with other fluids or partial exchange with carbonate rocks or CO2 at depth. The vertical 813C trend of the travertines is consistent with CO2 degassing during discharge and surface flow of the spring water. Light isotopes of carbon and oxygen preferentially fractionate into the gas during degassing, resulting in an appreciable increase in carbon isotope values of the dissolved inorganic carbon. Although the same phenomenon
5.3% 2.0%
occurs for the oxygen isotopes, the resulting shift in isotopic values is not large due to the overwhelming abundance of oxygen in the water. Similar trends have been documented in other studies of travertine deposits and associated spring waters (Fig. 9b and references therein). The positive increase in 818O and 813C values between veins and travertines from individual sites is most likely associated with increased isotopic fractionation due to cooling of water during upward migration and surface discharge. Assuming the water is cooling as it is ascending and there is an abundance of fluid in the system, then isotopic values of carbonates precipitating from cooler water will be higher (heavier) for both carbon and oxygen, resulting in a positive 818O-813C trend. Evaporation of the water could also have produced the positive trend. If a significant amount of water evaporates, then the water that remains becomes isotopically heavier (because water molecules with the lighter isotopes evaporate preferentially). Carbonates precipitated from water that has undergone significant evaporation will consequently have higher (heavier) oxygen isotopic values (and carbon values, assuming that CO2 is degassing at the same time). Either mechanism can produce positive trends.
Discussion In order to evaluate the effectiveness of the Little Grand Wash and Salt Wash faults as conduits for leakage of CO2 we need to know the source of the CO2, the volume of the gas reservoir, and the rate at
CO2 LEAKAGE THROUGH LOW-PERMEABILITY FAULTS
53
Fig. 8. Variation in sodium (dotted lines) and potassium (solid lines) concentrations from the Crystal Geyser during two separate eruptions (time from start of eruption in minutes along ;t-axis). The squares show data from an eruption that lasted 30 minutes, the diamonds show data from an eruption that lasted less than 15 minutes.
which this reservoir is being depleted. Although our work is still in its initial stages, the preliminary results can be used to build a working model for the source, pathways, and timing of flow in the Little Grand Wash and Salt Wash fault system.
Source of CO2 A number of processes can produce CO2 within basins including mantle degassing, metamorphism or decarbonation of carbonates, oxidation and/or bacterial degradation of organic matter, and maturation of hydrocarbons (Selley 1998). Helium isotope data of gas from the Crystal Geyser and a spring on the Salt Wash fault suggests only a minor component of mantle-derived helium (Heath et al 2002; Heath 2004), thus excluding a mantle source for the CO2. The gas in the Paradox basin is therefore likely to be sourced from one or more of the remaining processes, though this source cannot be unequivocally identified solely from the carbon and oxygen isotope data of the carbonates or water chemistry. The 813C values of +4 to +5%o for the veins are probably more closely associated with the subsurface fluid than the higher 813C values of the travertine samples. Barring the presence and dissolution of carbonates with unusually high 813C values in the basin, two probable sources of isotopically heavy CO2 in the field area are biologically mediated hydrocarbon generation and thermallyinduced decarbonation of carbonates. Many previous studies have focused on existing oil and gas deposits in the Paradox Basin (Hansley 1995; Nuccio & Condon 1996) and on the evidence for a palaeohydrocarbon play in and around the Moab fault (Chan et al 2000; Garden et al. 2001). The presence of the oil seep at the Little Grand Wash
Fig. 9. Stable isotope data for travertines and ray-crystal calcite/aragonite veins analysed in this study, and for travertines and associated spring waters from published data, (a) 813C (PDB) v. 818O (SMOW) for samples of veins (solid symbols) and travertines (unfilled symbols) at the Little Grand Wash and Salt Wash faults. Differences between veins and travertine at individual localities could be due either to decrease in temperature when fluids evulse onto land surface or different sources of fluid. Vertical arrays formed by travertine data are consistent with progressive CO2 degassing during surface flow of water, (b) Vertical data arrays of travertine and travertine-forming spring waters in 813C 818O space are common and result from downstream CO2 degassing of spring water, seasonal variations in isotopic values of source waters, and variable microbial influence in facilitating the precipitation of travertine carbonates. All six studies are from active travertine deposits: (1) Mammoth Hot Springs, Yellowstone (Friedman 1970), spring temperature in °C; (2) Durango, Colorado (Chafetz et al 1991£); (3) Oklahoma (Chafetz et al 19910); (4) Coast Range, California (Amundson & Kelley 1987); (5) near Florence, Italy (Guo et al. 1996); (6) Central Italy (Pentecost 1995).
54
Z.K.SHIPTONCTAL.
fault suggests that the fault is acting as a conduit for oil as well as CO2. The geochemistry of the oil may be consistent with it coming from Lower Permian Formations (P. Lillis pers. comm. 2001), which are the source for much of the oil in the Upper Palaeozoic rocks of the northern and central Rocky Mountains (Claypool et al 1978). Sanford (1995) concluded, however, that the Permian was an unlikely source rock for hydrocarbons in the Paradox Basin based on a study of palaeo-groundwater flow in the White Rim Sandstone, which has been bleached by hydrocarbon-bearing reducing fluids (Hansley 1995). Alternative hydrocarbon source rocks in the area include the late Proterozoic Chuar group, the Mississippian Chainman shale, the Lower Triassic Sinbad Limestone, and shales within the Paradox Formation. Although isotopically light 813C values of hydrocarbons and associated CO2 are the norm for hydrocarbon deposits (e.g. Schoell 1983), it is possible for hydrocarbon-associated CO2 to have isotopically heavy 813C values due either to biologically mediated reactions (e.g. Coleman et al 1988; Jenden et al. 1988) or to low-temperature carbon isotope exchange between hydrocarbons and CO2. Neither can be excluded with the present data set. A more probable source of isotopically heavy and abundant CO2 is the thermal decarbonation of carbonates (cf. Baumgartner & Valley 2001). Cappa & Rice (1995) presented evidence that some of the CO2 in the gas fields of southern Utah and Colorado was produced by high-temperature thermal decomposition of the Mississippian Leadville Limestone or decomposition of kerogen within the Leadville Limestone or the Paradox Formation. We suggest that this may have occurred within the contact aureoles of Tertiary intrusions (the La Sal and Henry mountains). Both the Leadville Limestone and the potential hydrocarbon source rocks lie within or below the Paradox Formation. Therefore, regardless of the source of CO2, it must have migrated through the salt.
at least as thick as those at the Crystal Geyser occur in areas where no drilling has occurred. The various ancient travertines are evidence that the migration pathways within and next to the fault zones have switched over time. Some of the ancient travertines are located up to 37m above the level of the present day springs. Assuming that they were not initially deposited at the tops of the buttes, the amount of incision could be used to infer the age of the travertines. Baer & Rigby (1978) suggested a date of 200000 years for the highest-level inactive travertines based on Colorado Plateau uplift rates. The uplift rate on the Colorado Plateau is currently a focus of debate; however, these spring systems have been active for a substantial amount of time.
Timing of fault activity Although we do not yet have geochronological constraints on the age of activity on the Little Grand and Salt Wash faults, structural relations between faults and travertine deposits are consistent with recent fault movement. Travertine filled fractures within the ancient travertine deposits could be related to seismic events (cf. Hancock et al. 1999). Some of the travertines appear to be nestled within a hanging wall half-graben, suggesting that a scarp existed at the time of travertine deposition. This could be the result of differential erosion of the relatively soft Mancos shale in the hanging wall, or due to syndepositional movement along the faults. Quaternary activity has been reported along the Salt Wash fault (Hecker 1993). These faults may be related to movement of the Paradox salt, which has resulted in numerous episodes of fault motion SE of our study area (Chan et al. 2000; Garden et al 2001). Pevear et al (1997) used K-Ar techniques on fine-grained illite of the Moab fault to show that this fault was active between 60-50 Ma consistent with one of the regional episodes of salt movement and dissolution (Chan et al 2000; Garden et al 2001).
Timing of fluid flow The ancient travertine terraces and veins preserved at Little Grand Wash and Salt Wash demonstrate that the faults have been the focus of CO2-charged waters for a substantial amount of time. Although many of the springs in the area are due to recent drilling activities, there continues to be flow from springs and gas seeps that are not associated with wells. The flow at the Crystal Geyser was active prior to 1935, when the Ruby 1-X well was drilled, and is likely to have produced the deposit that was observed by the Powell expedition in 1867. The rest of the drillingassociated springs have only poorly developed (incipient) travertine mounds. Deposits of travertine
Flow paths for CO2 and water All except one of the potential sources of CO2 are below the Paradox salt. The Paradox salt is an interbedded unit with halite beds up to several metres thick. Salt can deform by crystal-plastic deformation at high strain rates, which can result in rapid healing of faults. Therefore faults that cut salt are not usually considered to be conduits for fluid flow unless faulting forms a juxtaposition of non-salt rocks between the hanging wall and foot wall. From driller's records of wells in the study area, the thickness of the salt below the Little Grand Wash fault is
CO, LEAKAGE THROUGH LOW-PERMEABILITY FAULTS
of the order of 650-1300 m but the fault offset is only c. 200m. Thus flow along the faults is the likely pathway through the salt. Given the clay-rich nature of the fault zones within the Jurassic and Cretaceous sequences we would generally expect a low-permeability fault rock. However, the localization of springs and travertines, the presence of hydrocarbon seeps along the Little Grand Wash fault, and the close association of faulting and bleached sandstones (this study; Chan et al 2000; Garden et al 2001) suggests that there is a component of up-dip flow within or adjacent to these fault zones. Outcrops of the fault show that the foliated fault gouge has a strong fabric anisotropy, and that in places slip is localized onto discrete slip surfaces. This fabric anisotropy will result in anisotropy in fault gouge permeability. Specifically, the permeability is likely to be substantially higher along the fault surface than in the crossfault direction. Regional aquifers that could act as reservoirs for the CO2 include the Entrada Sandstone, the Navajo/Wingate sandstones and the White Rim sandstone. The water temperature (18°C) indicates that shallow aquifer waters are producing the springs, unless slow upwards flow rates allow thermal equilibration with the surrounding rocks. Heat flow values in the Green River area are consistent with geothermal gradients of 34 °C km"1 (Nuccio & Condon 1996) suggesting that the Navajo-Wingate aquifer (200-450m below the footwall to the Little Grand Wash fault) is the reservoir for shallow groundwater that is being charged with deep CO2. Several other CO2-charged springs occur in this region. The once spectacular Woodside Geyser, situated at the old town of Woodside approximately 40km north of the study area, now only erupts sporadically to a few metres from an abandoned drill hole. The Tumbleweed and Chaffin Ranch geysers erupt occasionally from drill holes to the south of the faults in this study (Fig. 2). These geysers, and the springs along the faults generally fall along the line of the regional north-plunging anticline axis (Fig. 2). This suggests that the flow of CO2 or CO2-charged fluids is focused along the anticline axis. The active springs along the Little Grand Wash and Salt Wash faults all occur in the footwall, apart from the Tenmile Geyser, where the host drill hole may penetrate the fault. The fault may be acting as a barrier to regional south-directed groundwater flow in the Navajo/Wingate aquifer (Hood & Patterson 1984), concentrating the CO2-charged fluid in a footwall reservoir. It would be contributing to the flow path through the impermeable layers above the aquifer (the Carmel limestone, Morrison Formation etc.). In this case, the fault would have higher up-dip permeability than cross-fault permeability. The travertines may be located at structural complexities along the
55
fault zone, such as at the ends of segments or at stepovers. This relationship is commonly observed in hydrothermal systems (Curewitz & Karson 1997; Hancock^ al 1999). The bleaching and euhedral aragonite cements that are found in the footwall host rock close to the faults are evidence that fluid infiltrated the host rock as well as simply being channelled up the fault. Chan et al (2000) and Garden et al (2001) suggest that such bleaching is the result of movement of a reducing fluid through the system. They argue that this was meteoric water driven by topographic flow to salt depth where the fluid became saline and reducing and reacted with hydrocarbons. They argue for fault-controlled upward flow and movement of the fluid out into the Jurassic aquifer sequence.
Implications for CO2 sequestration projects The preliminary results of this study have implications for the design of CO2 sequestration projects. These projects require that CO2 be isolated from the atmosphere on a timescale of hundreds to thousands of years (Rochelle et al 1999). The Little Grand Wash and Salt Wash faults have been conducting fluids for a substantial period of time; however, more work remains to be done on the Little Grand Wash and Salt Wash faults to evaluate the fraction of the total CO2 reservoir that is leaking from the faults. A relatively low volume leakage of a large reservoir may well be deemed acceptable in a large sequestration project. The volume of carbon that is trapped in carbonate deposition at the surface must be estimated from future mapping of the total volume of travertines and carbonates around the wells and springs. The fact that many of the springs in the region are centred on abandoned oil wells suggests that care should be taken with well-bore integrity during the design and monitoring of CO2 sequestration projects. As suggested by the correlation of CO2 degassing and earthquake faults, the stress state in and around faults may significantly influence whether a fault is conductive or not (Barton et al. 1995). Zoback and Townend (2001) suggested that many faults in the crust are critically stressed and are very near the point of failure as predicted from a Mohr-Coulomb failure analysis. Present-day seismicity in the Colorado Plateau is dominated by extension along NW to NNW striking normal faults (Wong & Humphrey 1989) and the present state of stress in the region is consistent with NNE extension (Mueller et al 2000). The Little Grand Wash and Salt Wash faults are therefore optimally oriented for failure in the present stress field and are likely conduits for the upward migration of fluids. In addition to mapping the occurrence of faults in the vicinity of potential
56
Z.K.SHIPTONCTAL.
CO2 disposal sites, the orientation of faults with respect to the current stress state should be taken into account. Given the possibility of Quaternary to Recent fault activity, it is possible that seismic activity has played a part in the migration of the CO2 to the surface; earthquake activity has often been correlated with outgassing of CO2 reservoirs (Irwin & Barnes 1980; Sugisaki et al 1983; Chiodini et al 1995; Guerra & Lombardi 2001) and there is often a link between earthquakes and the activity of geothermal geysers. The Little Grand Wash and Salt Wash faults cut rocks that are good analogues for low net-to-gross reservoir rocks such as those found in the North Sea. The sealing capacity of faults depends on the type of structures that occur in the fault zone, how they are arranged, the contribution of each structure type to flow, and geochemical processes in the faults, which may add or remove sealing capacity. Due to the poor preservation potential of faults in outcrops of shalerich rocks, few field analogues have been studied in detail. Consequently, there is little data with which to make predictions regarding the behaviour of these faults. Such faults are often accounted for in hydrocarbon reservoirs using simple shale smear or gouge algorithms (e.g. Freeman et al 1998). These make substantial assumptions about the fault, specifically that the seal is formed by physical smearing or mixing of the low permeability clays in the host rock with no diagenesis of the fault rocks. Fault-rock alteration and diagenesis is clearly taking place in this locality and would almost certainly take place within any CO2 sequestration schemes that included faults. Much further work needs to be done on such systems before the implications for long-term storage of CO2 can be fully understood.
Conclusions The Little Grand Wash and Salt Wash faults, eastcentral Utah, affect the present-day CO2-charged groundwater flow regime over reservoir-scale distances. The CO2 flows upward from one or more deep sources that include the thermal decarbonation of carbonates deep in the basin and/or microbially mediated hydrocarbon generation. Most of the potential CO2 source rocks are below or within the Pennsylvanian Paradox Formation, approximately 1.5km below the surface. Observations of fault-controlled cementation, bleaching, and hydrocarbon accumulation elsewhere in the basin suggest that faults have been a conduit for reducing fluids and hydrocarbons. It is therefore likely that faults have been a conduit for deep-sourced CO2 through the Paradox salt and other low-permeability rocks in the basin. The low temperature of the water that effuses at the springs and geysers suggests that it is sourced
from the shallow Navajo/Wingate aquifer. The units that provide top seals to this aquifer are the Jurassic Summerville and Morrison Formations. The Cretaceous Mancos Shale and Cedar Mountain Formation would provide a juxtaposition seal in the hanging wall to the faults. The location of well-bore-related springs in the area show that gas-charged groundwater is ponded at the apex of a regional north-plunging anticline. Both natural and drill hole-related spring and travertine activity is concentrated along the fault zones, specifically in the northern footwalls. This suggests that the faults have a relatively low crossfault permeability and form a seal to migration of gas up the plunge of the anticline. The faults must therefore be the conduit for the gas to move through the sealing units, and have relatively high hydraulic conductivity in the vertical (up-dip) direction. The gas either seeps upward continually or is transported during fault slip events. The springs and travertines are concentrated at structural complexities and linkage zones along the faults. The long-lived nature of the fluid flow is demonstrated by ancient travertines and vein systems that are localized along the fault zone. The faults cut rocks that are analogous to the North Sea reservoirs in which it is proposed to dispose of significant quantities of CO2. Even if a relatively small proportion of the gas reservoir is able to escape along fault zones, then over periods of hundreds to thousands of years the efficiency of CO2 sequestration projects could be called into question. We have determined the basic regional structure of the Little Grand Wash and Salt Wash fault system, making these faults an ideal candidate for a more detailed study of the interaction between fluids and low-permeability faults in clay-rich rocks at the oil-field scale. Thanks to Bill Shea for introducing us to the field site, and to Torrey Copfer for re-discovering some of the studied springs. This work was funded in part by the Carbon Capture Project, a joint industry-government consortium. Steve Laubach, Tim Needham and Richard Worden provided thoughtful reviews of the manuscript. We thank Scott Imbus for insights into the geochemistry of petroleum systems, and to John Gale and Shelagh Baines for organizing the 2001 session on CO2 at BUG. We thank the people of Green River, Utah who provided information on historical eruptions of the Crystal Geyser.
References ALLIS, R., CHIDSEY, T, GWYNN, W., MORGAN, C, WHITE, S., ADAMS, M. & MOORE, J. 2001. Natural CO2 reservoirs on the Colorado Plateau and southern Rocky Mountains: candidates for CO2 sequestration. In: Proceedings of the National Energy Technology Laboratory, First National Conference on Carbon Sequestration. World Wide Web address:http://www.netl.doe.gov/ publications/proceedings/01/carbon_seq/6a2.pdf.
CO? LEAKAGE THROUGH LOW-PERMEABILITY FAULTS AMUNDSON, R. G. & KELLY, E. 1987. The chemistry and mineralogy of a CO2-rich travertine depositing spring in the California Coast Range. Geochimica et CosmochimicaActa, 51,2883-2890. BAER, J. L. & RIGBY, J. K. 1978. Geology of the Crystal Geyser and the environmental implications of its effluent, Grand County, Utah. Utah Geology, 5, 125-130. BARTON, C. A., ZOBACK, M. A. & Moos, D. 1995. Fluid flow along potentially active faults in crystalline rock. Geology, 23,683-686. BAUMGARTNER L. R & VALLEY J. W. 2001. Stable isotope transport and contact metamorphic fluid flow: In: VALLEY, J. W. & COLE, D. R. (eds) Stable Isotope Geochemistry, Reviews in Mineralogy and Geochemistry, 43,415^68. BODINE, M. W. & JONES, B. F 1986. The Salt Norm: A quantitative chemical-mineralogical characterization of natural waters. US Geological Survey Water Resources Investigations, 86-4086,130 p. CAMPBELL, J. A. & BAER, J. L. 1978. Little Grand Wash fault-Crystal Geyser area. In: FASSET, J. E. & THOMAIDIS, N. D. (eds) Oil and gas fields of the Four Corners area. Four Corners Geological Society, 666-669. CAPPA, J. A. & RICE, D. D. 1995. Carbon dioxide in Mississippian rocks of the Paradox Basin and adjacent areas, Colorado, Utah, New Mexico and Arizona. US Geological Survey Bulletin 2000-H, 21pp. CHAFETZ, H. S. & FOLK, R. L. 1984. Travertines: Depositional morphology and the bacterially constructed constituents. Journal of Sedimentary Petrology, 54, 289-316. CHAFETZ, H.S., UTECH N. M. & FITZMAURICE S. P. 19910. Differences in the 818O and 813C signatures of seasonal laminae comprising travertine stromatolites. Journal of Sedimentary Petrology, 61,1015-1028. CHAFETZ, H. S., RUSH, P. & UTECH, N. \99lb. Microenvironmental controls on mineralogy and habit of CaCO3 precipitates: an example from an active travertine system. Sedimentology, 38,107-126. CHAN, M.A., PARRY, W. T. & BOWMAN, J. R. 2000. Diagenetic hematite and manganese oxides and faultrelated fluid flow in Jurassic sandstones, southeastern Utah. American Association of Petroleum Geologists Bulletin, 84, 1281-1310. CHIODINI, G., FRONDINI, F. & PONZIANI, F. 1995. Deep structures and carbon dioxide degassing in Central Italy. Geothermics, 24, 81-94. CLAYPOOL, G. E., LOVE, A. H. & MAUGHAN, E. K. 1978. Organic geochemistry, incipient metamorphism, and oil generation in black shale members of Phosphoria Formation, Western Interior United States. American Association of Petroleum Geologists Bulletin, 62, 98-120. COLEMAN, D. D., Liu, C.-L. & RILEY, K. M. 1988. Microbial methane in the shallow Paleozoic sediments and glacial deposits of Illinois, USA. Chemical Geology, 71,23^0. CUREWITZ D. & KARSON J. A. 1997. Structural settings of hydrothermal outflow; fracture permeability maintained by fault propagation and interaction. Journal of Volcanology and Geothermal Research, 79, 149-168. DOELLING, H. 1994. Tufa deposits in western Grand
57
County. Survey Notes - Utah Geological Survey, 26, 8-10, 13. DOELLING, H. 2001. The geologic map of the Moab and eastern San Rafael Desert 30X60' quadrangles. Utah Geological Survey Map M-180. FREEMAN, B., YIELDING, G., NEEDHAM, D. T & BADLEY, M. E. 1998. Fault seal prediction: the gouge ratio method. In: COWARD, M. P. DALTABAN, T. S., & JOHNSON, H. (eds) Structural Geology in Reservoir Characterization. Geological Society, London, Special Publications, 127,19-26. FRIEDMAN, I. 1970. Some investigations of the deposition of travertine from Hot Springs -1. The isotopic chemistry of a travertine-depositing spring. Geochimica et CosmochimicaActa,34,1303-1315. FOLK, R. L., CHAFETZ, H. S. & TIEZZI, P. A. 1985. Bizarre forms of depositional and diagentic calcite in hotspring travertines, Central Italy. In: SCHNEIDERMANN, N. & HARRIS, P. M. (eds) Carbonate Cements. Society for Sedimentary Geology Special Publication, 36, 349-379. GARDEN, I. R., GUSCOTT, S. C., BURLEY, S. D., FOXFORD, K. A., WALSH, J. J. & MARSHALL, J. 2001. An exhumed palaeo-hydrocarbon migration fairway in a faulted carrier system, Entrada Sandstone of SE Utah, USA. Geofluids, 1,195-214. GUERRA, M. & LOMBARDI, S. 2001. Soil-gas method for tracing neotectonic faults in clay basins: the Pisticci field (Southern Italy). Tectonophysics, 339, 511-522. Guo, L., ANDREWS, J., RIDING, R., DENNIS, P. & DRESSER, Q. 1996. Possible microbial effects on stable carbon isotopes in hot-spring travertines. Journal of Sedimentary Research, 66,468^73. HANCOCK, P. L., CHALMERS, R. M. L., ALTUNEL, E. & CAKIR, Z. 1999. Travitonics; using travertines in active fault studies. Journal of Structural Geology, 21, 903-916. HANSLEY, P. L. 1995. Diagenetic and burial history of the White Rim Sandstone in the Tar Sand Triangle, Paradox Basin, southeastern Utah. US Geological Survey Bulletin, 2000-1,41pp. HEATH, J. E. 2004. Hydrogeochemical characterization of CO2 charged fault zones in east-central Utah. M.S. Thesis, Utah State University, Logan. 166pp. HEATH J. E., LACHMAR T. E., SHIPTON Z. K., NELSON, S. & EVANS J. P. 2002. Hydrogeochemical analysis of leaking CO2-charged fault zones: the Little Grand Wash and Salt Wash fault zones, Emery and Grand counties, Utah. Geological Society of America Abstracts with Programs, 34 (6), p. 392. HECKER, S. 1993. Quaternary tectonics of Utah with emphasis on earthquake-hazard characterization. Utah Geological Survey Bulletin, 127, p. 157. HINTZE, J. F. 1993. Geologica History of Utah. 2nd edn. Brigham Young University Geology Studies, Special Publication 7.202p. HOOD, J. W & PATTERSON, D. J. 1984. Bedrock aquifers in the northern San Rafael Swell area, Utah, with special emphasis on the Navajo Sandstone. State of Utah Department of Natural Resources Technical Publication no. 78. IAEA (2001). GNIP Maps and Animations, International Atomic Energy Agency, Vienna. World Wide Web address: http://isohis.iaea.org.
58
Z.K.SHIPTON£TAL.
IRWIN, W. P. & BARNES, I. 1980. Tectonic relations of carbon dioxide discharges and earthquakes. Journal of Geophysical Research, 85, 3115-3121. JENDEN, P. D., KAPLAN, I. R., POREDA, R. J. & CRAIG, H. 1988. Origin of nitrogen-rich natural gases in the California Great Valley - evidence from helium, carbon, and nitrogen isotope ratios. Geochimica et Cosmochimica Acta, 52, 851-861. KIRSCHNER, D., ENCARNACION, J., & AGOSTA, F. 2000. Incorporating stable isotope geochemistry in undergraduate laboratory courses. Journal of Geoscience Education, 48, 209-215. MAYO, A. L., SHRUM, D. B. & CHIDSEY, T. C, JR. 1991. Factors contributing to exsolving carbon dioxide in ground-water systems in the Colorado Plateau, Utah. In: CHIDSEY, T. C, JR (ed.) Geology of east-central Utah. Utah Geological Association Publication, 19, 335-342. MCKNIGHT, E.T. 1940. Geology of area between Green and Colorado rivers, Grand and San Juan Counties, Utah. US Geological Survey Bulletin, 908,147 pp. MUELLER, B., REINECKER, J., HEIDBACH, O. & FUCHS, K. 2000. The 2000 release of the World Stress Map (available online at www.world-stress-map.org). Nuccio, V. F. & CONDON, S. M. 1996. Burial and thermal history of the Paradox Basin, Utah and Colorado, and the petroleum potential of the Middle Pennsylvanian Paradox Formation. US Geological Survey Bulletin, 2000-O,41pp. PETERSON, P. R. 1973. Salt Wash field. Utah Geological Survey oil and gas field studies, no 4. 3pp. PENTECOST, A. 1995. Geochemistry of carbon dioxide in six travertine-depositing waters of Italy. Journal of Hydrology, 167, 263-278. PEVEAR, D. R., VROLIJK, P. J. & LONGSTAFFE, F. J. 1997. Timing of Moab fault displacement and fluid movement integrated with burial history using radiogenic and stable isotopes. In: HENDRY, J., CAREY, P., PARNELL, J., RUFFEL, A. & WORDEN, R. (eds) Geofluids II1997 Extended abstract volume, 42-45. POWELL, J. W. 1895. The Canyons of the Colorado (now
published as The exploration of the Colorado River and its canyons. 1997. Penguin Books 416 pages.) ROCHELLE, C. A., PEARCE, J. M. & HOLLOWAY, S. 1999. The underground sequestration of carbon dioxide: containment by chemical reactions in the deep geosphere. In: METCALFE, R. & ROCHELLE, C. A. (eds) Chemical containment of waste in the lithosphere, Geological Society, London, Special Publications, 157,117-129. SANFORD, R. F. 1995. Groundwater flow and migration of hydrocarbons to the Lower Permian White Rim Sandstone, Tar Sand Triangle, Southeastern Utah. US Geological Survey Bulletin, 2000-J, 24pp. SCHOELL, M. 1983. Genetic characterization of natural gases. American Association of Petroleum Geologists Bulletin, 67,2225-2238. SELLEY, R. C. 1998. Elements of Petroleum Geology, 2nd edn, Academic Press. 470pp. SUGISAKI, R. IDO, M., TAKEDA, H. et al. 1983. Origin of hydrogen and carbon dioxide in fault gases and its relation to fault activity. Journal of Geology, 91, 239-258. TRUESDELL, A. H. AND JONES, B. F. 1974. WATEQ, a computer program for calculating chemical equilibria of natural waters. Journal of Research of the US Geological Survey, 2(2), 233-248. WILLIAMS, P. L. 1964. Geology, structure, and uranium deposits of the Moab Quadrangle. Colorado and Utah US Geological Survey Map I- 360. WILLIAMS, P. L. & HACKMAN, R. J. 1971. Geology, structure, and uranium deposits of the Salina Quadrangle, Utah. US Geological Survey Map 1-591. WONG, I. G. & HUMPHREY, J. R. 1989. Contemporary seismicity, faulting, and the state of stress in the Colorado Plateau. Geological Society of America Bulletin, 101, 1127-1146 ZOBACK, M. D. & TOWNEND, J. 2001. Implications of hydrostatic pore pressure and high crustal strength for the deformation of intraplate lithosphere. Tectonophysics, 336,19-30.
The long-term fate of CO2 in the subsurface: natural analogues for CO2 storage SHELAGH J.BAINES1 & RICHARD H. WORDEN2 1
BP Exploration and Production Company, Chertsey Road, Sunbury-on-Thames, TW16 7LN UK (e-mail:
[email protected]) 2 Department of Earth and Ocean Sciences, University of Liverpool, 4 Brownlow Street, Liverpool L69 3GP, UK Abstract: CO2 is a common gas in geological systems so that planned storage of CO2 in the subsurface may do no more than mimic nature. Natural CO2 has a wide number of sources that can be at least partly identified by carbon stable isotope geochemistry. Three pairs of case studies with different reservoir characteristics and CO2 contents have been examined to assess the natural impact of adding CO2 to geological systems. Carbonate minerals partially dissolve when CO2 is added simply because the CO2 dissolves in water and forms an acidic solution. Therefore, carbonate minerals in the subsurface are not capable of sequestering secondary CO2. The addition of CO2 to a pure quartz sandstone (or a sandstone in which the supply of reactive aluminosilicate minerals has been exhausted by excess natural CO2 addition) will have no consequences: the CO2 will simply saturate the water and then build up as a separate gas phase. The addition of CO2 to carbonate cemented sandstone without reactive aluminosilicate minerals will induce a degree of carbonate mineral dissolution but no solid phase sequestration of the added CO2. When CO2 is naturally added to sandstones it will induce combined aluminosilicate dissolution and carbonate cementation if the aluminosilicate minerals contain calcium or magnesium (or possibly iron). Examination of a CO2filled porous sandstone with abundant reactive aluminosilicate minerals that received a huge CO2 charge about 8000 to 100000 years ago reveals minimal evidence of solid phase sequestration of the added CO2. This indicates that either dissolution of reactive aluminosilicates or precipitation of carbonate minerals is relatively slow. It is very likely that the slow dissolution of reactive aluminosilicates is the rate-limiting step. Solid phase sequestration of CO2 occurs only when reactive aluminosilicates are present in a rock and when the system has had many tens to hundreds of thousands of years to equilibrate. The two critical aspects of the behaviour of CO2 when injected into the subsurface are (1) that the rock must contain reactive Ca and Mg aluminosilicates and (2) that reaction to produce carbonate minerals is extremely slow on a human timescale. The reactive minerals include anorthite, zeolite, smectite and other Fe- and Mg-clay minerals. Such minerals are absent from clean sandstones and limestones but are present in 'dirty' sandstones (lithic arenites which are mineralogically immature) and some mudstones. The analysis of geological analogues shows that injection of CO2 into carbonate-bearing rocks that do not contain reactive minerals will induce dissolution of the carbonate, whether it is a matrix cement, rock fragment, fault seal or part of a top-sealing mudstone.
Subsurface storage of CO2 is considered by both industry and national governments (e.g. UK DTI 2002; US DOE 2004) to be a key mechanism for reducing the emission of greenhouse gases to the atmosphere. CO2 captured either directly from a petroleum stream or from the exhaust products of electricity generating stations and industrial plants can be introduced into suitable subsurface reservoirs for long term, safe, storage. Simple logic suggests that since non-aqueous fluids and gases, such as petroleum accumulations, are naturally trapped in the subsurface for geological periods of time, so captured CO2 can also be injected and stored for the 1000 to 10000 year time period considered necessary for stabilization and remediation of the earth's atmosphere (Pacala pers. comm.). There are several options under investigation including injection into
deep, saline aquifers (e.g. Sleipner field, Baklid^^/. 1996), injection into former or current oil and gas accumulations and injection into coal beds. CO2 is already injected into oil fields during enhanced oil recovery thus reducing the issue to development of existing technology rather than invention of radically new technology. Two broad scenarios are envisaged here upon injection of the CO2 into the subsurface. The first is storage, or simple retention, similar to present day petroleum accumulations in which the CO2 is present as either a gas, aqueous or dense (supercritical) phase. This scenario retains the ability for the CO2 to migrate if the integrity of the site is breached over time, or if the geological framework is radically altered as has occurred in volcanically active areas, e.g. LakeNyos (Cameroon) and Mammoth Mountain
From: BAINES, S. J. & WORDEN, R. H. (eds) 2004. Geological Storage of Carbon Dioxide. Geological Society, London, Special Publications, 233,59-85.0305-8719/047$ 15.00 © The Geological Society of London 2004.
Fig. 1. Global distribution of high (>20%) CO2-content basins. The sites discussed in this study are indicated.
NATURAL ANALOGUES
(USA). The second is CO2 sequestration, in which the CO2 is locked into newly formed minerals that are stable under reservoir conditions. These approaches have one problem; CO2 is a reactive gas that has significantly more influence upon the host rocks and their formation waters than petroleum fluids. The long-term consequence of increasing the gross volume of CO2 in the subsurface is unknown, but is a key screening criterion in the evaluation of potential subsurface storage sites. Resorting to a modelling approach suffers from the perennial problem that the rates of the various component geochemical and physiochemical processes under geological conditions are not well known. Previous studies aimed at understanding the impact of CO2 on the subsurface environment have commonly been designed around forward-modelling, using geochemical software combined with short term experimental studies to replicate the expected conditions during CO2 injection and retention (e.g. Czernichowski-Lauriol et al. 1996; Gunter et al. 1997, 2004; Gupta & Sass 1999; Johnson et al. 2004; Rochelle et al. 2004). These use either measurements of present day physical and chemical conditions, or a range of values as primary inputs for forward modelling the final state of the CO2. The short-term experiments, necessarily performed at elevated temperatures to produce results on a measurable timescale, are expensive and thus limited, deal with only single variables instead of the huge complexity of nature and do not mimic nature. Other studies have used data from enhanced oil recovery (EOR) CO2 floods to study the interaction of CO2 with both the reservoir rock and the formation fluids (e.g. Wolcott et al. 1989; Smith 1998; Worden & Smith 2004). Both of these approaches, however, rely on generally incomplete datasets over relatively short timescales (months to one or two decades); at least two or more orders of magnitude less than those required for long-term storage. Many geological systems have abundant evidence that CO2 has been added in large volumes. Indeed there are numerous geological sites that today have naturally elevated CO2 concentrations (Fig. 1). The geological timescale is typically variable but can stretch from a few tens of thousands of years to many millions of years (and beyond). Thus the rock record, in association with current fluid compositional data, affords the opportunity for assessment of the long-term fate of CO2 in the subsurface. Rather than anthropogenic sources, the CO2 in question has natural sources.
CO2 in geology CO2 is an important gas in many geological formations and is responsible for, or involved in, many
61
natural geochemical processes. CO2 affects porosity since it can influence both mineral dissolution and growth in a given formation. The occurrence of high-pressure CO2 may lead to weakened fault rocks and may trigger faulting and thus earthquakes. Overpressured fluids, including CO2, in the subsurface can lead to seal failure and the escape of fluids to the surface. CO2 seepage via faults at the seabed from submarine fault zones can lead to deep-water reefs and bioherms. CO2 is one of the most common volatiles found in volcanic gases and is important in determining eruption cycles (e.g. Krauskopf 1979). There are a large number of possible sources of CO2 in the subsurface (e.g. Clayton 1991; Emery & Robinson 1993; Irwin et al 1977 Wycherley et al 1997; Table 1 & Fig. 2). The isotopic ratio of 13C to 12 C is commonly used to help type CO2. Both isotopes are non-radioactive and are not influenced directly by time. The ratio of 13C to 12C is conventionally compared to a standard (Peedee Formation belemnite) and the result is quoted as a per mil quantity (813C %0 PDB). It is commonly difficult to identify an exact source of any sample of CO2 based on either or both of geological location and carbon stable isotopes since there are ambiguities in both. Broadly, though, there are four main groups of CO2 sources (Fig. 2). (1)
(2)
Deep sources of CO2 are common in tectonically active regions of the earth's crust. Thermal metamorphism and juvenile sources of CO2 are associated with igneous processes (both intrusive and extrusive) prevalent at plate collision zones (e.g. the Andes) and plate rifting zones (e.g. the East African rift). High temperature metamorphism leads to devolatilization and loss of CO2 from minerals (e.g. Yardley 1989). Any carbonate-mineral bearing, prograde metamorphic, rock will tend to lose CO2 to the fluid phase that then will be free to ascend under normal buoyancy to the surface or a transient trapping structure. Deepseated faults have routinely been implicated in the release of very deep sources of CO2. Regional metamorphic terrains have their fluid phase dominated by CO2 and N2 so that release of fluid via deep faults (during seismicity) will tend to release CO2. Atmospherically derived CO2 in the subsurface has been suggested but only in regions where there are active groundwater systems with a constant flux of meteoric water. Such conditions are found in artesian basins. CO2 solubility in water is a function of many controls including temperature, pressure and salinity. The rate of passage of water through aquifers is sufficiently slow to permit equilibration of CO2 with the surrounding rock
62
S. J. BAINES & R. H. WORDEN
Table 1. Geological sources ofCO2 and their stable isotopic compositions. Data from Clayton (1991); Emery & Robinson (1993); Irwin et al. (1977); Wycherley et al (1997). CO9 source
Stable carbon isotope range (S 13 C%cPDB)
Main occurrence
Deep crust
-6%o
Near very deep faults
Juvenile (volcanic)
-6%c
Associated with plate spreading and subduction
Atmospheric CO2 exsolved from meteoric waters
Variable over geological time, approx. -6%c
In artesian sandstones and limestones, typically at the outlet of the aquifer
Fermentation of organics
-20%o
Very shallow buried organic rich sediments
Bacterial oxidation of organics
-23%o
Very shallow buried organic rich sediments Petroleum source rocks and coal beds
Kerogen and coal decarboxylation
-25%o
Burial diagenesis of impure limestones
Depends on limestone age, Most sedimentary basins, many clastic sediments approx -2 to +3%o
Burial diagenesis of carbonate cemented sandstones
Depends on 813C of carbonate cement
Most sedimentary basins, many clastic sediments
Thermal metamorphism of limestones
Depends on limestone age, approx -2 to +3%o
Contact metamorphism due to intrusion
Sulphate reduction of petroleum (petroleum oxidation)
-20to-40% 0
Bacterial = near surface, Thermochemical = > 120°C
Fig. 2. Schematic illustration of the natural origins of CO2 in geological systems (Clayton 1991; Emery & Robinson 1993; Irwin et al 1977; Wycherley et al. 1997)
(3)
matrix. Consequently this source of subsurface CO2 is only likely to be volumetrically significant in situations where water in relatively deep aquifers moves upwards rapidly before equilibrium can be achieved. The consequence is mineralized springs and geysers that are well known for their dissolved CO2 contents (e.g. Shiptonetal.2004). Organic sources of CO2 take many forms but all have a root cause: some sedimentary rocks, usually mudstones, contain organic detritus
when they are buried. Much of the organic matter that is initially buried with sediments undergoes decomposition within a few hundred metres burial (e.g. Irwin et al. 1977). The decomposition processes include fermentation, bacterial oxidation, and bacterial sulphate reduction (plus minor nitrate reduction, ferric iron reduction, and manganese reduction). CO2 from these sources can dominate gases in the local subsurface and typically has low 813C values, reflecting the isotopic composition of the organic matter. Methanogenesis is a bacterial process that synthesizes hydrocarbon gas from CO2 resulting in isotopically enriched CO2 although inevitably reducing the quantity of CO2. Organic matter that survives initial burial is known as kerogen. Kerogen undergoes thermal breakdown processes, known as maturation, as a function of time and temperature. Kerogen decarboxylation (loss of C-O functional groups) occurs relatively early in the maturation process and produces CO2 with fairly low 813C values (e.g. Tissot & Welte 1984). Petroleum, also produced by kerogen thermal breakdown, can undergo oxidation by reaction with sulphate or ferric iron. Sulphate and ferric iron reduction, whether bacterial or thermochemical, produces abundant CO2 although much is locally reprecipitated as carbonate minerals. Within inverted (uplifted) oil fields, ingress of meteoric water may introduce sulphate-reducing bacteria and possibly dis-
63
NATURAL ANALOGUES Table 2. End member case study locations and accumulation type. End member type
Carbonate
Quartz arenite
Arkose
High CO2 case (% C02)
Blue Whale, Da Nang Basin (>70%)
Miller field (-28%)
Bravo Dome (100%)
Accumulation type (high CO2 case)
Offshore gas
Offshore oil and gas
Onshore CO2
Low CO2 case (% CO2)
Dolphin, Da Nang Basin (<1%)
Magnus field (2%)
Vert le Grand, Paris Basin (2.5-2.8%)
Accumulation type (low CO2 case)
Offshore gas
Offshore oil (with associated gas)
Onshore oil (with associated gas)
(4)
solved oxygen that may lead to CO2 generation due to petroleum degradation (e.g. Ehrenberg &Jacobson2001). Burial diagenesis of rocks that contain carbonate and clay minerals can lead to both CO2 generation and natural sequestration (e.g. Smith & Ehrenberg 1989; Hutcheon et al 1993; Coudrain-Ribstein et al 1998; Hutcheon & Desrocher 2003). Such rocks include both impure limestones and carbonate cemented clay-bearing sandstones; the reactions are the same. Carbonate and aluminosilicate minerals tend to react to produce CO2 and high temperature sheet silicates upon heating to temperatures of about 120°C. These reactions continue through high temperature diagenesis and into low-grade metamorphism and produce CO2 with 813C values characteristic of the indigenous carbonate minerals. Thus impure limestones would produce CO2 with 813C values close to 0%o whereas sandstones produce CO, with 813C values typically somewhat lower than 0%o (reflecting the kerogen decarboxylation source of much carbonate cement in sandstones).
The most volumetrically abundant sources of CO2 are seemingly the deep crustal sources (with 813C values of about — 6%o) and kerogen maturation sources (with 813C values of about -20%o; Wycherley et al. 1997). However, most accumulations of CO2 are likely to be multi-sourced and it is naive to hunt for one source in any one CO2 accumulation (past or present).
Selection of case studies Analysis of the rock record and fluids contained within porous rocks permits the recognition and quantification of the fate of CO2 in rocks on a geological timescale. The fate of CO2 when added to a rock-fluid system depends on a large number of
factors such as the pre-CO2 mineralogy and rock fabric, the geochemical composition of the formation water, shear strength, fluid pressure, effective stress and temperature of the CO2 reservoir and the amount of CO2 and the rate at which it was added into the system. In order to address the question of what happens to CO2 in the geological context, three pairs of accumulations have been chosen, representing three compositional end-members (Table 2), by employing various selection criteria. The following criteria were used: (1)
(2) (3)
Pairs of sites with fundamentally similar geology, with regard to tectonic regime and sediment composition, but with contrasting CO2 concentrations in the gas phase were chosen to reveal critical controls on the fate of CO2 in the long term. Examples were selected that were reasonably well understood in terms of geology and sources of CO2. Only sites with a broad base of petrological and geochemical data were selected.
The approach adopted here included extensive data collection followed by synthesis and identification of the post-CO2 mineral assemblage and rock geochemistry and then extrapolation of the pre-CO2 mineral assemblage and rock geochemistry. The diagenetic consequences of CO2 addition to the system and tentative reaction schemes were thus inferred. These reactions were then tested using site-specific geochemical modelling of the fluid-rock reactions. Five of the cases examined are petroleum accumulations that have undergone CO2 input (Table 2). These have the advantage that there are abundant data on rock mineralogy and geochemistry, water and gas geochemistry and isotopes and data on the thermal evolution of the rocks. The other case study is a commercial CO2 accumulation, exploited to assist improved oil recovery and also has a reasonable database. Data from so-called 'deep saline
64
S. J. BAINES & R. H. WORDEN
Fig. 3. Carbonate end-member: map showing the tectonic setting and location of study wells on the Tri Ton Horst, offshore Vietnam.
aquifers' have not been included here although the water zone of oil fields is broadly indistinguishable in geochemical terms from such aquifers. One pair of cases includes carbonates (Table 2) thus testing the value of such reservoirs for future CO2 storage. Note that the bulk of the world's recoverable petroleum sits within carbonate reservoirs (e.g. Middle East, SE Asia) so that these could be important future CO2 storage sites. A second pair of cases includes relatively clean (Upper Jurassic) sandstones representative of many marine sands around the world including the deltaic, submarine fan and turbidite deposits of the North Sea, Gulf of Mexico and West Africa basins. The last pair of cases includes sandstones that contain abundant clay and feldspar minerals as well as quartz. They are broadly typical of fluvial sand accumulations and immature sands deposited in technically active regions such as back-arc basins, inter-montane thrusts and subduction zones. One of this last pair of examples received the CO2 charge recently whereas the other received it many millions of years ago.
Carbonate reservoirs SE Asian petroleum systems are well known for having a high risk of CO2 (e.g. Eraser et al 1997). Variable concentrations of CO2 have been discovered in Miocene platform carbonates in the Da Nang Basin, a strike-slip basin, offshore Vietnam. The carbonates sit on the Tri Ton horst that forms the eastern flank to the Quang Ngai rift (Fig. 3). Rifting and development of this structure commenced between the Upper Eocene and Oligocene with
major extensional fault movement leading to the formation of the Tri Ton Horst that then became the site of extensive carbonate platform development during the Lower to Middle Miocene. The platform is overlain by pelagic mudstones, deposited during thermal subsidence and eustatic sea level rise. Two accumulations in adjacent closures in the same gross structure have quite different CO2 concentrations (Table 3). The Blue Whale accumulation contains more than 70% CO2 by volume whereas the adjacent Dolphin accumulation contains less than 1% CO2 by volume. Both accumulations are developed in permeable Miocene platform carbonates. These reservoir formations are predominantly bioclastic, foraminiferal, and rhodolithic limestones containing negligible clastic material, which have been charged with methane-dominated petroleum gas. Isotopic data (o13Cc02 of -2.2 %0; Table 3) suggest that the free CO2 in the high-CO2 accumulation (Blue Whale) is predominantly a mixture of carbonate rock-sourced and mantle-derived CO2. Unequivocal isotopic indicators of organically-derived CO2 have not been identified suggesting a significantly lower contribution compared to the inorganic CO2. Helium isotope data also support a mixed origin but suggest significant mantle contribution (3He/4He ratio range: 2.53-2.59; Table 3). In contrast, carbon isotopic data from gas samples from the low-CO2 Dolphin accumulation are much lower (813CCO of —22 to —31 %o), suggesting an organic origin for the CO2, probably from kerogen decarboxylation associated with gas charge (Tables 1 & 3). The origin of the CO2 in Blue Whale is not fully understood; however, seismic and magnetic data suggest abundant shallow Neogene intrusive and volcanic activity within the vicinity of the accumulation. Significantly, neither intrusive nor volcanic geological features have been found near to the low CO2 Dolphin accumulation. Seismic data suggest the presence of a geological inversion structure, possibly related to the development of the Tri Ton horst, between the high CO2 (Blue Whale) and low CO2 (Dolphin) accumulations may have prevented migration of CO2 between the reservoirs. Emplacement of the inorganic CO2 into Blue Whale is considered to have occurred at least 5 Ma during the local intrusive activity with CO2 ultimately derived from the thermal decomposition of carbonates deeper in the section. Figure 4 illustrates the key burial diagenetic events identified in both accumulations. Blue Whale and Dolphin both show well-developed early marine carbonate cements (isopachous grain-fringing cement) which partially infill the primary porosity (Fig. 5). Karstification during early meteoric ingress led to the development of mouldic and vuggy porosity through the upper parts of the carbonate reservoir intervals although there was localized meteoric calcite cement
65
NATURAL ANALOGUES Table 3. Fluid and gas composition data for the carbonate end-member case studies: Dolphin and Blue Whale. RFT water analysis (mg/1) High CO2 (Blue Whale) Low CO2 (Dolphin) RFT gas analysis (mol %) High CO2 (Blue Whale) Low CO2 (Dolphin)
Na
K
Ca
Mg
Ba
Sr
Cl
SO4
HCO3
8402
6314
514
220
0.3
10
12700
480
930
14116
7950
299
125
0
4
27300
2200
-
C1/(C1-C5)
CH4
13
8 C (%o)
C02
8 C (%0)
R/Racorr*
0.96
19.6
-31.5
75.1
-2.2
2.53
0.93
80
-40.16
0.82
-31.3
0.97
13
* He/Ne ratio: corrected value assumes all Ne is atmosphere derived.
formation filling pores. Thus the early diagenetic history of the two contrasting structures was essentially identical. In low-CO2 Dolphin, the early diagenetic vugs are filled by dull cathodoluminescent (CL) calcite cement (o13Ccalcite of +1.2 to +1.7%0). Dull luminescence is characteristic of Fe-bearing calcite and is thus typically assumed to be the result of burial diagenesis. High-CO2 Blue Whale also contains porefilling, blocky ferroan calcite cement (813Ccalcite of —2.1 to — 0.6%c) which appears to have precipitated after the initial stages of compaction. Although postdating the onset of compaction, both isotopic data and fluid inclusion data from the ferroan calcite suggest precipitation from fresh to brackish (0-2 wt % NaCl equivalent) pore fluids. Late stage, burialrelated compaction is found in both accumulations, but late diagenetic cement textures are only developed in low-CO2 Dolphin. Late, post-vug filling, and brightly luminescent calcite ($13Ccalcite of -13.9 to —9.5%c), interpreted to have precipitated from a more evolved fluid (S18Ocalcite of -1.9 to -2.7%c PDB) during burial, has filled much of the remaining porosity in Dolphin (Fig. 5). In contrast, a late-stage dissolution event can be recognized in high CO2 Blue Whale that has developed vuggy and mouldic porosity that post-dates all early cements (Fig. 5) and opened up sutured stylolites. This late dissolution event has not been identified in low-CO2 Dolphin. Petrophysically- and petrographicallydetermined porosity in the high-CO2 Blue Whale accumulation is significantly higher than low-CO2 Dolphin. The Dolphin accumulation carbonates also show evidence for minor dolomitization (Sl3Ccalcite of -1.8 to +4.0%c). High-CO2 Blue Whale contains no late diagenetic dolomite cement. Top-seal pelagic carbonates and mudstones, stratigraphically and structurally above high-CO2 Blue Whale, locally show signs of extensive diagenetic dissolution. Pelagic foraminifera in the overlying mudstones are commonly ragged (due to partial
dissolution) whereas early marine calcite cements are commonly etched and partially dissolved. These features are not well developed in low-CO2 Dolphin. To summarize, Blue Whale with abundant CO2 has extensive signs of late diagenetic dissolution and minimal late diagenetic cement growth after the onset of initial burial compaction. Dolphin with low CO2 has only minor signs of late stage dissolution but has well developed late diagenetic calcite and dolomite cements with 813C isotopic signatures consistent with a mixed kerogen decarboxylation and marine carbonate origin. The 513C and 3He/4He values confirm that the high-CO2 content in Blue Whale is predominantly derived from crustal (mantle plus alteration of carbonate) sources whereas the 813C values of the late carbonate cements confirm that the low-CO2 Dolphin structure is unlikely to have ever received a large volume of crustal CO2; i.e. the late cements in Dolphin have not 'mopped-up' a significant volume of crustal CO2. It is possible to surmise that the high partial pressure of CO2 in Blue Whale has inhibited carbonate mineral growth and has induced local dissolution and enhancement of porosity. The addition of significant CO2 into carbonate rocks has thus induced mineral dissolution and inhibited mineral cementation. Sequestration of the geological CO2 has not occurred in these limestones. Moreover, the influx of the acid gas seems to have caused mineral dissolution.
Quartzose sandstone reservoirs The quartzose to sublithic and subarkosic endmember case study is represented by the permeable, Upper Jurassic submarine fans of the Miller and Magnus fields in the UK North Sea (Table 2; Fig. 6) from the South and North Viking Grabens (respectively). Both reservoirs received petroleum derived from Kimmeridge Clay Formation shale during the early Tertiary with petroleum generation being
66
S. J. BAINES & R. H. WORDEN
Fig. 4. Paragenetic sequence and key diagenetic events in the carbonate Blue Whale and Dolphin reservoirs.
essentially complete by the Miocene. Organicallyderived CO2 is present in both fields in low volumes, but the Miller field also received a relatively large volume of inorganic CO2 at some point during its more recent burial history. The Miller field (Fig. 6) sits within UK North Sea licence blocks 16/7 and 16/8b. The reservoir units are part of a late Jurassic (J60 to J70) submarine fan system, one of several deposited along the western margin of the South Viking Graben. The Miller area of the North Sea is characterized by petroleum reservoirs containing high volumes of CO2 (Fig. 7), e.g. South and Central Brae (<35% CO2 in oil-associated gas), Miller (c. 28% CO2), Sleipner (c. 17% CO2), and the Tiffany and Thelma fields (c. 15-20% CO2; James 1990). CO2 is present within all compartments of the Miller field. The black oil phase contains 15-25 mol% CO2 whereas up to 60-70 mol% CO2 has been measured from separator gas during water testing. Carbon isotope data (613CCO of -8.2%0; Table 4) for CO2 from separator samples taken at the platform give values similar to those from other high CO2 fields in the area (513CCO of -3.0to -8.0%o; James 1990). TheCO 2 ispossibfy a mixture of organically-derived CO2 from thermal maturation (decarboxylation) of organic matter, with a component of inorganic CO2 derived from thermal alteration of deeper marine carbonates, possibly Zechstein dolomites (cf. Macaulay et al 1992) that are juxtaposed against the Jurassic reservoir sandstone along the graben. The presence and timing of emplacement of such high CO2 concentrations in this area of the North Sea are not fully understood. Organically-derived CO2 most likely entered the system immediately prior to, or during, early petro-
leum charging, although the timing of the input of the inorganic component is unclear. It is possible that CO2 migrated from depth along conduits associated with the deep-seated Brae Fault system or other deep structures that are unique to this part of the basin. The variable CO2 content of each reservoir compartment (Fig. 6) suggests that the accumulation of both organic and inorganic CO2 was controlled either by barriers to vertical flow in the field which are known to impact production (C. Smalley pers. comm.), or was due to sample type, with CO2 partitioning between the oil, gas and water phases. The Miller sandstone represents the more distal part of the conglomeratic and coarse-grained Brae field fan and currently sits at 3970-4090 m subsea. The main Miller reservoir can be subdivided into two stacked units (Garland 1993; Fig. 6). The main producing reservoir, Unit 2, consists of thick variably well-connected high flow density turbidite packages. The sandstones are generally mediumgrained, moderately sorted quartzose to sub-lithic arenites. This unit is underlain by a much less sandrich package (Unit 3) of chaotic to remobilized turbidite sediments that are slightly finer-grained with greater volumes of detrital clay than Unit 2 sands. The Miller field is also highly compartmentalized to both petroleum and CO2 by both regional and local mudstones that have controlled the petroleum filling history of the reservoir units. The detrital and authigenic mineral compositions of Unit 2 and 3 sandstones from three wells are given in Table 4. Miller sandstone is typically dominated by detrital quartz with very minor potassium feldspar and quartzose lithic fragments (Gluyas et al. 2000). The clay content is minor (<5%) in the clean, main reservoir
NATURAL ANALOGUES
67
Fig. 5. Thin section micrographs illustrating the key diagenetic features attributed to the impact of CO2 in the reservoirs, (a) The high CO2 Blue Whale accumulation. The bioclastic packstone reservoir shows well developed intergranular porosity, partially occluded by early diagenetic grain-fringing cements. No late burial cements are observed and large secondary pores (V) are present, (b) The low CO2 Dolphin accumulation. In contrast, Dolphin packstones contain abundant late sparry calcite cement which occlude the intergranular porosity (P) and early diagenetic dissolution vugs, (c) Miller field. Brae Formation sandstones with minor carbonate cement (C) and extensive secondary porosity after K-feldspar (S) and carbonate dissolution, (d) Magnus field, Magnus Sandstone Member. Ferroan dolomite rhombs sit within primary and secondary macropores and are post-dated by a late-stage non-ferroan calcite cement, (e) Bravo Dome Tubb Formation sandstone. The dolomitized volcanic grains (V) show signs of initial dissolution whilst fibrous bundles of zeolite (laumontite) infill primary and secondary macropores. Microcline (M) crystals show evidence for overgrowths of adularia. Quartz overgrowth development (arrowed) is minor, (f) Vert le Grand field, Chaunoy Formation. Scanning electron micrograph of a quartz and dolomite cemented sample with dolomite sitting on top of reactive clay grains within partially dissolved detrital feldspar grains.
68
S. J. BAINES & R. H. WORDEN Fig. 6. Map of the North Sea Basins showing the locations of Miller and Magnus fields. The insets show cross-sections of each. The Miller field (b) is compartmentalized with respect to CO2 content of the oil and gas with high CO2 content in the lower Unit 3 sands (cross-section after Rooksby 1991).
NATURAL ANALOGUES
69
quartz, K-feldspar and plagioclase and more minor lithic fragments. The detrital lithic grain population is dominated by illitic rock fragments, with additional glaucony grains and organic matter. The authigenic mineral assemblage (Table 4) in the Magnus and Miller reservoir sandstones can be subdivided into pre- and post-hydrocarbon emplacement stages. Mineral parageneses detailing the diagenetic events are given in Figure 8. The early, pre-petroleum diagenetic histories of the reservoirs are similar including the precipitation of an early diagenetic assemblage of framboidal pyrite, and non- to weakly-ferroan calcite. In Magnus, early non-ferroan calcite cement occurs as small nodules and a weakly-ferroan calcite cement is present as a pore-filling phase which also exhibits minor leaching or dissolution textures (Fig. 8). In Miller, large early carbonate concretions are present near the base of reservoir compartments (Marchand et al. 1997). Cold cathodoluminescence (carbonate CL) microscopy of Miller concretions suggests that there are at least two different generations of late-stage (burial diagenetic) calcite cements towards the outer edges of the concretions. Calcite cements from the margins Fig. 7. CO2 distribution in the area around the Miller field of concretions luminesce differently to the earlier showing average CO2 content of adjacent Upper Jurassic calcite and show significant evidence for leaching oil and gas reservoirs. and dissolution near the margins of the concretions. A later stage of calcite cement, described as having a sands, but increases significantly in the poorer recrystallized texture, sits within grain fractures and quality Unit 3 sandstone where grain-rimming and may replace outer concretionary calcite. In Miller pore-filling illitic clays abound. Miller can thus be early diagenetic cores of concretions have high 813C classified as a very clean sublithic arenite bordering values of up to +4%o PDB whereas the edges give on being a quartz arenite. significantly lighter 813C values of < -6 %0 PDB. The onset of quartz overgrowth precipitation postThe Magnus Field sits within UKCS block 21 l/12a and 21 l/17a in the North Viking Graben of dated calcite cement in Miller and Magnus. the Northern North Sea (Fig. 6). It contains a rela- Petroleum fluid inclusions are observed throughout tively small volume (2%) of isotopically light CO2 the overgrowths indicating that quartz cement grew (813C -13 to - 16%o PDB) in the associated gas cap at the same time as the initial migration of petroleum (Barclay & Worden 2000; Macaulay et al. 1993; into the reservoir sandstone in both Miller Table 4). The isotopic composition is strongly sug- (Marchand et al. 1997, 2000) and Magnus gestive of an organic source for the CO2 through (Haszeldine et al 1998; Worden & Barclay 2000, thermal maturation of organic matter. Addition of 2003). Feldspar dissolution in both Magnus and CO2 into the Magnus sandstone is probably related Miller occurred either concomitantly with, or to the petroleum generation and migration from off- shortly after, the onset of quartz precipitation, structure Kimmeridge Clay Formation. In contrast although the dissolution event was significantly to Miller, there is no significant volume of inorgani- more extensive in the Miller field reservoirs. cally-derived CO2 present in the hydrocarbon phase Significant secondary porosity is present in Miller in Magnus. due to the dissolution of pre-existing minerals (up to The main Magnus reservoir unit is the Upper 6% secondary porosity in Unit 3 and up to 15 % in Jurassic Magnus Sandstone Member (Fig. 6), a Unit 2 sands). Evidence for K-feldspar dissolution is thick, sand-dominated interval of high density tur- observed (Figs 5 & 8), particularly in Unit 3, but bidite sandstones with subordinate, more thinly- typical products of feldspar dissolution, such as bedded and mud-rich, low density turbidite deposits. kaolinite, are not commonly observed in samples Stratigraphically, the sandstone forms part of the from either Miller reservoir unit, or further up dip in Kimmeridge Clay Formation. The sandstones com- the South Brae reservoirs (McLaughlin et al. 1994). prise fine to medium-grained, poorly to moderately- In Magnus, the well-defined partial dissolution of Ksorted sublithic to subarkosic sandstone (Table 4; feldspars was followed by precipitation of kaolinite De'Ath & Schuyleman 1981), dominated by detrital and illite (Worden & Barclay 2003). Both Miller and
Table 4. Rock, fluid and gas composition data for the quartz arenite to sub-lithic/arkose end-member case studies: Miller and Magnus. Detrital grains (%) Mineralogical composition (mean %) High CO2 (Miller) Unit 2: HOT* Unit 3: LOT* Sandstone Classification
Quartz
63 61
Feldspar
4 6
Authigenic cements (%)
Lithic Grains
Mica
Clays
Organic Matter
Quartz
Carbonate
11 9
tr 1
5 13
1
3
1 1
1
2
6
1 1.5
1 18
0.5 3
2 3
1 2 2
Clay
Porosity (%) Other**
Primary
Secondary
1
2
1
4
n/a 2
n/a 10
4 6
1 3
13 3
tr tr
SO4
HCO3
Quartzosesublithic
Low CO2 (Magnus)
HOT LOT
62
16 8
2.5 2.5
Sandstone Classification
52.5 Subarkosic
Fluid analysis (mgl-1)
Na
K
Ca
Mg
Ba
Sr
Fe
Cl
27080 27650
1340 1280
735 645
105 105
775 780
180 72
175 47
43660 44260
5 5
2520 2470
10100 9800 11450 13000
165 125 225 230
130 70 220 240
25 38 47 55
110 110 82 117
28 18 36 50
-
16250 14600 18000 20650
13 20 30 38
1400 1175 790 800
High CO2 (Miller) Sample 1 Sample 2 Low CO2 (Magnus) Sample 1 Sample 2 Sample 3 Sample 4 Gas analysis High CO2 (Miller) Hydrocarbon zone Water zone Low CO2 (Magnus) Hydrocarbon zone
CO2(mol%)
8 13 C(%oPDB)
15.95% 64.3%
n/a
2%
-13 to -16%o
-8.2%o
* HOT: High density turbidite sands; *LDT: Low density turbidite sands ** 'Other' cements include pyrite, feldspar overgrowths Mineralogical data are derived from mean point count values from a study of 30 (Miller) and 5 (Magnus) samples respectively. Porosity data are not included in the mineral count. Magnus fluid analyses from Warren & Smalley (1994).
NATURAL ANALOGUES
71
Fig. 8. Paragenetic sequence and key diagenetic events in the Miller and Magnus field reservoirs.
Magnus fields contain small quantities of late-stage ferroan dolomite or ankerite cement, present as isolated rhombs that sit within secondary macropores (Fig. 5). In Magnus, this late burial ferroan dolomite has 813C values of -7.7 to -12.8%0 PDB. In Miller, the ferroan dolomite is post-dated by a patchy-, nonferroan calcite cement that encloses well-developed quartz overgrowths and occludes both primary and secondary macropores. These later-stage, porefilling carbonate cements outside the concretions have negative 813C values of -7 to -8%o PDB (Marchandetal. 1997). To summarize, Miller has abundant CO2 derived from a minor organic (petroleum source rock) origin and a dominant inorganic (deep crustal) source. Magnus has a low present-day CO2 concentration derived from a petroleum source rock. Miller sandstones tend to be cleaner than Magnus sandstones and can be classified as a marginal quartz arenite. Magnus is a subarkosic arenite. Magnus and Miller both underwent early diagenetic calcite cementation; early carbonate cements in Miller have an isotopic signature consistent with early diagenesis in marine sands. Miller, with its high CO2 content, has undergone extensive mineral dissolution and the creation of secondary porosity followed by precipitation of minor volumes of late ferroan dolomite and calcite cement with isotopic values close to those of the CO2 in the reservoir hydrocarbons. It is likely that the dissolved phase in Miller was a feldspar mineral (Fig.5).
Magnus with its low CO2 content has undergone less extensive feldspar replacement, compared to Miller, with minor ferroan dolomite cementation preserving an isotopic signature indicative of an organic origin (Table 2), suggesting that CO2 sequestration has been effective in this reservoir. The influx of CO2 in Miller may have been responsible for mineral dissolution and seems to have led to some late-stage carbonate cement. However, not all the CO2 has been sequestered effectively to form carbonate minerals in Miller because there was a shortfall of an appropriate source of divalent cations in the rock. Thus in Miller minor sequestration occurred initially through local precipitation of carbonate minerals although subsequent CO2 addition led to the accumulation of the gas once all the reactive minerals (feldspars) had been exhausted.
Red bed sandstone reservoirs (feldspar-lithic-rich sandstones) The two arkosic sandstones in this study, Bravo Dome and Vert le Grand, come from disparate geographic locations (Fig. 1) yet are strongly related in terms of depositional origin and mineralogy. Indeed, at the time of deposition, these sandstones were both on the Pangean supercontinent and were considerably closer together than they are now. They developed in arid continental conditions with sediment
72
S. J. BAINES & R. H. WORDEN
derived from Hercynian erogenic belts leading to similar mineral populations. The high CO2 arkosic end-member is represented by Bravo Dome in northeastern New Mexico which contains a pure CO2 accumulation. Bravo Dome is a spur of the Sierra Grande uplift with the field sitting in an anticlinal nose on the structure. CO2 in Bravo Dome is reservoired in the Permian Tubb Formation arkosic sandstones and is trapped by a combination of fold structure, stratigraphic pinch-out and possibly hydrodynamic forces (Broadhead 1990; Table 5). The Tubb Formation sandstone is interpreted to have formed during sand-dominated alluvial, fluvial and aeolian deposition. The topseal to the nonaqueous fluid is the Cimarron Anhydrite consisting of shallow marine, evaporitic sediments and arkosic muds (Fig. 9). The low CO2 arkosic end-member, Vert le Grand, sits in the Paris Basin, an intracratonic Permian basin in NW Europe. In Vert le Grand, petroleum is found in permeable Triassic continental red bed arkosic sandstones (Chaunoy Formation) in the centre of the basin where they are sealed by Lower Jurassic shales that also acted as the petroleum source rock (Worden et al 1999; Fig. 10). The Triassic Chaunoy Formation is a synrift sandstone formed in arid to semi-arid conditions and occurs as conglomerates, fluvial sandstones, dolocretes and mudstones. Bravo Dome contains an estimated 8 TCP of pure (99.97%) CO2 (Roberts & Godfrey 1994). All porous intervals tested contain free CO2, however, no petroleum is associated with the CO2. 813C values (Table 5) of the CO2 are consistent with an inorganic mantle or volcanic origin for the CO2 having values of between -4 to -5%c PDB (Lang 1959; Johnson 1983). The migration pathway for the CO2 is assumed to have been volcanic vents associated with the Rio Grande rift system. High-pressure compartments have been found associated with deep seated faults that cut through the fractured crystalline basement beneath the Tubb Formation sandstones (Fig. 9). Volcanic activity in the area occurred between 100000 and 8000 years ago (Broadhead 1998), suggesting that the onset of CO2 emplacement occurred relatively recently compared to the other study sites. Migration of CO2 into the reservoir may still be ongoing. Pressure measurements from a shut-in well in a compartment (sealed by a Plio-Pleistocene fault that forms an intra-reservoir barrier pressure barrier) have increased significantly (200 psi) over the past decade suggesting active filling of the compartment. The Bravo Dome Tubb Formation reservoir comprises vertically and laterally stacked lenticular sandstones interbedded with thin, discontinuous shaly sands and dolostones. Sand texture and mineralogy varies between depositional environments with mineralogically-immature, coarse-grained and
poorly-sorted sands associated with alluvial and fluvial deposits and mature, well-sorted, fine- to medium-grained sands being deposited by aeolian and paralic processes. Compositionally, the Tubb sands are predominantly arkosic (25% to 65% Kfeldspar and plagioclase) with quartz, lithic grains and minor clay material (Table 5). Vert le Grand is typical of the Paris Basin oil accumulations in having small quantities of CO2 in the associated gas phase with concentrations varying from 2.5-2.8% (Pinti & Marty 1995). Dissolved CO2 from the formation water has 813C values of -7 to -9%c (Worden & Matray 1998; Worden et al. 1999). The Lower Jurassic marine mudstones sourced the petroleum and most likely also sourced the CO2. Source rock maturation occurred for several million years before basin inversion in the Alpine orogeny (c. 25 Ma). The addition of CO2 to the structure probably ceased from about 25 Ma. The Vert le Grand Chaunoy Formation comprises fluvial and fan sandstones with common pedogenic dolocrete horizons testifying to the arid depositional environment. Minor anhydride shales are present but are not laterally continuous. At the time of deposition, the sandstones were arkosic to subarkosic with a variety of detrital K-feldspar minerals (orthoclase and microcline), perthite and plagioclase grains. There are a variety of dissolution fabrics developed especially in K-feldspars although the sandstones still contain up to several tens of percent of K-feldspar. The plagioclase grains (now pure albite plagioclase) are distinctly turbid, containing abundant sericite inclusions and appear to be extensively weathered in contrast to the clear albite overgrowths on plagioclase. The authigenic mineral assemblage in the high CO2 Bravo Dome Tubb Formation sandstone is described below (Fig. 11) and has previously been employed by Pearce et al. (1996) as a comparison for laboratory experiments on CO2-rock-water systems. The Tubb Formation sandstone had an early burial diagenetic phase dominated by oxidizing conditions that resulted in the authigenesis of hematite, smectite-illite, gypsum and anhydrite. Minor mesogenetic quartz and albite overgrowths were post-dated by pore-filling anhydrite cement that occludes much of the primary intergranular porosity. Poikilotopic dolomite cements show evidence of widespread dissolution. Secondary porosity was also generated through the leaching of dolomitized lithic grains, sulphate minerals and evaporites and widespread dissolution of detrital plagioclase and K-feldspar. The final cements to precipitate were ferroan dolomite (replacing anhydrite and sitting within relict feldspar grains), siderite, kaolinite, chlorite, gibbsite and Ca-zeolite (laumontite) that sit within primary and secondary macropores (Fig. 5). Calcite and ferroan calcite are rarely
NATURAL ANALOGUES
present as intergranular cements. The late stage cements, gibbsite and laumontite are spatially associated with faulted zones. Unfortunately, no carbon isotopic data are available for the dolomite and ferroan carbonate cements. The Chaunoy Formation in Vert le Grand was initially arkosic having a varied feldspar population including alkali feldspar, plagioclase and various types of perthite. Early diagenetic dolomite was a common product of arid environment pedogenesis and had 813C values from +2 to -6%o PDB. (Spot! et al 1993; Worden et al 1999). Burial diagenesis resulted in a variety of carbonate cements including ferroan dolomite, saddle dolomite and ferroan calcite (Figs 5 & 11). Quartz cement grew at much the same time as the ferroan dolomite cement. Feldspars were extensively corroded resulting in some secondary porosity although most of the newly formed porosity was in-filled with ferroan dolomite cement. The later carbonate cements had 813C values from —7 to — 13%o PDB suggesting that a source of carbonate with low 813C values must be invoked. The overlying organic rich Lower Jurassic mudstones are an obvious candidate implying that petroleum and added CO2 have a common source. In synthesis, Bravo Dome in New Mexico is filled with CO2 but also retains plentiful potentially reactive minerals such as plagioclase, zeolite and clays. The dominant CO2 seems to have a volcanic origin and is a recent, and possibly on-going, addition to the structure. Vert le Grand in the Paris Basin has a relatively low CO9 concentration from, in part, an organic source but has undergone extensive addition of CO2 in the past as witnessed by the dominant carbonate cements with their relatively low 813C values. The late carbonate cement has the same S13C value as the minor CO2 found in Vert le Grand and appears to have grown at the same time as feldspar dissolution. Although Bravo Dome and Vert le Grand were superficially similar in terms of mineralogy, the Paris Basin sandstone has sequestered the added CO2 as carbonate cements (probably following feldspar dissolution) whereas Bravo Dome contains a still-reactive mixture of CO2 and plagioclase, zeolite and clay. The early authigenic cements in Bravo Dome show evidence for a distinct dissolution event during the later stages of burial which affected early dolomite, sulphate and aluminosilicates. In Bravo Dome, late-stage CO2, detrital calcic plagioclase and diagenetic calcic zeolite remain in contact even though they are mutually reactive.
Natural reactions in rocks due to CO2 addition There have been several studies of the origin of the relationship between the increasing fugacity
73
(approximately equivalent to partial pressure) of CO2 and depth and temperature in sedimentary basins (e.g. Smith & Ehrenberg 1989; Hutcheon et al. 1993;Coudrain-Ribstein^<2/. 1998; Hutcheon & Desrocher 2003). These studies have attempted to relate the increasing partial pressure of CO2 to sudden influxes of CO2 and fluid-mineral reactions in reservoir or aquifer formations. Partial pressure is equivalent to the mole fraction of the gas phase in question multiplied by the total fluid pressure. Reanalyses of at least some of these data sets reveals an uncomfortable conclusion that although CO2-partial pressure typically increases with depth in sedimentary basins, it is not so much the quantity (mole fraction) of CO2 but the fluid pressure that increases with depth and temperature. Thus the studies that have concluded that there is an influx of CO2, or that CO2 has been produced, resulting in increasing CO2 partial pressure may be wrong. Models and ideas that assume that the quantity of CO2 in the subsurface increases with depth may be based on a false premise since, in at least some basins, it is the increase in fluid pressure with depth that leads to increasing CO2-partial pressure with increasing depth. However, clearly CO2 is both produced and consumed by a number of processes in the subsurface so what happens when CO2 is naturally added to reservoir rocks in the subsurface? What is the range of possible reactions that could happen once a large quantity of CO2 is added to a subsurface formation? In this section the reactions that seem to have occurred in the case studies will be considered. The addition of CO2 to formation water tends to cause an overall increase in acidity since CO2 dissolves in water (Rl), and then partially dissociates into bicarbonate anions and protons (R2a) and into the subordinate carbonate anions (R2b).
In general there are two classes of reaction: those that occur in rocks that are not buffered to changes in acidity (non-pH buffered) and those that occur in rocks that are mineralogically buffered to changes in acidity (pH buffered). The term buffering refers to a rock's capacity to counteract any change in fluid composition by mineral-rock interaction. However, the dissolution of CO2 into water results in an acidic solution (R2 & R3). Such an acidic solution is capable of causing carbonate mineral dissolution by reactions of the type:
Table 5. Mineralogical composition ofarkosic end-member sandstones from Bravo Dome and Vert le Grand fields. A: Mineralogical modal analyses from 17 samples from the Middle Tubb Formation, Bravo Dome field. B: Mineralogical modal analyses from the Chaunoy Formation, Vert le Grand field (a) Mineralogical modal analyses from 17 samples from the Middle Tubb Formation, Bravo Dome field. Porosity
Authigenic
Detrital Sample
Framework Grains (%)**
Matrix Grains
Cement (Total %) (%)***
Haematite
Quartz
Adularia
Anhydrite
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
55 54 45 55 56 52 67 49 60 54 53 49 53 58 51 52 48
4 1 13 2 13 3 3 1 2 3 6 2 4 2 2 3 9
22 17 22 15 13 11 8 42 19 24 28 35 27 23 25 34 19
24 44 8 43 76 14 70 5 3 4 19 12 7 4 8 17
10 23 2 14
12 16 11 14 9 18 13 3 9 14 6 21 19 1 4 21 21
1
3 17 21 20 7 6 22 20 5 5
** Framework category contains all rigid detrital grains. ***Matrix category contains all ductile detrital grains and clays. Cement abundance given as % of total cement tr - trace
5 1 2 4
Gypsum
4 2
-
2 14 19 7 7 _ 14 56 29
Dolomite
Chlorite
28 30 5 2 29 92 17 19 27 77 60 9 30 2 -
Ferroan carbonate
Siderite
Kaolinite
Zeolite
%of total vol.
2 _ 2 7 2 _ 5 7 11 1 3
1 3 2 2 1 2 5 -
tr
18 9 48 19 12 9 tr tr 46 23 15
20 28 20 29 18 34 23 8 19 19 13 13 16 16 22 10 24
1 -
4 53 12 3 21
(b) Mineralogical modal analyses from samples from the Chaunoy Formation, Vert le Grand field Fluid analysis (mgl"1)
Na
K
Ca
Mg
Ba
Sr
Vert le Grand Sample 1 Sample 2
21800 22200
375 406
3397 3814
490 105
0.9 1.1
202 195
Gas analysis
CO2 (mol%)
5 13 C(%oPDB)
Vert le Grand Sample 1 Sample 2
4.89% 2.09%
-8.5%o
Fe
Cl
SO,
HCO3
39996 41907
780 700
221 219
76
S. J. BAINES & R. H. WORDEN
Fig. 9. Map and cross-section of geological setting and reservoir of the pure CO2 field Bravo Dome, NE New Mexico, USA.
This reaction can be combined with reactions 1 and 2 to demonstrate that adding CO2 to a pure carbonate rock will induce mineral dissolution:
It has been assumed that the addition of CO2 into a saline aquifer in the subsurface would result in carbonate mineral precipitation although it transpires that, under thermodynamic equilibrium conditions, this will only happen when there is an inexhaustible supply of divalent cations. A generic example of such a reaction is given below:
The formula M(II)CO3 represents any divalent metal carbonate. Note that reactions Rl, R2a and R2b are implicit in the above reaction (R4). Such reactions rely upon a continuous supply of aqueous calcium, or other divalent cation, although a carbonate mineral cannot be the route of this supply since carbonate minerals are also at the end of this reaction. Rocks that are pH-buffered to the addition and subsequent aqueous dissociation of CO2 bear minerals that contain divalent cations capable of creating neoformed carbonate minerals. The main cations are calcium, magnesium and, to a lesser extent, iron. A generic reaction could thus be written:
The resulting Al-silicate (clay) mineral is most likely to be kaolinite. Non-carbonate minerals that contain divalent cations include the anorthite component in plagioclase, Fe-, Ca- and Mg-bearing clay minerals
(such as chlorite, saponite and other smectite minerals) and some zeolite minerals (e.g. laumontite and heulandite). There is also a wide range of detrital high temperature minerals that could be present in mineralogically immature sandstones, such as olivine, pyroxene and hornblende. Sulphate minerals such as anhydrite also contain divalent cations although sulphate minerals are stable in the presence of CO2 so they cannot sequester the gas to produce carbonate minerals Thus the natural addition of CO2 to a subsurface formation could theoretically lead to both carbonate mineral precipitation or dissolution depending on whether the rock contains reactive minerals (i.e. is pH buffered). The reactive minerals must contain divalent cations and include Ca-feldspars, Mg and Fe clays and zeolites. Rates of reactions R1-R5 have not yet been considered. Thermodynamic equilibrium has been addressed, but the speed at which component processes occur has not yet been dealt with. All geochemical mineral reactions involve a number of steps including dissolution (of gas and mineral), transport (by diffusion or flow) and precipitation (whether homogeneous including nucleation, or heterogeneous simply requiring growth on a suitable substrate). Dissolution and precipitation are known as surface reactions. The slowest step in a chain dictates the overall rate of the reaction. A critical observation is that silicate minerals, being structurally complex and containing sparingly soluble alumina and silica, have surface reaction rates that are many orders of magnitude slower than those for carbonate minerals. Although equilibrium considerations tell us what a rock should evolve into, they do not reveal how, or if, it will get there.
NATURAL ANALOGUES
77
Fig. 10. Map and cross-section of geological setting of the Triassic sandstones of the Paris Basin, France, (a) The locations marked represent the sites of sampled wells, (b) The Triassic Chaunoy Formation sits beneath the Lower Jurassic source rock and away from any (other) sources of CO2 (e.g. intrusions, carbonate rocks, etc).
Natural reactions in the case studies due to CO2 addition Porous rock formations typically contain high density minerals and a volumetrically much smaller quantity of formation fluids (predominantly water). Formation waters can have a wide range of compositions from being nearly fresh water through to being ultra-saline brines with total dissolved concentrations of up to 300 000mg I"1. Salinity is typically dominated by sodium and chloride. Generalization about formation water composition is dangerous although the gross controls on salinity (e.g. Na- and Cl-concentrations) can be understood on a basin scale since the main controls on salinity are: seawater evaporation, evaporite dissolution, meteoric dilution and to a much lesser extent porous rock-formationwater interaction (i.e. carbonate-water, silicate-
water interaction) (Worden 1996). The cations that influence carbonate mineral precipitation and dissolution (alkaline earth elements and divalent transition metals; Mg, Ca, Sr, Fe, Mn), are probably more locally controlled and will be directly influenced by the minerals in the porous rock unit. In general, formation waters may play a role in influencing CO2 and its consequences since large-scale water advection systems, the exception in basins, could theoretically supply CO2 to a porous formation or possibly remove cations resulting from mineral dissolution. The following discussion will not dwell on the flow of formation water since it is the exception rather than the rule; reactive cations are likely to be locally controlled (buffered) prior to CO2-addition; and CO2 movement can happen in the gas phase or even as a gas dissolved in petroleum. Formation water is certainly important in the reactions that result from CO2
78
S. J. BAINES & R. H. WORDEN
Fig. 11. Paragenetic sequence and key diagenetic events in the: (a) Bravo Dome field and (b) Vert le Grand field.
addition since water is a crucial catalyst for mineral reactions. However, water is omnipresent in the subsurface, even in petroleum fields (as a residual water phase; e.g. Archer & Wall 1992), so that the role of an aqueous medium for reactions may be of a second rank.
In the carbonate end-member pair of cases, Blue Whale has abundant CO2 and has undergone significant carbonate mineral dissolution. Dolphin has lower CO2 and has undergone only minor late-stage dissolution. These carbonate accumulations are ostensibly free of aluminosilicate minerals. The only
NATURAL ANALOGUES
minerals that contain divalent cations are the carbonate host minerals. They are not pH buffered so that addition of CO2 cannot lead to CO2 sequestration. Some of the added CO2 will dissolve until the water is fully saturated (Rl). The dissolved CO2 will partially dissociate, create acidity (R2) and induce a degree of dissolution of carbonate minerals (R3). This dissolution will be limited by the amount of CO2 in aqueous solution, itself controlled by salinity, pressure and temperature. The association of high CO2 and rock dissolution fabrics in Blue Whale is thus unlikely to be coincidental. Dolphin has seemingly never received abundant CO2 and has thus not undergone significant dissolution. Geological data tell us that natural addition of CO2 to limestones and dolostones will lead to mineral dissolution and not bulk sequestration, unless there is an independent supply of divalent cations from exotic sources. The long-term scenario is, therefore, one of host rock dissolution and CO2 in the gas phase. Magnus and Miller both seem to have received abundant CO2 although possibly from slightly different sources. The occurrence of percentage levels of late diagenetic carbonate cement in Magnus with 813C values as low as — 12.8%c suggests that there might have been a supply of CO2 that was subsequently scrubbed from the formation water (by divalent cations) to form carbonate cement. Macaulay et al (1998) interpreted 813C values of -12.8%o as being the result of organic acid influx; however, it is likely that Magnus sandstones have undergone significant natural CO2 sequestration as witnessed by the carbon isotopes in the late ferroan dolomite cement (an example of R5) and the low CO2 content of its gas. There is seemingly little evidence of dissolution of carbonate minerals in Magnus confirming that the influx of isotopically distinct CO2 did not lead to low pH and carbonate dissolution. In contrast, Miller sandstones seem not to have led to significant CO2 sequestration, given the much greater volume of CO2 present over time, since there are only minor (typically less than 2%; rare maximum of 10%) amounts of diagenetically late carbonate cement. Miller sandstones were probably, in the first case, much cleaner and more quartzose than Magnus sandstones. It seems that Magnus, with its high feldspar content, has buffered the pH and allowed carbonate minerals to form. Miller seems to have had less initial feldspar than Magnus. However, the elevated secondary porosity in Miller may represent a feldspar dissolution event, thus implying some degree of pH buffering and suggesting that Miller sandstones are diagenetic (as opposed to detrital) quartz arenites. For Miller, the current relative lack of feldspar seems to suggest that the buffering capacity was exhausted before the supply of CO2 was exhausted. Quartz arenites do not have a pH buffering capacity and thus cannot naturally
79
sequester CO2 unless there is an exotic source of divalent cations. Thus it may be that the primary pH buffering capacity of the rock, rather than the absolute volume of CO2, had the greatest control on the accumulation of gas phase CO2 (i.e. R4 and R5 could not occur in Miller as successfully as they have done in Magnus). Miller has a high CO2 content since the initial CO2 exhausted the buffering capacity of the sandstone and the subsequent CO2 had nothing else to react with. The pH buffering capability of Magnus is also witnessed in Vert le Grand where there has been extensive growth of diagenetically-late ferroan dolomite with characteristic stable isotopes implying an organic source of CO2 although the CO2 concentration in the associated gas is now relatively low. Vert le Grand has a wealth of signs of pH buffering including feldspar dissolution fabrics with ferroan dolomite growth spatially associated with feldspar dissolution. There is little evidence of diagenetically-late dissolution of carbonate minerals in Vert le Grand confirming that the influx of isotopically-distinct CO2 did not lead to a low pH and carbonate dissolution. Since Bravo Dome and Vert le Grand are mineralogically similar, the critical question concerns why the addition of CO2 in Bravo Dome has not resulted in its sequestration as diagenetically-late and isotopically-distinct carbonate cement. The pHbuffering capacity is certainly present in the reservoir with the abundance of calcic plagioclase, calcic zeolites, clays, etc. Bravo Dome filled with CO2 between 8000 and 100000 years ago and it seems that there has simply not been enough time for thermodynamic equilibrium to be achieved. CO2 wants to react with, for example, plagioclase, but the rate of plagioclase dissolution is seemingly so slow that the reaction is incomplete. The rate of plagioclase dissolution must be too slow to allow this mineral to scrub the CO2 from the fluid phase. Note that, despite the reactive minerals that could produce calcite from plagioclase, the rate of carbonate dissolution is so fast that the rock may have undergone some degree of carbonate dissolution. The rock may thus be transiently non-pH buffered, not because the mineralogy is inappropriate but because the rate of pH buffering (by silicate mineral dissolution) is so much slower than the rate of carbonate mineral dissolution.
Modelling and simulation of CO2 addition to geological systems Geochemical modelling Geochemist's Workbench has been used (Bethke 1994) with a standard thermodynamic database of
80
S. J. BAINES & R. H. WORDEN
minerals (Delaney & Lundeen 1990) to help understand and illustrate the reactions that have taken place in geological systems following the natural geological addition of CO2 to the different geological systems described above (Table 6) The data are cast in terms of the fugacity of CO2 where fugacity is similar to partial pressure but accounts for the nonideality of gases (partial pressure multiplied by a fugacity coefficient equals the fugacity). Similarly aqueous species are cast in terms of activity where activity is similar to concentration but accounts for the non-ideality of aqueous solution. The data in Table 6 represent the equilibrium state and thus do not inform us how long (or even if) a reaction will occur. They tell us how a rock should evolve to reach thermodynamic equilibrium, not whether it will evolve. The conditions to be defined during geochemical modelling include: temperature, mineralogy, mineral proportions, water composition, pH and gas fugacity. In general, the models have been designed for formation water at 80°C, with 1 molar solution of NaCl with 0.2 moles of aqueous calcium and 0.01 moles of aqueous bicarbonate. The models have all been built around a rock with approximately 20% porosity (c. 4000cm3 of minerals to c. 1000cm3 of water) that has had lOOg of CO2 progressively added (thus mimicking the geological addition of CO2). The activity (—concentration) of calcium is an important variable along with the fugacity (partial pressure) of CO2. Models for different starting compositions in saline aquifers reveal the minimum yCO2 under which carbonates grow depends on mineralogy and formation water chemistry. Each reservoir is unique in terms of mineralogy, formation water geochemistry, temperature and pressure so that it is not possible to erect universally applicable models.
Addition of CO2 to carbonate rocks The carbonate end-members (Blue Whale and Dolphin) are not particularly interesting for geochemical modelling since the addition of CO2 (as represented by increasing CO2 fugacity) simply causes carbonate mineral dissolution (Table 6). Addition of CO2 does not lead to growth of new carbonate minerals; there is no solid-phase sequestration of this gas. Indeed, the modelling confirms that the diagenetically-late secondary porosity in Blue Whale carbonate could plausibly be due to CO2 influx since the addition of CO2 in the model led to minor solid volume decrease. Note that addition of CO2 causes an increase of aqueous calcium due to calcite dissolution. Flushing the system with progressively more CO2 would lead to progressively greater quantities of secondary porosity.
Addition ofCO2to quartz-rich sandstones Miller is represented by a quartz-rich sandstone endmember, modelled here with a minor quantity of calcite cement, that responds in a similar way to the carbonate rocks - simply because quartz is immune to the effects of CO2. Modelling the addition of CO2 (Table 6) shows that it accumulates in the gas phase (/CO2 increases) and leads to minor rock dissolution (note the increase in aqueous bicarbonate). Thus Miller has high CO2 because CO2 has been added to the reservoir and because the clean sandstone rock cannot sequester the gas as a solid phase.
Addition ofCO2to arkosic (feldspar-rich) sandstones For the purposes of modelling, Magnus is represented by feldspar-rich sandstone end-members. However, the models for feldspar are extremely sensitive to the precise mineralogy of the feldspars. Anorthite cannot be added to an equilibrium model since it is unstable (and so should react totally) under all conditions found in (and on) sedimentary basins. K-feldspar-rich arkosic sandstones (that here are modelled to contain minor calcite) have negligible capacity to sequester CO2 (Table 6). This is shown by the fact that the final fugacity of CO2 is nearly as high as in the carbonate and quartz arenite cases. Kfeldspar reacts in an acidic environment (induced by CO2 dissolution and dissociation) to produce muscovite (illite). There is growth of a small quantity of a carbonate mineral though, in this case the little-known sodium aluminium hydroxy carbonate mineral: dawsonite (NaAlCO3(OH)2). This mineral has been reported from a number of CO2-bearing basins worldwide (e.g. Baker 1991) although it has not been reported in any of the case studies presented here. Dawsonite grows because aluminium is available via K-feldspar and sodium is available from the formation water. In the case of the K-feldspar model, dawsonite represents less than 0.5% of the rock. Dawsonite may be more common than has thus been recognized since it has many petrographic characteristics in common with illite. Albite-rich arkosic sandstones (that here are modelled to contain minor calcite) have fair capacity to sequester CO2 (Table 6). The final fugacity of CO2 is much lower than in the case of carbonates, quartz arenite or K-feldspar-rich arkose suggesting that the rock has buffered the system to the effects of added CO2. There is growth of a significant quantity of dawsonite (NaAlCO3(OH)2) and a much smaller quantity of calcite In the case of the CO2-albite model, dawsonite represents more than 3.0% of the rock. The model suggests that the geological addition of CO2 to albite-rich sandstones should lead to
Table 6. Results of geochemical modelling of the addition ofCO2 to different initial rock types at 80 °C, containing a 1 molar solution ofNaCl with 0.2 moles of aqueous calcium and 0.01 moles of aqueous bicarbonate. The models have all been built around a rock with c. 25% porosity that has had 100 g ofCO2 added progressively. The top row of the table has negligible capacity to sequester CO2 whereas the lower row has ample capacity to sequester CO2. lithology
limestone
initial
final
dolomitic limestone
initial
final
calc-arenite
initial
final
K-spar arkose
initial
final
mineral volumes (initial) cm3
calcite
3690
3689
calcite dolomite
1845 1745
1843 1744
quartz calcite
3800 740
3776 739
quartz calcite K-feldspar dawsonite muscovite
3021 739 391 0 0
3067 744 296 19 25
volume change -0.027% Ca + + mol Na+ mol K + mol Mg + + mol Fe ++ mol HC
Albite arkose quartz calcite albite dawsonite paragonite
volume change -0.028%
0.003 1.000 0.103 6.00 0.62
0.031 1.000 2.404 4.72 2.25
initial
final
3021 741 367 0 6.5
3188 745 119 127 21
volume change 1.40% Ca ++ mol Na+ mol K+mol Mg + + mol Fe + + mol HCO3 mol pH logfCO 2
-0.100 1.099 _ 0.001 6.53 -1.95
volume change -0.022'7c
0.003 1.000 0.103 6.00 0.64
0.031 1.000 2.450 4.77 2.27
zeolite lithic sandstone
initial
final
quartz calcite laumonite albite kaolinite
3011 369 833 47 0.5
3114 453 364 44 226
volume change 1.36% 0.002 1.317
volume change 0.024%
0.154 1.000 0.054 5.20 0.53
0.177 1.000 2.350 4.31 2.22
Mg-clay lithic sandstone
initial
final
quartz saponite dolomite kaolinite magnesite
2672 759 0.3 264 0
2726 664 12.1 276 53
volume change 1.00% 0.420 0.560
0.430 0.536
0.107 1.000 0.088 0.007 5.20 0.52
0.020 0.686 0.680 1.880 4.90 2.11
Fe-clay lithic sandstone
initial
final
quartz Fe-chlorite calcite dolomite siderite kaolinite
3021 625 366 7.5 0 0.4
3031 524 366 2.6 65 45
volume change 0.50%
0.010 1.000
0.190 1.000
_ -
_ _
-
_ -
-
0.031 6.43 -0.13
0.000 6.73 -2.96
0.000 6.74 -2.96
0.001 5.91 -1.37
0.008 5.58 -0.43
_
0.234
0.049
0.230 1.000
0.230 1.000 -
-
0.009 0.003 0.002 5.91 -1.06
0.009 0.007 0.005 5.74 -0.69
82
S. J. BAINES & R. H. WORDEN
the growth of dawsonite. This is puzzling because there are few reports in the literature. The dawsonite may have been misidentified by petrographers (it has a resemblance to illite in thin section and SEM) or its thermodynamic properties may have been incorrectly defined (i.e. it may not be as stable and likely to form as the models show).
Addition ofCO2 to lithic (clay-rich) sandstones Bravo Dome and Vert le Grand are here represented by models of a variety of lithic-rich sandstone endmembers. The lithic component of the sandstones has been modelled by the addition of Ca-zeolite (laumontite), Mg-clay (saponite) and Fe-clay (daphnite). As with the albite-rich sandstone, these display the ability to sequester geologically-added CO2 with the commensurate growth of calcite, magnesite and siderite (respectively) (Table 6). Each of the modelled lithic sandstone examples has the capacity to keep the CO2 fugacity low through the growth of new carbonate minerals. Modelling different idealized lithic sandstones is fine but in reality lithic sandstones typically contain Fe-Mg-rich lithic grains so that Fe-Mg-rich carbonates are more likely than pure siderite or magnesite. Note that ferroan dolomite is a very common late diagenetic mineral in many sandstones buried to temperatures of 100-120 °C. In the case of Vert le Grand, ferroan dolomite is the last diagenetic mineral to form (before minor inversion). The Triassic sandstones in Vert le Grand are also feldspar-rich with primary K-feldspar and plagioclase representing up to 30% of the rock. The detrital plagioclase is now highly altered (especially in contrast to the transparent albite overgrowths) suggesting that the anorthite component in the detrital plagioclase has reacted with CO2 thus helping to generate the late Fe-dolomite cement. Bravo Dome is a lithic sandstone with a rich stew of zeolite, clay and feldspars. However, despite the massive flux of CO2, these minerals remain and the rock is not at equilibrium. The minerals and gas are unstable together and solid phase sequestration has not occurred.
Rate of CO2-induced reactions The one system that has received the CO2 relatively recently (Bravo Dome) has not reached equilibrium since the mixture of zeolites, feldspar and clays should react to produce dolomite and other divalent cation carbonate minerals (±dawsonite). Although traces of very late ferroan dolomite are present in partially dissolved feldspar grains, the (metastable)
co-existence of CO2 and the mixture of zeolites, feldspar and clays leads to the conclusion that sluggish rates of reaction have inhibited reaction. Rates of geochemical reactions are relatively poorly known. General sources of rate data include: experiments, typically performed at artificially high temperatures on simplified systems to allow reaction to occur on a laboratory timescale, and large geological databases where rates have been inferred using samples from supposedly similar rocks that have experienced a range of thermal histories (for example quartz cementation, Walderhaug 1994; smectite to illite conversion, Elliott et al 1991). Kinetic data have been produced, in many cases from high-temperature experiments (extrapolated to low temperature) performed on finely divided minerals at unnaturally elevated fluid-rock ratios, for a variety of relevant processes: anorthite dissolution (Sverdup 1990), calcite precipitation (Plummer et al. 1978) and dolomite precipitation (Busenburg & Plummer 1982). In general, aluminosilicate surface reactions (e.g. dissolution) are many orders of magnitude slower than carbonate surface reactions (e.g. precipitation) so that aluminosilicate mineral dissolution will tend to be rate-limiting. Crucially though, the absolute rate of dissolution depends not only on kinetic constants but also on the physical state of the rock. Minerals with a high surface area react (dissolve) more quickly than minerals with a low surface area. Reactive minerals that are physically isolated from the reactive fluid will not react on any timescale (e.g. Bildstein et al. 2001). Thus an anorthite crystal that is totally enclosed within a quartz grain, for example in a myrmekite lithic grain, is effectively isolated from the reactive fluid (CO2) and will not react under any circumstances. Interestingly, models that employ anorthite dissolution rate data at conditions relevant to Bravo Dome and assuming that the anorthite is exposed to the CO2 seem to show that anorthite should be consumed within a few years. Its actual preservation in Bravo Dome leads to the conclusion that either the rate data are wrong by several orders of magnitude or that the anorthite is not in contact with the reactive fluid. Gunter et al (1997) suggested that such anorthite-CO2 reactions should occur over a few tens to hundreds of years. The persistence of anorthite (and other reactive minerals) in Bravo Dome suggests that the rate data may be in error since CO2 has sat in the structure for many thousands of years (despite being reactive to the rock). Rate data for complex divalent metal minerals such as zeolites and smectite clays are even less well known (especially under geological conditions) than for anorthite. This lack of fundamental kinetic data represents a serious gap in the knowledge base and profoundly limits the ability to predict rates of CO2-rock reaction. The small amount of experimen-
NATURAL ANALOGUES
tal rate data that exist typically have been derived from cleaned and finely-divided pure mineral samples that in no way mimic real rocks. The Bravo Dome case indicates that a modelling approach using currently available, experimentallyderived, rate data may be irrelevant to real geological systems. By extension it is likely to be irrelevant to predictive studies of CO2 injection into the subsurface. It may be significantly more fruitful to examine large geological datasets, or data from enhanced oil recovery CO2 injection studies (e.g. Worden & Smith 2004)
Conclusions (1) (2)
(3)
(4)
(5)
(6)
Natural CO2 is a common gas in many sedimentary basins and has a large number of possible sources. Studies of natural CO2-water-rock interaction can help to project what will happen, on a long timescale, when CO2 is injected into the subsurface. Such projection is difficult using either a modelling or an experimental approach due to limited kinetic data and the problems of getting reactions to proceed in experiments at the low temperatures required for realistic simulation. Geological data show that when CO2 is added to carbonate rocks or carbonate cemented pure quartz sandstones, carbonate minerals will partially dissolve due to the decrease in pH following CO2 dissolution and dissociation. Once the formation water is saturated with CO2, CO2 will simply remain as a gas and lead to an increase in fluid pressure. Geological data show that when CO2 is added to pure quartz sandstones, once the formation water is saturated with CO2, CO2 will simply remain as a gas and increase fluid pressure. Geological data show that when CO2 is added to sandstones containing reactive minerals that can act as a pH buffer, then the CO2 will tend to be sequestered, over geological timescales (millions of years) as a solid carbonate cement following dissolution of the reactive minerals. Such reactive minerals include anorthite and other feldspars, zeolite minerals and Fe-Mg clay minerals. When the reactive minerals have all been utilized (buffer capacity exhausted) and the solid sequestering capability exceeded, then CO2 will simply remain as a gas and increase fluid pressure. The addition of CO2 to sandstones that contain reactive minerals only leads to sequestering the CO2 as a carbonate cement on a timescale greater than tens to hundreds of thousands of years. The sequestration rate will probably be limited by the dissolution rate of the reactive
(7)
83
minerals and the physical state of the rock. When reactive minerals have limited surface area exposure to CO2-bearing water, the rate of sequestration will be slow. If such reactive rocks contain pre-existing carbonate cements, these cements may transiently dissolve due to the reduced pH since carbonate dissolution tends to be many of orders of magnitude faster than aluminosilicate dissolution. It is likely that the injection of CO2 into the subsurface to reduce greenhouse gas emissions will result in carbonate mineral dissolution and an increase in the fluid pressure. The combination of dissolution and decreasing effective stress lead to concerns about rock strength, especially of top-seals and fault-seals that are carbonate mineral-bearing.
The authors thank BP Exploration and Production for access to data from the study sites discussed and for permission to publish the work. R. Miller, A.D. Horbury, D.C. Holland, H. Nicholson, N.H. Oxtoby and J. Thrasher contributed to internal BP reports from which data were extracted and re-interpreted. Reviews by Stuart Haszeldine and Ross McCartney helped to improve the manuscript.
References ARCHER, J. S. & WALL, P. G. 1992. Petroleum Engineering. Principles and Practice. Graham and Trotman. 362 pp. BAKER J. C. 1991. Diagenesis and reservoir quality of the Aldebaran sandstone, Denison Trough, east-central Queensland, Australia. Sedimentology, 38, 819-838. BAKLID, A., KORBOL, R. & OWREN, G. 1996. Sleipner Vest CO2 Disposal, CO2 injection into a shallow underground aquifer. 1996 SPE Annual Technical Conference and Exhibition, Denver Colorado, USA. SPE paper 36600,1-9. BARCLAY, S. A. & WORDEN, R. H. 2000. Geochemical modeling of diagenetic reactions in a sub-arkosic sandstone. Clay Minerals, 35, 57-67. BETHKE, C. M. 1994. The Geochemists Workbench, version 2.4, A Users Guide to Rxn, Act2, Tact, React and Gtplot. Hydrodology Program, University of Illinois, USA. BILDSTEIN, O., WORDEN R. H. & BROSSE, E. 2001, Appraisal of the effect of anhydrite dissolution kinetics on the rate of thermochemical sulphate reduction. Chemical Geology, 176,173-189. BROADHEAD, R. F. 1998. Natural accumulations of carbon dioxide in the New Mexico region - Where are they, how do they occur and what are the uses for CO2? Lite Geology, 20, 2-6. BROADHEAD, R. R 1990. Bravo Dome Carbon Dioxide Gas Field. In: BEAUMONT, E.A. & FOSTER, N.H. (eds) Atlas of Oil and Gas Fields. American Association of Petroleum Geologists, Tulsa, 213-232. BUSENBURG, E. & PLUMMER, L. N. 1982. The kinetics of dissolution of dolomite in CO2-H2O systems at 1.5°C to 65°C and 0 to 1 arm Pco . American Journal of Science, 282,45-78.
84
S. J. BAINES & R. H. WORDEN
CLAYTON, C. 1991. Carbon isotope fractionation during natural gas generation from kerogen. Marine and Petroleum Geology, 8,232-240. COUDRAIN-RlBSTEIN, A., GOUZE, P. & DE MARSILY, G.
1998. Temperature-carbon dioxide partial pressure trends in confined aquifers. Chemical Geology, 145, 73-89. CZERNICHOWSKI-LAURIOL, I., SANJUAN, B., ROCHELLE, C., BATEMAN, K., PEARCE, J. & BLACKWELL, P. 1996. Analysis of the geochemical aspects of the underground disposal of CO2. In: APPS, J.A. & TSANG, C.F. (eds) Deep injection disposal of hazardous and industrial waste. Academic Press, Inc, 565-583. DE'ATH, N. G. & SCHUYLEMAN, S. F. 1981. The Geology of the Magnus Field. In: ILLING, L.V. & HOBSON, G.D. (eds) Proceedings of the Second Conference of Petroleum Geology of the Continental Shelf of NW Europe. Institute of Petroleum Geology, London, 342-351. DELANEY, J. M. & LUNDEEN, S. R. 1990. The LLNL thermochemical database. Lawrence Livermore National Laboratory Report UCRL-21658. EHRENBERG, S. N. & JACOBSEN, K. G. 2001. Plagioclase dissolution related to biodegradation of oil in Brent Group sandstones (Middle Jurassic) of Gullfaks Field, northern North Sea. Sedimentology, 48,703-722. ELLIOTT W. C., ARONSON, J. L., MATISOFF, G. & GAUTIER, D. L. 1991. Kinetics of the smectite to illite transformation in the Denver basin: clay mineral, K-Ar data and mathematical model. American Association of Petroleum Geologists Bulletin, 75,436-462. EMERY, D & ROBINSON, A. C. 1993. Inorganic geochemistry: applications to petroleum geology. Blackwells, Oxford, 254p. FRASER, A. J., MATTHEWS, A. J. & MURPHY, R. W. (eds) 1997. Petroleum Geology of S E Asia. Geological Society, London, Special Publications, 126. GARLAND, C. A. 1993. Miller Field: reservoir stratigraphy and its impact on development. In: J.R. PARKER (ed.) Petroleum Geology of Northwest Europe. Proceedings of the 4th Conference, 401^14. Geological Society, London. GLUYAS, J., GARLAND, C., OXTOBY, N. H. & HOGG, A. J. C. 2000. Quartz cement: the Miller's tale. In: WORDEN, R. H. & MORAD, S. (eds) Quartz cement in sandstones. International Association of Sedimentologists, Special Publication, 29, 199-218. Blackwells, Oxford, UK. GUNTER, W. D., WIWCHAR, B. & PERKINS, E. H. 1997. Aquifer disposal of CO2-rich greenhouse gases: extension of the time scale of experiment for CO2sequestering reactions by geochemical modelling. Mineralogy and Petrology, 59,121-140. GUNTER, W. D. ETAL. 2004. The role of hydrostratigraphic and geochemical trapping in secure geological storage of carbon dioxide. In: BAINES, S. J. & WORDEN, R. H. (eds) Geological Storage of Carbon Dioxide, Geological Society, London, Special Publications, 233,129-145. GUPTA, N. & SASS, B. 1999. Hydrodynamic and geochemical modeling for carbon dioxide sequestration in deep saline formations. U.S.DOE report for project DEAF26-98FT00874. HASZELDINE, R. S., MACAULAY, C., FALLICK, A. E.,
MARCHAND, A. & SWENNAN, R. 1998. Enhanced porosity from diagenesis of deepwater sandstones: Brae Oildfield, North Sea. American Association of Petroleum Geologists Annual Meeting Abstracts A280. HUTCHEON I, & DESROCHER, S. 2003. Silicate-carbonate reaction in sedimentary rocks: fluid composition control and potential for generation of overpressure. In: WORDEN, R. H. & MORAD, S. (eds) Clay cement in sandstones. International Association of Sedimentologists, Special Publication, 34,161-176. Blackwells, Oxford, UK. HUTCHEON I., SHEVALIER M. & ABERCROMBIE H. J. 1993. pH buffering by metastable mineral-fluid equilibria and evolution of carbon dioxide fugacity during burial diagenesis. Geochimica et Cosmochimica Acta, 57, 1017-1027. IRWIN, H., CURTIS, C. D. & COLEMAN, M. L. 1977. Isotopic evidence for source of diagenetic carbonate formed during the burial of organic rich sediments. Nature, 269,209-213. JAMES, A. T. 1990. Correlation of reservoired gases using the carbon isotopic compositions of wet gas components. American Association of Petroleum Geologists Bulletin, 74,1441-1458. JOHNSON, J. W., NITAO, J. J., STEEFAL, C. I. & KNAUSS, K. G. 2004. Reactive transport modeling of CO2 storage in saline aquifers to elucidate fundamental processes, trapping mechanisms and sequestration partitioning. In: BAINES, S. J. & WORDEN, R. H. (eds) Geological Storage of Carbon Dioxide, Geological Society, London, Special Publications, 233, 107-128. JOHNSON, R.E. 1983. Bravo Dome Carbon Dioxide area. In: FASSETT, J. E. (ed.) Oil and Gas Fields of the Four Corners Area, v.III. Four Corners Geological Society, p. 745-748. KRAUSKOPF, K. B. 1979. Introduction to Geochemistry. McGraw-Hill Kogakusha Ltd, Tokyo. LANG, W. B. 1959. The origin of some natural carbon dioxide gases. Journal of Geophysical Research, 64, 127-131. MACAULAY C. L, HASZELDINE R. S. & FALLICK A. E. 1992. Diagenetic pore waters stratified for at least 35 million years: Magnus oil field, North Sea. American Association of Petroleum Geologists Bulletin, 76, 1625-1634. MACAULAY, C. L, HASZELDINE, R. S & FALLICK, A. E. 1993. Distribution, chemistry, isotopic composition and origin of diagenetic carbonates: Magnus sandstone, North Sea. Journal of Sedimentary Petrology, 63, 33-43 MACAULAY, C. L, FALLICK, A. E., MCLAUGHLIN, O. M., HASZELDINE, R. S & PEARSON, M. J. 1998. The significance of 813C of carbonate cements in reservoir sandstones: a regional perspective from the Jurassic of the northern North Sea. In: MORAD, S. (ed.) Carbonate Cementation in Sandstone. Special Publication of the International Association of Sedimentologists, 26, 395-408. MARCHAND, A., SWENNEN, R., MACAULAY, C., HASZELDINE, S. & FALLICK, A. 1997. Diagenetic research on core samples of the Miller oil reservoir (North Sea). Final report - Katholieke Universiteit Leuven. MARCHAND, A. M. E, HASZELDINE, R. S., MACAULAY, C. L, SWENNEN, R., & FALLICK, A. E. 2000. Quartz cemen-
NATURAL ANALOGUES tation inhibited by crestal oil charge: Miller deep water sandstone, UK North Sea. Clay Minerals, 35, 201-210. MCLAUGHLIN, O. M., HASZELDINE, R. S., FALLICK, A. E. & ROGERS, G. 1994. The case of the missing clay. Aluminium loss and secondary porosity, South Brae oilfield, North Sea. Clay Minerals, 29, 651-663. PEARCE, J. M., HOLLOWAY, S., WACKER, H., NELIS, M. K., ROCHELLE, C. & BATEMAN, K. 1996. Natural occurrences as analogues for the geological disposal of carbon dioxide. Energy Conversion and Management, 37,1123-1128. PINTI, D. L. & MARTY, B. 1995. Noble gases in crude oils from the Paris Basin, France: implications for the origin of fluids and constraints on oil-water-gas interaction. Geochimica et Cosmochmica Acta, 59, 3389-3404. PLUMMER, L. N., PARKHURST, D. L. & WIGLEY, T. M. L. 1978. The kinetics of calcite dissolution in CO2-water systems at 5°C to 60°C and 0.0 to 1.0 atm CO2. American Journal of Science, 278,179-216. ROBERTS, J. W. & GODFREY, J. A. JR. 1994. Historu of Bravo Dome CO2 - discovery and early development. In: AHLEN, J., PETERSON, J. & A. L. (eds.) Geologic activities in the 90's. New Mexico Bureau of Mines and Mineral Resources, Bulletin 150, 9-12. ROCHELLE, C., CZERNICHOWSKI-LAURIOL, I. & MILODOWSKI, A. E. 2004. How chemical reactions can affect CO2 sequestration in geological formations. In: BAINES, S. J. & WORDEN, R. H. (eds) Geological Storage of Carbon Dioxide. Geological Society, London, Special Publications, 233, 87-106. ROOKSBY, S. K. 1991. The Miller Field, Blocks 16/7B, 16/8B, UK North Sea. In: ABBOTTS, I. L. (ed.) United Kingdom Oil and Gas Fields. 25 Years Commemorative Volume, Geological Society Memoir No. 14, 159-164. SHIPTON Z., EVANS, J. P., KIRSCHNER, KOLESAR, P. T., WILLIAMS, A. P. & HEATH, J. 2004. Analysis of CO2 leakage along faults from natural reservoirs in the Colorado Plateau, US. In: BAINES, S. J. & WORDEN, R. H. (eds) Geological Storage of Carbon Dioxide, Geological Society, London, Special Publications, 233,43-58. SMITH J. T. & EHRENBERG S. N. 1989. Correlation of carbon dioxide abundance with temperature in clastic hydrocarbon reservoirs: relationship to inorganic chemical equilibrium. Marine and Petroleum Geology, 6, 129-135. SMITH L. K. 1998. Carbonate cement dissolution during a cyclic CO2-enhanced oil recovery treatment. In: MORAD, S. (ed.) Carbonate cements in sandstones. International Association of Sedimentologists Special Publication, 26,483-499. SPOTL, C., MATTER, A. & BREVART, O. 1993. Diagenesis and pore water evolution of the Keuper reservoir, Paris Basin (France). Journal of Sedimentary Petrology, 63,909-928. SVERDRUP, H. U. 1990. The Kinetics of Base Cation Release due to Chemical Weathering. Lund University Press, 2450.
85
TISSOT, B. P. & WELTE, D. H. 1984. Petroleum Formation and Occurrence. 2nd edn. Springer Verlag, Berlin UNITED KINGDOM DEPARTMENT OF TRADE AND INDUSTRY (DTI). 2002. Our energy future - creating a low carbon economy. Energy White Paper. http://www. dti.gov.uk/energy/whitepaper UNITED STATES DEPARTMENT OF ENERGY (DOE). 2004. Reigning in CO2 Emissions, http://www.fe.doe.gov WALDERHAUG, O. 1994. Precipitation rates for quartz cement in sandstones determined by fluid inclusion microthermometry and temperature history modelling. Journal of Sedimentary Research, 64, 324-333. WARREN, E. A. & SMALLEY, P. C. 1994. North Sea formation water atlas. Geological Society, London, Memoir 15. WOLCOTT, J. M., MONGER, T. G., SASSEN, R. & CHINN, E. W. 1989. The effects of CO2 flooding on reservoir mineral properties. 1989 Society of Petroleum Engineers International Symposium, Houston, Texas, SPE Paper, 18467,101-109. WORDEN, R. H. 1996. Controls on the halogen content of sedimentary formation waters. Mineral Magazine, 60, 259-274. WORDEN, R. H. & BARCLAY, S. A. 2000. Internally-sourced quartz cement due to externally-derived CO2 in subarkosic sandstones, Northern North Sea, UKCS. Journal of Geochemical Exploration, 69-70, 649-653. WORDEN, R. H. & BARCLAY, S. A. 2003. The effect of oil emplacement on diagenetic clay mineralogy: Upper Jurassic Magnus sandstone, North Sea. In: WORDEN, R. H. & MORAD, S. (eds) Clay cement in sandstones. International Association of Sedimentologists, Special Publication, 34,453-469. Blackwells, Oxford, UK. WORDEN, R. H. & MATRAY, J. M. 1998. Carbonate cement in the Triassic Chaunoy Formation of the Paris Basin, France: distribution and effect on flow properties. In: MORAD, S. (ed) Carbonate cement in sandstone reservoirs. Special Publication of the International Association of Sedimentologists, 26,163-177. WORDEN, R. H. & SMITH, L. K. 2004. CO2 injection into oil fields for enhanced oil recovery (EOR): lessons for the geological sequestration of CO2 in the subsurface. In: BAINES, S. J. & WORDEN, R. H. (eds) Geological Storage of Carbon Dioxide, Geological Society, London, Special Publications, 233, 211-224. WORDEN, R. H., COLEMAN, M. L. & MATRAY, J. M. 1999. Basin scale evolution of formation waters: a diagenetic and formation water study of the Triassic Chaunoy formation, Paris Basin. Geochimica CosmochimicaActa, 63, 2513-2528 WYCHERLEY, H., FLEET, A., SHAW, H. & WILKINSON, J. 1997. Origins of large volumes of carbon dioxide accumulations in sedimentary basins. In: HENDRY, J., CAREY, P., PARNELL, J., RUFFELL, A. & WORDEN, R. H. (eds) Geofluids II '97 Extended abstracts, 264-268. Belfast, Northern Ireland. YARDLEY, B. W D. 1989. An introduction to metamorphic petrology. Longman New York, 248pp.
This page intentionally left blank
The impact of chemical reactions on CO2 storage in geological formations: a brief review C. A. ROCHELLE1,1. CZERNICHOWSKI-LAURIOL2 & A. E. MILODOWSKI1 1
British Geological Survey, Kingsley Dunham Centre, Keyworth, Nottingham NG12 5GG, UK (e-mail:
[email protected]) 2 BRGM, Avenue Claude Guillemin, BP 6009, 45060 Orleans cedex 2, France Abstract: The sequestration of CO2 in the deep geosphere is one potential method for reducing anthropogenic emissions to the atmosphere without a drastic change in our energy-producing technologies. Immediately after injection, the CO2 will be stored as a free phase within the host rock. Over time it will dissolve into the local formation water and initiate a variety of geochemical reactions. Some of these reactions could be beneficial, helping to chemically contain or 'trap' the CO2 as dissolved species and by the formation of new carbonate minerals; others may be deleterious, and actually aid the migration of CO2. It will be important to understand the overall impact of these competing processes. However, these processes will also be dependent upon the structure, mineralogy and hydrogeology of the specific lithologies concerned and the chemical stability of the engineered features (principally, the cement and steel components in the well completions). Therefore, individual storage operations will have to take account of local geological, fluid chemical and hydrogeological conditions. The aim of this paper is to review some of the possible chemical reactions that might occur once CO2 is injected underground, and to highlight their possible impacts on long-term CO2 storage.
If sequestration of CO2 is to be a practicable largescale disposal method, the CO2 must remain safely underground, and not return to the atmosphere within relatively short timescales (e.g. thousands of years), so that natural buffering processes (e.g. oceanic and forestry sinks) have sufficient time to reduce global atmospheric CO2 levels to environmentally acceptable levels. Indeed, acceptable performance will need to be demonstrated in order to satisfy operational, regulatory and public acceptance criteria. The track record of CO2-assisted enhanced oil recovery (EOR) operations and purposedesigned underground storage of natural gas shows that underground storage can be practicable and leakage minimized over 'industrial' time periods (e.g. tens of years). However, there is much less information for longer-term processes, and these must be understood, especially as CO2 is more chemically reactive than methane. The injection of a relatively reactive substance such as CO2 into the deep subsurface will result in chemical disequilibria and the initiation of various chemical reactions. It is important to understand the direction, rate and magnitude of such reactions both in terms of their impact upon the ability of a host formation to contain the injected CO2 safely, and in terms of the longevity of CO2 containment (e.g. Rochelle et al. 1999). Some reactions, such as the precipitation of CO2 in secondary carbonate minerals, may be beneficial and aid containment. However, other reactions may result in mineral dis-
solution - facilitating the formation of migration pathways and so act to reduce containment. The aim of this paper is to highlight some of these processes, and to illustrate their possible impact on long-term CO2 storage. It is hoped that this will aid other researchers (be they computer modellers, experimentalists, aqueous geochemists or mineralogists) in identifying key geochemical processes occurring within the systems that they are studying. A typical underground storage operation might involve liquid CO2 being pumped from surface installations into a deep host formation. On its descent down the borehole, pressure and temperature will increase past the point where CO2 becomes supercritical (approximately 31 °C, 74 bar); it is this phase that will be injected into the host formation (Fig. 1). The density of supercritical CO2 varies depending on pressure and temperature, but being less dense than water, it will rise until it reaches an overlying aquiclude where it will be physically trapped (Bachu et al 1994) in the form of a 'bubble'. During this process, the CO2 will be able to react with the host formation and local pore-water, and also the overlying seal (e.g. aquiclude). The extent of such reaction will be dependent upon factors such as the composition of the pore-water, the composition of the rocks and minerals it encounters, as well as the in-situ pressure and temperature. One of the first reactions of the CO2 will be with the formation water, namely its dissolution. This reaction is important because it can form a relatively
From: BAINES, S. J. & WORDEN, R. H. (eds) 2004. Geological Storage of Carbon Dioxide. Geological Society, London, Special Publications, 233, 87-106.0305-8719/047$ 15.00 © The Geological Society of London 2004.
88
C. A. ROCHELLE ETAL.
Fig. 1. CO2 phase diagram (modified after Atkins 1982).
rapid and large sink for CO2. Once dissolved, CO2 migration will no longer be driven by buoyancy (leading to relatively fast movement), but by relatively slow regional-scale deep ground water flow patterns (Gunter et al 1997). Bachu et al. (1994) called this process 'hydrodynamic trapping', and in their studies of the Alberta Basin, Canada, suggested that diffusion and dispersion would be the main transport mechanisms for dissolved CO2, leading to an increased sweep efficiency and hence greater storage potential for CO2. Although hydrodynamic trapping could be a relatively fast process and a good mechanism for retarding CO2 migration, subsequent pressure reduction might then liberate dissolved CO2 as a free phase. Longer-term 'mineral trapping' (Bachu et al 1994) could be achieved by precipitation reactions, and in particular those locking up CO2 in the form of carbonate minerals. Although mineral trapping reactions are likely to be slower than CO2-water reactions, they might provide a more permanent sink for CO2 in the form of carbonate minerals. Indeed, over very long timescales, mineral trapping may be the key trapping mechanism for CO2 (e.g. Baker et al 1995; Gunter et al 1997, 2004). The advantages of this trapping mechanism are that many carbonate minerals are stable, and that the CO2 would effectively be immobilized for geologically-important timescales. However, the possibility that rapid reactions may occur, producing metastable minerals in some parts of the disposal system, cannot be overlooked. The types and magnitude of reactions that will occur depend upon a variety of factors; for example, the mineralogical composition of the host rock, formation water chemistry, in-situ pressure and temperature, groundwater flow rates, and the relative rates of the dominant reactions. Unfortunately, not all reactions may be as beneficial as those detailed above. It is possible that dissolution/precipitation reactions might alter porosity
and permeability in such a way as to either hinder the actual injection of CO2, or to aid its migration out of the storage volume. For example, excessive rapid calcite precipitation may block flow pathways needed to maintain high injection rates. Conversely, dissolution of minerals in the caprock might result in the formation of flow pathways that might aid CO2 migration. The same could be true for borehole completions, where relatively rapid reactions might be expected between (acidic) CO2 and (alkaline) cement, and where steel borehole casings might corrode. Although borehole cementing techniques are well-established, one of the most neglected aspects of borehole completion is an understanding of the long-term effect of the geological environment on the durability and integrity of the borehole cement (Robins & Milodowski 1986). The design life expectancy of an oil well does not usually exceed 25 years, but borehole sealing for CO2 sequestration requires a considerably longer lifetime, and premature failure of the well completion system may result in cross-formational leakage or even direct release of CO2 to the surface. The following sections consider different parts of the CO2 containment system, and highlight some of the possible reactions that may occur in each of them. Figure 2 gives a simple schematic representation of these and their relationship to each other. The relative position of the following sections is broadly in line with the order in which the injected CO2 will encounter the different parts of the system. However, as the reactions will proceed at different rates, many of them may occur at the same time, but in different parts of the storage system.
Reactions within the host formation Reactions between supercritical CO2 and aqueous fluids Injection of industrial quantities of CO2 into a host rock will probably result in a plume of CO2, which will ascend to form a supercritical CO2 'bubble' overlain by the caprock (e.g. Johnson et al 2001, 2004; Fig. 2). The injected CO2 is likely to be relatively dry as this reduces its corrosivity towards infrastructure (steel tubing, pumps, etc.) (e.g. EFCP 1994). In colder environments this also reduces the possibility of CO2 hydrates forming within pipework. Initial reactions are therefore likely to involve dissolution of CO2 into water, and vice versa (e.g. King et al 1992). The reaction sequence could include the following:
although significant CO32~ is only likely to form in highly alkaline environments (e.g. cement pore-
IMPACT OF CHEMICAL REACTIONS ON CO2 STORAGE
89
Fig. 2. Simple schematic diagram of different parts of the storage system highlighting CO2 injection and interaction with formation water, interaction with host rock, interaction with caprock and interaction with borehole completions.
water). This series of linked reversible reactions is influenced by in-situ temperature and pressure, which control the solubility of CO2. Reactions involving the generation of H + are also dependent upon the ability of the host aquifer to buffer pH. For example, additional reactions that consume H + ions will tend to drive reaction (1) to the right, and so cause more CO2 to dissolve. Previous work with deionized water (van Eldik & Palmer 1982), albeit at relatively low pressures, has shown that about 99% of the total dissolved CO2 is in the form of the dissolved gas, CO2/aq), rather than as true carbonic acid, H2C03°. Dissolution of CO2 can be a fast process where CO2 and formation water have good contact and are well mixed. Indeed, studies at pressures below 90 bar indicate that equilibration can be achieved in under 24 hours (Ellis & Golding 1963; Stewart & Munjal 1970). Other studies at 80 °C and 200bar pressure (Czernichowski-Lauriol et al. I996a) show equilibration in distilled de-ionized water, or 0.55M NaCl solution, in about 5 hours, and give an initial (maximum) dissolution rate in the order of 7 X 10~5 molcm~ 2 s~ 1 . This rate is far faster than most fluid-mineral reactions (for quartz and K-feldspar this is in the order 10~ 15 -10~ 15 molcm~ 2 s~'; Knauss & Wolery 1986, 1988; Bevan & Savage 1989), and therefore for many geochemical modelling applications, the simplification of assuming CO2-water equilibria is reasonable given the relatively fast reaction kinetics. However, there are other factors that will act to reduce the rate of CO2 dissolution into the formation water. The speed of fluid mixing within a porous medium will be far slower than for the free phases, and it is very likely that there will be a decreasing CO2 concentration gradient over several metres away from any free supercritical CO2 phase
Fig. 3. Solubility of CO2 in pure water (based upon data Wiebe & Gaddy (1940) and Wiebe (1941).
(e.g. Johnson et al 2001, 2004). Once such a gradient has been established, dissolution of CO2 into the formation could be reduced greatly. Drawdown of CO2 into the pore-water is then likely to be driven by factors such as diffusional processes, buoyancydriven fluid mixing, viscous fingering and mixing driven by regional-scale deep groundwater flow. For supercritical CO2 at temperatures between 37 and 100 °C and pressures below 300bars, its solubility in water decreases with increasing temperature, but increases with increasing pressure (Fig. 3). This relatively simple relationship is likely to be applicable to purpose-designed CO2 storage operations, as financial constraints are likely to limit injection to shallower depths where temperatures are lower. However, certain CO2-EOR (enhanced oil recovery) operations might involve higher temperature regimes, and at higher temperatures and pressures CO2 solubility actually increases with increasing temperature (e.g. Ellis & Golding 1963; Fig. 3). A further factor controlling CO2 solubility is the specific composition of the formation water. In general, there is a trend of decreasing CO2 solubility with increasing ionic strength, irrespective of the exact composition of the aqueous phase (Enick & Klara 1990, 1992) (Fig. 4). This 'salting-out' effect (Garrels & Christ 1965) will tend to reduce the effectiveness of hydrodynamic trapping in saline formation waters. Relationships such as those described above can be used to predict CO2 solubility over a range of pressure, temperature and salinity. For example work in Alberta (Guntem al 1993,1997; Perkins & Gunter 1995; Hitchon 1996; Law & Bachu 1996) focused on CO2 migration studies that might occur
90
C. A. ROCHELLE ETAL.
influenced by chemical interaction with the host formation, and this will exert a control on CO2 solubility. Consequently, the rates of mineral reaction (and related factors such as mineral abundance/distribution and reactive surface area) will also have an impact on CO0 solubility.
Reactions between dissolved CO2 and aquifer minerals
Fig. 4. CO2 solubility changes with salinity relative to that in pure water (based upon data at 20-250 °C, 3-85MPa, Enick and Klara 1990).
during the injection of CO2 into the Cretaceous Glauconite Sandstone aquifer within the Alberta Basin in western Canada. Law & Bachu (1996) reported that a significant amount of CO2 would dissolve into the formation water (up to 25% by weight), with the rest remaining as a supercritical phase. The CO2 (as either a dissolved or supercritical phase) was also predicted to migrate less than 5km from the injection point over a 30-year time period. As a further example of how important dissolved CO2 can be in the overall CO2 budget, consider work by the Lawrence Livermore National Laboratory (Johnson et al 2001, 2004) which focused on coupled modelling simulations of CO2 injection into the Miocene Utsira formation at Sleipner, North Sea (IEA 1998). The results from this study indicate a CO2 solubility in the order of 4-5 weight %, and that over ten years of CO2 injection at least 15% by weight (results being dependent on the physical model used for the Utsira Formation) of the total injected CO2 would dissolve into the formation water. Over this time period the CO2 was predicted to migrate less that 300m from the injection point, though a perfectly horizontal top to the reservoir was assumed. In summary, the degree to which the CO2 reacts with the formation water (dominated by its solubility) will vary according to factors such as; pressure, temperature, fluid chemistry, and degree of CO2-water mixing or contact surface with CO2 and water. Such factors will be site specific, and thus any detailed studies of individual storage operations will require knowledge of local in-situ conditions prior to CO2 injection in order to quantify storage potential. However, formation-water chemistry will also be
General considerations The practicalities of injecting industrial quantities of CO2 deep underground will require a host formation with good permeability and high porosity. As a consequence, some of the most likely potential host rocks will be either carbonate, or sandstone formations. However, the chemical response, and rate of reaction, of each of these will be very different due to their different mineralogical compositions. This is important in terms of direct mineral dissolution/ precipitation, and in terms of pH buffering. As CO2 solubility decreases with decreasing pH, more dissolved CO2 can be 'trapped' in an aquifer that can maintain (buffer) the pH of the formation water compared to one where formation water pH decreases. Previous studies (Perkins & Gunter 1995; Czernichowski-Lauriol et al. 1996a, b; Gunter et al 1993,1997,2000; Law & Bachu 1996) show that sequestration of CO2 into siliciclastic rock types (e.g. sandstones) is preferable to that in carbonate rock types (e.g. chalk) because of the greater potential for pH buffering, solution of CO2, and net carbonate mineral precipitation. Examples of reactions that could immobilize CO2 could involve feldspar minerals. For Ca-rich feldspar (Gunter et al 1997) trapping could be in a mineral phase (calcite): anorthite
(2) calcite
kaolinite
For K-rich feldspar in saline solutions Johnson et al (2001) have postulated the precipitation of a different carbonate mineral, dawsonite: K-feldspar
(3) dawsonite quartz/chalcedony /cristobalite However, other silicate mineral reactions might trap CO2 as a dissolved species such as bicarbonate (G\aA&ietal 1997):
IMPACT OF CHEMICAL REACTIONS ON CO, STORAGE
91
albite
(4) quartz which is a similar trapping species to that involved when minerals within a carbonate host formation dissolve (Gunteref al 1993):
(5) calcite Modelling investigations By way of a demonstration of such differing responses, Czernichowski-Lauriol et al. (19960) report a modelling study considering two typical host rock systems. These were chosen because of their broad applicability to possible storage systems. Their use is for illustrative purposes only, and is not meant to address the suitability (or not) of each formation for actual CO2 storage operations. The first system was based upon the Dogger aquifer in the Paris basin (carbonate formation), and the second a typical sandstone (without hydrocarbons) from the North Sea. The former was represented by calcite and disordered dolomite, a salinity equivalent to 0.4M NaCl, a temperature of 78 °C, and a depth of 1600m (assumed hydrostatic head of 160bar). The latter was represented by quartz, K-feldspar, Na-feldspar, kaolinite, illite and calcite, a salinity equivalent to 0.7 M NaCl, a temperature of 98 °C, and a depth of 2500m (assumed hydrostatic head of 250bar). Calculations were performed with the simulator 'CO2ROCK', and assumed a closed system with initial thermodynamic equilibrium between the formation fluid and the selected minerals. CO2 was progressively added into this system, and all the resulting homogeneous and heterogeneous geochemical reactions were assumed to occur at thermodynamic equilibrium. Thermodynamic data were taken from the DataO.com.R10 database of the EQ3/6 geochemical software package with the aqueous solution model based on an extended Debye-Huckel formalism, similar to that in EQ3/6 (Wolery 1992; Wolery & Daveler 1992). The activity coefficients of charged species were calculated using the B-dot equation, those of dissolved CO2 and neutral non-polar species using an expression after Drummond (1981), and those of neutral polar species set to unity. The activity of water was calculated by an expression which was quasi-consistent with the Bdot equation. Changes in the amount of solvent water throughout the simulation were accounted for in the water model through the mass balance equations for each chemical element (O and H in this case). Laws of mass action expressed thermodynamic equilib-
Fig. 5. Variation in CO2 partial pressure (PCO2) in bars and pH, with the number of moles of CO2 added (NCO2) to a carbonate and sandstone formation, normalized to 1 kg of water (redrawn after Czernichowski-Lauriol et al 19960). Carbonate formation from the Paris Basin, 78 °C, 160bar, 0.4 M NaCl. Sandstone formation from the North Sea, 98 °C, 250bar, 0.7M NaCl.
rium between all aqueous species. The fugacity of CO2 was computed from the partial pressure, using the fugacity coefficients given by Duan et al. (1992). Further details of the simulator are given in Czernichowski-Lauriol et al (19960). Results from some of the simulations are presented in Figure 5 and show the final geochemical state of the reservoir depending on the amount of CO2 added. Assuming that gases other than CO2 are negligible, the maximum amount of CO2 that will be taken up by the formation water and mineral reactions (i.e. with no 'free' CO2 phase) will be reached when the partial pressure of CO2 equals the formation pressure (the water vapour pressure being negligible). This was predicted to be l.Hmolkg" 1 water for the carbonate formation (equivalent to approximately 50g CO2 per kg of water), and 2.25molkg~1 water for the sandstone formation (equivalent to approximately lOOg CO2 per kg of water). Without mineral reactions the corresponding CO2 solubility under the same conditions was predicted to be 1.12molkg"' water (equivalent to approximately 49g CO2 per kg of water) and l^lmolkg" 1 water (equivalent to 58 g CO2 per kg of water). If both of these cases are studied at 200bar pressure (i.e. the effect of CO2 solubility variation as a function of pressure is removed) the maximum amount of CO2 trapped would have been 1.42molkg"1 water (equivalent to approximately 62 g CO2 per kg of water) for the carbonate aquifer, and l.S^molkg"1 water (equivalent to approximately 83 g CO2 per kg of water) for the sandstone formation, taking mineral reaction into account in both cases.
92
C. A. ROCHELLE E7AL.
Therefore, even though the sandstone formation was slightly hotter and more saline than the carbonate formation (CO2 being less soluble at this higher temperature and salinity) it was still able to take up more CO2 compared to the carbonate formation. A main controlling reason for this is that the silicate minerals are better at buffering pH at a higher value than the carbonate system. As a result, the dissolved CO2 in the sandstone formation was predicted to be present as aqueous CO2, bicarbonate ions, and bicarbonate complexes. Conversely, for the carbonate formation, the predicted pH decrease caused dissolution of calcite and an increase of Ca2+ in solution, but this time virtually all of the CO2 was predicted to be present as aqueous CO2. Modelling such as that described above, concurs with previous studies (e.g. Gunter et al. 1993; Hitchon 1996). These show that from a geochemical standpoint, sandstone formations are likely to be able to store more CO2 compared to pure carbonate formations, due to reactions with basic silicate or aluminosilicate minerals. However, this does not mean that carbonate formations are unsuited to CO2 storage, as they may safely contain CO2 as a free phase and as dissolved CO2. Regardless of the type of formation being considered, CO2 storage potential will vary depending on the specific mineralogy of the lithology concerned, and so will be sitespecific. As a consequence, a good understanding of baseline geochemistry (i.e. the geochemistry of the host formation) prior to CO2 injection is important in order to make realistic predictions of potential CO2 storage. The North Sea sandstone example described above can be further used to investigate how the front of dissolved CO2 and mineral dissolution/precipitation processes would propagate within an open reservoir during CO2 injection. However, this time the coupled reaction-transport code 'CATCO2' was used. This consists of sequential calls to the transport and chemistry subroutines for each time step in each grid block (Fabriol & Czernichowski-Lauriol 1992; Kervevan et al 1994). The transport module was based on a random walk algorithm and accounted for the movement of dissolved species by advection, dispersion and diffusion. The chemistry module was the CO2ROCK simulator. Further details of the simulator are given in CzernichowskiLauriol et al. (19960). Two differing cases were run: Case A There was an excess of all key minerals in the system (i.e. none of them were exhausted during the simulation). These were: quartz, Kfeldspar, albite, kaolinite, illite and calcite. CaseB Realistic abundances of minerals were used (86.2% quartz, 5.9% K-feldspar, 0.6% albite, 3.5% kaolinite, 1.3% illite and 0.5%
calcite; the 2.0% of more minor minerals were not considered in this modelled system). There were some limitations of the coupled modelling. First, mineral reactions were controlled by thermodynamic equilibrium alone (i.e. reaction kinetics were not considered). Second, the CO2 injection rate was low compared to regional flow, so that all the injected CO2 would be able to dissolve 'instantaneously' in the formation water and not disturb the regional flow. Typical values for porosity (14.1%) and regional flow velocity (1.1 ma" 1 ) were chosen. A 150m portion of the sandstone formation was simulated and modelled as 20 cells 7.5m long and 4.5m wide. CO2 was injected at 2.76kgday"1 (maximum rate given the regional flow velocity and that all the CO2 had to dissolve in the formation water), the dispersivity was set to 1m, and a time step chosen (350.7 days) so that the Courant number was unity. For Case A, mineral reactions act to buffer pH at approximately 6 and the partial pressure of CO2 at approximately 130 bar. The latter is below the formation pressure of 250bar, so in theory, CO2 injection rates could be increased without the formation of a supercritical CO2 phase. However, the calculations were performed assuming fluid-mineral equilibrium, rather than with realistic reaction kinetics and surface areas. Incorporation of these into the model would act to slow down reactions and modify the efficiency and reaction time of the buffer system, and hence be a major control on the uptake of CO2. Based upon thermodynamic considerations alone, reaction for this mineralogical assemblage includes dissolution of albite and illite, and precipitation of K-feldspar, kaolinite, quartz and calcite. The resulting overall change in mineral volume leads to a porosity increase close to the injection point, from 14.1% to 14.3% after 38 years. Although small, this could have advantages for the injection operation, so long as the formation permeability did not decrease. Results of the modelling for Case B are more complex. Significant differences were observed compared to Case A in that first albite, followed by illite completely dissolved (Fig. 6). As a consequence, the buffering provided by the formation was modified and was far less efficient, and resulted in a 'stationary state' pH of about 5, with a CO2 partial pressure of about 230bar (only slightly below reservoir pressure). Precipitation of some quartz and calcite was also predicted. As the simulation progressed, reaction 'fronts' moved away from the injection point, with the zone of albite removal ahead of that of illite removal. The region between these two fronts was a complex transition zone between a forward zone of kaolinite/K-feldspar dissolution and illite precipitation, and a back zone with progressive removal of illite and consequent precipitation of kaolinite/K-feldspar. The overall trend in
IMPACT OF CHEMICAL REACTIONS ON CO, STORAGE
93
Fig. 6. Changes in the numbers of moles of albite and illite during the simulated injection of CO2 into a typical sandstone formation. Each cell is 7.5m long, so that the overall total simulated length is 150m. Note the progression of reaction fronts with increasing time (redrawn after Czernichowski-Lauriol et al. I996a).
porosity was a slight reduction over the entire flooded zone, from 14.10% to 14.01% after 48 years. Therefore, in this case, mineral reactions barely enhanced the CO2 storage capacity of the reservoir once the two reaction zones had passed. However, as for Case A, incorporation of realistic reaction kinetics would be important as they could markedly alter the relative position of the predicted reaction zones. Although theoretical, the Case A and Case B calculations as presented illustrate that the potential for CO2 trapping by chemical reactions is dependent on the precise nature and proportions of the mineralogical assemblage of the aquifer. Previous modelling studies (Gunter et al. 1993, 1997, 2000; Perkins & Gunter 1995; Hitchon 1996; Law & Bachu 1996) simulated reactions that might occur during the injection of CO2 into the Cretaceous Glauconite Sandstone aquifer within the Alberta Basin in western Canada. In summary, the modelling predicted that albite, biotite (as a proxy for glauconite) and K-feldspar would be dissolved, whereas calcite, dolomite, kaolinite, muscovite, quartz and siderite would be precipitated. The major CO2-trapping reactions were the precipitation of calcite and siderite, and the formation of aqueous bicarbonate ions. Although there is some uncertainty in the rates of reaction, it appears that CO2-trapping reactions under likely in-situ conditions (54 °C and 260bar) would take hundreds of years to complete. However, as the residence time of fluids within deep aquifers in the Alberta Basin is measured in timescales that are 2-3 orders of magnitude larger
than this, there would be sufficient time for these mineral-trapping reactions to occur. In their modelling study of the Utsira formation at Sleipner, Johnson et al (2001, 2004) predicted the precipitation of dawsonite (reaction (3) above) and calcite-group Ca/Mg/Fe carbonate minerals. This study simulated a 20-year time period, 10 years of CO2 injection followed by 10 years without injection. Within the Utsira sand, dawsonite is predicted to form mainly within the region of the CO2 plume, whereas the calcite-group Ca/Mg/Fe carbonates are predicted to form at the periphery of the plume. The mechanism for forming dawsonite was dissolution of K-feldspar (reaction (3) above) whereas Ca and Mg were sourced from dissolution of plagioclase feldspar and phlogopite respectively. Although the simulations detailed above are, by necessity, relatively simple representations of reality, they are useful in demonstrating that chemical reactions between CO2, formation water and formation minerals can have an impact on CO2 storage and are highly site-specific. Such simulations also show the benefit of geochemical modelling in helping to scope potential water-rock-CO2 reactions, and as a result help constrain the most appropriate conditions for storage operations. The development of more complex models will be necessary to investigate a wider range of processes, and the inclusion of a realistic treatment of reaction kinetics, and possibly more solid phases, would be useful future developments. However, acceptable performance of a CO2 storage operation will need to be demonstrated in
94
C. A. ROCHELLE ETAL
order to satisfy operational, regulatory and public acceptance criteria. This is likely to involve reducing uncertainties to a minimum, and comparing predictive calculations against real data. The value of such theoretical modelling will therefore be greatly increased if its output can be tested against direct observations, from laboratory experiments (shortterm), demonstration projects (medium-term), or natural analogues (very long-term). By and large, there is currently relatively limited geochemical information from direct observational studies of CO2-water-rock interaction. That being said, increasing amounts of information are starting to come from demonstration projects, which include those currently being undertaken at Sleipner, North Sea (Baklid et al 1996; IEA 1998; Arts et al 2004; Zweigel etal. 2004) and at Weyburn, Saskatchewan, Canada (IEA 2001; Malik & Islam 2000; Moberg 2001). More information is available from laboratory studies, though this is limited to investigations of shorter-term processes (see the following section). Information on longer-term processes using natural analogues is somewhat limited at the moment, though projects are now underway to address this (e.g. Pearce et al. 2004, see later). Observations from laboratory experiments There have been a variety of experimental studies that have examined the effect of CO2 injection upon host-rock mineralogy, though most have been connected to CO2-EOR rather than CO2 storage. The studies have considered both sandstones and carbonates. Gunter et al. (1997) and Hitchon (1996) reported experiments that simulated reactions that might occur during the injection of CO2 into the Glauconite Sandstone aquifer within the Alberta Basin in western Canada. Samples of crushed Glauconite Sandstone were reacted with CO2-saturated synthetic formation water in 'batch' equipment for 1 month. A temperature of 105 °C and pressure of 90bar were used in the experiments. This temperature is above the in-situ temperature of the Glauconite Sandstone (54 °C), but was used to enhance the rate of reaction so that any significant changes occurring could be observed over a reasonable timescale. To enhance CO2-trapping reactions, some of the experiments were 'spiked' with selected pure minerals that were already present within the sandstone. Mineralogical analysis of the reacted solids revealed no new secondary minerals, or changes in the primary minerals. However, the fluid chemistry of the synthetic formation water did change, so at least minor reaction had occurred. Increased alkalinity indicated that fluid-mineral reactions were proceeding slowly, possibly controlled by the relatively sluggish reaction kinetics of silicate mineral dissolution at the already elevated experimental temperature.
Other studies (Czernichowski-Lauriol et al 1996^) used a variety of rock types from UK onshore material from the Lower Triassic Sherwood Sandstone Group, which is a sequence of continental red beds that forms important aquifers and petroleum reservoirs both onshore in the UK and offshore beneath the North Sea. These were used in 'batch' and 'flow-through' experiments conducted at 80°C and 200bar pressure, conditions similar to those in some North Sea aquifers. Most of the batch experiments used small monoliths of sandstones partially submerged in 0.55 M NaCl solution (approximate salinity of seawater), with the remaining volume filled with supercritical CO2. Flow-through experiments used cores of sandstone (15cm long and 5cm diameter) flushed with NaCl solution previously equilibrated with supercritical CO2. Batch experiments had durations of up to 8 months, flow experiments only two months. Reaction of all sandstone samples resulted in K-feldspar corrosion, though the nature of this differed slightly between experiments. In particular, corrosion tended to be concentrated on K-feldspar grains which had previously undergone dissolution during diagenesis. A secondary Na-KCa-aluminium silicate precipitate (possible clay) was identified, tentatively associated with Kfeldspar dissolution. The sandstones also contained varying amounts of dolomite cement, which showed corrosion during the experiments. Small amounts of calcite formed during the experiments in the form of a minor coating on the external surface of sandstone blocks in the batch experiments, particularly at the CO2-water interface. Secondary calcite developed only within a sandstone sample rich in anhydrite cement, suggesting that these samples can favour mineral trapping. Clay minerals originally present within the sandstones showed reaction with the saline fluid, with authigenic illite-smectite clays losing Ca from the exchangeable sites of the smectite interlayers and gaining Na from solution. Although such reaction might be due to an overly simplistic pore fluid being used in the experiments, it is noteworthy that Na-smectites have greater swelling characteristics compared to Ca-smectites, and if this process occurred extensively, pore throats might become blocked, with a consequent reduction in permeability. Furthermore, release of calcium into solution might contribute to the precipitation of calcite. More investigations are needed to study the possible role of ion-exchange reactions on mineral trapping and permeability reduction. Shiraki & Dunn (2000) reported experiments that simulated injection of CO2 into the Tensleep Sandstone which forms reservoirs in northern Wyoming, USA. Cores of this dolomite- and anhydrite-cemented sandstone were flushed with CO0-saturated synthetic formation water for approximately 1 week. A temperature of 80 °C and pressure
IMPACT OF CHEMICAL REACTIONS ON CO? STORAGE
of 166bar were used in the experiments. Mineral reactions were observed, and included dissolution of dolomite and K-feldspar, with the later initiating kaolinite precipitation. These observations suggest a buffering of pH, and that reactions such as the following were taking place:
K-feldspar
(6) kaolinite
dolomite
(7) The behaviour of anhydrite cement varied between the experiments, either dissolving or precipitating. Although dissolution of authigenic carbonate cement was observed, permeability actually reduced, probably due to the formation of kaolinite in pore throats. Bowker & Shuler (1991) reported experiments that simulated injection of CO2 into the PennsylvanianPermian Weber Sandstone which forms a reservoir at Rangely field, Colorado, USA. Cores of this carbonate-cemented sandstone were flushed with carbonated brine for approximately 1 week. Experiments were run at room temperature and pressures of approximately 138 and 200bar. Observed mineral reactions included dissolution of ferroan dolomite which was also reflected in changes in fluid chemistry. The permeability of the sandstone samples remained largely unchanged throughout the experiments even though carbonate dissolution was occurring. Migration of clays and consequent reduction in permeability was thought to be the mechanism that offset any permeability increases due to carbonate dissolution. The possible mobilization of clay phases is noteworthy. Although these might not take part in chemical reactions with CO2, their movement may impact on the mobility of CO2, and hence on the injection process. Standard geochemical models do not account for such physical processes, which could lead to further uncertainties in predictive calculations. Ross et al (1981) undertook experiments that flooded cores calcite- and dolomite-cemented sandstones and oolitic limestone with CO2-saturated brine. Experiments were run at 20 and 80 °C and pressures of approximately 70-170bar. Permeability increases were observed by monitoring dissolved Ca, and also by the direct measurement of subsamples of the reacted core. Dissolution of the cement appeared non-uniform, with removal of constrictions in the larger pores. Dissolution was also concentrated at the inlet end of the sample, with an
95
apparent permeability front moving through the core as the experiment progressed. The observation of a possible reaction front (as indicated by permeability change) within a small-scale laboratory experiment is noteworthy. It showed that geochemical processes can generate these relatively quickly, at least over distances measurable in centimetres. This serves to increase confidence in modelling activities that predict them over distances measurable in tens or hundreds of metres (e.g. Czernichowski-Lauriol et al 1996a; Fig. 6). Although Ross et al (1981)postulated the migration of particulate matter into pore throats as a mechanism for reducing injectivity, this process was not observed. Omole & Osoba (1983) conducted experiments reacting liquid (as opposed to supercritical) CO2 with 0.1N KC1 solution and cores of dolomite. Experiments were run at approximately 27 °C and pressures of approximately 70-170bar. The degree of permeability increase was directly related to higher CO2 pressures. However, it was also linked to experiments with smaller pressure differences across the cores. These permeability increases were presumed to be due to dissolution of carbonate. However, for experiments with a higher differential pressure across the core, permeability decreases were observed. These were related to subsequent reprecipitation of carbonate phases. Omole & Osoba (1983) concluded that carbonate mineral dissolution would occur in the vicinity of injection wells, but that these might be re-precipitated as fluids move down a pressure gradient. This indicates the sensivity of reaction (5) (via reaction (1)) to CO2 partial pressure - higher pressures moving the reaction to the right, and lower pressures moving it to the left. Observations from natural systems Natural accumulations of CO2 occur in various parts of the world, but some of the best described are within the USA. Several of these are exploited commercially, with the CO2 being used in enhanced oil recovery operations. One such site is the Bravo Dome CO2 field in northeastern New Mexico, which is contained within the Permian Wolfcampian-Leonardian Tubb Sandstone (Broadhead 1989; Pearce et al. 1996), and was developed in the early 1980s to supply CO2 to the west Texas oilfields. This field is thought originally to have contained about 12 trillion cubic feet (3.4 X 10um3) of CO2 (Amoco P.C. 1990). The host Tubb Sandstone is sealed by the 6m thick Cimarron Anhydrite which, although flexed and folded, still retains good sealing characteristics across the field. The CO2 is thought to have been generated from the degassing of magma, and to have entered the reservoir less than 50000 years ago. The Tubb Sandstone has been subjected to dissolution of early anhydrite, dolomite and detrital plagioclase (Nelis 1994; Pearce et al 1996), which has been
96
C. A. ROCHELLE ETAL
attributed to the early introduction of CO2-rich formation waters possibly via early Pennsylvanian and/or Tertiary faults (Nelis 1994). K-feldspar grains are corroded, and there are traces of highly corroded gypsum cement. Kaolinite, zeolites and gibbsite occur as late-stage cementing minerals (Nelis 1994; Pearce et al 1996), and are closely associated with Tertiary faults, through which the CO2 is thought to have migrated. Although zeolites and gibbsite have not been observed as reaction products in laboratory experiments this may be due to the shorter duration of the experiments compared to reaction timescales in natural systems. There are other natural systems that show evidence of significant reaction with CO2-rich fluids in the past. For example in a study of Upper Jurassic subarkosic sandstones of the Magnus oil field, Worden & Barclay (2000) identified CO2-mineral reactions that occurred immediately prior to, or during, the early stages of oil filling. The organically-derived CO2 was from a source outside the sandstone, and initiated K-feldspar dissolution and quartz, kaolinite and ankerite (Mg-Ca-Fe carbonate) precipitation. The reaction led to bulk loss of K, probably into the surrounding mudrocks. This sequence of reactions has many similarities with those occurring within the Bravo Dome CO2 field. Another example of CO2-mineral reactions is that occurring in the Bowden-Gunnedah-Sydney Basin system of eastern Australia (Baker et al. 1995). Here dawsonite is found as a widespread cavity-filling cement of Permo-Triassic sandstones, which can make up 13.5% of the bulk sandstone rock. Its formation is thought to be due to widespread influx of magmatic CO2 during the Tertiary. This caused dissolution of aluminosilicate minerals (principally plagioclase feldspar, but also possibly K-feldspar and illite) and precipitation of dawsonite from the fairly alkaline sodium-bicarbonate-rich pore-waters, probably at 30-75 °C. Other authigenic minerals within dawsonite-rich formations in the basin include; kaolinite, quartz, norstrandite (A1(OH)3), alumohydrocalcite (CaAl2(CO3)2(OH)4.3H2O) and various Ca-Mg-Fe carbonates. Summary In summary therefore, there is a certain degree of agreement between predicted mineral reactions and those observed in experiments and natural systems. There appears to be reasonably good evidence for the CO2-enhanced dissolution of feldspars followed by precipitation of kaolinite, other clay minerals, a silica phase, and sometimes Ca-Mg-Fe carbonates. For rocks with a saline pore fluid, dawsonite can also be an important secondary phase. However, for many carbonate-rich rocks, mineral reactions appear to be dominated by dissolution reactions rather than precipitation, though fluid movement along a
decreasing pressure gradient (i.e. away from an injection well) would appear to facilitate carbonate precipitation away from the site of dissolution. One process that is apparently not presently covered by standard geochemical modelling packages, is the physical mobilization of clay minerals as a result of dissolution reactions or injection pressure. Although this process may not participate directly in the geochemical trapping of CO2, it might impact on the mobility of CO2 and the efficiency of the injection process.
Reactions with the caprock General considerations The long-term containment of CO2 in the deep subsurface will be crucially dependent on the performance of the materials sealing the host formation. Acceptable performance will need to be demonstrated in order to satisfy operational, regulatory and public acceptance criteria. The success of CO2EOR operations in several parts of the world (e.g. Texas, USA) tends to suggest that formation seals can effectively contain CO2 for 'industrial' time periods (e.g. tens of years). However, many details of performance are assumed (e.g. based upon history matches), and there remains scope to assess in detail exactly how effective these seals really are. Similarly, the occurrence of natural hydrocarbon fields and CO2 accumulations demonstrates the containment potential of caprock materials over 'geological' time periods. Indeed, natural accumulations of CO2 show that reactions over longer timescales do not necessarily lead to wholesale dissolution and breakdown of the overlying aquicludes (i.e. if the seal was not effective the CO2 would have escaped long ago). However, in terms of data, poor seals would tend to be under-represented as they would not be exploration targets, and fewer samples would be taken for study. Thus introducing a potentially reactive phase into contact with a caprock not previously exposed to large amounts of CO2, could potentially degrade it and cause subsequent CO2 migration. Similarly, manmade breaches in the caprock and other overlying formations might provide potential fast return pathways for the CO2 to the surface. It is important, therefore, to assess the long-term performance of the 'engineered' seals in the caprock, and in particular the behaviour of borehole completions in the presence of CO2. Most of the studies considering CO2-water=rock interaction have focused on the formation hosting the CO2 (e.g. sandstone or carbonate) rather than the caprock. This reflects the interests of the hydrocarbon industry in trying to maximize oil production
IMPACT OF CHEMICAL REACTIONS ON CO2 STORAGE
during CO2-EOR operations. However, underground CO2 storage is a relatively 'young' topic, and there is a tendency towards focusing on the host rock as the first lithology studied when considering potential CO2 storage sites. Information on caprock reactions tends to be more limited than those of the host rock. The exact make-up of a particular caprock will be site-specific. However, in general terms they are likely to fall into two main types; argillites (clays, shales etc.) and evaporites. Previous work (Czernichowski-Lauriol et al. \996a; Pearce el al 1996) studying the interaction of caprock materials with CO2, used samples of mudstones and anhydrite from UK onshore material from the late Triassic Mercia Mudstone Group, which is broadly equivalent to material beneath the North Sea. The work focused on reacting samples with both dry supercritical CO2 and CO2-saturated solutions of varying salinity at 200bar and 80 °C, for up to 8 months. In essence these 'batch' experiments replicated (on a small scale) possible conditions at the formation water-CO2 'bubble' boundary. They contained small monoliths of rock partly submerged in CO2-saturated solutions, with the rest of the rock being exposed to supercritical CO2. Some experiments were also carried out with the rock sample only in contact with supercritical CO2. Lack of in-situ pore-water data resulted in simple fluids being used (de-ionized water and 0.55 M NaCl solution), so that small amounts of dissolution features would be expected as the caprocks equilibrated with the fluids. Although interpretation of the results was complicated by this initial disequilibrium, the experiments were still considered useful for understanding processes that would occur.
Reactions between dry supercritical CO2 and the caprock Injected CO2 is very likely to be undersaturated with water for the actual in-situ pressure and temperature conditions (for <40°C data see King et al 1992). Indeed, drying of the CO2 may be a necessary process to reduce corrosion of surface infrastructure during its separation and transport. Although dry CO2 will pick up some water when injected into its host formation, long-term injection might result in a 'chimney' of relatively dry CO2 surrounded by a halo of progressively wetter CO2, which in turn is surrounded by a halo of formation water enriched in CO2. The potential exists for relatively dry CO2 to contact argillaceous caprocks. This may have important consequences, because if the chemical potential of water in the CO2 is low enough, it might initiate water loss from the overlying caprocks. Dehydration
97
processes of mudrocks are well-known routes for the formation of shrinkage cracks (Fleureau et al. 1993), which if extensive, might allow for enhanced CO2 migration into, or through, the caprock. In the experiments mentioned above (rock samples and dry supercritical CO2) no obvious physical or mineralogical changes were observed. Both anhydrite and mudstone remained unaltered when exposed to supercritical CO2, either completely, or just the upper half of the sample). However, the study used samples that had been in store for some time. Although the argillaceous samples were relatively compact and would have had a relatively low water content, they could have undergone some degree of water loss prior to the experiments. In general terms, such an apparent low chemical reactivity between dry CO2 and aquiclude samples serves to increase confidence in the capability of the overlying aquiclude to maintain a good seal over long timescales.
Reactions between dissolved CO2 and the caprock In the experiments mentioned above, CzernichowskiLauriol et al (19960) reported that, although mudstone and anhydrite samples remained unaltered in contact with supercritical CO2, significant reactions were observed when in contact with CO2-saturated pore-water. The main reaction process identified between mudstones and CO2-saturated fluids was dissolution of dolomite, which was locally associated with clay mineral precipitation. Minor K-feldspar dissolution also occurred in the form of submicron pitting of the grain surfaces. Albite, chlorite and muscovite dissolution may have been additional minor reactions, as inferred from changes in fluid chemistry. Secondary carbonate precipitation was not observed although suspected in some areas in association with dolomite dissolution according to SEM-EDXA analyses. Ion exchange reactions between smectites and solution were suspected. Physical changes also occurred as the mudstone blocks were found to be fragmented, often along curved surfaces not present before the experiments, probably due to mechanical deformation and swelling of the smectite clays within the clay matrix. This observed disintegration was probably not a result of the effect of CO2, but an effect of immersing dry samples in aqueous fluids. The main reaction process identified between anhydrite and CO2-saturated fluids was the dissolution of anhydrite. This reaction was relatively rapid, with apparent anhydrite saturation obtained in less than four months, with Ca and SO4 concentrations reflecting the increasing solubility of anhydrite with increasing salinity (Bock 1961; Marshall etal 1964; Blount & Dickson 1969,1973). Dissolution appears
98
C. A. ROCHELLE ETAL.
Even if reactions occur between dissolved CO2 and the caprock, their effect might be minimized, or even over-ridden, if they cause extensive precipitation of secondary phases. In a predictive modelling study by Johnson et al (2001, 2004), two reactions have been suggested that could cause porosity reduction within the base of shales directly overlying any CO2 'bubble'. The first is dawsonite precipitation as detailed in reaction (3) above, which can give a solid volume increase for the reaction of between 17-25.4% depending on the silica phase used in the calculation. The second, more important, reaction is Fe-Mg-carbonate precipitation, as represented by magnesite: Fig. 7. SEM image of radiating needles of calcite growing on the surface of corroded anhydrite. Sample reacted for 8 months at 20MPa and 80 °C.
to have been controlled by crystal boundaries and cleavage planes, forming elongate voids up to 100|mm long. Dissolution penetrated about 5mm into the samples over the duration of the experiments (approximately 8 months). Associated with this dissolution was secondary precipitation of calcite, which tended to be concentrated at the interface between the aqueous and CO2 phases - though it also occurred in smaller quantities across the whole of the anhydrite surface in contact with the aqueous phase (Fig. 7). The calcite formed stubby crystals up to lOjjLm long, occasionally forming as radiating clusters, together with blocky crystals with a poorlydeveloped cross section. A secondary CaSO4 phase was also observed, and was presumed to be anhydrite. Thermodynamic calculations suggested that the effect of CO2 on anhydrite dissolution is small. The calculated solubility of anhydrite in CO2-saturated water is only slightly higher than for much lower CO2 fugacity solutions (e.g. corresponding to atmospheric CO2 pressure), due to the increase in concentration of the CaHCO3+ complex. However, the presence of CO2 might favour calcite precipitation from the Ca released into solution. It is possible that the observed anhydrite dissolution features result from the use of relatively simple (Ca and SO4poor) fluids in the experiments. Observations of samples from the Bravo Dome CO2 field in northeastern New Mexico (Broadhead 1989; Pearce et al 1996) show that the Cimmarron Anhydrite forms an effective seal for the CO2, despite evidence for anhydrite cement dissolution in the underlying Tubb Sandstone and in the experiments mentioned previously. The conclusions from the anhydrite batch experiments were that addition of CO2 had a very small effect on anhydrite solubility. Indeed, Bravo Dome may have contained CO2 for time periods of up to 50 000 years. Hence, it appears unlikely that a large evaporite seal would be breached.
K-feldspar
Mg-chlorite
muscovite magnesite
kaolinite quartz/chalcedony
(8)
This reaction has a solid volume increase of approximately 18.5%, and in the simulation by Johnson et al (2001), has the potential to reduce the shale porosity by 50% (though in their 20-year simulation, slow reaction kinetics resulted in only 15% completion of the reaction). Although CO2 trapping by mineral precipitation was predicted to be relatively minor, the potential for secondary mineral precipitation in shales could be highly significant if it increased caprock integrity. With respect to anhydrite caprocks, the relatively rapid precipitation of secondary anhydrite and calcite as observed in the anhydrite batch experiments would be advantageous for the containment of CO2 if it acted as a 'self sealing' mechanism - as per the study by Johnson et al (2001) detailed above. However, calcite has a molar volume only some 87% of that of anhydrite, consequently porosity generation might be expected. Thus, if such a self-sealing process did occur, then other mechanisms might also be needed, such as the morphology of the secondary calcite acting so as to increase its effective volume. Previous studies have shown that carbonate mineral morphology is strongly influenced by solution chemistry and growth kinetics (Folk 1974; Lahann 1978; Given & Wilkinson 1985; Braithwaite & Heath 1989; Gonzalez et al 1992; Kimbell & Humphrey 1994; Milodowski et al 1995, 1997). In more saline solutions, or in the presence of high Sr2+, Mg 2+ , SO42~, or under increasing degree of saturation, or with rapid precipitation, carbonate minerals tend to produce more elongate crystal forms. The precipita-
IMPACT OF CHEMICAL REACTIONS ON CO2 STORAGE
tion of very elongate or fibrous crystal forms may increase pore tortuosity and hence reduce permeability similarly to that observed with fibrous illite (e.g. McHardy et al. 1982). At present, there is no geological evidence for calcite sealing in the anhydrite caprock of the Bravo Dome CO2 field.
Summary There are some indications of the types of reactions that might take place in caprocks, though these are less well constrained compared to host rocks. There is some evidence for the initial dissolution of dolomite, K-feldspar and possibly sheet silicates, whereas anhydrite may be barely affected. Such dissolution reactions may reduce the ability of the caprock to retain CO2, and might aid its migration, as could mudrock dehydration reactions. However, dissolution may be followed by precipitation of secondary minerals such as Ca-Mg-Fe carbonates and dawsonite, which may be extensive enough to achieve an overall reduction in porosity within the caprocks, and so improve sealing capacity and CO9 containment potential.
physical properties. Indeed, most wells are only partly cemented, and there is usually no cement behind the borehole casing between the top of the production interval and the uppermost few hundred metres of the well. As a consequence, failure of these cement seals (especially the lower one) could provide the CO2 with fast migration pathways to the surface or overlying formations, regardless of the sealing potential of the caprock. Thus, the long-term stability of steel and cement are of key interest. Cements are a complex mixture containing much poorly crystalline and gel-like, hydrated calcium silicate (CSH) and calcium aluminosilicate (CASH) phases. They are also highly alkaline, with porewaters buffered by the dissolution of portlandite (Ca(OH)2) and the slower dissociation of Ca(OH)2 from CSH and CASH phases. CO2 on the other hand is an acidic gas, which under supercritical conditions has a relatively low viscosity and relatively high density. There is much potential for CO2 to access even relatively small pore spaces and initiate carbonation reactions. These reactions could have important consequences in terms of: (1)
Reactions between supercritical CO2 and borehole completions General considerations The occurrence of natural hydrocarbon fields and CO2 accumulations demonstrates the containment potential of caprock materials over 'geological' time periods. However, exploitation of these resources, together with CO2 injection operations, necessitates breaches in these caprock formations. The performance of the 'engineered' seals in the caprock (together with well abandonment procedures) are thus of key importance for the long-term containment of CO2. Many oil well completions, particularly in older fields, typically have a design life of the order of 25 years, sufficient for the operational lifetime of the field. Consideration of the stability, alteration and longer-term performance of the cement and steel casing in the geochemical environment, following well abandonment, is often neglected (Robins & Milodowski 1986). In a typical borehole completion, an Ordinary Portland Cement (OPC)based cement is used to form a seal between the steel casing and the host formations. A good cement-toformation or cement-to-casing seal relies strongly on a physical rather than chemical bond. In this respect, the cement-formation and the cement-steel interfaces represent particularly important features, and may represent zones of preferential fluid movement should chemical changes in the cement alter its
99
(2)
Changing the physical nature of the cement. For example reaction might cause shrinkage of the cement, and it might 'pull away' from the borehole casing or along the interface between the cement and the surrounding caprock. Other reactions can potentially result in large volume increases, which might enhance sealing. Alternatively they may cause deformation of the casing and compromise the integrity of borehole completion. This might allow CO2 migration and either reaction with the next piece of cement, or escape into/through overlying formations. Changing the reactivity of the cement. For example initial carbonation reactions might act to 'armour' the remaining cement, and prevent further ingress of CO2. A 'self-sealing' reaction such as this might aid CO2 containment.
Observations from various industries Steel corrosion has been identified in oil and gas wells in Texas and Louisiana since the early-mid 1940s (Crolet 1994). Occurrences of high concentrations of CO2 (and other acid gases) in other oil and gas fields, led to numerous investigations of the corrosion properties of the various steels used in the production infrastructure (e.g. Schremp & Roberson 1975; Crolet 1983, 1994; EFCP 1994). This has allowed for the development of speciality steels more resistant to CO2 corrosion - at least over the likely operational timescale of a hydrocarbon production facility. These include Cr-containing stainless steels
100
C.A.ROCHELLECTAL.
Fig. 8. SEM image of hydrated and set Class G Oilwell Cement after autoclaving for 24 hours at 7MPa and 30 °C. Note euhedral portlandite crystals in low macroporosity CSH gel matrix.
(Crolet 1983). Although corrosion rates of less that 0.1 mm/year may be acceptable over an assumed 30year lifetime of a pipe or an old completion (Crolet 1994), localized corrosion may present specific problems. This could include erosion-corrosion caused by fast-flowing fluid streams within the pipes, especially at the junctions between lengths of pipe. Although stainless steels have good general corrosion resistance, they are susceptible to pitting corrosion as a result of localized reduction in Cr:Fe ratio around MnS inclusions (Ryan et al. 2002). In summary, although CO2 can corrode steel, there is a reasonable degree of understanding of the corrosion behaviour of various steels over 'industrial' time periods (i.e. up to a few tens of years). Acceptable performance over longer timescales (possibly measurable in thousands of years) will need to be demonstrated in order to satisfy operational, regulatory and public acceptance criteria. This will require quantitative information about longer-term corrosion processes. Carbonation of cement has been studied in the engineering and construction industries with regard to mechanical properties of structures. For example, cement-based materials can become significantly stronger when they are exposed to high pressures of CO2 (Hashida et al 1996; Gibbs 1996). However, very little of this considers interactions within the deep geological environment. Schremp & Roberson (1975) tested the reactivity of cement-lined pipes at approximately 22 °C and 136bar CO2 pressure. A small increase in calcite was identified by X-ray analysis, but there was no evidence of deterioration or separation from the pipe wall. However, the test was only of 25 days duration. In contrast, Robins & Milodowski (1982) demonstrated very rapid reaction, with major physical and mineralogical
Fig. 9. SEM image of set Class G Oilwell Cement after reaction with liquid CO2 for 24hours at 8.5MPa and 40 °C (from Robins & Milodowski 1982) Note high macroporosity, complete loss of portlandite and alteration of CSH hydrated cement paste to calcite and aragonite. Baryte is observed as drilling fluid was included during sample preparation.
changes, in composite rock-cement blocks of Type G OPC-based oil-well cement (which had previously been cured by autoclaving for 24 hours at 30 °C and 7MPa) on reaction with supercritical CO2 for 24 hours at 40 °C and 85 atmospheres pressure. These authors showed that the portlandite and tight, dense gel-like CSH matrix of the original cement (Fig. 8) had been extensively altered to a porous and friable crystalline matrix consisting of needles of aragonite and rhombs of calcite (Fig. 9). As a result there was a complete loss of bonding across the interface between the cement and the formation test sample. One possible explanation for the difference in degree of reaction between the above two studies could be the amount of water that was present within the cement pores - the Schremp & Roberson (1975) cement being relatively water-rich. Totally dry cement would not react as some water has to be present to initiate the CO2-cement reactions. Similarly, pores full of water hinder the access of CO2 and hence also lead to relatively low reaction. However, partly-saturated pores would allow CO2 ingress and facilitate reaction (Klemm & Berger 1972). The petroleum industry literature contains some information on borehole cement properties, but these generally consider only short operational time scales. In recent years, detailed investigations of the longterm stability and interaction of cements in the geological environment have been carried out with regard to the sealing of geological repositories for radioactive wastes (e.g. Alexander 1992; Linklater 1998; Smellie 1998; Glasser 2001). However, most of these studies concentrate primarily on trace
IMPACT OF CHEMICAL REACTIONS ON CO, STORAGE
element speciation, solubility, migration and retardation in high pH groundwaters, and alkali fluid-rock interaction processes, the evolution of cement porefluids and pH buffering of alkali pore fluid systems. However, examination of natural CSH minerals and gels similar to those found in hydrated OPC-based cements shows that they carbonate readily on reaction with bicarbonate groundwaters or atmospheric CO2 to form a range of minerals, including: calcite, aragonite, vaterite, scawtite and hydrotalcite (Alexander 1992; Linklater 1998; McConnell 1960; Milodowski et al 1989; Smellie 1998). Some of these minerals may only be metastable (e.g. vaterite, aragonite). In contrast to the rapid and extensive reaction with supercritical CO2 observed by Robins and Milodowski (1982), the slower reaction of natural CSH minerals with bicarbonate groundwaters or atmospheric CO2 often produces a dense carbonate reaction rim which appears to have restricted reaction, allowing the CSH mineral to be preserved beneath the carbonate rim.
Potential reactions within a CO2 storage operation Within a CO2 storage operation, the effectiveness of long-term containment will depend on factors such as; the relative rates at which reactions occur, and which types of reactions dominate overall. The type of reactions that occur will potentially vary greatly within different regions of the system (e.g. lower pH conditions at the edge of the cement, or higher pH conditions within the core of a mass of cement). The different processes may result in different rates of reaction, variations in the removal of cement phases, and differences in volumetric changes (e.g. isovolumetric changes as opposed to isochemical changes). By way of an example to highlight the complexities of possible solid volume changes during reactions, consider the reaction of CO2 with tobermorite (one member of a family of calcium silicate hydrate [CSH] phases that can be an important constituent of cements):
tobermorite
(9) calcite
amorphous silica
Consider first molar volume data for tobermorite calculated from X-ray diffraction measurements (JCPDS 1986: approximately 287cm3mor1), with calcite and amorphous silica molar volume data from the 'dataO.cmp.v8.R6' database of the EQ3/6 geochemical modelling package (Wolery 1992; Wolery & Daveler 1992: calcite 36.9cm3mor1,
101
amorphous silica 29.0 cm3 mol J ). These data would give a solid volume increase of 30% for reaction (9), which could indicate that reaction of CO2 might improve the integrity of borehole seals. However, natural tobermorite formed over long time periods (and hence giving a good X-ray diffraction pattern) can be considerably more ordered than relatively rapidly-formed tobermorite in man-made cement. Indeed, freshly-formed tobermorite can be relatively amorphous and gel-like (e.g. Lee 1970). Cement minerals also have variable degrees of hydration, and as a result, variable molar volumes. This will make it more difficult to estimate the potential for porosity increases or decreases upon reaction with CO2. Consequently, the molar volume of tobermorite in borehole cement can be considerably higher than given above. Consider the case of the tobermorite molar volume being almost twice that of the above case (500cm3 mol"1, as taken from the 'dataO.cmp.v8.R6' database of the EQ3/6 geochemical modelling package). These data would give a solids volume decrease of 25% for reaction (9). This reaction of CO2 could be deleterious to the integrity of borehole seals, possibly causing pervasive porosity increase or even cracking. Other studies suggest a tobermorite molar volume intermediate between the two values above (369 cm3 mol"1, Glasser et al 1999). This suggests a solids volume increase for reaction (9), but of only 1 %. To complicate these calculations even further, aragonite (34.2cm3mol"1, from 'dataO.cmp.v8.R6' of EQ3/6, Wolery 1992; Wolery & Daveler 1992) may form instead of, or as well as, calcite. A similar situation exists for SiO2, in that quartz, or more likely chalcedony (22.7cm3mor1, from 'dataO.cmp.v8.R6' of EQ3/6, Wolery 1992; Wolery & Daveler 1992) may form in preference to amorphous silica. There are other complicating factors that impinge on calculations such as those given above. The first is that all of the Ca and Si released from tobermorite dissolution are presumed to be precipitated within the cement. Alkali cement in contact with acidic CO2 would have steep pH gradients over relatively small distances. This could greatly influence the local solubility of both calcite and silica phases. It is possible that not all of the Ca and Si would be precipitated, causing a further increase in porosity. Another factor is that CSH phases may form poorly ordered structures that are sheet-like, rope-like or even threedimensional networks of fibres (e.g. for images of CSH phases precipitated around, and cementing grains together, see Bateman et al 1995, 1999; Rochelle et al 1998, 2001; and examples in Figs 10-12). The effectivemolar volume (i.e. the overall volume of cement phases plus associated pore space) of such structures could be much larger than the actual molar volume of the solids themselves. If these structures underwent dissolution, followed by
102
C. A. ROCHELLE ETAL
Fig. 10. Secondary electron photomicrograph of fibrous CSH formed during the reaction of a Ca hydroxide fluid with quartz and muscovite at 70 °C, 0.1 MPa over approximately 5 months (from Bateman et al. 1995). Note high degree of porosity and high surface area of CSH.
Fig. 11. Secondary electron photomicrograph of CSH gel in which relict fibrous CSH is enclosed (from Bateman et al. 1995). Experiment reacted quartz and muscovite with Ca/Na/K hydroxide fluid at 70 °C, 0.1 MPa over approximately 5 months. Note high degree of porosity and high surface area of CSH.
precipitation of relatively dense calcite and silica phases, then effective porosity increases might be enhanced even further. There is clearly much scope to assess and quantify in detail the nature and extent of CO2-borehole cement reactions. The CO2 storage operation at Sleipner, North Sea, has used a highly inclined well for CO2 injection (Baklid et al. 1996). Consequently, the borehole completion (and especially that at the base of the caprock) is a significant lateral distance from the CO2 injection point. Given suitable hydrogeological
Fig. 12. Secondary electron photomicrograph of secondary fibrous K-rich CSH gel coating sand-sized grains of rock (from Bateman et al. 1999). Experiment reacted grains of metamorphosed volcanic basement rock with Ca/Na/K hydroxide fluid at 70 °C, 0.1 MPa over approximately 6 months. Note blockage of pore spaces, and high degree of porosity and high surface area of CSH.
and structural conditions, inclined wells (such as that at Sleipner) may make it possible to keep the borehole completion away from any CO2 'bubble' or significant concentrations of dissolved CO2. This might serve to minimize the potential for reactions between CO2 and borehole completions. It is clearly desirable to obtain much more detailed information on cement carbonation and the resulting porosity changes. However, several natural CO2 fields have wells that have been completed or plugged with cement, or have cemented wells that have been used in conjunction with CO2 injection for EOR. That these boreholes have apparently not failed, would tend to indicate that the borehole completions have not been totally compromised by reaction with CO2, although a detailed and rigorous assessment of the performance of these wells remains to be done. An empirical observation such as this only considers 'industrial' time periods measurable in tens of years. Acceptable performance over longer timescales (possibly measurable in thousands of years) will need to be demonstrated in order to satisfy operational, regulatory and public acceptance criteria.
Summary During underground sequestration operations the presence of supercritical CO2 will result in chemical disequilibria and hence the initiation of reactions. These reactions may actually help in the chemical containment of CO2, either as a free phase, a dissolved phase, or when precipitated as carbonate minerals. It is important to understand the direction,
IMPACT OF CHEMICAL REACTIONS ON CO? STORAGE
rate and magnitude of such reactions, both in terms of their impact upon the ability of the aquifer to contain the injected CO2 safely, and in terms of the longevity of CO2 containment. Four broad areas of reaction can be considered: (1)
(2)
(3)
(4)
Reaction with the formation water. This will be important because it forms a relatively rapid and large sink for CO2. The amount of reaction will vary according to factors such as pressure, temperature, fluid chemistry, and degree of CO2-water mixing. Reaction with the host aquifer. This is likely to be slower than with the formation water, but it might enhance trapping in two ways. It may buffer formation water pH and so allow for more CO2 to be stored in solution. Siliciclastic aquifers appear to offer more advantageous buffering compared to carbonate aquifers, though they are likely to react more slowly. Second, reactions with the host rock may provide a more permanent sink for CO2 in the form of carbonate minerals such as calcite. Reaction with the caprock. This will be important as any degradation of the overlying seal could lead to CO2 migration. Equally, a selfsealing mechanism could occur which might aid containment. Although the reaction of CO2 with caprocks is somewhat less well constrained compared to host rocks, there is evidence that both dissolution and precipitation reactions could occur. The properties of the caprock therefore, will depend crucially on which of these processes dominates the overall solid volume change. Reaction with borehole materials. This is particularly important as degradation of borehole infrastructure could lead to CO2 migration. Indeed, a leaky borehole could allow CO2 to bypass the caprock and migrate to shallower formations or directly to the surface. Although steel well linings and borehole cements may provide suitable containment over a few tens of years, their behaviour over longer timescales is less certain. Acceptable performance over longer timescales must be demonstrated in order to satisfy operational, regulatory and public acceptance criteria.
There are a variety of CO2-water-rock chemical reactions that can help in the chemical containment of CO2, and have the potential to trap it for geologically-important timescales. Individual formations used to store CO2 will vary in structure, mineralogy, and hydrogeology, and each storage operation will have to take account of local geological, fluid chemical, and hydrogeological conditions. It will be necessary to ascertain these prior to CO2 injection in order
103
to quantify storage potential and the long-term behaviour of CO2. Although the geological sequestration of CO2 has many advantages, much work remains to be done to understand the complex interrelationships between the various chemical reactions. This is especially true for the caprock and borehole seals that prevent rapid migration of CO2 back to the atmosphere. Some of the results presented in this paper formed part of a much larger study into the feasibility of underground CO2 disposal (Holloway I996a, b), and the authors gratefully acknowledge comments from other members of the project during that study. The BGS and BRGM gratefully acknowledge funding from the Environment Division of the UK Department of Trade and Industry, from the French Ministry of Industry, and from the European Commission as part of their Joule II Programme (contract no. JOU2CT92-0031). The authors thank S. Hovorka, R. Worden, and an anonymous reviewer for their useful comments that helped to improve this paper. This paper is published with the permission of the Director of the British Geological Survey (NERC).
References ALEXANDER, W. R. (ed.) 1992. A natural analogue study of the Maqarin hyperalkaline groundwaters. I Source term description and thermodynamic testing, Nagra Technical Report 91-10. AMOCO P. C. 1990. Bravo Dome Update - Carbon dioxide production in New Mexico. Bravo Dome Operations Center, Amoco Production Company. ARTS, R; EIKEN, O., CHADWICK, A., ZWEIGEL, P., VAN DER MEER, B. AND KIRBY, G. 2004. Seismic monitoring at the Sleipner underground CO2 storage site (North Sea). In: BAINES, S. J. & WORDEN, R. H. (eds) Geological Storage of Carbon Dioxide, Geological Society, London, Special Publications, 233, 181-191. ATKINS, P. W. 1982. Physical Chemistry (2nd edn). Oxford University Press. BACHU, S., GUNTER, W. D. & PERKINS, E. H. 1994. Aquifer disposal of CO2: hydrodynamic and mineral trapping. Energy Conversion and Management, 35, 269-279. BAKER, J. C., BAI, G. P., HAMILTON, P. J., GOLDING, S. D. & KEENE, J. B. 1995. Continental-scale magmatic carbon dioxide seepage recorded by dawsonite in the BowenGunnedah-Sydney basin system, eastern Australia. Journal of Sedimentary Research, A65, 252-530. BAKLID, A., KORB0, L. R. & OWREN, G. 1996. Sleipner Vest CO2 disposal, CO2 injection into a shallow underground aquifer. Society of Petroleum Engineers, 36600,269-277. BATEMAN, K., COOMBS, P., PEARCE, J. M. & WETTON, P. 1995. Nagra/Nirex/SKB column experiments: Fluid chemical and mineralogical studies. British Geological Survey, Report WE/95/26,104p. BATEMAN, K., COOMBS, P., PEARCE, J. M., NOY, D. J. & WETTON, P. 1999. Fluid-rock interaction in the Disturbed Zone: Nagra/Nirex/SKB column experiments - Phase II. British Geological Survey, Report WE/99/5,130p.
104
C. A. ROCHELLE ETAL.
BEVAN, J. & SAVAGE, D. 1989. The effect of organic acids on the dissolution of K-feldspar under conditions relevant to burial diagenesis. Mineralogical Magazine, 53,415^25. BLOUNT, C. W. & DICKSON, F. W. 1969. The solubility of anhydrite (CaSO4) in NaCl-H2O from 100 to 450°C and 1 to 1000 bars. Geochimica et Cosmochimica Acta, 33,227-245. BLOUNT, C. W. & DICKSON, F. W. 1973. Gypsum-anhydrite equilibria in systems CaSO4-H2O and CaSO4-NaClH2O. American Mineralogist, 58, 323-331. BOCK, E. 1961. On the solubility of anhydrous calcium sulphate and of gypsum in concentrated solutions of sodium chloride at 25°C, 30°C, 40°C and 50°C. Canadian Journal of Chemistry, 39,1746-1751. BOWKER, K. A. & SHULER, P. J. 1991. Carbon dioxide injection and resultant alteration of the Weber Sandstone, Rangely Field, Colorado. American Association of Petroleum Geologists Bulletin, 75,1489-1499. BRAITHWAITE, C. J. R. & HEATH, R. A. 1989. Inhibition, distortion and dissolution of overgrowth cements on pelmatazoan fragments. Journal of Sedimentary Petrology, 59,267-271. BROADHEAD, R. F. 1989. Bravo Dome CO2 Field - U.S.A. New Mexico. Treatise of Petroleum Geology, Atlas of Oil and Gas Fields, The American Association of Petroleum Geologists, A-011. CROLET, J.-L. 1983. Acid corrosion in wells (CO2, H2S): Metallurgical aspects. Journal of Petroleum Technology, August 1983,1553-1558. CROLET, J.-L. 1994. Which CO2 corrosion, hence which prediction? In 'A working party report on: Predicting CO2 corrosion in the oil and gas industry', European Federation of Corrosion Publications number 13, The Institute of Materials, London, ISSN 1354-5116, 1-29. CZERNICHOWSKI-LAURIOL, I., SANJUAN, B., ROCHELLE, C.,
BATEMAN, K., PEARCE, J. & BLACKWELL, P. 19960. Inorganic geochemistry. In: S. HOLLOWAY (ed.) The Underground Disposal of Carbon Dioxide Final Report of Joule II Project Number CT92-0031. CZERNICHOWSKI-LAURIOL, I., SANJUAN, B., ROCHELLE, C., BATEMAN, K., PEARCE, J. & BLACKWELL, P. 1996&. Analysis of the geochemical aspects of the underground disposal of CO2. In: APPS, J. A. & TSANG, C.-F. (eds) Deep Injection Disposal of Hazardous and Industrial Wastes, Scientific and Engineering Aspects Academic Press, 565-583. DRUMMOND, S. E. JR 1981. Boiling and mixing hydrothermal fluids: chemical effects on mineral precipitation. PhD thesis, The Pennsylvania State University, USA. DUAN, Z., MOLLER, N. & WEARS, J. H. 1992. An equation of state for the CH4-CO2-H2O system: I. Pure systems from 0 to 1000°C and 0 to 8000 bar. Geochimica et Cosmochimica Acta, 56,2605-2617. EFCP 1994. A working party report on: Predicting CO2 corrosion in the oil and gas industry, European Federation of Corrosion Publications number 13, The Institute of Materials, London, 173p. ELLIS, A. J. & GOLDING, R. M. 1963. The solubility of carbon dioxide above 100°C in water and in sodium chloride solutions. American Journal of Science, 261, 47-60. ENICK, R. M. & KLARA, S. M. 1990. CO2 solubility in water
and brine under reservoir conditions. Chemical Engineering Communications, 90,23-33.. ENICK, R. M. 1992. Effects of CO2 solubility in brine on the compositional simulation of CO2 floods. SPE Reservoir Engineering, May 1992, 253-258 (SPE 20278). FABRIOL, R. & CZERNICHOWSKI-LAURIOL, I. 1992. A new approach to geochemical modelling with an integrated simulator generation system. In: KHARAKA, Y. K. & MAEST, A. S. (eds) Proceedings of the 7th International Symposium on Water-Rock Interaction -WRI-7 213-216, Balkema, Rotterdam. FLEUREAU, J., KHEIRBEK-SAOUD, S., RIA SOEMITRO, R. & TAIBI, S. 1993. Behavior of clayey soils on dryingwelting paths. Canadian Geotechnical Journal, 30, 287-296. FOLK, R. L. 1974. The natural history of crystalline calcium carbonate: effect of magnesium content and salinity. Journal of Sedimentary Petrology, 44,40-53. GARRELS, R. M. & CHRIST, C. L. 1965. Solutions, minerals, and equilibria. Freeman-Cooper, San Francisco, 450pp. GIBBS, W. W. 1996. Pressure to change, supercritical carbon dioxide to toughen common materials. Scientific American, November 1996,27-28. GIVEN, R. K. & WILKINSON, B. H. 1985. Kinetic control of morphology, composition, and mineralogy of abiotic sedimentary carbonates. Journal of Sedimentary Petrology, 55,109-119. GLASSER, F. P., 2001. Mineralogical aspects of cement in radioactive waste disposal. Mineralogical Magazine, 65,621-633. GLASSER, F. P., TYRER, M., QUILLIN, K., Ross, D., PEDERSEN, J., GOLDTHORPE, K., BENNETT, D. & ATKINS, M. 1999. The chemistry of blended cements and backfills intended for use in radioactive waste disposal. UK Environment Agency R&D Technical report P98. GONZALEZ, L. A., CARPENTER, S. C. & LOMANN, K. C. 1992. Inorganic calcite morphology: roles of fluid chemistry and fluid flow. Journal of Sedimentary Petrology, 62, 382-399. GUNTER, W. D., PERKINS, E. H. & MCCANN, T. J. 1993. Aquifer disposal of CO2-rich gases: reaction design for added capacity. Energy Conversion Management, 34,941-948. GUNTER, W. D., WIWCHAR, B. & PERKINS, E. H. 1997. Aquifer disposal of CO2-rich greenhouse gases: extension of the time scale of experiment for CO2sequestering reactions by geochemical modelling. Mineralogy and Petrology, 59, 121-140. GUNTER, W. D., PERKINS, E. H. & HUTCHEON, I. 2000. Aquifer disposal of acid gases: modelling of waterrock reactions for trapping of acid wastes. Applied Geochemistry, 15, 1085-1095. GUNTER, W. D., BACHU, S. & BENSON, S. 2004. The role of hydrological and geochemical trapping in sedimentary basins for secure storage of CO2. In: BAINES, S. J. & WORDEN, R. H. (eds) Geological Storage of Carbon Dioxide. Geological Society, London, Special Publications, 233,129-145. HASHIDA, T., TERAMURA, S., HA, J. C. & TAKAHASHI, H. 1996. Development of a CO2-solidification method for recycling concrete wastes. Proceedings of the 4th
IMPACT OF CHEMICAL REACTIONS ON CO? STORAGE Materials Engineering Conference (Materials for the New Millennium), November 10-14, Washington, DC, USA, 764-683. HITCHON, B. (ed.) 1996. Aquifer Disposal of Carbon Dioxide, Hydrodynamic and Mineral Trapping Proof of Concept. Geoscience Publishing Ltd, Alberta, Canada, 165p. HOLLOWAY, S. (ed.) 19960. The Underground Disposal of Carbon Dioxide, Final Report of Joulle II Project Number CT92-0031, 355p, published by the British Geological Survey. HOLLOWAY, S., HEEDERIK, J. P., VAN DER MEER, L. G. H., CZERNICHOWSKI-LAURIOL, I., HARRISON, R., LlNDEBERG, E., SUMMERFIELD, I. R., ROCHELLE, C, SCHWARTKOPF, T., KAARSTAD, O. & BERGER, B. I996b.
The underground disposal of carbon dioxide, Summary of final report of project number JOU2-CT92-0031, 24p, published by the British Geological Survey. IEA 1998. Sleipner aquifer storage of CO2. Greenhouse Issues, 34,4. IEA 2001. IEA GHG meetings in Canada. Greenhouse Issues, 54,2. JCPDS 1986. Mineral powder diffraction file data book. JCPDS International Center for Diffraction Data (publishers), USA. JOHNSON, J. W., NITAO, J. J., STEEFEL, C. I. & KNAUSS, K. G. 2001. Reactive transport modelling of geologic CO2 sequestration in saline aquifers: the influence of intraaquifer shales and the relative effectiveness of structural, solubility, and mineral trapping during prograde and retrograde sequestration. In: Proceedings of the First National Conference on Carbon Sequestration, Washington, DC, May 14-17, 2001. JOHNSON, J. W., NITAO, J. J. & KNAUSS, K. G. 2004. Reactive transport modelling of CO2 storage in saline aquifers to elucidate fundamental processes, trapping mechanisms, and sequestration partitioning. In: BAINES, S. J. & WORDEN, R. H. (eds) Geological Storage of Carbon Dioxide. Geological Society, London, Special Publications, 233, 107-128. KERVEVAN, C., FABRIOL, R., GUILLEN, A. & SAUTY, J. P. 1994. A sequential-parallel approach for coupling chemistry and transport in groundwater quality modeling. In: GENTZSCH, W. & HARMS, U. (eds) High Performance Computing and Networking, Lecture Notes in Computer Science, 796, 335-339, SpringerVerlag, Berlin. KIMBELL, T. N. & HUMPHREY, J. D. 1994. Geochemistry and crystal morphology of aragonite cements of mixing-zone origin, Barbados, West Indies. Journal of Sedimentary Research, A64, 604-614. KING, M. B., MUBARAK, A., KIM, J. D. & BOTT, T. R. 1992. The mutual solubilities of water with supercritical and liquid carbon dioxide. Journal of Supercritical Fluids, 5,296-302. KLEMM, W. A. & BERGER, R. L. 1972. Accelerated curing of cementitious systems by carbon dioxide. Part I. Portland cement. Cement and Concrete Research, 2, 567-576. KNAUSS, K. G. & WOLERY, T. J. 1986. Dependence of albite dissolution kinetics on pH and time at 25 °C and 70 °C. Geochimica et Cosmochimica Acta, 50, 2481-2497. KNAUSS, K. G. & WOLERY, T. J. 1988. The dissolution
105
kinetics of quartz as a function of pH and time at 70°C. Geochimica et Cosmochimica Acta, 52, 1493-1501. KUK, M. S. & MONTAGNA, J. C. 1983. Solubility of oxygenated hydrocarbons in supercritical carbon dioxide. In: Chemical Engineering at Supercritical Fluid Conditions, 101-111, Ann Arbor Science, Ann Arbor, USA. LAHANN, R. W 1978. A chemical model for calcite crystal growth and morphology control. Journal of Sedimentary Petrology, 48, 337-344. LAW, D. H.-S. & BACHU, S. 1996. Hydrogeological and numerical analysis of CO2 disposal in deep aquifers in the Alberta sedimentary basin. Energy Conversion and Management, 37, 1167-1174. LEE, F. M. 1970. The chemistry of cement and concrete. Edward Arnold, London. LINKLATER, C. M. (ed.) 1998. A natural analogue study of cement-buffered,hyperalkaline groundwaters and their interaction with a repository host rock, phase II, UK Nirex Ltd. Science Report, S/98/003. MALIK, Q. M. & ISLAM, M. R. 2000. CO2 injection in the Weyburn field of Canada: Optimization of enhanced oil recovery and greenhouse gas storage with horizontal wells. Society of Petroleum Engineers, 59327, 16pp. MARSHALL, W. L., SLUSHER, R. & JONES, E. V. 1964. Solubility and thermodynamic relationships for CaSO4 in NaCl-H2O solutions from 40°C to 200°C, 0 to 4 molal NaCl. Journal of Chemical Engineering Data, 9,187-191. MCCONNELL, J. D. C. 1960. Vaterite from Ballycraigy, Larne, Northern Ireland. Mineralogical Magazine, 32,535-545. MCHARDY, W J., WILSON, M. J. & TAIT, J. M. 1982. Electron microscope and X-ray diffraction studies of filamentous illitic clay from sandstones of the Magnus Field. Clay Minerals, 17, 23-39. MILODOWSKI, A. E., NANCARROW, P. H. A. & SPIRO, B. 1989. A mineralogical and stable isotope study of natural analogues of Ordinary Portland Cement (OPC) and CaO-SiO2-H2O (CSH) compounds. United Kingdom Nirex SafetyStudies Report, NSS/R240. MILODOWSKI, A. E., GILLESPIE, M. R., SHAW, R. P. & BAILEY, D. E. 1995. Flow zone characterisation: mineralogical and fracture orientation characteristics in the PRZ and Fleming Hall Fault Zone area boreholes, Sellafield. United Kingdom Nirex Limited Report, S A/95/001. MILODOWSKI, A. E., GILLESPIE, M. R., & METCALFE, R. 1997. Relationship between mineralogical transformations and groundwater chemistry at Sellafield, N.W. England: a tool for studying Quaternary palaeohydrogeology. In: HENDRY, J. P., CAREY, P. F, PARNELL, J., RUFFELL, A. H. & WORDEN, R. H. (eds) GEOFLUIDS II '97: Contributions to the Second International Conference on Fluid Evolution, Migration and Interaction in Sedimentary Basins and Orogenic Belts (Belfast,10-14 March 1997), University of Belfast. 30-33. MOBERG, R. 2001. The Weyburn CO2 monitoring and storage project. Greenhouse Issues, 57,2-3. NELIS, M. K. 1994. Bravo Dome II petrographic study.
106
C. A. ROCHELLE ETAL.
Report of the Amoco Production Company. OMOLE, O. & OSOBA, J. S. 1983. Carbon dioxide-dolomite rock interaction during CO2 flooding process. Petroleum Society of Canada Institute of Mining and Metallurgy, paper 83-34-17,13p. PEARCE, J. M., HOLLOWAY, S., WACKER, H., NELIS, M. K., ROCHELLE, C. A. & BATEMAN, K. 1996. Natural occurrences as analogues for the geological disposal of carbon dioxide. Energy Conversion and Management, 37,1123-1128. PEARCE, J. M., CZERNICHOWSKI, I., LOMBARDI, S., BRUNE, S., NADOR, A., BAKER, J., PAUWELS, H., HAZIYANNIS, G., BEAUBIEN, S. & FABER, E. 2004 A review of natural CO2 accumulations in Europe as analogues for geological sequestration. In: BAINES, SJ. & WORDEN, R. H. (eds) Geological Storage of Carbon Dioxide, Geological Society, London, Special Publications, 233,29-41. PERKINS, E. H. & GUNTER, W. D. 1995. Aquifer disposal of CO2-rich greenhouse gases: Modelling of water-rock reaction paths in a siliciclastic aquifer. In: KHARAKA, Y. K. & CHUDAEV, O. V. (eds) Proceedings of the 8th International Symposium on Water-Rock Interaction WRI-8,895-898, Balkema, Rotterdam. ROBINS, N. S. & MILODOWSKI, A. E. 1982. Design and evaluation of the Harwell borehole cement systems. Institute of Geological Sciences, Technical Report, ENPU 82-9. ROBINS, N. S. & MILODOWSKI, A. E. 1986. Borehole cements and the downhole environment - a review. Quarterly Journal of Engineering Geology, London, 19,175-181. ROCHELLE, C. A., BATEMAN, K., MACGREGOR, R., PEARCE, J. M., PRIOR, S. V., SAVAGE, D. & P. WETTON, P. 1998. The evaluation of chemical mass transfer in the disturbed zone of a deep geological disposal facility for radioactive wastes. VI. The hydrothermal aging of disks of bulk rock and near-fracture Borrowdale Volcanic Group lithologies with a synthetic alkaline near-field groundwater. UK Nirex Ltd. Science Report, NSS/R390, 34p. ROCHELLE, C. A., PEARCE, J. M. & HOLLOWAY, S. 1999. The underground sequestration of carbon dioxide: containment by chemical reactions. In: Chemical Containment of Waste in the Geosphere, Geological Society, London, Special Publications, 157,117-129. ROCHELLE, C. A., BATEMAN, K., COOMBS, P., PEARCE, J. M. & WETTON, P. 2001. The evaluation of chemical mass transfer in the disturbed zone of a deep geological disposal facility for radioactive wastes. IX. Further reactions involving discs of Borrowdale Volcanic Group lithologies with synthetic evolved near-field groundwater. British Geological Survey, Report WE/94/45, 30p. Ross, G. D., TODD, A. C. & TWEEDIE, J. A. 1981. The effect of CO2 flooding on the permeability of reservoir
rocks. In: F. J. PAYERS (ed.) Enhanced oil recovery, Developments in Petroleum Science volume 13, Elsevier, Amsterdam, 351-366. RYAN, M. P., WILLIAMS, D. E., CHATER, R. J., HUTTON, B. M. & McPHALL, D. S. 2002. Why stainless steel corrodes. Nature, 415,770-774. SCHREMP, F. W. & ROBERSON, G. R. 1975. Effect of supercritical carbon dioxide (CO2) on construction materials. Society of Petroleum Engineers Journal, June 1975,227-233. SHIRAKI, R. & DUNN, T. L. 2000. Experimental study on water-rock interactions during CO2 flooding in the Tensleep Formation, Wyoming, USA. Applied geochemistry, 15, 265-279. SMELLIE, J. A. T. (ed.) 1998. Maqarin natural analogue study, phase III, SKB Technical Report TR-98-04. STEWART, P. B. & MUNJAL, P. K. 1970. The solubility of carbon dioxide in pure water, synthetic sea water and synthetic sea-water concentrates at -5 to 25°C and 10 to 45 atm pressure. Journal of Chemical and Engineering Data, 1591, 67-71. VAN ELDIK, R. & PALMER, D. A. 1982. Effects of pressure on the kinetics of the dehydration of carbonic acid and the hydrolysis of CO2 in aqueous solution. Journal of Solution Chemistry, 11, 339-346. WIEBE, R. 1941. The binary system carbon dioxide-water underpressure. Chemical Reviews, 29,475-481. WIEBE, R. & GADDY, V. L. 1940. The solubility of carbon dioxide in water at various temperatures from 12 to 40° and at pressures to 500 atmospheres. Critical phenomena. Journal of the American Chemical Society, 62,815-817. WOLERY, T. J. 1992. EQ3NR, A Computer Program for Geochemical Aqueous Speciation-Solubility Calculations: Theoretical Manual, User's Guide, and Related Documentation (Version 7.0), Lawrence Livermore National Laboratory Report UCRL-MA110662 PT III, 246p. WOLERY, T. J. & DAVELER 1992. EQ6, A Computer Program for Reaction Path Modeling of Aqueous Geochemical Systems: Theoretical Manual, User's Guide, and Related Documentation (Version 7.0), Lawrence Livermore National Laboratory Report UCRL-MA-110662 PT IV, 338p. WORDEN, R. H. & BARCLAY, S. A. 2000. Internally-sourced quartz cement due to externally-derived CO2 in subarkosic sandstones, North Sea. Journal of Geochemical Exploration, 69-70, 645-649. ZWEIGEL, P., ARTS, R., LOTHE, A. E. & LINDEBERG, E. B. G. 2004. Reservoir geology of the Utsira Formation at the first industrial-scale underground CO2 storage site (Sleipner area, North Sea). In: BAINES, S. J. & WORDEN, R. H. (eds) Geological Storage of Carbon Dioxide, Geological Society, London, Special Publications, 233,165-180.
Reactive transport modelling of CO2 storage in saline aquifers to elucidate fundamental processes, trapping mechanisms and sequestration partitioning JAMES W. JOHNSON*, JOHN J. NITAO & KEVIN G. KNAUSS Lawrence Livermore National Laboratory, Environmental Sciences Division, L-221, P.O. Box 808, Livermore, CA 94550, USA *( e-mail:
[email protected]) Abstract: The ultimate fate of CO2 injected into saline aquifers for environmental isolation is governed by three interdependent yet conceptually distinct processes: CO2 migration as a buoyant immiscible fluid phase, direct chemical interaction of this rising plume with ambient saline waters, and its indirect chemical interaction with aquifer and caprock minerals through the aqueous wetting phase. Each process is directly linked to a corresponding trapping mechanism: immiscible plume migration to hydrodynamic trapping, plume-water interaction to solubility trapping, and plume-mineral interaction to mineral trapping. In this study, reactive transport modelling of CO2 storage in a shale-capped sandstone aquifer at Sleipner has elucidated and established key parametric dependencies of these fundamental processes, the associated trapping mechanisms, and sequestration partitioning among them during consecutive ten-year prograde (active-injection) and retrograde (post-injection) regimes. Intra-aquifer permeability structure controls the path of immiscible CO2 migration, thereby establishing the spatial framework of plume-aquifer interaction and the potential effectiveness of solubility and mineral trapping. Inter-bedded thin shales, which occur at Sleipner, retard vertical and promote lateral plume migration, thereby significantly expanding this framework and enhancing this potential. Actual efficacy of these trapping mechanisms is determined by compositional characteristics of the aquifer and caprock: the degree of solubility trapping decreases with increasing formation-water salinity, whereas that of mineral trapping is proportional to the bulk concentration of carbonateforming elements, principally Fe, Mg, Ca, Na and Al. In the near-field environment of Sleipner-like settings, 80-85% by mass of injected CO2 remains and migrates as an immiscible fluid phase, 15-20% dissolves into formation waters, and less than 1% precipitates as carbonate minerals. This partitioning defines the relative effectiveness of hydrodynamic, solubility, and mineral trapping on a mass basis. Seemingly inconsequential, mineral trapping has enormous strategic significance: it maintains injectivity, delineates the storage volume, and improves caprock integrity. Four distinct mechanisms have been identified: dawsonite [NaAlCO3(OH)2] cementation occurs throughout the intra-aquifer plume, while calcite-group carbonates [principally (Fe,Mg,Ca)CO3] precipitate via disparate processes along lateral and upper plume margins, and by yet another process within inter-bedded and caprock shales. The coupled mineral dissolution/precipitation reaction associated with each mechanism reduces local porosity and permeability. For Sleipner-like settings, the magnitude of such reduction for dawsonite cementation is near negligible; hence, this process effectively maintains initial CO2 injectivity. Of similarly small magnitude is the reduction associated with formation of carbonate rind along upper and lateral plume boundaries; these processes effectively delineate the CO2 storage volume, and for saline aquifers anomalously rich in Fe-Mg-Ca may partially self-seal the plume. Porosity and permeability reduction is most extreme within shales, because their clay-rich mineralogy defines bulk Fe-Mg concentrations much greater than those of saline aquifers. In the basal caprock shale of our models, these reductions amount to 4.5 and 13%, respectively, after the prograde regime. During the retrograde phase, residual saturation of immiscible CO2 maintains the prograde extent of solubility trapping while continuously enhancing that of mineral trapping. At the close of our 20year simulations, initial porosity and permeability of the basal caprock shale have been reduced by 8 and 22%, respectively. Extrapolating to hypothetical complete consumption of Fe-Mg-bearing shale minerals (here 10 vol.% Mg-chlorite) yields an ultimate reduction of about 52 and 90%, respectively, after 130 years. Hence, the most crucial strategic impact of mineral trapping in Sleipner-like settings: it continuously improves hydrodynamic seal integrity of the caprock and, therefore, containment of the immiscible plume and solubility-trapped CO2.
Sufficient curbing of projected anthropogenic CO2 emissions to achieve a stabilized 'safe' atmospheric concentration ranks high among the grand challenges of this century. In the near term, significant emissions reduction can only be achieved through innovative capture/isolation strategies applied to
point-source waste streams. Among currently proposed storage techniques, injection into confined geological formations - in particular saline aquifers, given their immense storage capacity and widespread geographic distribution - represents one of the most promising alternatives. Successful
From: BAINES, S. J. & WORDEN, R. H. (eds) 2004. Geological Storage of Carbon Dioxide. Geological Society, London, Special Publications, 233,107-128. 0305-8719/047$ 15.00 © The Geological Society of London 2004.
108
J.W. JOHNSON ETAL
implementation of this approach hinges on our ability to predict the relative effectiveness of subsurface CO2 migration and sequestration processes (isolation performance) as a function of key aquifer and caprock properties (screening criteria), which will enable us to identify optimal sites and forecast their long-term security. Specific requirements of this predictive capability can be obtained by subdividing these competing processes into their fundamental components. In terms of migration, injected CO2 moves by volumetric displacement of formation waters, with which it is largely immiscible; by gravity segregation, which causes the immiscible plume to rise owing to its relatively low density; and by viscous fingering, which causes it to migrate preferentially into local highpermeability zones owing to its relatively low viscosity. In terms of sequestration, some fraction of the rising plume will dissolve into formation waters (solubility trapping); some fraction will react with formation minerals to precipitate carbonates (mineral trapping); and the remaining fraction will reach and become isolated beneath the caprock (hydrodynamic trapping), migrate up-dip along this interface, and accumulate in any local topographic highs (structural trapping). Numerical simulation of these interdependent migration and sequestration processes requires a computational capability that explicitly represents and couples multiphase flow and kmetically controlled geochemical processes within porous media characterized by physical and compositional heterogeneity. We have developed a unique computational package that implements this capability, and in earlier studies (Johnson etal 2001; Johnson & Nitao 2002) used it to address a series of key technical issues regarding CO2 storage in saline aquifers. In these investigations, we quantified the dependence of migration/sequestration balance on permeability structure and composition, the relative effectiveness of hydrodynamic, solubility, and mineral trapping mechanisms, and the isolation performance of a typical shale caprock. Further, we introduced the concept of prograde (active-injection) and retrograde (post-injection) storage regimes, and demonstrated that residual saturation of immiscible CO2 during the latter maintains or enhances prograde trapping mechanisms. In the present contribution, we first provide an overview of the reactive transport modelling approach and our simulation capabilities. We then use this methodology and computational package to elucidate and describe the fundamental processes, associated trapping mechanisms, and sequestration partitioning that characterize prograde and retrograde CO2 storage regimes in Sleipner-like settings. Based on this description, we propose screening criteria that portend optimal isolation performance in
such environments. Finally, we outline a series of important future investigations that are posed by the results of this study.
Reactive transport modelling: methodology and simulation tools Reactive transport modelling is an advanced computational method for quantitatively predicting the long-term consequences of natural or engineered perturbations to the subsurface environment (Johnson et al. 1999). Because these predictions typically involve space, time and system complexity scales that preclude development of direct analytical or experimental analogues, they often represent a unique forecasting tool. The necessary point of departure for predictive investigations of this kind is established by successful application of the method to simulate well-constrained laboratory experiments (e.g.Bertrandtffl/. 1994; Johnson
REACTIVE TRANSPORT MODELLING OF CO7
Fig. 1. Schematic depiction of interdependent subsurface processes that redistribute mass and energy in response to the disequilibrium state (lateral 7", P, pf gradients) imposed by natural or engineered perturbation events. Porosity and permeability are the key variables that link thermal-hydrological (blue), geomechanical (maroon), and geochemical (aqua) sectors of the diagram.
tional heterogeneity. The package implements an integrated finite-difference, spatial discretization to solve the flow and reactive-transport equations, using the Newton-Raphson method to solve the resulting non-linear systems at each time step. Explicit account is taken of multiphase advection, diffusion, and dispersion; of relative permeability and capillary pressure, using an extended Van Genuchten formulation (Parker et al. 1987); and of kinetically controlled fluid-mineral reactions, using rate laws from transition state theory (e.g. Lasaga 1998). Moreover, explicit account is also taken of coupling between these transport and geochemical processes through the dependence of permeability on porosity changes due to mineral precipitation/dissolution, using a normalized Kozeny equation (Scheidegger 1974), and through the dependence of fluid-phase volumetric saturations on gas (e.g. CO 2(} ) generated or consumed by fluid-mineral reactions. The equation-of-state and viscosity formulations implemented in NUFT for supercritical CO2 are those developed by Span & Wagner (1996) and Fenghour & Wakeman (1998), respectively. The corresponding formulations implemented for H2O are those presented by Meyer et al. (1993). Chemical
109
interaction between and within distinct fluid phases is governed by inter- and intra-phase equilibrium constraints. Activity coefficients for charged aqueous solutes are represented using an extended form of the Debye-Huckel equation (B-dot formulation: Helgeson 1969), those for non-polar neutral solutes are represented using the Drummond (1981) model, and those for polar neutral solutes are taken to be unity. The GEMBOCHS system (Johnson & Lundeen 1994a,&, 1995) integrates a comprehensive relational thermodynamic/kinetic database and dedicated software library that together facilitate generation of application-specific thermodynamic/ kinetic datafiles for use with a variety of geochemical modelling codes and reactive transport simulators. The thermodynamic database covers about 3200 distinct chemical species, spanning 86 elements of the periodic table; its core component is the current version of the SUPCRT92 database (Johnson et al 1992; Shock 1998), which covers about 1550 species, spanning 82 elements. Custom datafiles are generated using Jewel, a GUI-driven software package that extrapolates reference-state properties to elevated P-T conditions using a number of standard algorithms, the core set of which are those encoded with the SUPCRT92 software package (Johnson et al 1992). These include global- and critical-region equations of state and a dielectric formulation for H2O (Johnson & Norton 1991) that are explicitly integrated with equations of state for both aqueous solutes (Tanger & Helgeson 1988; Shock et al 1992) and minerals/gases (Helgeson etal 1978).
CO2 storage at Sleipner: overview and model definition Statoil's North Sea Sleipner facility is the world's first saline-aquifer CO2 storage site. Here, excess CO2 from a natural-gas production stream is removed by amine absorption, then stripped from the amine, and subsequently injected into the Utsira formation 1000m below the seabed for the purpose of environmental isolation (Gregersen et al 1998). Since October 1996, this process has diverted roughly one million tons of CO2 annually from atmospheric release. At Sleipner, the CO2 injection well extends 3^-km horizontally before reaching the expulsion zone - a screen length of 100m near the base of the Utsira. As a result of this lengthy transport, the thermal perturbation associated with CO2 injection is negligible. In addition, the 200m-thick Utsira is laterally extensive and consists of extremely permeable unconsolidated sandstone, capped by the several hundred metre thick Nordland shale. As a consequence, CO2 injectivity is high and injection-induced pressure anomalies along
110
J. W. JOHNSON ETAL.
the caprock interface are minimal. Hence, at Sleipner thermal and geomechanical processes represent second-order effects, whereas interdependent multiphase flow and geochemical processes dominate integrated system response to the perturbation event. Numerical simulation of this scenario - which typifies desirable saline-aquifer storage sites - involves constructing representative spatial domains, assigning to these the appropriate initial and boundary conditions based on hydrological and compositional data, and compiling the thermodynamic and kinetic data required to represent compositional evolution of the system.
Simulation domains XSH, CSH, and DSH In this study, all of the reactive transport simulations have been conducted within a single spatial domain, which represents the near-field environment of CO2 disposal at Sleipner (Fig. 2a), and over a single time frame, which encompasses equal-duration prograde and retrograde regimes. In the common physical setting, an Utsira-like saline aquifer (200m thick) is confined by a shale caprock (25 m), which itself is overlain by a thin confined saline aquifer (25 m) to facilitate evaluation of caprock performance. In the common 20-year time frame, an injection rate of 10 000 tons per year CO2 is first maintained for 10 years (prograde regime), then ramped down to zero over three months (prograde-retrograde transition phase), and finally maintained at zero for another 9.75 years (retrograde regime). This spatial domain and injection rate correspond to a one metre-thick cross-section through and perpendicular to the actual 100m screen length of CO2 injection at Sleipner. Within this domain, three distinct injection scenarios - models XSH, CSH and DSH - have been evaluated. Model XSH (Fig. 2b) examines CO2 injection into a shale-capped homogeneous sandstone aquifer. Models CSH and DSH impose into XSH four thin (3 m thick) intra-aquifer shales, which are separated from the caprock and each other by 25m. Model CSH (Fig. 2c) examines the effect of imposing laterally-continuous microfractured shales, whose assigned permeability (3 mD) equates to a continuum representation of 100 (mm fractures spaced roughly 30m apart. Model DSH (Fig. 2d) examines the effect of imposing laterally-discontinuous shales, which are bridged by lateral facies change to sandstone; assigned permeability of these shales (3 |mD; same as the caprock) reflects typical shale integrity (Freeze & Cherry 1979). Each model contains about 4000 variably sized grid cells, which range in width-by-height from 0.5 X 1.0m for the injection well to 25 X 5m near lateral boundaries. The base and top of the domains
Fig. 2. (a) CO2 injection into a saline aquifer at Statoil's North Sea Sleipner facility; (b-d) the spatial domains adopted for reactive transport simulation of the near-field environment (region within a few 100m of the injection well) in models XSH, CSH, and DSH.
are impermeable to fluid flow and mass transfer. Constant hydrostatic head and geochemical conditions are maintained within a column of vanishingly thin lateral boundary cells, which therefore serve as an infinite sink for outward migration of both immiscible CO2 and aqueous phases; i.e. lateral domain boundaries can be viewed as permeable to outward fluid flow and mass transfer. An ambient flow field has not been imposed within the saline aquifer, nor has any degree of tilt or non-planar topography been imposed on the aquifer-caprock interface.
Hydrological and compositional data Pressure-temperature conditions, porosities, and permeabilities assigned to the saline aquifer are consistent with those reported for the Utsira formation; such data are presently unavailable for inter-bedded
111
REACTIVE TRANSPORT MODELLING OF CO,
Table 1. Pressure-temperature conditions, porosities, and permeabilities reported for the saline aquifer, shale caprock, and intra-aquifer shales at Sleipner and those adopted in this study. Parameter
Utsira Formationa
Saline aquifer5
Shale caprockb'c
Intra-aquifer shales (CSH)b'<
Intra-aquifer shales (DSH)b c
T(°C) P (bars) poros. (%) perm, (m2) thickness (m)
37 80-110 35-40 (1-8)(10-12) 150-250
37 90-110 35 3(10-12) 200
37 87.5-90 5d 3(10-18)d 25
37 — 5d 3(10-15)e 3
37 — 5d 3(10-18)d 3
a b c d e
Gregersen^a/. (1998) adopted in this study data from or directly representative of Sleipner are unavailable average values (Freeze & Cherry 1979) modified from average value (Freeze & Cherry 1979) per adopted density and aperture of microfractures; see text
and caprock shales at Sleipner, and have been estimated (Table 1). Compositional data adopted for the saline aquifer are based on mineralogy reported for a representative North Sea formation (Table 2), which specifies an impure quartz sand, and a fluid analysis reported for the Utsira itself 200km north of Sleipner at Oseberg, which defines a seawater-like aqueous phase (Table 3). Analogous data for interbedded and caprock shales directly representative of those at Sleipner are currently unavailable; these units have been assigned an average shale mineralogy (Table 2) and a fluid composition identical to that of the saline aquifer (Table 3). Our approach of incorporating Fe-Mg solid solutions as representative Mg end-member components (Table 2) reflects the necessity of removing O2(aq) and total Fe from the fluid analysis (see Table 3 footnotes). Note that within the system Fe-C-O-H for the P—T conditions and injection CO2 fugacity at Sleipner (see below), siderite is stable over a wide range of O 2 ( } fugacity - from the lower limit of H2O stability (10~832) to hematite-siderite equilibrium (10~ 527 ) - roughly centred about the magnetitehematite buffer (10~690). Hence, incorporating Fe-Mg solid solutions as Fe end-member components would have little effect on the present study beyond replacing magnesite (MgCO3) with siderite (FeCO3) as the relevant calcite-group carbonate. In all likelihood, reality lies within the middle ground of siderite-magnesite solid solutions. The minor calcite fraction (5%) reported for the Utsira proxy (Table 2) was excluded in the simulations because the actual Utsira formation is distinctly unconsolidated (Gregersen et al. 1998), which reflects a lack of carbonate cement, whereas the adopted shale mineralogy excludes carbonates because this constraint is most typical of such lithologies (Blatt et al 1972). Further, in the context of the present modelling work, it is advantageous to
restrict mineral-trapped CO2 to the form of injection-triggered carbonate precipitation, which simplifies account of sequestration partitioning into mineral-trapping mechanisms. These first-order approximations are made while recognizing that many uncemented sandstones do contain small fractions of clastic or biogenic calcite, carbonaceous shales are far from uncommon, and that these complications need to be addressed in future work. The adopted waste stream composition is pure CO2, and it is injected under supercritical conditions (37 °C, 110.5 bar) at the base of the saline aquifer. At this pressure and temperature, the injection CO2 fugacity is 61.05 bar in the context of the adopted CO2 fugacity coefficient (see below). Although aqueous solute concentrations are explicitly accounted for (Table 3), PVT properties of the aqueous phase are here taken to be those of pure H2O instead of an actual seawater-like fluid in the absence (ambient conditions) or equilibrated presence (prograde/retrograde regimes) of an immiscible CO2 phase. At the relevant P-T conditions, density differences introduced by this approximation are less than 3% with respect to pure seawater (Fofonoff & Millard 1983), and less than 4% with respect to seawater equilibrated with immiscible CO2 (Ennis-King & Paterson 2003). Hence, this simplification has near-negligible impact on prograde buoyancy-driven immiscible CO2 migration (and dependent aqueous flow), where the density contrast between immiscible CO2 and aqueous phases is 30-40%. However, it could be significant during the retrograde phase, when aqueous convection may be governed by very small density contrasts (on the order of 1%) between aqueous phases that are and are not equilibrated with residual immiscible CO2 (Ennis-King & Paterson 2002, 2003; Lindeberg & Bergmo 2002). As is appropriate, the aqueous phase is taken to be
J.W. JOHNSON ETAL.
112
Table 2. Saline-aquifer and shale mineralogies reported at Sleipner and those adopted in this study; abundances are given in volume percent. Mineral
Miocene sanda
Saline aquiferb
Shaleb'c
Quartz K-feldspar Plagioclase 'Mica' calcite 'Clay' Muscovite Phlogopite1 Mg-chlorite'J Grain diameterk (mm)
79 6 3 5d 5e 1 0 0 0 unreported
80f 10f 5f 0 0 0 3 2 0
35 5 0 0 0 0 50h 0 10h
0.251
0.025m
a
provided by Tore Torp (Statoil) and Niels Springer (GEUS) as a proxy for an analysis from the Utsira Formation (pers . comm.)
Table 3. Aqueous concentrations (in molality) reported for seawater, the Utsira formation at Oseberg, and those adopted in this study. Component
Seawater3
Osebergb
This study0
Sodium Potassium Calcium Magnesium Strontium Barium Aluminum Silica Iron [total] Chloride Bicarbonate Carbonate Sulphate Oxygen [O2(aq>] (ppb)
0.48 0.010 0.010 0.054 unreported unreported unreported unreported unreported 0.56 0.0024 2.7(10-4) 0.028 unreported
0.4520 0.0053 0.0106 0.0259 1.14(10-4) 3.64(10-6) unreported unreported 3.58(10-5) 0.5213 0.0116 unreported undetected
Oseberg Oseberg 0.00742d 0.01813d
5
Ok
pH
8.15
7.1
7.0-7.21
Oe Oe 1.3(10-8)f 1.664(10~4)8
Oh
Oseberg 0.00232' —i Oseberg (0)
b QrJr»r»tc«/-l in tlii<2 etiiHt/ c
data from or directly representative of Sleipner are unavailable; adopted mineralogy (60% clay minerals, 35% quartz, 5% feldspar) is based on average shale compositions (Blatt et al 1972) d incorporated as 3% muscovite and 2% phlogopite, the latter as a proxy for trace glauconite and biotite, which have been reported for the Utsira (Gregersen et al 1998) e eliminated per unconsolidated nature of the Utsira, which implies a lack of carbonate cement f augmented per elimination of calcite (see previous footnote) h representing the 60% clay fraction with this muscovite/ Mg-chlorite ratio preserves the typical K^O^MgO+FeO) ratio of shales (Blatt et al. 1972), while permitting avoidance of more realistic illite, smectite, and montmorillonite solid solutions, for which thermodynamic and kinetic data are lacking 1 Mg end-member component used to represent Fe-Mg solid solutions (see text) Jclinochlore-14A k nominal value based on the Udden-Wentworth size grade scale 1 consistent with 'fine to medium grained sand', which has been reported for the Utsira formation (SACS 2000) m consistent with 'mixture of very fine sand, silt, and clay particles', which typifies shale lithologies (Blatt et al. 1972)
the wetting fluid. Hence, chemical interaction of the immiscible CO2 fluid and kinetically-reacting mineral grains occurs exclusively through a grainsurrounding aqueous-phase film, whose thickness decreases with increasing immiscible CO2 saturation. The residual saturation of immiscible CO2 - the limit below which it is no longer a contiguous (advectively mobile) phase - is taken to be 0.05, which falls at the low end of values reported for various sandstones (Holtz 2002, 2003). Irreducible
a
average surface seawater at 25°C (Garrels & Christ 1965) from Gregersen et al. (1998), who report concentrations in g/L; this analysis is from the Utsira at Oseberg, about 200km north of Sleipner c adopted for both the saline aquifer and shale units d 70% of Oseberg value; reduced to obtain undersaturation with respect to calcite and magnesite per the absence of carbonate cement in the Utsira e trace concentrations removed f between K-feld + muscovite and kaolinite + quartz equilibrium (37 °C, 100 bar) s quartz equilibrium (37 °C, 100 bar) h removed because ferrous/ferric partitioning cannot be constrained (see footnote k) 1 20% of Oseberg value (roughly equivalent to average seawater); reduced to obtain undersaturation with respect to calcite and magnesite per the absence of carbonate cement in the Utsira; ambient bicarbonate concentration (reduced or not) is completely overwhelmed by the effect of CO2 injection •i in this study, 'bicarbonate' concentration combines both bicarbonate and carbonate contributions k removed because the Oseberg value cannot represent an in-situ concentration; at 37 °C, HObar, it defines an equilibrated O2(g) fugacity of 10~3-8 (cf. the magnetite-hematite buffer under these conditions: 10~69-°); hence, the reported O2(aq) cannot be used to constrain the ambient oxidation state and dependent ferrous/ferric partitioning 1 range encompasses extra-plume pH evolution within both the saline aquifer and all shales over the 20-year simulation b
water saturation is taken to be 0.20. Adopted Van Genuchten parameters are as follows: m (sorting parameter) is taken to be 0.4 (moderately well sorted) for both sandstone and shale, whereas a (inversely proportional to bubble pressure) is taken to be 66 and 0.18 bar"1 for sandstone and shale, respectively.
113
REACTIVE TRANSPORT MODELLING OF CO, Table 4. Adopted kinetic data for primary (boldface) and observed (precipitated) secondary minerals. Mineral
Pr-Tr diss/pptn rate constanta-b (mol m~ 2 s"1)
Activation energy3 (kJmol- 1 )
Reactive specific surface area' (m2 m-3)mineral
Surface area scaling factor'1
Quartz K-feldspar Plag-Ab80 Muscovite Phlogopite Mg-chloriteP Magnesite Dawsonite Calcite Cristobalite Chalcedony Kaolinite Talc Paragonite Pyrophyllite Gibbsite
1.035(10-'4)e 1. 778(10- 10)f 5.623(10-13)f 1. 000(10" 13)8 4.000(10- 13)§ 3.000(10- 13)8 1.000(10-9)h 1.000(10-7)a 1.500(10-6}J 3.450(10- 13)k 3.450(10- 13)im 4.000(10- 13)8 1.000(10-12)8 1. 000(10- 13)in 4.000(10-13)io 3.000(10- 13)8
87.7e 51.7f 80.3f 22.08 29.0s 88.08 62.81 62.81 62.81 62.81 62.81 29.0§ 42.08 22.0in 29.0io 62.81
5.741(104) 1.113(105) 8.752(104) 1.246(106) 1.246(106) 1.246(106) — — — — — — — — — —
2.39 4.64 3.65^ 51.90 51.90" 51.90" — — — — — — — — — —
a
adopted data are those for the relevant acidic conditions (pH 4-5.5), which characterize the intra-plume aqueous phase during both prograde and retrograde regimes b dissolution and precipitation rate constants are taken to be of equal magnitude, recognizing that this approximation is crude in some cases c specific surface areas (SSA) are those of representative spheres (see Table 2) multiplied by the surface area scaling factor; reactive SSA are SSA multiplied by the fraction of SSA presumed in contact with the aqueous phase; reactive SSA for each mineral is equivalent in the aquifer and shales (see text) d factor of increase observed for lOOfjum grains whose SSA has been measured using BET methods by Knauss & Copenhaver (1995), Knauss & Wolery (1986,1988,1989),andStfflings*fa/. (1996) e Tester etal. (1994);f Blum & Stillings (1995); 8 Nagy (1995) h Pokrovsky & Schott (1999);' estimated in the present study J Jordan & Rammensee (1998);k Renders et al. (1995) 1 estimated as intermediate to calcite and magnesite m presumed equivalent to cristobalite " presumed equivalent to muscovite 0 presumed equivalent to kaolinite P clinochlore-14A 1 based on observed scaling factors for albite and anorthite
Thermodynamic and kinetic data Equilibrium thermodynamic characterization of the adopted compositional framework (Tables 2 & 3) is specified within a data file for the 10-component system K-Na-Ca-Mg-Al-Si-C-O-H-Cl that includes hydrolysis constants for 36 aqueous solutes, 2 gases, and 70 minerals at 100bar and 20-90 °C. This equilibrium reference frame is derived almost entirely from those data - and using exclusively those equations of state - contained in the SUPCRT92 software package (Johnson et al 1992; Shock 1998). Adopted thermodynamic data beyond those contained in SUPCRT92 include Debye-Huckel ion-size parameters for aqueous solutes (Johnson & Lundeen 1994Z?), a fugacity coefficient (0.55) for CO2(g) (Garrels & Christ 1965), and reference-state properties for dawsonite (Robie et al. 1978). In addition, data for three plagioclase
solid-solution compositions (plag-Ab80, plagAb50, plag-Ab20) were incorporated; these were derived using an ideal site-mixing model applied to SUPCRT92 data for albite and anorthite. Finally, dolomite and antigorite were suppressed from precipitating, owing to the well known (but poorly quantified) extreme discrepancy between lowtemperature dissolution and precipitation rates of the former (e.g. Berner 1971), and problematic stoichiometry of the latter. Requisite kinetic data are also specified within the data file; specifically, reference-state mineral dissolution and precipitation rate constants, activation energies, and reactive specific surface areas (Table 4). In general, these parameters are markedly less certain than the related thermodynamic data. The reactive specific surface area of dissolving and precipitating minerals in the subsurface is a particularly important parameter that is poorly quantified at
114
J.W. JOHNSON ETAL.
present. Here, the following approach has been used to assign values to this key variable. First, the specific surface area is calculated for spherical grains of the relevant size fraction, presuming the nominal grain diameter is equivalent for all minerals within a given lithology. Adopted diameters are 0.25 mm for the saline aquifer and 0.025 mm for the shales (Table 2); hence, idealized specific surface areas are ten times larger in the shales than in the aquifer. These idealized values are then multiplied by a scaling factor derived from experimentally observed factors of increase between specific surface areas of lOOjjim spheres and BET-measured specific surface areas of lOOjuim grains for quartz, albite, anorthite, Kfeldspar, and muscovite (Table 4). Scaling factors for primary minerals outside this group are assigned based on closest structural analogy. In the highly porous and permeable saline aquifer, all of the mineral surface area is considered to be available reactive surface area, i.e. in contact with the aqueous phase. However, within typical shales only a small fraction of the total mineral surface area is in such contact; specifically, those grain surfaces along and within diffusion distance of fracture walls. As a first approximation, this reactive fraction of the total surface area is taken to be 10%. Because the specific surface area of shales was taken to be 10 times that characterizing the aquifer, in this continuum model reactive specific surface area in the shales and aquifer is identical (Table 4). For precipitation of secondary minerals in both the aquifer and shales, all reactive surface area of primary minerals is considered to represent available substrate; i.e. 100% and 10% of the total surface area per bulk volume for the aquifer and shales, respectively. Within the adopted implementation of transition state theory rate laws, mineral dissolution and precipitation rate constants are assigned equivalent absolute magnitude, precipitation onset does not require any degree of aqueous-phase supersaturation, and the potential limiting effect of armouring on precipitation is not represented. These common approximations, taken together with the foregoing description of reactive specific surface areas, provide a first-order estimate of mineral trapping effectiveness that probably falls near the high end of reality.
Fundamental processes and associated trapping mechanisms The ultimate fate of CO2 injected into saline aquifers for environmental isolation is governed by three interdependent yet conceptually distinct processes: CO2 migration as a buoyant immiscible fluid phase, direct chemical interaction of this rising plume with ambient saline waters, and its indirect interaction with aquifer and caprock minerals through the
aqueous phase. The first process is directly linked to hydrodynamic trapping, the second to solubility trapping and pH evolution, and the third to mineral trapping (and pH evolution). In this section, we quantify and compare the relative effectiveness of these three trapping mechanisms for models XSH, CSH, and DSH during prograde and retrograde CO2 storage.
CO2 immiscible migration and hydrodynamic trapping During prograde storage, initial ascent of the immiscible CO2 plume towards the caprock is governed by four constraints. The first two, density contrast between CO2 and formation waters and the absolute formation permeability, effectively determine the injection overpressure required to acheive a given influx rate, which eventually translates to a corresponding pressure anomaly along the caprock interface. In model XSH, the injection-point density contrast is only about 30%, but permeability is extremely high (3D); hence, injection overpressure amounts to less than one bar. The second pair of constraints - saturation-dependent relative permeability of the formation to immiscible CO2 and pressuredependent volumetric expansion of this phase during ascent - effectively controls dynamic plume configuration, causing concomitant lateral telescoping of its main conduit and lateral expansion of its leading edge (Fig. 3a; 2.4-, 6.8-, and 12-day insets). Interplay of these two processes creates a zone of prograde residual CO2 immiscible saturation, which marks the wake of initial plume ascent. In model XSH, immiscible CO2 reaches the caprock and lateral domain boundaries within 15 and 50 days, respectively. The plume cap, column, and prograde residual saturation zones all attain near steady-state geometric plume configuration within one year (Fig. 3a). At this time, CO2 saturation has reached 0.50-0.58 within the uppermost 25 m of the aquifer (plume cap), whereas in the plume column it increases from 0.13 just below the cap to 0.22 at 25 m above the injection well. Hence, a steep vertical gradient connects distinct saturation profiles in the plume cap and column. Subsequent prograde evolution of CO2 saturation is limited to a gradual increase in the plume cap, from 0.50-0.58 at one year to 0.60-0.67 after 10 years (Fig. 3b). Owing to 18% specific-volume expansion of the immiscible CO2 fluid as it rises 200 m in the aquifer (20bar pressure drop), once steady-state geometric configuration of the plume is attained, the intraplume aqueous phase is displaced primarily downward within and slightly outward from the plume column and cap during prograde storage (Fig. 3b). Velocities of this injection-induced aqueous flow are
REACTIVE TRANSPORT MODELLING OF CO2
Fig. 3. CO2 immiscible saturation in model XSH. (a) Prograde residual saturation tracks initial ascent of the plume, whose steady-state configuration obtains within one year; (b) cap-zone saturation slowly increases during prograde migration; (b,c) injection-induced aqueous flow reverses direction at the prograde/retrograde transition; (d) residual saturation drives retrograde trapping mechanisms.
15-575 cm a ' at one year, with maximum values localized within the lower two-thirds of the plume cap. The maximum values decrease with time to llOcma" 1 at two years, ^-Ocma"1 at five years, and 20 cm a"1 at ten years. Such velocities are of the order of those typically encountered in unperturbed saline aquifers. Just nine months into the retrograde phase, continued upward and outward plume migration has reduced the plume cap to a new steady-state thickness of 5 m (single horizontal layer of domain cells), within which CO2 saturation has dropped to 0.52 (Fig. 3c). This thin veneer now caps a zone of
115
uniform residual saturation that encompasses the prograde plume column and residual saturation zones. During the retrograde regime, aqueous-phase flow is directly opposite to that of the prograde regime, now primarily upward and slightly inward within the relict plume as the aqueous phase migrates into the void left by rising CO2; while, flow velocities are similar in absolute magnitude. Ten years into the retrograde phase, plume-cap saturation has diminished to 0.15 (Fig. 3d). During the prograde regime of model XSH, roughly 85% by mass of injected CO2 remains and migrates as an immiscible fluid phase. Hence, gravity segregation is the dominant migration process and hydrodynamic trapping represents the most important potential trapping mechanism. In this model, the 25m (3|xD) shale caprock provides a very secure seal. In particular, CO2 does not migrate completely through the 5 horizontal layers of 5m-thick shale grid cells. Within the uppermost 20m of shale, the maximum CO2 saturation attained varies from 0.000015 in the upper 5m to 0.0001 in the lowest 5 m, and saturation is declining at 20 years. In the 5 m of shale immediately overlying the aquifer, maximum saturation of 0.047 (less than the residual limit of 0.05) is attained at 14 years and declines thereafter to 0.042 at 20 years. Hence, the 20-year isolation performance of this shale caprock is both excellent and improving with time. The presence of thin intra-aquifer shales has a profound influence on immiscible-plume migration and dependent sequestration mechanisms. For the case of laterally continuous microfractured shales (model CSH), the most obvious, and significant, effects are vertical bifurcation of the plume cap and a tremendous resultant increase in the spatial extent of plume-aquifer interaction (Fig. 4a,b), which greatly enhances the potential effectiveness of solubility and mineral trapping. When laterally-discontinuous tight shales are imposed (model DSH), these effects are somewhat less pronounced; however, the lateral breaks lead to spatial recursion of focused vertical CO2 migration paths and lateral saturation gradients throughout the near-field domain (Fig. 4c,d), further enhancing the potential effectiveness of mineral trapping. In model CSH, maximum steady-state CO2 saturation within the four relatively permeable intraaquifer shales (from 0.42 in the lowest to 0.25 in the highest) is attained within 5-6 years during prograde storage. During the first year of the retrograde regime, saturation within all four decreases to the residual limit of 0.05, and by 18 years has been reduced to zero (Fig. 4b). In model DSH, saturation in the four relatively impermeable shales increases slowly throughout the prograde regime (reaching maximum values of 0.06-0.10). It then decreases slowly throughout the retrograde phase, eventually
116
J.W. JOHNSON ETAL
Fig. 4. CO2 immiscible saturation during prograde anc retrograde plume migration for models CSH and DSH Spatial extent of the steady-state plume for CSH and DSH exceeds that of XSH by factors of 3.2 and 1.4, respectively.
reaching zero at 18 years (Fig. 4d). Ultimate depletion of immiscible CO2 in the intra-aquifer shales of CSH and DSH reflects its eventual consumption by carbonate-precipitating fluid-mineral reactions (described below). CO2 saturation profiles for models XSH, CSH, and DSH at 3 years are shown together with the 3year seismic profile of immiscible CO2 accumulations at Sleipner in Figure 5. Comparison of these four images strongly suggests the presence of thin intra-aquifer shales in the upper half of Utsira Formation. Moreover, close correspondence of model CSH and the seismic data suggests that these shales are predominantly contiguous laterally and have microfracture permeability on the order of CSH values (3mD). The presence of lateral facies
Fig. 5. Plume configuration after 3 years in models XSH, CSH, and DSH as compared to that observed seismically at Sleipner, which strongly suggests that the Utsira combines elements of models CSH and DSH.
changes to sand would result in a much greater proportion of injected CO2 reaching the caprock after 3 years, as in model DSH, where the lateral breaks are quite restricted (10-25m in width) relative to typical field settings.
Solubility trapping andpH evolution Chemical interaction of the immiscible CO2 plume with saline formation waters causes a dramatic increase in total aqueous carbon concentration, primarily as CO2(aq) and HCO3~, and a substantial decrease in pH. The coupled reaction can be expressed as:
REACTIVE TRANSPORT MODELLING OF CO2
117
Fig. 6. CO2 aqueous concentrations (composite molality of all carbon-bearing aqueous solutes) and pH in model XSH during prograde (1 and 10 years) and retrograde (20 years) regimes.
C02(g) + H20^±C02(aq) + H2O^HC03- + H + . (1) In model XSH, formation waters equilibrate on contact with the ascending CO2 plume, creating a coincident zone of solubility-trapped CO2. For the adopted aqueous-phase composition (Table 3), injection CO2( ) fugacity, and vertical pressure gradient, the CO2 aqueous solubility limit varies from about 1.2 molal near the injection well to about 1.1 molal at the aquifer-caprock interface (Fig. 6a), accounting for roughly 15% of injected CO2 mass within the near-field environment. This solubility limit will decrease with increasing ionic strength, owing to the 'salting-out-effect' (Garrels & Christ 1965). Hence, the effectiveness of solubility trapping will decrease with increasing formation water salinity; this compositional dependence is accounted for in simulation studies by adoption of an appropriate activity-coefficient model for dis-
solved-gas aqueous solutes (e.g. Drummond 1981). Because solubility-limited CO2 aqueous concentration is maintained by the presence of CO2(g), the composition-dependent degree of solubility trapping is sustained throughout the prograde regime (Fig. 6a,b) by continuous CO2 injection (Fig. 3a,b) and maintained during the retrograde phase (Fig. 6c) by residual CO2 saturation (Fig. 3c,d). The presence of intra-aquifer shales retards vertical but promotes lateral migration of the immiscible CO2 plume, thereby expanding the spatial extent of plume-aquifer interaction and reducing volume averaged CO2 saturation within the near-field environment. The expanded interaction volume increases the total mass of CO2 trapped within the near-field aqueous phase, while the reduced average saturation increases the mass so trapped per mass of immiscible CO2, increasing the relative effective-
118
J. W.JOHNSON ET AL.
Fig. 7. CO2 aqueous concentrations (composite molality of all carbon-bearing aqueous solutes) and pH in models CSH and DSH at the close of prograde (10 years) and retrograde trapping (20 years).
ness of solubility trapping. In models CSH and DSH, the total mass of solubility-trapped CO2 exceeds that for model XSH by a factor of 3.2 and 1.4, respectively, which is equivalent to the factorof-increase in spatial extent of the solubility-limit CO2 aqueous concentration (cf. Figs 6b,c & 7a-d). The mass ratio of solubility-trapped to immiscible CO2 in models CSH and DSH exceeds that of XSH by 32 and 44%, respectively, which increases
sequestration partitioning into the aqueous phase from 15 to 17 and 19%, respectively. CO2 aqueous concentration has dropped below even the ambient value within all intra-aquifer shales of CHS and DSH 10 years into the retrograde regime (Figs 7b,d). This reflects complete consumption of CO2 from the intra-plume aqueous phase (and therefore from the immiscible plume itself) by carbonateprecipitating fluid-mineral reactions.
119
REACTIVE TRANSPORT MODELLING OF CO2
Intra-plume pH evolution is governed by two opposing processes: aqueous solubility of CO2(g), which dramatically decreases pH, and dependent kinetic dissolution of silicate minerals, which increases it. CO 2() solubility lowers pH from the ambient value of 7.1 to about 3.4 upon initial plume-formation water contact, whereas resultant initial silicate dissolution increases pH to roughly 4.5 throughout the plume by the time its steady-state configuration is attained (Fig. 6d). Continued silicate dissolution throughout the prograde regime affects continued increase in pH from 4.5 to 4.9 in the plume column and residual saturation zone, and from 4.5 to 5.2 in the plume cap (Fig. 6e). During the retrograde phase, residual CO2 saturation maintains the solubility-induced background pH, which continues to drive silicate dissolution, and in turn causes further pH increase, to about 5.3 throughout the relict plume at 20 years (Fig. 6f). The presence of intra-aquifer shales has two important effects on aquifer pH distribution and dependent mineral trapping mechanisms. First, the expanded region of low-pH fluids (Fig. 7e-h) and silicate minerals under acid attack represents a similarly expanded source region of potential carbonateforming cations. Second, spatial recursion of focused vertical CO2 migration paths in model DSH leads to similar recursion of steep lateral gradients in pH (Fig. 7c,d,g,h), which play an important role in mineral trapping. The relatively high pH values observed within intra-aquifer shales of CSH and DSH after 20 years (Fig. 7f,h) are reached rapidly between 16 and 18 years as the final vestige of CO2is consumed by local carbonate precipitation.
Mineral trapping In saline aquifers, mineral trapping will occur primarily in the form of dawsonite [NaAlCO3(OH)2] and the calcite-group carbonates, most significantly siderite [FeCO3], magnesite [MgCO3], calcite [CaCO3], and their solid solutions. In the present models, we evaluate the contributions of dawsonite, magnesite (representing siderite-magnesite solutions), and calcite, which precipitate by four distinct mechanisms: dawsonite cementation occurs throughout the intra-aquifer plume, while magnesite and calcite precipitate via disparate processes along lateral and upper plume margins, and also by another process within inter-bedded and caprock shales. Intra-plume dawsonite cementation. Dawsonite precipitation results from high ambient Na and injection-induced CO2 concentrations in the aqueous phase together with acid-induced kinetic dissolution
of K-feldspar. The coupled reaction (in both the aquifer and shale environments) can be expressed as:
K-feldspar
(2) dawsonite
silica
The adopted reference-state dissolution rate constant of K-feldspar is 103 times slower than the dawsonite precipitation rate constant, but exceeds by a factor of 103-104 the precipitation rate constants of the three most stable silica polymorphs (quartz, chalcedony and cristobalite). Hence, in the presence of excess aqueous Na and CO2 (as is the case here), K-feldspar dissolution is the rate-limiting step for dawsonite precipitation, and aqueous silica concentrations quickly reach and maintain supersaturation with respect to all three silica polymorphs, which co-precipitate together with dawsonite. Because K-feldspar dissolution is represented kinetically, in contrast to the equilibrium treatment of injection-induced aqueous CO2 concentrations, initial dawsonite precipitation in model XSH is not coincident with initial plume ascent. Dawsonite cementation is first realized within the shale caprock (basal 5m layer only) at about 0.25 years; intraplume dawsonite cementation begins at about 0.5 years. Once initiated, these processes continue throughout the prograde and retrograde regimes (Fig. 8a,b). Reaction (2) proceeds to the right with an increase in solid-phase volume that varies from 17% (quartz and chalcedony) to 25.4% (cristobalite). However, K-feldspar comprises only 6.5% of the initial aquifer volume and reaction (2) has achieved only about 2.5% completion after 20 years; hence, aquifer porosity has been reduced by a factor of only 0.1% during this time, and would ultimately be reduced by a maximum factor of only 4%. As a result, for Sleipner-like settings the coupled mineral dissolution/precipitation process that results in dawsonite cementation has the desirable effect of maintaining initial CO2 injectivity. The presence of intra-aquifer shales dramatically increases the spatial extent (but not concentration) of dawsonite cement in the near-field environment (Fig. 8c,d). Dawsonite cementation is unique among mineraltrapping mechanisms in that it is enabled by the composition of ambient formation waters - the high Na concentration of saline fluids (here, 0.45 molal). Hence, dawsonite cement is likely to form as a result of CO2 injection into any saline aquifer that contains Al-bearing silicates (in particular K-feldspar). In fact, natural analogues for this process have been documented: widespread dawsonite cement in the BowenGunnedah-Sydney Basin, Eastern Australia, which has been interpreted to reflect magmatic CO2 seepage
120
J. W.JOHNSON ET AL.
Fig. 8. Intra-plume dawsonite cementation at the close of (a) prograde mineral trapping for model XSH and (b-d) retrograde trapping for models XSH, CSH, and DSH.
on a continental scale (Baker et al 1995), and sporadic dawsonite cement in the clastic Springerville-St Johns CO2 reservoir (Moore etal 2003). Plume-bounding precipitation of carbonate rind. Calcite-group carbonates (MCO3) precipitate via distinct mechanisms along lateral and upper plume margins, effectively delineating the CO2 storage volume. Both mechanisms can be described in the context of the following general precipitation reaction: (3) for which (4) Once the immiscible CO2 plume has attained steadystate configuration, intra-plume CO2( } concen-
trations (which have increased, favouring MCO3 precipitation) and pH (which has decreased, favouring MCO3 dissolution) are maintained at nearly constant values. Given this near-constant contribution of -(log aco +2 pH) to log Q (the actual activity product) for equation (4), typical saline-aquifer aqueous M2+ concentrations, such as those for Mg2+ and Ca2+ in this study, are insufficient to realize saturation with respect to any MCO3. However, these bulk ambient concentrations are continuously increased by silicate dissolution during the storage process. The rate of increase in intraplume aqueous concentration for a specific M2+ depends on the abundance, M-concentration, and dissolution rate of M-bearing formation minerals; it also varies significantly with CO2 saturation. Here, Mg2+ is available exclusively from phlogopite (2 solidvol.%, 3 moles-Mg/mole-phlogopite), while Ca2+ is available exclusively from plag-Ab80 (5% solidvol.%, 0.2 moles-Ca/mole-plag-Ab80). Hence, the source reservoir for Mg is roughly six times that of Ca. Moreover, the adopted specific surface area of phlogopite exceeds that of plag-Ab80 by a factor of almost 15, and its dissolution rate constant is only 30% smaller. As a result, in this model the bulk release rate for Mg from silicate dissolution will be roughly 60-70 times that of Ca. Because of this much larger release rate, and the larger background concentration of Mg (factor of 2.4, Table 3), aqueous saturation with respect to magnesite is attained far more readily than that with respect to calcite, despite the slightly higher solubility of the former. While intra-plume aqueous M2+ concentrations are increasing during prograde disposal in model XSH, the aqueous flow direction is downward and slightly outward (Fig. 3b). As the aqueous phase migrates across lateral plume boundaries along this path, H + concentration decreases by roughly 2.5 orders of magnitude (Fig. 6b) while the corresponding decrease in CO 2(a} concentration is only about half of that value (Fig. 6e). This creates a background gradient in -(log aco^ + 2 pH) that is highly conducive to MCO3 precipitation. Hence, lateral carbonate 'rind' begins to precipitate along an outer shell (after about 0.25 years in model XSH), which marks the initial achievement of sufficient M2+ concentration. Then, as this concentration within migrating fluids continues to increase with time, this rind becomes more concentrated while growing upward and slightly inward from its outer shell, directly opposite to the flow direction. In model XSH, prograde magnesite rind forms by this process (Fig. 9a), whereas Ca2+ concentrations are not sufficient to permit formation of coincident calcite rind (Fig. 9e). During the retrograde regime, aqueous-phase flow directions are reversed. Hence, carbonateundersaturated fluids are now migrating through
REACTIVE TRANSPORT MODELLING OF CO2
121
Fig. 9. Magnesite and calcite components of carbonate rind that delineates lateral and upper plume boundaries at the close of (a,e) prograde mineral trapping for model XSH and (b-d, f-h) retrograde trapping for models XSH, CSH, and DSH. Intra-aquifer and basal caprock shales have magnesite >0.05 vol.% (white regions) and calcite typically <0.005 vol.% (trace concentrations in g & h).
prograde lateral rind along a gradient in -(log ac02 + 2 pH) that favours carbonate dissolution. However, along this traverse M2+ concentrations increase rapidly because of carbonate, and continued silicate, dissolution within outer portions of the prograde rind. As a result, the fluid quickly becomes saturated with respect to magnesite, which begins to precipitate just inside the outer shell and continues
beyond this point. The net result is slow migration of the prograde lateral magnesite rind upwards and inward while its concentration increases (Fig. 9b). This will continue as long as the Mg2+ source region is not exhausted. Continued plag-Ab80 dissolution during this time results in the initial development and subsequent growth of a retrograde lateral calcite rind (Fig.9f).
122
J.W. JOHNSON ETAL.
The influence of intra-aquifer shales on lateral rind formation is highly dependent on their lateral continuity. When such continuity exists (model CSH), the shales have little influence other than vertical restriction of near-field rind development to beneath the lowest shale inter-bed (Fig. 9c,g). In contrast, the presence of laterally discontinuous shales (model DSH) has two dramatic effects. First, spatially recursive lateral gradients in CO2 aqueous solubility (Fig. 7c,d) and pH (Fig. 7g,h) that develop around each column of focused vertical CO2 migration (Fig. 4c,d), lead to similarly recursive precipitation of lateral carbonate rinds. During the prograde phase, they precipitate as 'blooms' just above and to either side of each lateral break. During the retrograde regime, these blooms migrate away from the breaks, coalesce (from either side), and appear as mounds above each interior shale segment (Fig. 9d). The second dramatic effect of lateral discontinuity is to create a highly irregular outer boundary to the residual saturation zone; this flattens the otherwise steep CO2 aqueous solubility and pH gradients, leading to precipitation of much broader lateral rind (cf. Fig. 9b,d). As a result of these two effects, the total mass of magnesite precipitated in lateral rind in model DSH is 75% greater than in models XSH and CSH. Although a carbonate rind forms along lateral plume margins as a consequence of aqueous flow across steep gradients in pH and CO2(aq) concentration, a similar rind also forms along upper plume boundaries in an environment of near-constant pH and CO 2(a} concentration. Here, carbonate precipitation is enabled by its indirect dependence on CO2 saturation. Specifically, for a given rate of M 2+ hydrolysis from silicate dissolution the associated increase in bulk aqueous M2+ concentration is inversely proportional to thickness of the aqueous wetting fluid, i.e. proportional to CO2 saturation. Hence, aqueous M2+ concentrations will increase far more rapidly within the prograde plume cap (where CO2 saturation reaches 67%) than elsewhere, and may equilibrate with MCO3 in the context of local pH and CO 2(a) concentration. Moreover, because CO2 saturation is maximized at the caprock interface, this is where MCO3 cementation will initiate, and then expand downward into the aquifer. In model XSH, this process is exemplified by the formation of an upper magnesite rind, which begins to precipitate at the caprock interface after about eight years, and then expands downward henceforth (Fig. 9a,b). The identical process occurs in model CSH, where an upper magnesite rind develops beneath the caprock and all intra-aquifer shales (Fig. 9c), and in model DSH, where it develops only beneath the caprock, but here to the greatest extent observed in the three models (Fig. 9d). An upper calcite rind does not form owing to insufficient Ca2+ concentrations (Fig. 9e-h).
Fig. 10. Saline aquifer porosity at the close of (a) prograde mineral trapping for model XSH and (b-d) retrograde trapping for models XSH, CSH, and DSH.
Porosity evolution within the near-field environment reflects the integrated effect of the three distinct mineral dissolution/precipitation processes examined above. For Sleipner-like settings, this evolution is virtually negligible, although the individual contributions of intra-plume dawsonite cementation, lateral plume-bounding magnesite/calcite rind, and upper plume-bounding magnesite rind are readily distinguished (Fig. 10). Hence, in these environments the coupled mineral dissolution/precipitation process that results in dawsonite cementation has the desirable effect of maintaining initial CO2 injectivity, whereas those that form upper and lateral carbonate rind effectively delineate the CO2 storage volume. Carbonate cementation of intra-aquifer and caprock shales. The physical and compositional environment within which mineral trapping occurs in shales is
REACTIVE TRANSPORT MODELLING OF CO,
123
distinct from that examined above for the saline aquifer. Chemical interaction of the immiscible plume and shale mineralogy is primarily localized to shale interfaces with the underlying aquifer, but it also occurs along those intra-shale microfractures that permit some degree of immiscible CO2 penetration. Here, the aqueous phase moves by advection and diffusion rates considerably slower than those characterizing the aquifer, whereas immiscible CO2 interacts through this aqueous phase with a claydominated mineral assemblage whose bulk Fe-Mg concentration is much greater than that of the aquifer. This setting is far more conducive to Fe-Mg-carbonate cementation, because much larger aqueous Fe-Mg concentrations can be achieved through silicate dissolution. In the present models, Fe-Mg-rich clays are represented by Mg-chlorite, Fe-Mg carbonates are represented by magnesite, and the coupled mineral dissolution/precipitation reaction that leads to magnesite cementation can be expressed as:
(5) whose stoichiometry follows from the relative magnitude among participating solids of both molar volumes and dissolution/precipitation volume fractions, the latter being predicted by the simulations. Within the 25 m shale caprock of models XSH, CSH, and DSH, magnesite precipitation is limited to the basal 10m (represented by two 5m layers of grid cells), where minimal immiscible CO2 penetration occurs. Such precipitation is virtually limited to the basal 5 m layer, where near uniform magnesite concentration of 0.57 vol.% is obtained by the close of the prograde regime (Fig. 11 a); the analogous concentration in the overlying 5m layer is 0.0023 vol.%. During the retrograde phase, immiscible CO2 trapped beneath the caprock continues to promote carbonate cementation in the basal 5m layer, where near uniform magnesite concentration approaches 1.0 vol.% after 20 years (Fig.llb-d). This concentration is 20-30 times larger than those achieved within plume-bounding magnesite rinds (cf. Figs. 9 & 11), reflecting the enhanced Mg concentrations and diminished aqueous flow rates within shale. In models CSH and DSH, magnesite cementation of intra-aquifer shales is identical to that of the basal caprock over the first 16-18 years. At this point, residual CO2 saturation within these shales has been com-
Fig. 11. Magnesite precipitation within basal caprock and intra-aquifer shales at the close of (a) prograde mineral trapping for model XSH and (b-d) retrograde trapping for models XSH, CSH, and DSH. The 20-30 fold increase in magnesite concentration from plumebounding rind (Fig. 9) to shale primarily reflects that of initial bulk Mg.
pletely consumed through the aqueous phase by the cementation process (reaction 5), which therefore terminates. As a result, final magnesite concentrations within inter-bedded shales are slightly lower than in the basal caprock (Fig. 1 lc,d). Still, the presence of these shales causes the total mass of mineraltrapped CO2 in models CSH and DSH to exceed that of model XSH by a factor of 3.0 and 1.8, respectively. However, even in model CSH composite mineral trapping accounts for only about 0.3% by mass of injected CO2. Although seemingly negligible, mineral trapping has enormous strategic significance because it continuously reduces shale porosity and permeability,
124
J.W. JOHNSON ETAL.
Ladbroke Grove natural gas field, where postaccumulation CO2 influx has converted Fe-rich chlorite to Fe-rich dolomite (ankerite), kaolinite, and silica (Watson^al 2002).
Screening criteria for identifying optimal storage sites in Sleipner-like settings
Fig. 12. Interdependent porosity-permeability evolution due to kinetic progression of reaction (5) within the basal shale caprock of model XSH (a) at the close of the retrograde phase, and (b) as extrapolated to completion, which requires 130 years.
and thereby continuously improves the isolation performance of shale caprocks. This reduction process is exemplified by reaction (5), which proceeds to the right with a solid-phase volume increase of 18.5% (magnesite accounting for 47 vol.% of the product assemblage), and conveniently illustrated by porosity-permeability evolution within the basal caprock of model XSH. Here, kinetic reaction (5) has attained roughly 15% completion after 20 years, reducing porosity by 8% (from 5 to 4.6%). Owing to adopted cubic dependence of the current-to-initial permeability ratio on that of porosity, this translates to a 22% reduction in permeability (Fig. 12a). Moreover, this reaction will proceed to completion unless the underlying immiscible CO2 is completely exhausted, one of the reactant minerals is entirely consumed (or effectively armoured by product phases against further dissolution), or shale porosity is fully sealed. Presuming the first two events do not occur, the third can be evaluated by extrapolating reaction (5) to hypothetical completion, which requires about 130 years. In this extreme case, final porosity of the basal caprock shale has been reduced by 52% from its original value, whereas permeability has been reduced by 90% - an order of magnitude (Fig. 12b). A natural analogue to reaction (5) has recently been documented in the
The simulation results examined above for models XSH, CSH, and DSH suggest preliminary screening criteria that can be used to identify optimal saline aquifer storage sites in Sleipner-like settings (shalecapped sandstone systems). These criteria fall into three fundamental groups: constraints on achieving optimal caprock performance and CO2 injectivity, maximizing the spatial framework of plume-aquifer interaction, and enhancing efficacy of solubility and mineral trapping within this framework. Optimal isolation performance ultimately hinges on the caprock creating an effective hydrodynamic seal against vertical migration of the buoyant immiscible CO2 plume. A 25m-thick shale of typical porosity and permeability (5% and 3|xD, respectively: Freeze & Cherry 1979) met this requirement in the simulations. However, trace immiscible CO2 did penetrate this shale to 10m during a relatively brief 10-year prograde regime; hence, this set of properties might be considered minimally effective; increased thickness and reduced permeability would be preferred. High concentrations of Fe-Mg-rich clays are also advantageous to maximize local carbonate cementation (e.g. reaction 5) and thereby continuously improve seal integrity. Lateral extent of the cap rock must be commensurate with the anticipated injection volume and topography of the aquifer-caprock interface. The underlying aquifer should be highly permeable, as in the simulations, to maximize CO2 injectivity and minimize injection overpressures, which upon translation to the caprock interface in less permeable (or laterally confined) systems may lead to geomechanical deformation of this seal (Johnson etal 2003). Given a secure caprock and sufficient injectivity, the spatial framework of plume-aquifer interaction (i.e. CO2 storage capacity) is maximized - and outward migration of the plume from the near-field environment delayed - by the presence of multiple inter-bedded lithologies that retard vertical and promote lateral plume migration, such as the thin shales in models CSH and DSH. By expanding the plume-formation water interface and source region for carbonate-forming cations, these inter-beds increase potential effectiveness of solubility and mineral trapping, the latter being further enhanced if such units are laterally discontinuous (DSH). Actual effectiveness of these trapping mechanisms is maximized by the presence of specific composi-
REACTIVE TRANSPORT MODELLING OF CO7
tional constraints. Formation-water salinity should be moderate, as in the simulations; high enough to promote intra-plume dawsonite cementation, which roughly maintains initial CO2 injectivity in the context of concomitant K-feldspar dissolution (reaction 2), but not so concentrated as to reduce the effectiveness of solubility trapping significantly (reaction 1). High concentrations of Fe-Mg-Ca-bearing minerals, uncharacteristic of the simulations, are required to maximize carbonate precipitation within upper and lateral plume-bounding rinds (reaction 3). Average elemental concentrations in unconsolidated sand aquifers (and in shales) are Fe > Mg > Ca (Blatt et al 1972); for settings anomalously rich in Fe-Mg, the formation of such rinds may partially self-seal the plume. Ca concentrations are relatively small in both unconsolidated aquifers and shales (although the opposite is true for many carbonate-cemented sandstones); hence, in Sleipner-like settings, calcite is expected to play a minor role in mineral trapping relative to siderite, magnesite, and their solid solutions.
Conclusions Reactive transport modelling provides a unique methodology for identifying optimal geological CO2 storage sites and forecasting their long-term isolation performance. We have used this approach to elucidate the fundamental processes, associated trapping mechanisms, and sequestration partitioning that will characterize saline-aquifer storage in Sleipnerlike settings (shale-capped sandstone systems). Based on this analysis, we have proposed a preliminary set of physical and compositional constraints that portend optimal performance in these environments. Critical future investigations include quantifying sensitivity of predictive results, such as those reported here, to gaps and uncertainties in process models and parameters, quantifying sensitivity of isolation performance to site-specific variations in system characteristics, benchmarking key model predictions against well-constrained laboratory and large-scale field experiments, and extending the space-time framework to encompass far-field environments and millennia. Extensive literature documents the gaps that exist within, and uncertainty limits that surround, many of the key process models and associated data adopted in this study. Particularly important are current gaps in the kinetic rate law for mineral precipitation (e.g. representation of various nucleation processes) and uncertainties in the relevant parameters (e.g. precipitation rate constants, activation energies, supersaturation thresholds, and reactive specific surface areas). Accurate representation of multiphase flow processes hinges on uncertainties in several key parameters, such as residual gas and irreducible
125
water saturations and the Van Genuchten sorting and bubble-pressure parameters, all of which are used to determine relative permeability and capillary pressure curves. Accurate representation of aqueous convection due to density instabilities that result from solubility trapping requires a rigorous equation of state for H2O-NaCl-CO2, which is approximated by that for H2O in this study. Although we have made every effort to incorporate best-available process models and data in the simulations presented here, quantifying sensitivity of predictive results to the myriad uncertainties and approximations inherent to reactive transport modelling, such as those noted above, is critical for assigning to such results an appropriate level of confidence. Sensitivity analyses of this kind are also an effective means of prioritizing further refinement of key process models and parameters. A number of potentially significant, but likely second-order, features and processes were not addressed in this study, and their presumed modifying influence on the fundamental processes and trapping mechanisms must be assessed. Particularly important to evaluate are the effects of intra-aquifer heterogeneity (i.e. viscous fingering), ambient flow, density-driven aqueous convection, and solid-solution compositions on the long-term stability and spatial distribution of solubility- and mineraltrapped CO2, and the effect of aquifer-caprock interface topography on the spatial distribution of hydrodynamically- and structurally-trapped CO2. Explicit account of these effects, which can be accomplished with our current modelling capability, is required to assess their potential impact. Identification of optimal saline-aquifer CO2 storage sites requires quantifying sensitivity of isolation performance to the wide range of compositional, hydrological, structural, and depth variations that characterize these environments and proposed injection scenarios. Important compositional variations include those associated with waste-stream impurities (e.g. SOX, NOx, CH4, H2S), aquifer and caprock mineralogy (ranging from the presence of carbonate cements and silicate solid solutions in Sleipner-like settings to radically different lithologies such as anhydrite-capped carbonate systems), and the associated ambient aqueous phases (e.g. salinity and the concentrations of potential carbonate-forming cations). Important hydrological variations include CO2 influx and ambient flow rates, aquifer and caprock porosity and permeability (including their degree and style of heterogeneity), and residual gas and irreducible water saturations (Holtz 2002, 2003). Important structural variations include lateral continuity, which together with CO2 flux rate and aquifer permeability determine injection overpressures that translate to and may cause geomechanical deformation of the caprock (Johnson
126
J.W. JOHNSON ETAL.
et al 2003), and topography of the aquifer-caprock interface, which governs the up-dip migration path of underlying immiscible CO2. Finally, system depth and the local geothermal gradient determine pressure-temperature conditions, which control phase relations within the equilibrium reference frame for mass transfer processes (e.g. solubility and mineral trapping) as well as the fluid-phase density contrasts that control buoyancy-driven immiscible CO2 migration. Analysing sensitivity of isolation performance to all of the above is an effective means of identifying key screening criteria, optimal storage sites, and determining the accuracy with which various system parameters need to be measured. Several of the predictive results from this study have important positive implications for the longterm efficacy of saline-aquifer CO2 storage in Sleipner-like (and other) settings. These include the effect of residual CO2 saturation to maintain or further enhance prograde trapping efficiency during the retrograde regime, and the effect of mineral trapping to maintain CO2 injectivity (reaction 2), to delineate and partially self-seal the CO2 storage volume (reaction 3), and to improve the hydrodynamic seal integrity of shale caprocks (reaction 5). This set of predictions can be assessed and refined through a series of integrated laboratory and modelling studies in which appropriately designed plugflow and batch reactor experiments are conducted and simulated. They can also be used as a benchmark against relevant data obtained from field demonstrations, such as the US Frio project (Hovorka et al. 2002). Experimental and field corroboration of key predictive results is crucial because it represents the necessary point of departure for defensible extension of the reactive transport modelling approach to forecast the long-term isolation performance of commercial-scale saline aquifer CO2 storage sites. This work was performed under the auspices of the US Department of Energy by the University of California, Lawrence Livermore National Laboratory under Contract W-7405-Eng-48. It is a pleasure to thank Etienne Brosse, Daniel Garcia, and Richard Worden for their thorough reviews of an earlier version of this manuscript; each contributed many insightful comments that were invaluable in preparing the revised contribution.
References BAKER, J. C, BAI, G. P., HAMILTON, P. J., GOLDING, S. D. & KEENE, J. B. 1995. Continental-scale magmatic carbon-dioxide seepage recorded by dawsonite in the Bowen-Gunnedah-Sydney Basin, Eastern Australia. Journal of Sedimentary Research, Section A Sedimentary Petrology and Processes, 65,522-530. BERTRAND, C., FRITZ, B. & SUREAU, J. F. 1994.
Hydrothermal experiments and thermo-kinetic modelling of water-sandstone interactions: Chemical Geology, 116, 189-192. BERNER, R. A. 1971. Principles of Chemical Sedimentology. McGraw-Hill, New York. 240pp. BLATT, H., MIDDLETON. G. & MURRAY, R. 1972. Origin of Sedimentary Rocks. Prentice-Hall, Englewood Cliffs, NJ. 634pp. BLUM, A. E. & STILLINGS, L. L. 1995. Feldspar dissolution kinetics. In: WHITE, A. F. & BRANTLEY, S. L. (eds). Chemical Weathering Rates of Silicate Minerals. Reviews of Minerology, 31, Mineralogical Society of America, Washington, DC, 291-351. DAVELER, S. A. 1998. XTOOL 7.15 user's manual: Lawrence Livermore National Laboratory, UCRLMA-130654, 32pp. DRUMMOND, S. E. 1981. Boiling and mixing of hydrothermal fluids. Chemical effects on mineral precipitation. Unpublished Ph.D. dissertation, Pennsylvania State University. EIKEN, O., BREVIK, L, ARTS, R., LINDEBERG, E. & FAGERVIK, K. 2000. Seismic monitoring of CO2 injection into a marine aquifer. Proceedings of the Society of Exploration Geophysicists International Conference 70th Annual Meeting, 4pp. ENNIS-KING, J. & PATERSON, L. 2002. Rate of dissolution due to convective mixing in the underground storage of carbon dioxide. In: GALE, J. & KAYA, Y. (eds) Proceedings of the 6th International Conference on Greenhouse Gas Control Technologies (GHGT-6), Kyoto, Japan, Oct 1-4, 2002,1, 507-510. ENNIS-KING, J. & PATERSON, L. 2003. Role of convective mixing in the long-term storage of carbon dioxide in deep saline formations. Proceedings of the Society of Petroleum Engineers Annual Technical Conference, Denver, CO, October 5-8,2003,12pp. FENGHOUR, A. & WAKEMAN, W. A. 1998. The viscosity of carbon dioxide. Journal of Physical and Chemical Reference Data, 27, 31-44. FOFONOFF, P. & MILLARD, R. C. JR. 1983. Algorithms for computation of fundamental properties of seawater. Unesco Technical Papers in Marine Sciences 44, 53pp. FREEZE, R. A. & CHERRY, J. A. 1979. Groundwater. Prentice-Hall, Englewood Cliffs, NJ, 604pp. GARRELS, R. M. & CHRIST, C. L. 1965. Solutions, Minerals, and Equilibria. Freeman-Cooper, San Francisco, 450pp. GREGERSEN, U., JOHANNESSEN, P. N., M0LLER, J. J., KRISTENSEN, L., CHRISTENSEN, N. P., HOLLOWAY, S., CHADWICK, A., KIRBY, G., LINDEBERG, E. & ZWEIGEL, P. 1998. Saline Aquifer CO2 Storage S.A.C.S Phase Zero 1998. Internal report. GEUS (DK), BGS (UK), IKU (N). (provided courtesy of SACS). HELGESON, H. C. 1969. Thermodynamics of hydrothermal systems at elevated temperatures and pressures. American Journal of Science, 267, p. 729-804. HELGESON, H. C., DELANY, J. M, NESBITT, H. W. & BIRD, D. K. 1978. Summary and critique of the thermodynamic properties of rock-forming minerals. American Journal of Science, 278-A, 229 p. HOLTZ, M. H. 2002. Residual gas saturation to aquifer influx. A calculation method for 3-D computer reservoir model construction. Proceedings of the Society of Petroleum Engineers Gas Technology Symposium,
REACTIVE TRANSPORT MODELLING OF CO7 Calgary, Alberta, Canada, April 30-May 2, 2002, 10pp. HOLTZ, M. H., 2003. Pore-scale influences on saline aquifer CO2 sequestration. American Association of Petroleum Geologists Annual Meeting, Salt Lake City, UT, May 11-14, 2003,12, A79. HOVORKA, S. D., KNOX, P. R., JACKSON, J. A. & JACKSON, K. G. 2002. Frio brine sequestration pilot in the Texas gulf coast. In: GALE, J. & KAYA, Y. (eds) Proceedings of the Sixth International Conference on Greenhouse Gas Control Technologies (GHGT-6), Kyoto, Japan, Oct 1-4,2002,1,583-587. JOHNSON, J. W. & LUNDEEN, S. R. I994a. Jewel: A graphical-user interface for generating custom GEMBOCHS thermodynamic datafiles for use with geochemical modelling software. LLNL-YMP Milestone report MOL63, 23 p. JOHNSON, J. W. & LUNDEEN, S. R. I994b. GEMBOCHS thermodynamic datafiles for use with the EQ3/6 software package. LLNL-YMP Milestone report MOL72, 99 p. JOHNSON, J. W. & LUNDEEN, S. R. 1995. Facet: A graphical-user interface for viewing and updating the GEMBOCHS thermodynamic database. LLNL-YMP Milestone report MOL208,28pp. JOHNSON, J. W. & NITAO, J. J. 2002. Reactive transport modelling of geologic CO2 sequestration at Sleipner. In: GALE, J. & KAYA, Y. (eds) Proceedings of the 6th International Conference on Greenhouse Gas Control Technologies (GHGT-6), Kyoto, Japan, Oct 1-4, 2002,1, 327-332. JOHNSON, J. W. & NORTON, D. 1991. Critical phenomena in hydrothermal systems: State, thermodynamic, electrostatic, and transport properties of H2O in the critical region. American Journal of Science, 291,541-648. JOHNSON, J. W., OELKERS, E. H. & HELGESON, H. C. 1992. SUPCRT92: A software package for calculating the standard molal thermodynamic properties of minerals, gases, aqueous species, and reactions from 1 to 5000 bars and 0 to 1000°C. Computers and Geosciences, 18, 899-947. JOHNSON, J. W., KNAUSS, K. G., GLASSLEY, W. E., DELOACH, L. D. & TOMPSON, A. F B. 1998. Reactive transport modelling of plug-flow reactor experiments: quartz and tuff dissolution at 240°C. Journal of Hydrology, 209, 81-111. JOHNSON, J. W., NITAO, J. J., TOMPSON, A. F. B. ETAL. 1999. 21st-century tools for modelling reactive transport in dynamic geologic systems of economic and environmental significance: In: Earth and Environmental Sciences 1998 Annual Report, p. 7-11, LLNL. JOHNSON, J. W., NITAO, J. J., STEEFEL, C. I. & KNAUSS, K. G. 2001. Reactive transport modelling of geologic CO2 sequestration in saline aquifers: the influence of intraaquifer shales and the relative effectiveness of structural, solubility, and mineral trapping during prograde and retrograde sequestration. Proceedings of the 1st National Conference on Carbon Sequestration, Washington, DC, May 14-17, 2001, 60 p. www.netl .doe.gov/publications/01/carbon_seq/P28.pdf JOHNSON, J. W., NITAO, J. J., MORRIS, J. P. & BLAIR, S. C. 2003. Reactive transport modelling of geohazards associated with offshore CO2 injection for EOR and geologic sequestration. Proceedings of the Offshore
127
Technology Conference, Houston, TX, May 5-8, 2003, 9pp. JORDAN, G. & RAMMENSEE, W. 1998. Dissolution rates of calcite (1014) obtained by scanning force microscopy: Microtopography-based dissolution kinetics on surfaces with anisotropic step velocities. Geochimica et CosmochimicaActa, 62, 941-947. KNAUSS, K. G. & COPENHAVER, S. A. 1995. The effect of malonate on the dissolution kinetics of albite, quartz and microcline as a function of pH at 70°C. Applied Geochemistry, 10, 17-33. KNAUSS, K. G. & WOLERY, T. J. 1989. Muscovite dissolution kinetics as a function of pH and time at 70°C: Geochimica et CosmochimicaActa, 53, 1493-1501. KNAUSS, K. G. & WOLERY, T. J. 1988, The dissolution kinetics of quartz as a function of pH and time at 70°C. Geochimica et CosmochimicaActa, 52,43-53. KNAUSS, K. G. & WOLERY, T. J. 1986. Dependence of albite dissolution kinetics on pH and time at 25 °C and 70 °C. Geochimica of CosmochimicaActa, 50,2481-2497. LASAGA, A. C. 1998. Kinetic Theory in the Earth Sciences. Princeton Univ. Press, Princeton, NJ, 811pp. LINDEBERG, E. & BERGMO, P. 2002. The long-term fate of CO2 injected into an aquifer. In: GALE, J. & KAYA, Y. (eds) Proceedings of the 6th International Conference on Greenhouse Gas Control Technologies (GHGT-6), Kyoto, Japan, Oct \-4, 2002,1,489-494. MEYER, C. A., MCCLINTOCK, R. B. & SILVESTRI, G. J. 1993. ASME Steam Tables: Thermodynamic and Transport Properties of Steam (comprising tables and charts for steam and water, calculated using the 1967 IFC formulation for industrial use, in conformity with the 1963 international skeleton tables, as adopted by the Sixth International Conference on the Properties of Steam), 6th edn, ASME, New York, 436pp. MOORE, J., ADAMS, M., ALLIS, R., LUTZ, S. & RAUZI, S. 2003. Investigations of CO2 mobility in natural reservoirs beneath the Colorado Plateau and Southern Rocky Mountains. Proceedings of the 2nd Annual Conference on Carbon Sequestration, May 5-8,2003, Alexandria, Virginia, 22 p. NAGY, K. L. 1995. Dissolution and precipitation kinetics of sheet silicates. In: WHITE, A. F. & BRANTLEY, S. L. (eds). Chemical Weathering Rates of Silicate Minerals. Reviews of Mineralogy, 31, Mineralogical Society of America, Washington, DC, 173-233. NITAO, J. J. 1998a. Reference manual for the NUFT flow and transport code, version 2.0. Lawrence Livermore National Laboratory, UCRL-MA-130651,55 p. NITAO, J. J. 1998/7. User's manual for the USNT module of the NUFT Code, Version 2.0 (NP-Phase, NC-component, Thermal). Lawrence Livermore National Laboratory, UCRL-MA-13065 3,76pp. NORTON, D. 1984. Theory of hydrothermal systems. Annual Review of Earth & Planetary Sciences, 12, 155-177. PARKER, J. C., LENHARD, R. J. & KUPPUSAMY, T. 1987. A parametric model for constitutive properties governing multiphase flow in porous media. Water Resources Research, 23, 618-624. POKROVSKY, O. S. & SCHOTT, J. 1999. Processes at the magnesium-bearing carbonates solution interface. II. Kinetics and mechanism of magnesite dissolution. Geochimica et CosmochimicaActa, 63, 881-897.
128
J.W. JOHNSON ETAL.
RENDERS, P. J. N., GAMMONS, C. H. & BARNES, H. L. 1995. Precipitation and dissolution rate constants for cristobalite from 150 to 300°C. Geochimica et CosmochimicaActa, 59,77-85. ROBIE, R. A., HEMINGWAY, B. S. & FISHER, J. R. 1978. Thermodynamic properties of minerals and related substances at 298.15 K and 1 bar (105 Pascals) pressure and at higher temperatures. United States Geological Survey Bulletin 1452,456pp. SACS. 2000. Final Technical Report 'SACS' - Saline Aquifer CO2 Storage, 34 p. (available on the SACS web site: http://www.ieagreen.org.uk/sacshome.htm) SCHEIDEGGER, A. E. 1974. The Physics of Flow through Porous Media (3rd edn). University of Toronto Press, Toronto, 353pp. SHOCK, E. L. 1998. An updated and augmented version (slop98.dat) of the original SUPCRT92 database (sprons92.dat) is available on the Dr Shock's website: http://geopig.asu.edu SHOCK, E. L., OELKERS, E. H., JOHNSON, J. W., SVERJENSKY, D. A. & HELGESON, H. C. 1992. Calculation of the thermodynamic properties of aqueous species at high pressures and temperatures: Effective electrostatic radii, dissociation constants, and standard partial molal properties to 1000C and 5 kb. Journal of the Chemical Society (London) Faraday Transactions, 88, 803-826.
SPAN, R. & WAGNER, W. 1996. A new equation of state for carbon dioxide covering the fluid region from the triple-point temperature to 1100 K at pressures up to 800 MPa. Journal of Physical and Chemical Reference Data, 25,1509-1596. STILLINGS, L. L., DREVER, J. L, BRANTLEY, R. L., SUN, Y. & OXBURGH, R. 1996. Rates of feldspar dissolution at pH 3-7 with 0-8 mM oxalic acid. Chemical Geology, 132,79-89. TANGER, J. C. IV & HELGESON, H. C. 1988. Calculation of the thermodynamic and transport properties of aqueous species at high pressures and temperatures: Revised equations of state for the standard partial molal properties of ions and electrolytes. American Journal of Science, 288,19-98. TESTER, J. W., WORLEY, W. G., ROBINSON, B. A., GRIGSBY, C. O. & FEERER, J. R. 1994. Correlating quartz dissolution rates in pure water from 25 to 625 °C: Geochimica et CosmochimicaActa, 58, 2407-2420. WATSON, M. N., ZWINGMANN, N. & LEMON, N. M. 2002. The Ladbroke Grove-Katnook carbon dioxide natural laboratory: A recent CO2 accumulation in a lithic sandstone reservoir. In: GALE, J. & KAYA, Y. (eds) Proceedings of the 6th International Conference on Greenhouse Gas Control Technologies (GHGT-6), Kyoto, Japan, Oct 1-4, 2002,435-^40.
The role of hydrogeological and geochemical trapping in sedimentary basins for secure geological storage of carbon dioxide WILLIAM D. GUNTER1, STEFAN BACHU2 & SALLY BENSON3 1
Alberta Research Council, Edmonton, Alberta, T6N1E4, Canada (e-mail:
[email protected]) 2 Alberta Geological Survey, Edmonton, Alberta, T6B 2X3, Canada ^Lawrence Berkeley National Laboratories, Berkeley, California, 94720, USA Abstract: Sedimentary basins throughout the world are thick piles of lithified sediments that, in many cases, are the hosts for fossil fuel resources. They may become even more important in the future if they are used for the storage of anthropogenic carbon dioxide. The efficiency of CO2 geological storage is determined by the structure of the sedimentary basins, which have an intricate plumbing system defined by the location of high and low permeability strata that control the flow of fluids throughout the basin and define 'hydrogeological' traps. The most secure type of hydrogeological trapping is found in 'stratigraphic' and 'structural' traps in oil and gas reservoirs that have held oil and gas for millions of years. Another form of hydrogeological trapping is 'hydrodynamic' trapping which has been recognized in saline aquifers of sedimentary basins that have extremely slow flow rates. A volume of carbon dioxide injected into a deep hydrodynamic trap may take millions of years to travel by buoyancy forces updip to reach the surface before it leaks back into the atmosphere. Moreover, as the carbon dioxide migrates towards the surface, it dissolves in the surrounding brine ('solubility' trapping) and may react geochemically with rock minerals to become permanently trapped in the sedimentary basin by 'ionic' or 'mineral' trapping. The efficiency of the CO2 geological storage in sedimentary basins depends on many factors, among the most important being CO2 buoyancy, formation water density, lithological heterogeneity and mineralogy. A risk analysis must be completed for each site chosen for the geological storage of CO2 to evaluate the trapping security.
Co-location of sedimentary basins and CO2 sources There is a natural association of sedimentary basins and fossil fuels. There are more than 800 sedimentary basins around the world (Fig. 1), of which about 90 produce from or have discoveries of giant-size oil and/or gas fields, around 180 produce hydrocarbons from sub-giant fields, and 540 are non-productive with fair or poor potential (St John et al 1984). Therefore, we should expect a relationship between the location of sedimentary basins, the exploitation of its fossil fuels, and the resulting greenhouse gas emissions. The coincidence of the source of fossil fuels and the locations in which it is consumed is fortuitous for the use of sedimentary basins for the geological storage of carbon dioxide (Hitchon et al. 1999), which involves mainly the injection and longterm or permanent storage of CO2 in the pore space of rocks within basins. This includes injection into: hydrocarbon reservoirs, to enhance oil and gas recovery; depleted oil and gas reservoirs; and deep brine-saturated formations (Koide et al. 1993). A further development is the concept of injecting carbon dioxide, from the burning of fossil fuels, into deep unmineable coal-beds to remove methane and permanently store CO2 (Gunter et al. 1997'a). In
addition, CO2 can be stored in salt caverns (Dusseault et al 2001, 2002), similarly to the storage of natural liquid gas and wastes. The various ways of storing CO2 in sedimentary basins are illustrated in Figure 2. Within the past decade geological storage of carbon dioxide has evolved from an idea to one of the leading options for mitigating the build-up of greenhouse gases in the atmosphere. Today, geological storage of carbon dioxide is taking place in an undersea aquifer at Sleipner Vest in the North Sea, a largescale CO2 monitoring demonstration is taking place at the Weyburn oil field in Canada, and experiments to evaluate CO2 enhanced coal-bed methane production are taking place in Canada and the United States. Also, although driven by the need to avoid atmospheric emissions of H2S, significant amounts of CO2 are currently re-injected into deep aquifers and hydrocarbon reservoirs in the Alberta basin in Canada and several others in the United States (see companion paper by Bachu & Gunter 2004). At the same time, the international research community has become deeply engaged in a host of research projects aimed at revealing better descriptions of fundamental processes, developing cost-effective monitoring technologies (Gunter et al 1999), improving flow and transport models, addressing safety aspects of
From: BAINES, S. J. & WORDEN, R. H. (eds) 2004. Geological Storage of Carbon Dioxide. Geological Society, London, Special Publications, 233,129-145.0305-8719/047$ 15.00 © The Geological Society of London 2004.
Fig. 1. Global distribution and types of sedimentary basins (after St John et al. 1984).
GEOLOGICAL TRAPPING OF CO9
131
Table 1. Storage capacity in sedimentary basins (from IEA 1995; Gunter et al. 1998; Stevens et al. 1999)
Depleted oil reserve Depleted gas reserve Brine formation Unminable coal
Global capacity (GTC)
US capacity (GTC)
Canada capacity (GTC)
40-190 140-310 87-2700 5^0
10-14 20-30 1-130 4-5
0.6 4 >10 4
GTC, Giga Tonnes carbon Fig. 2. Various ways of storing CO2 in sedimentary basins (from Bachu 2001).
geological storage (Holloway 1991b) and developing a risk assessment framework for CO2 storage in geological formations. The goal of this paper is to synthesize these issues in the context of the geological framework created by sedimentary basins that are both host to the hydrocarbon fuels that produce greenhouse gases and the preferred location for geological storage of carbon dioxide.
Capacity of sedimentary basins for CO2 storage Several studies have been conducted over the past decade to estimate the amount of CO2 that can be stored in sedimentary basins (Bergman & Winter 1995; IEA 1995; Holloway 1997a; Gunter et al 1998; Stevens et al 1999; Benson, 2001; Bachu 2002). These studies have attempted to estimate how much pore volume is likely to be available for CO2 storage in depleted oil and gas reservoirs, unmineable coal seams and deep brine-filled formations. More recently, Bachu & Adams (2003) and Bachu et al (2003) have estimated that the ultimate CO2 sequestration capacity in solution in formation water in two deep saline aquifers in western Canada is of the order of 170Gt. Since there is no generally accepted method to estimate the storage capacity of a particular formation, each of these studies has developed a different approach. Nevertheless, as shown in Table 1, there is general agreement on several important issues. First, global storage capacity is greater than lOOOGtC (gigatonnes of carbon). Second, brine-filled formations have the greatest capacity, followed by oil and gas reservoirs, and then, unmineable coal seams. Third, potentially suitable formations are broadly distributed throughout the world. Hepple & Benson (2003) have calculated that in the order of lOOOGtC may need to be sequestered
over a 300-year period to keep atmospheric concentrations below 550ppm (twice the pre-industrial level of 270ppm). In the longer term, as the use of carbon-based fossil fuels decreases due to diminishing reserves and replacement by other non-carbonbased energy forms, sequestration of CO2 will not be necessary. One thousand GtC is the same order of magnitude as the estimated global capacity for storage of CO2 in sedimentary basins, and suggests that they have the capacity to store most, if not all, of the CO2 needed to prevent the build-up of harmful levels CO2 in the atmosphere. However, the capacity needed for storage is greater than the capacity of depleted oil and gas reservoirs. Therefore storage in brine-filled formations will eventually be required if geological sequestration is to play an important role in climate change mitigation.
Structure and characteristics of sedimentary basins The geological storage of CO2 requires access to large subsurface volumes, either in the rock pore space or in caverns, which can act as sealed pressurized containers. The pore space is initially occupied by geofluids such as brines, hydrocarbon and other gases (e.g. H2S and CO2), which have to be displaced in CO2 storage operations. The pressure to keep CO2 at liquid-similar densities is found at depths usually below 800m, which is also below potable groundwater sources. However, it is not possible to inject and store CO2 everywhere in the Earth's crust. Cratonic platforms, such as the Canadian and Brazilian shields, are generally unsuitable for geological sequestration of CO2 because their crystalline or metamorphic rocks lack the porosity needed for storage space and the permeability needed for injection. Orogenic belts (mountain ranges) are not suitable either because of lack of continuous seals as a result of extensive faulting and fracturing during mountain forming. Only sedimentary basins, which hold the largest pore-based volumes, are generally suitable for geological
132
W.D.GUNTERCTAL.
sequestration of CO2 because they possess the right type of porous and permeable rocks for storage and injection, such as sandstones and carbonates, and the low permeability-to-impermeable rocks needed for sealing, such as shales and evaporitic beds. Not all sedimentary basins are similarly-suitable for CO2 sequestration. Sedimentary basins are of several types, depending on their origin (St John et al 1984). Convergent basins (forearc, backarc, California- and Pannonian-type) along active tectonic margins (Fig. 1), such as the circum-Pacific basins, are usually subject to volcanism, earthquakes and active faulting. Thus, CO2 storage in such basins poses higher risks because of a higher probability of accidental release of large quantities of CO2 along open faults and fractures as a result of local catastrophic events such as earthquakes. Divergent basins on the rigid lithosphere, like on passive margins (Atlantic type), intracratonic (e.g. Williston, Michigan) and on cratonic margins, and foredeep basins in the foreland of mountain ranges, such as Rocky Mountain and Andean basins, are located in tectonically stable areas (Fig. 1) and are much less prone to earthquakes and other significant hazardous Earth events which may lead to CO2 release back into the atmosphere (Bachu 2001). Sedimentary basins are formed by thick accumulation of sediments that undergo burial, compaction, lithification and uplift over millions of years. Depending on the sediment source and type, the raw materials deposited are detrital sand, silt and mud, calcareous bioclasts, chemically precipitated carbonate minerals, evaporites in shallow seas, and organic matter. As the sediments are buried, they start to compact and dewater in the earlier stages of the lithification process. Eventually the grains consolidate as a result of increasing temperature and pressure, cement together as a result of various physical and geochemical processes, and become rock (e.g. sandstone, siltstone, shale, limestone, halite). The coarser-grained sedimentary rocks, such as sandstone, usually have highly-interconnected pore space and the ability for fluid flow (permeability), although carbonate rocks, particularly reefs, may also have significant permeability. The finer grained sedimentary rocks, such as shales, have poor interconnected pore space and much lower permeability. Evaporitic rocks (halite and anhydrite) have extremely low permeability. As a result of various tectonic, depositional and erosional processes, sedimentary basins have a plumbing system defined by the superposition of these high and low permeability strata that control the flow of fluids throughout the basins. The flow of formation fluids (water, brine, hydrocarbons and other gases) takes place along bedding in the highpermeability strata (aquifers). The flow of fluids in the low-permeability strata (aquitards) takes place
across bedding between two adjacent aquifers, at rates several orders of magnitude slower than the flow in aquifers. Although the flow rate is much smaller across aquitards than in aquifers, volumes are quite significant because of the large areal extent of the former. Evaporitic beds have a permeability several orders of magnitude smaller than that of aquitards, and are considered to be impervious to fluid flow (aquicludes). Salt beds in particular constitute good aquicludes because of the 'self-healing' of potential fractures as a result of salt creep. Protruding reefs, faults and open fractures across aquitards and aquicludes constitute conduits for vertical flow across aquitards (cross-formational flow). There is a close link between the type of a sedimentary basin and the flow of formation waters. In basins located on the marine shelf and in subduction zones (Atlantic-type, craton-margin, forearc and backarc, Fig. 1) the flow of formation waters is driven by compaction, vertically out of shales (aquitards), and laterally outward toward the basin margin in the intervening aquifers (Fig. 3a). The shaly aquitards are usually considerably overpressured (pressure greater than hydrostatic), and so are the aquifers, although to a lesser degree. In basins adjacent to active orogenic belts (foredeep, Fig. 1) formation water in deep aquifers may be driven by tectonic compression laterally out in the basin and towards the margin (Fig. 3b). Waters expelled from underneath mountain building (orogenies) are usually overpressured, hot and very saline (Oliver 1986). In foreland and intracratonic basins that have undergone significant uplift and erosion since the Miocene, and/or where thick ice was present during the Pleistocene glaciations, such as the Alberta basin in Canada, flow may be driven by erosional rebound vertically into thick shales and laterally inward in thin adjacent aquifers (Bachu 1995) (Fig. 3c). In such cases, the aquitards and adjacent aquifers are significantly underpressured (pressures less than hydrostatic). Finally, most flow systems in continental basins (foredeep, intracratonic and intramontane, Fig. 1) are driven by topography from recharge areas at high elevations to discharge areas at low elevations (Fig. 3d). Aquifer pressures are usually close to hydrostatic, slight over- or underpressuring being controlled by permeability distributions. The geothermal regime in sedimentary basins depends on: basin type, age and tectonism; proximity to crustal heat sources; basement heat flow; thermal conductivity of the rock fill and heat production in the sedimentary succession; and surface temperature. The latter varies from -2°C at the base of permafrost for continental arctic basins, and 3-4 °C at the bottom of the sea for offshore basins, to close to 30 °C for low-altitude sedimentary basins in equatorial regions. Geothermal gradients vary from less
GEOLOGICAL TRAPPING OF CO9
133
Fig. 3. Various flow-driving mechanisms in sedimentary basins (after Bachu 2000): (a) compaction; (b) tectonic compression; (c) erosional rebound; and (d) gravity.
than 20°Ckm ! in basins on old (Archean) continental lithosphere to more than 80 °Ckm~' in young, technically active areas and near crustal hot spots. The flow of formation waters may 'cool' a sedimentary basin in recharge areas and 'warm' it in discharge areas, as water picks up terrestrial heat as it flows through the plumbing system of the basin. The pressure, hydrodynamic and geothermal regimes in a sedimentary basin have a significant impact on its suitability for geological storage of CO2 (Bachu 2000, 2002). Rocks in younger basins have, in general, higher permeability than in old basins. Compacting basins and basins in tectonically active areas, including erogenic regions, are less favourable for CO2 storage than stable foredeep and intracratonic basins (Hitchon et al. 1999). Leakage from foredeep and cratonic sedimentary basins is mainly at aquifer outcrop, which would normally force a CO2 plume to travel long distances laterally. However, the rate of leakage through lateral flow to outcrop is insignificant on a human timescale of decades to centuries. On the other hand, CO2 leakage in basins in technically active areas may occur vertically along open faults and fractures. In all basins, vertical flow and leakage of CO2 may occur through improperly completed and/or abandoned wellbores (Celia & Bachu 2003). The vertical CO2 leakage may occur on a timescale and in amounts that are relevant to human life and to currently-increasing amounts of CO2 in the atmosphere. Thus, tectonically-active regions and parts of sedimentary basins prone to leakage, such as transmissive faults, should not be targeted for the geological storage of CO2.
Security of CO2 storage Hydro geological trapping First, the CO2 must be trapped below low permeability rock, such as shale or salt beds, to avoid rapid migration of CO2 to the surface. If the top of the trap is closed, such as is the case with most oil and gas reservoirs, the CO2 could be expected to remain in the trap for geological time periods. Sedimentary basins have many such closed, physically-bound traps, called reservoirs, in which the fluid is static; some of which are occupied by oil and gas and the remainder by aqueous fluids. These oil and gas occurrences were initially also filled by aqueous fluids. Moreover, such closed traps are very small compared to deep aquifers, which are unconfined and eventually discharge their waters to the surface on a geological timescale. However, these closed traps have held fluids securely over geological time and, obviously, would be the first targets for geological storage. In addition, the production of oil and natural gas from sedimentary basins creates 'void' or low-pressure storage space which can be refilled/repressurized with CO2. Thus, these depleted reservoirs would be even more attractive for storage. Hence, the most secure type of hydrogeological trapping is likely to be found in oil and gas reservoirs in the form of 'structural' or 'stratigraphic' traps, termed 'hydrostratigraphic' traps, which have held oil and gas for millions of years. Examples are traps bounded by unconformities, facies change, anticlines and nontransmissive faults (Fig. 4). Obviously, these are
134
W. D. GUNTER ETAL
Fig. 5. Migration paths in hydrodynamic (open) traps in deep saline aquifers: (a) in regional-scale systems, and (b) in interconnected systems through faults and fractures.
Fig. 4. Hydrostratigraphic (closed) traps: (a) at unconformities, (b) caused by facies change; (c) in anticlines; and (d) at non-transmissive faults.
very attractive for CO2 storage due to their long history of containment. A second type of hydrogeological trapping in sedimentary basins is provided in deep saline aquifers (Fig. 2) that characteristically have slow flow rates of the order of centimetres per year (Bachu et al 1994). The pore space is filled with saline formation water. Carbon dioxide can be injected into these deep aquifers by displacing the saline formation water (Hitchon 1996). If the trap is not completely sealed, CO2 is expected to migrate under the force of buoyancy towards the surface. The pathway that the CO2 takes is determined by the complex plumbing of the sedimentary basin. Buoyant CO2 will seek out the interconnected high permeability pathways,
including interconnected aquifers, faults and fractures that will carry it upwards where it may eventually discharge at the surface (Fig. 5). A volume of CO2 injected into such a deep open hydrogeological trap can take over a million years to travel updip to reach the surface and be released into the atmosphere; for this reason this storage mechanism is referred to as a 'hydrodynamic' trap (Bachu et al. 1994). Flow rates in regional-scale flow systems in the deeper part of the sedimentary basins are extremely small, of the order of centimetres per year or less, whereas distances from the deep injection sites to discharge at outcrop can be of the order of hundreds of kilometres, as in the case of the Alberta basin (Fig. 6). Therefore, the time needed for a volume of fluid to reach the surface from the deep basin is measured on a geological timescale (more than 105 years), while the timeframe needed to stabilize CO2 atmospheric concentrations is of the order of 102 years.The efficiency of hydrodynamic trapping is significantly enhanced when the flow of formation waters is driven downdip by erosional rebound, as in the Cretaceous aquifers in the Alberta basin (Fig. 6; Bachu 1995). The point is that all sedimentary basins are leaky over geological, but not necessarily human, timescales. During such a long time period, geochemical trapping can take place to store CO2 permanently in geological media (Bachu et al. 1994), assuming that no future metamorphic or volcanic events affect it.
Geochemical trapping While the CO? is trapped in the subsurface, it will undergo a sequence of geochemical interactions
GEOLOGICAL TRAPPING OF CO7
135
Fig. 6. Cross-section through the Alberta basin, showing main flow types and systems (after Bachu 1995).
with the host rock controlled by pH (Fig. 7), formation water composition, pressure, temperature and rock mineralogy that will further increase the storage security. First, it will dissolve in the water, acidifying it, in so-called 'solubility' trapping, then it will form ionic species ('ionic trapping') as the host rock dissolves accompanied by a rise in the pH, and, finally, some fraction of the CO2 may be converted to stable mineral phases ('mineral trapping'), such as carbonates and dawsonite (Hitchon 1996; Knauss et al 2001). In geological settings with closed traps, geochemical reactions are not a necessary condition for safe and secure CO2 storage. In fact, many such CO2 reservoirs are in existence today, either filled with nearly pure CO2 or mixed with hydrocarbons (Baines & Worden 2001). For hydrodynamic traps, geochemical trapping is an important and perhaps an essential step for the secure storage of CO2. For example, once the CO2 is dissolved in brine (or oil), it will no longer migrate upwards as a separate phase. Once dissolved, it will travel at the same rate as the in-situ fluids, which may remain underground for millions of years before discharging at the surface. Short-term (decades) numerical simulations by Law & Bachu (1996), Doughty et al (2001) and Johnson et al (2001) have shown that of the order of 10-20% of the CO2 will dissolve in the pore fluid shortly after it is injected underground. Longer-term simulations have shown that even long after the initial injection, CO2 will continue to dissolve as a result of diffusion and local convection driven by the small density differences caused by dissolution of CO2 into the brine (Lindeberg et al 2003; Ennis-King & Patterson 2003). More important, but not yet studied in detail, is the ongoing dissolution into unsaturated brine that will occur as CO2 migrates through the sedimentary basin under the influence of buoyancy. Contacting new brine that is normally under-saturated with respect to CO2 will provide additional opportunities
for solubility trapping. Mass balance calculations using solubility-trapping rates determined from the simulations cited above suggest that 90% of the plume may dissolve after migrating distances of the order of ten times their original size, whereas basinscale simulations indicate that all the injected CO2 may ultimately dissolve (McPherson & Cole 2000). Solubility trapping is the dissolution of the CO2 in the formation water by the chemical reaction: Carbon dioxide solubility in pure water increases sharply with increasing pressure up to approximately 7 MPa, at which point it starts to level off at concentrations over 1 molal (i.e. 5% by weight) (Fig. 8). However, the CO2 solubility in formation water, whose temperature and most often salinity increase with depth, decreases (Fig. 8). Nevertheless, at storage depths characteristic of sedimentary basins, a significant amount of the CO2 will dissolve and be carried in the aqueous phase. Once the CO2 is dissolved in water, the second type of geochemical trapping can occur, that is ionic trapping. As it flows through the pores, the dissolved CO2 acts as a weak acid and will react with the sodium and potassium basic silicate (e.g. feldspars, clay minerals) or calcium, magnesium and iron carbonate, and silicate minerals (e.g. feldspars, clay minerals, olivines) in the reservoir or aquifer, and becomes neutralized by forming bicarbonate ions or carbonate ions (Gunter et al 1993) by chemical reactions of the form:
The aluminum silicate mineral neutralizes the weak acid (and at the same time increases the salinity of the formation water), as shown by the homogeneous reaction (where OH~ is contributed by mineral dissolution):
136
W.D. GUNTER £TAL.
Fig. 8. Dependence of CO2 solubility in water on pressure and salinity (from Koide et al. 1993). Fig. 7. Dependence of solubility, ionic and mineral trapping of CO2 on the chemical characteristics of the host formation water at equilibrium. The phase boundary along which carbonate minerals would precipitate is indicated by solid lines. This phase boundary may shift vertically up or down depending on pressure, temperature and the total concentration of the carbonate species (i.e. H2C03 + HC03-(aq) + C032 (aq)). Below the phase boundary only the aqueous phase is present which contains various coexisting carbonate species, each dominating over select pH ranges.
which then could continue depending on the availability of hydroxide ions (and therefore pH) to: Consequently, the carbonate species in the formation water forming from solution and reaction of the CO2 are functions of pH. The boundary between solubility and ionic trapping coincides with the changeover in dominance of the carbonate species (Fig. 7). Whereas in solubility trapping the reaction reverses and CO2 is released when the pressure drops, in ionic trapping, once the bicarbonate ion is formed, a drop in pressure releases considerably less CO2. The exact amount is dependent on the pH of the formation water. Reaction times vary. Solubility trapping is rapid near the CO2-water interface and is limited by the diffusion of the dissolved CO2 away from the interface. Reaction of the dissolved CO2 with minerals is rapid (days) in the case of carbonate minerals, but slow in the case of silicate minerals (hundreds to thousands of years, Gunter et al 1991 b). Carbonate minerals dissolve in the presence of high levels of dissolved CO2 and would create additional porosity.
However, the extent of reaction of dissolved CO2 with carbonate minerals before reaching equilibrium is small (Emberley et al 2003), so that the strength of the rock layer is not threatened in carbonate reservoirs for a closed system (i.e. assumes that the fluid flux is not large/continuous across the mineral interfaces). This is not the case when mixed gases such as acid gas (i.e. a mixture of H2S and CO2) are injected into a storage formation (Bachu & Gunter 2004; and discussion below). The H2S will strip the iron from the carbonate minerals to produce iron sulphide minerals, and the CO2 pressure, instead of decreasing, will build up, in which case fracturing may occur depending on the extent of the reaction (determined by the ratio of H2S : CO2 and mineralogy) and pressure dissipation rate. The most secure form of geochemical trapping is by formation of carbonate minerals from continued reaction of the bicarbonate ions with the calcium, magnesium and iron basic silicate minerals present in the rock matrix, such as clay minerals, micas, chlorites and feldspars (Gunter et al 1993; 1997&) by reactions of the form:
This happens most strongly when the formation water is located in the dissolved-carbonate dominance field, but typically it happens in the bicarbonate field (Fig. 7), as normally bicarbonate ions dominate over carbonate ions in formation waters. The basic silicate minerals contribute both the cation (e.g. Ca2+) to the carbonate minerals precipitating and the hydroxide ion to neutralize any proton production.
GEOLOGICAL TRAPPING OF CO2
Fig. 9. Changes in mineral amounts with time due to geochemical reactions that would take place in a typical siliciclastic aquifer in the case of: (a) a CO2-charged environment; and (b) a H2S-charged environment (after Gunter et al 2000).
Figure 9a shows the reactions that would take place in a typical siliciclastic (glauconitic) aquifer (the assumptions, uncertainties and details of these calculations are found in Gunter et al 2000), where the breakdown of basic iron minerals (in this case an iron-rich mica) combine with the injected CO2 to form siderite and muscovite, and dissolved bicarbonate in the formation water. Albite and potassium feldspar also breakdown to form muscovite and neutralize the CO2 by forming bicarbonate ion, but this trapping of CO2 is partially offset by the acid clay kaolinite reacting with the bicarbonate ion to produce CO2 and muscovite. If kaolinite were not present, then the formation water would become increasingly saline and could more efficiently move towards becoming a bicarbonate brine. The trapping is controlled by the silicate reactions, with mineral trapping dominating in the first fifty years by the formation of siderite, and then, over the next 1000 years (the length of time required to reach equilibrium), ionic trapping predominates before equilibrium is reached. Normally, if the basic aluminosilicate min-
137
erals are in excess, the geochemically-trapped CO2 will be in the most secure forms, distributed between the ionic and mineral trapping mechanisms. In a companion paper on acid gas re-injection, Bachu & Gunter (2004) describe the industry established around geological storage of acid gases in the Alberta basin. In this industry, geological storage of significant quantities of CO2 has already occurred, driven by regulatory limits on atmospheric H2S release. As the H2S and CO2 mixtures are injected into deep geological formations, they react with the reservoir rock. In a siliciclastic reservoir containing accessory carbonate minerals, the carbonate minerals will break down initially (because of the rapid surface reaction kinetics of the carbonates compared to silicates) to form iron sulphides and increase the CO2 pressure (Gunter et al. 2000). However, eventually, the slower-reacting silicates will reverse the reaction (e.g. carbonate minerals can react on the timescale of days whereas silicates are slower reacting, typically by five orders of magnitude). Iron released by the breakdown of iron-rich mica will combine with the CO2 to form siderite, and the other carbonate reactions with albite, potassium feldspar and kaolinite will continue (Fig. 9b), much the same as for the pure CO2 injection case discussed above. However, if no carbonate minerals are present, the breakdown of the iron-rich mica would result directly in iron sulphide mineral formation. The relative amounts of, and the composition of, the acid gas compared to the basic aluminosilicate minerals will determine the exact reaction path followed and the extent to which the injected acid gas can be stored in secure geochemical traps. Other factors that affect the rate of mineral trapping of the acid gas components are the interface between the mineral grains and the fluids, the flow rate of the fluids past the interface, and the pressure and temperature. Increasing the interface, temperature and pressure will increase reaction rates. The size of the interface is controlled by grain size (i.e. smaller grains have a larger surface area per unit volume), grain contacts (i.e. if a grain is in contact with another grain, the surface area of the contact is occluded from being in contact with the fluid and any reaction with the fluid must take place by diffusion through the mineral) and pore space (i.e. this is where the mineral-fluid contact occurs). The flow rate is controlled by the permeability (e.g. pore connectivity, size of pore throats) and pressure gradients. Creation of these carbonate minerals reduces the pore space and may lead to plugging of the flow paths in the case of sedimentary rocks containing large amounts of basic silicate mineral assemblages. However, if the basic silicate minerals represent less than 5% of the total mineral assemblage in an aquifer, the volume increase would be less than 1 % if they reacted completely to form carbonate minerals,
W. D. GUNTER ETAL.
138
which would reduce the porosity by less than 10%. As the flow rate is much faster than the reaction rate for silicate minerals, injectivity is unlikely to be impeded by silicate reactions. Trapping of CO2 by carbonate minerals is also possible at the surface as part of an industrial process (Goff et al 1997) involving mining of serpentinites and reacting them with CO2, Serpentine + 3CO2 -> 3 magnesite 4- quartz + H2O. A similar reaction could be written for calcium silicates such as wollastonite to form calcite and quartz. Here, there is a time constraint as long reaction times are not economically feasible. The rates of the carbonate mineral-forming reactions have been improved by reducing the grain size of the reactant minerals to maximize the reactive surface area and by artificially increasing the temperature above that commonly found in sedimentary basins. In North America, serpentine, the rock mineral needed for the above reaction, is found in the necessary quantities only on the east and west coasts of the continent (Goff et al. 1997), away from the hydrocarbonproducing mid-continent sedimentary basins. Wollastonite does not exist in the concentration and quantities necessary for significant mineral trapping amounts. Thus, there are two geochemical pathways for trapping CO2 inertly in the form of carbonate minerals. When sedimentary basins are near the source of emissions, geological storage will be favoured, but as distances increase between the source and the sedimentary basin, inorganic surface storage may be preferred.
Hydrogeology and efficiency of CO2 storage Efficient storage of carbon dioxide in geological formations has been discussed by van der Meer (1993), and depends on many factors that can be grouped into three main categories, namely: fluid properties, such as the density, viscosity and solubility of CO2; geochemical properties, such as the water-rock reactivity with CO2-saturated formation waters; and geological variables, such as the availability of sufficient pore space, the presence of a low permeability seal on the top of the storage formation to prevent the rapid upward migration of CO2 due to buoyancy forces, and sufficient permeability to allow injection at reasonable rates. The higher the density of CO2, then the more efficiently the pore space can be used. The fluid properties of CO2 are strongly dependent on pressure,
Fig. 10. Carbon dioxide unary phase diagram for the range of conditions found in sedimentary basins in general, and in the Alberta basin in particular.
temperature and presence of other gases. At normal atmospheric conditions, CO2 has a higher density than air, with a density of 1.872kgm~3 at 15°C and 101.325kPa. For temperatures greater than 31.1°C and pressures greater than 7.38MPa (critical point), CO2 is in the supercritical state (Fig. 10), behaving like a gas by filling all the available volume, but having a 'liquid' density that increases with pressure to values reaching that of oil and close to that of water (Fig. 1 la). Subcritical CO2 is either a gas or a liquid, depending on temperature and pressure (Fig. 10). Assuming hydrostatic pressure gradients (although in some cases basins or portions thereof may be abnormally pressured) and typical geothermal gradients varying between 20-60 °Ckrrr], a general envelope of CO2 phase behaviour in sedimentary basins can be identified (Fig. 10). Geothermal gradients vary not only between, but also within sedimentary basins, leading to basin-specific CO2-phase envelopes, as illustrated in Figure 10 for the Alberta basin. The 'cold' or 'warm' nature of a sedimentary basin has significant effects on the phase and flow behaviour of CO2 (Bachu 2002). Generally, cold basins are a better target for CO2 storage because greater CO2 density, hence storage efficiency, can be achieved for the same pressure (i.e. depth). In addition, if all other conditions are the same, buoyancy effects are smaller in cold basins than in warm basins as a result of higher CO2 density, hence the updip migration of a CO2 plume will be slower than in a warm basin. CO2 density and viscosity increase with pressure and decrease with temperature (Fig. 11). Since the temperature and pressure increase with depth in a sedimentary basin, the density of CO2 will generally increase as a result of pressure effects, but then it will level off as a result of strong temperature effects, as shown in Figure lla for the case of a hydrostatic pressure gradient and a geothermal gradient of
GEOLOGICAL TRAPPING OF CO,
139
Fig. 11. Relevant CO2 properties as a function of temperature and pressure: (a) density; and (b) viscosity. The black irregularly dashed line shows actual variation of density and viscosity in a sedimentary basin assuming a hydrostatic pressure gradient and a geothermal gradient of 30 °C km"'.
30 °Ckm l. Generally the pressure is expected to be sufficiently high for CO2 to be in liquid or supercritical phase at depths greater than 800m (equivalent to 8MPa hydrostatic column), where it should have sufficient density for efficient storage. However even at these depths, CO2 is considerably less dense than brine (600-800 kg m"3, depending on temperature, compared to more than lOOOkgm"3 for brine) leading to buoyancy effects. In reality, the depth at which CO2 has significant density for storage efficiency and safety varies between and within sedimentary basins as a result of variations in geothermal gradients, hence temperature (Bachu 2002, 2003). For example, the depth at which CO2
reaches supercritical conditions varies in the Alberta basin from more than 1200m in the south to less than 700m in the north (Bachu & Stewart 2002). Depending on the geothermal regime in a sedimentary basin or parts thereof, the density and viscosity of CO2 continue to increase only slightly with increasing depth, as illustrated in Figure 11 for the Alberta basin, but not enough to justify deeper storage on that basis alone. In fact, the increase in temperature associated with the natural geothermal gradient counteracts most of the density increase that would be expected due to increasing pressures with depth. In other cases, after reaching first liquid phase, the density of CO2 actually decreases slightly
140
W.D.GUNTERCTAL.
with increasing depth as it changes phase to supercritical (Bachu 2003). The lack of significant increase in the density of CO2 below certain depths suggests that the maximum storage depths will be determined by other factors such as drilling and monitoring costs, which may limit maximum depths to several kilometres. The density of CO2, and consequently the storage efficiency, also depends strongly on the degree to which other gases are present. For example, at a depth of 1.5km and a temperature of 40 °C, Oldenburg & Benson (2002) have shown that as little as 5% methane can decrease the density of CO2 mixtures by as much as 20%, thus significantly decreasing the storage efficiency. Similarly, high concentrations of nitrogen or air will not only lower the density of CO2, but also limit the amount of pore space available for CO2 storage. Looking to the future, one of the most important decisions to be made is what is the optimal composition of the sequestered CO2. This decision will hinge on a number of factors, including the efficiency (and cost-effectiveness) of CO2 separation technologies and the possibility of co-disposing of CO2 and other priority air pollutants such as SOz, NO^, H2S and mercury. The storage efficiency also depends on multiphase flow processes that determine the extent to which CO2 can displace the fluids that originally occupy the rock pore space. The viscosity of CO2 (Fig. 1 Ib) is lower by up to two orders of magnitude than that of pure water, which is ImPa-s at 20 °C, and it is much smaller than that of oil. Carbon dioxide has a lower surface tension than water. Both of these factors limit the fraction of the original pore fluids that can be displaced by CO2, and consequently, limit the storage efficiency. For example, the Buckley-Leverett theory (Buckley & Leverett 1942) can be used broadly to predict the fractional flow (fCO2) (Fig. 12a) and saturation distribution (rs) of CO2 (Fig. 12b) under steady state injection in water, neglecting the effects of solubility of one fluid in another, capillary forces, gravitational forces, and extreme mobility ratios:
where q is flow rate (specific discharge), JJL is viscosity, p is density, kr is relative permeability, S is saturation, ()> is porosity, h is aquifer thickness, t is time, and the subscripts CO2 and w stand for carbon dioxide and water, respectively.
Fig. 12. Advancement of a CO2 plume as predicted by Buckley-Leverett theory: (a) fractional flow, and (b) CO2 saturation as a function of distance from the injection well. Sf indicates the CO2 saturation (Sco) at the CO2 flow front of approximately 0.2 whereas the fractional flow of CO2 (/c0l) is approximately 0.8 compared to water. The front is located approximately 130 m (rm) from the injection well at time t.
Application of Buckley-Leverett theory indicates that, for typical relative permeability (kr) curves, the saturation at the front of the plume will be about 20% and the fractional flow of CO2 approaches 100% when the saturation of CO2 approaches about 30% (see Fig. 12). Consequently, the average saturation in the CO2 plume will be in the range of 20-30%. Thus, at best, only 20-30% of the pore space can be used for CO2 storage in brine-saturated deep formations. In oil reservoirs, similar efficiency is likely, due to the fact that most oil reservoirs undergo water flooding for secondary oil recovery before they are used for CO2 enhanced oil recovery or storage. The storage efficiency in gas reservoirs is likely to be higher because, compared to methane, CO2 is more viscous
GEOLOGICAL TRAPPING OF CO,
141
Fig. 13. The effects of: (a) CO2 buoyancy in horizontal and dipping aquifers; and (b) changing from homogeneous to medium lithological heterogeneity in a horizontal aquifer, on the flow of CO2 in a brine-saturated formation (after Doughty et al. 2001). Low permeability areas indicated by diamond pattern.
and has a higher density. In addition, CO2 is miscible with natural gas. Numerical simulations by Oldenburg et al. (2001) suggest that nearly all of the pore space in depleted gas reservoirs may be available for CO2 storage. Low displacement and storage efficiency in brinefilled and oil reservoirs is compounded by two other important factors, namely, the effect of buoyancy, which drives CO2 towards the top of the formation (gravity override), and geological heterogeneity, which leads to preferential flow paths and controls the migration of the CO2 plume. Figure 13 shows the effects of CO2 buoyancy and medium lithological heterogeneity. Figure 14 presents results of numerical simulations of injecting CO2 into an aquifer in the Alberta basin (Law & Bachu 1996) that show the advancement of the CO2 plume and illustrates buoyancy effects (gravity override at the top of the aquifer) and reduced sweep efficiency of the aquifer as a result of the density contrast between the injected CO2 and the brine in the host formation. Doughty et al (2001) performed threedimensional simulations of twenty years of CO2 injection into a heterogeneous rectangular block (Fig. 15a) to assess the importance of buoyancy and heterogeneity on CO2 storage efficiency (Fig. 15b) on the efficiency of CO2 geological storage. There are several important conclusions to be drawn from these simulations. First, subsurface heterogeneity and gravitational forces (buoyancy) both significantly reduce storage efficiency, in this case by a factor of about 0.3 and 0.5, respectively. Second, reservoir heterogeneity, in particular the presence of low permeability layers within the reservoir may counteract the influence of buoyancy and lead to more efficient storage. Finally, after the end of the
twenty-year injection period, the plume continues to migrate out of the simulated aquifer volume, thus further decreasing the fraction of pore volume occupied by CO2. This mimics the slow migration of the plume that would occur if the hydrogeological trap were not completely closed. Two other factors will increase the efficiency of the storage of CO2. First, dissolution of CO2 into the aquifer brine under the appropriate pressure-temperature conditions leads to a higher density brine phase (Bachu & Adams 2003), which would flow downdip, being displaced by the lighter CO2-unsaturated brine, which in turn helps to counteract the rising of the lighter CO2 plume. Second, during the transport of CO2, there are continuous reactions between the CO2-charged brine and the minerals it contacts. Geochemical modelling of CO2 dispersion in the Utsira aquifer for the CO2 storage project at Sleipner Vest by Johnson et al (2001) indicated a significant contribution to the storage efficiency through mineral precipitation. In summary, the storage efficiency is influenced by many factors that depend on the physical, chemical and geological characteristics of the storage site. Storage efficiency is likely to be highly site specific. During the injection phase of the CO2 storage project, up to 30% of the pore volume may be occupied by CO2, but most likely, it will be in the range of one third to one sixth. The deeper the formation, the greater the density of CO2 and, consequently, the higher the storage efficiency. However, the incremental benefits of increasing density with greater depths diminish below 1-2 km depth due to the effects of the geothermal gradient. Solubility, ionic and mineral trapping also increase storage efficiency, but more importantly, increase storage security.
142
W.D.GUNTERE7AL.
Fig. 14. Simulated advancement of a CO2 front in an aquifer in the Alberta basin (after Law & Bachu 1996). Age contours (years) are for 50% CO2 saturation.
Discussion and conclusions The global energy mix, dominated by fossil fuels, hydro and biomass, and to a lesser extent, wind and solar power, is becoming increasingly environmentally-constrained. It is our opinion that the conversion of fossil fuels to the secondary energy forms consisting of electricity, hydrogen, heat and pressure will evolve in a way to minimize release of pollutants. End of the pipe solutions for fossil fuel conversion will gradually be replaced by cleaner conversion technologies. These, in turn, will be replaced by renewable and other forms of energy. However, until then fossil fuels will continue to play an important role (Bajura 2001), and the geological storage of carbon dioxide will, accordingly, play a significant role during the transition period in this and next century. In preparation for this, geological mapping of sedimentary basins needs to be revisited with more focus on the plumbing system and the safe storage of CO2 than has been the traditional approach of the energy industry (Bachu 2000, 2001). We think that integration of oil and gas recovery operations with energy conversion (Gunter et al 19970, Hitchon et al 1999; Wong effl/. 2001) will be an early part of the future route leading to a world energy supply based largely on renewables. Injection and storage technologies, developed by the oil and gas industry, are fairly mature. The volume of storage depends on the current and ultimate pressures of the reservoir or aquifer. Experience in injection of CO2 has been gained from repressurizing oil reservoirs using CO2 in enhanced oil recovery, from acid-gas re-injection, and similar technology is being developed for production of methane from coal beds (i.e. coalbed methane or CBM). The ultimate capacity of geological storage of carbon dioxide is likely to be huge, contingent upon identifying secure traps in sedimentary basins. Over longer time periods, mineral trapping may become a long-term stable sink for CO2. The extent
Fig. 15. The effect of CO2 buoyancy and heterogeneity on the efficiency of CO2 geological storage as indicated by numerical simulation of 20 years of CO2 injection into a heterogeneous aquifer: (a) simulation grid showing the heterogeneous structure of the subsurface; and (b) storage efficiency (as defined by capacity) (after Doughty et al. 2001).
and rate at which this occurs depends on the mineralogy and brine chemistry of the sedimentary rocks contacted by CO2. The security of hydrogeological and geochemical storage, namely, the ability to store CO2 safely and effectively in the pore space of sedimentary basins, is illustrated in Figure 16. The most secure hydrogeological traps are closed stratigraphic or structural traps, which have been well characterized during their exploitation for oil and gas. Although the capacity of these traps for CO2 storage is small relative to open hydrodynamic traps in deep sedimentary basins, they are likely to be used first as they are known to be secure, having held oil and gas for geological time. Storage of CO2 as carbonate minerals is the most secure form of storage, but the reactions that trap the CO2 in carbonate minerals are slow on the human timescale, but relatively fast on a
GEOLOGICAL TRAPPING OF CO,
Fig. 16. The relation between the security of geological storage of CO2 and the main mechanisms of CO2 trapping in sedimentary basins.
geological timescale. As the capacity of closed traps is exhausted and more is learned about the rates of geochemical trapping, the large storage capacity available in open hydrodynamic traps will be utilized, where the security of the geological storage of CO2 can be enhanced by geochemical reactions of the CO2 with basic silicate minerals. References BACHU, S. 1995. Synthesis and model of formation water flow in the Alberta basin. American Association Petroleum Geologists Bulletin, 79,1159-1178. BACHU, S. 2000. Sequestration of carbon dioxide in geological media: Criteria and approach for site selection. Energy Conversion and Management, 41,953-970. BACHU, S. 2001. Geological sequestration of anthropogenic carbon dioxide: Applicability and current issues, In: GERHARD, L., HARRISON W. E. & HANSON, B. M. (eds) 2001. Geological Perspectives of Global Climate Change, AAPG Studies in Geology 47, American Association of Petroleum Geologists, 285-304. BACHU, S. 2002. Sequestration of CO2 in geological media in response to climate change: roadmap for site selection using the transform of the geological space into the CO2-phase space. Energy Conversion and Management, 43, 87-102. BACHU, S. 2003. Screening and ranking of sedimentary basins for sequestration of CO2 in geological media in response to climate change. Environmental Geology, 44,277-289. BACHU, S. & ADAMS, J. J. 2003. Sequestration of CO2 in geological media in response to climate change: Capacity of deep saline aquifers to sequester CO2 in solution. Energy Conversion and Management, 44, 3157-3175. BACHU, S. & GUNTER, W. D. 2004. Acid gas injection in the Alberta basin, Canada: A CO2 storage experience. In: BAINES, S. & WORDEN, R. H. (eds) Geological Storage of Carbon Dioxide, Geological Society, London, Special Publications, 233,225-234.
143
BACHU, S. & STEWART, S. 2002. Geological sequestration of anthropogenic carbon dioxide in the Western Canada Sedimentary Basin: Suitability analysis. Journal of Canadian Petroleum Technology, 41, 32-40. BACHU, S., GUNTER, W. D. & PERKINS, E. H. 1994. Aquifer disposal of CO2: Hydrodynamic and mineral trapping. Energy Conversion and Management, 35,269-279. BACHU, S., MICHAEL, K. & ADAMS, J.J. 2003. Effects of insitu conditions on aquifer capacity for CO2 sequestration in solution. In: Second Annual Conference on Carbon Sequestration, May 5-8, Alexandria, VA, DOE National Energy Technology Laboratory. BAINES, S. J. & WORDEN, R. H. 2001. Geological CO2 disposal: Understanding the long term fate of CO2 in naturally occurring accumulations. In: WILLIAMS, D. J., DURIE, R. A., MCMULLAN, P., PAULSON, C. A. J. & SMITH, A. Y (eds) Fifth International Conference on Greenhouse Gas Control Technologies (GHGT-5), CSIRO Publishing, Collingwood, Vic, Au, 311-316. BAJURA, R. A. 2001. The role of carbon dioxide sequestration in the long term energy future; In: WILLIAMS, D. J., DURIE, R. A., MCMULLAN, P., PAULSON, C. A. J. & SMITH, A. Y (eds) Fifth International Conference on Greenhouse Gas Control Technologies (GHGT-5), CSIRO Publishing, Collingwood, Vic, Au, 52-58. BENSON, S. M., 2001. Comparison of three options for geologic sequestration of CO2 - A case study for California. In: WILLIAMS, D. J., DURIE, R. A., MCMULLAN, P., PAULSON, C. A. J. & SMITH, A. Y (eds) Fifth International Conference on Greenhouse Gas Control Technologies (GHGT-5), CSIRO Publishing, Collingwood, Vic, Au, 299-304. BERGMAN P. D. & WINTER E. M. 1995. Disposal of carbon dioxide in aquifers in the US Energy Conversion and Management, 36, 523-526. BUCKLEY, S. E. & LEVERETT, M. C., 1942. Mechanisms of fluid displacement in sand. In: Transactions of the American Institute of Mechanical Engineers, 146, 107-116. CELIA, M. A. & BACHU, S. 2003. Geological sequestration of CO2: is leakage unavoidable and acceptable? In: GALE, J. & KAYA, Y. (eds) Proceedings of the 6th International Conference on Greenhouse Gas Control Technologies, October 1-5, 2002. Pergamon, Kyoto, Japan, 477-482. DOUGHTY, C., PRUESS, K., BENSON, S. M., HOVORKA, S., KNOX, P. & GREEN, C. 2001. Capacity investigation of brine bearing sands on the Frio Formation for geologic sequestration of CO2. In: First National Conference on Carbon Sequestration, May 14-17, Washington DC, DOE National Energy Technology Laboratory. DUSSEAULT, M. B., BACHU, S. & DAVIDSON, B. 2001. Carbon dioxide sequestration in Alberta salt solution caverns. In: Solution Mining Research Institute, Fall 2001 Technical Meeting, Albuquerque, NM, Oct. 8-10. DUSSEAULT, M. B., BACHU, S. & ROTHENBURG, L. 2002. Sequestration of CO2 in salt caverns. Paper 2002-237, CIM Petroleum Society 53rd Annual Technical Meeting, Canadian International Petroleum Conference. Calgary, AB, June 11-13,2002,1 Ip.
EMBERLEY, S., HUTCHEON, I. SHEVALIER, M. DUROCHER,
144
W. D. GUNTER ETAL.
K., GUNTER, W. D. & PERKINS, E. H, 2003. Geochemical monitoring of fluid-rock interaction and CO2 storage at the Weyburn CO2-injection enhanced oil recovery site, Saskatchewan, Canada, In: GALE, J. & KAYA, Y. (eds) Proceedings of the 6th International Conference on Greenhouse Gas Control Technologies, October 1-5, 2002, Kyoto, Japan. Pergamon, 365—370. ENNIS-KING, J. & PATERSON, L. 2003. Rate of dissolution due to convective mixing in the underground storage of carbon dioxide. In: GALE, J. & KAYA, Y. (eds) Proceedings of the 6th International Conference on Greenhouse Gas Control Technologies, October 1-5, 2002, Pergamon, Kyoto, Japan. 507-370. GOFF, R, GUTHRIE, G. & LACKNER, K. 1997. Carbon dioxide sequestering potential of ultramafic rocks, In: 23rd International Technical Conference on Coal Utilization and Fuel Systems, March 16-19, Clearwater, Florida, 603-613. GUNTER, W. D., PERKINS, E. H. & HUTCHEON I. 2000. Aquifer disposal of acid gases: Modelling of water-rock reactions for trapping acid wastes. Applied Geochemistry, 15, 1085-1095. GUNTER, W. D., CHALATURNYK, R. J. & SCOTT, J. D. 1999. Monitoring of aquifer disposal of CO2. Experience from underground gas storage and enhanced oil recovery, In: REIMER, P., ELIASSON, B. & WOKAUN, A. (eds.) Proceedings of the 4th International Conference on GHG Control Technologies, Interlaken, Switzerland. Pergamon, 151-156. GUNTER, W. D., WONG, S., CHEEL, D. B. & SJOSTROM, G. 1998. Large CO2 sinks: Their role in the mitigation of greenhouse gases from an international, national (Canadian) and provincial (Alberta) perspective. Applied Energy, 61, 209-227. GUNTER, W..D., GENTZIS, T., ROTTENFUSSER, B. A. & RICHARDSON, R. J. H. 1997a. Deep coalbed methane in Alberta, Canada: A fuel resource with the potential of zero greenhouse gas emissions, Energy Conversion and Management Supplement, 38, S217-S222. GUNTER, W. D., WIWCHAR, B. & PERKINS, E. H. 1997&. Aquifer disposal of CO2-rich Greenhouse gases: extension of the time scale of experiment for CO2sequestering reactions by geochemical modelling. Mineralogy and Petrology, 59,121-140. GUNTER, W. D., PERKINS, E. H. & MCCANN, T. J. 1993. Aquifer disposal of CO2-rich gases: reaction design for added capacity. Energy Conversion and Management, 34,941-948. HEPPLE R. & BENSON, S. M. 2003. Implications of surface leakage on the effectiveness of geologic storage of carbon dioxide as a climate change mitigation strategy. In: GALES, J. & KAYA, Y. (eds) Proceedings of the 6th International Conference on Greenhouse Gas Control Technologies, Kyoto, Japan, Sept.30-Oct.4, 2002. Pergamon, 261-266. HITCHON, B. (ed.) 1996. Aquifer Disposal of Carbon Dioxide: Hydrodynamic and Mineral Trapping -Proof of Concept, Geoscience Publishing Ltd., Sherwood Park, Alberta, 165p. HITCHON, B., GUNTER, W. D., GENTZIS, T. & BAILEY, R. 1999. Sedimentary basins and greenhouse gases: A serendipitous association. Energy Conversion and Management, 40, 825-843.
HOLLOWAY, S. 1997'a. An overview of the Joule II project: The underground disposal of carbon dioxide. Energy Conversion and Management, 38S, S193-S198 HOLLOWAY, S. 1997£>. Safety of the underground disposal of carbon dioxide. Energy Conversion and Management, 38S,S241-S245 IEA.1995. Carbon Dioxide Disposal from Power Stations. IEA Greenhouse Gas R&D Programme. Publication ISBN 1898373 07 8. JOHNSON, J. W., NITAO, J. J, STEEFEL, C. I. & KNAUSS, K. 2001. Reactive transport modeling of geologic CO2 sequestration in saline aquifers: The influence of intra-aquifer shales and the relative effectiveness of structural, solubility, and mineral trapping during prograde and retrograde sequestration. In: Proceedings of the 1st National Conference on Carbon Sequestration, May 14-17, Washington DC, DOE National Energy Technology Laboratory 60p. KNAUSS, K. G., JOHNSON, J. W, STEEFEL, C. I. & NITAO, J. J. 2001. Evaluation of the impact of CO2, aqueous fluid, and reservoir rock interactions on the geologic sequestration of CO2, with special emphasis on economic implications. In: Proceedings of the 1st National Conference on Carbon Sequestration, May 14-17, Washington DC, DOE National Energy Technology Laboratory lip. KOIDE, H. G., TAZAKI, Y, NOGUCHI, Y, IIJIMA, M., ITO, K. &. SHINDO, Y 1993. Underground storage of carbon dioxide in depleted natural gas reservoirs and useless aquifers. Engineering Geology, 34,175-179. LAW, D. H-S. & BACHU, S.1996. Hydrogeological and numerical analysis of CO2 disposal in deep aquifers in the Alberta sedimentary basin. Energy Conversion and Management, 37, 1167-1174. LINDEBERG, E., BERGMO, P. & MOEN, A. 2003. The longterm fate of CO2 injected into an aquifer. In: GALE, J. & KAYA, Y. (eds) Proceedings of the 6th International Conference on Greenhouse Gas Control Technologies, October 1-5, 2002, Kyoto, Japan. Pergamon, 489-494. MCPHERSON, B. J. O. L. & COLE, B. S. 2000. Multiphase CO2 flow, transport and sequestration in the Powder River basin, Wyoming, USA. Journal of Geochemical Exploration, 69-70,65-69. OLDENBURG C. M. & BENSON, S. M. 2002. CO2 injection for enhanced gas production and carbon sequestration. SPE Paper 74367. In: Society of Petroleum Engineers International Conference and Exhibition, Mexico City, February 10-12,2002. OLDENBURG, C. M., PRUESS, K. & BENSON, S. M. 2001. Process modeling of CO2 injection into natural gas reservoirs for carbon sequestration and enhanced gas production. Energy and Fuels, 15,293-298. OLIVER, J. 1986. Fluids expelled tectonically from orogenic belts, their role in hydrocarbon migration and other geologic phenomena. Geology, 14,99-102. ST JOHN, B., BALLY, A. W. & KLEMME, H. D. 1984. Sedimentary provinces of the world-hydrocarbon productive and nonproductive. The American Association of Petroleum Geologists, Tulsa, OK. 35 pp. STEVENS, S., KRUUSKRAA, V. & SPECTOR, D. 1999. CO2 sequestration in deep coal seams: Pilot results and worldwide potential, In: REIMER, P., ELIASSON, B. & WOKAUN, A. (eds) Proceedings of the 4th International
GEOLOGICAL TRAPPING OF CO2 Conference on GHG Control Technologies, Interlaken, Switzerland, Pergamon, 175-180. VAN DER MEER, L. G. H. 1993. The conditions limiting CO2 storage in aquifers. Energy Conversion and Management, 34,959-966. WONG, S., GUNTER, W. D., LAW, D. H.-S. & MAVOR, M. J. 2001. Economics of flue gas injection and CO2
145
sequestration in coalbed methane reservoirs, In: WILLIAMS, D. J., DURIE, R. A., MCMULLAN, P., PAULSON, C. A. J. & SMITH, A. Y. (eds) Fifth International Conference on Greenhouse Gas Control Technologies (GHGT-5), CSIRO Publishing, Collingwood, Vic, Au, 543-548.
This page intentionally left blank
The impact of geological heterogeneity on CO2 storage in brine formations: a case study from the Texas Gulf Coast SUSAN D. HOVORKA1, CHRISTINE DOUGHTY2, SALLY M. BENSON2, KARSTEN PRUESS2 & PAUL R. KNOX1 Bureau of Economic Geology, BoxX, Jackson School of Geosciences, The University of Texas at Austin, Austin TX 78713, USA (e-mail:
[email protected]) 2 Lawrence Berkeley National Laboratory, 1 Cyclotron Road Mailstop 90-1116 Berkeley, CA 94720, USA Abstract: Geological complexities such as variable permeability and structure (folds and faults) exist to a greater or lesser extent in all subsurface environments. In order to identify safe and effective sites in which to inject CO2 for sequestration, it is necessary to predict the effect of these heterogeneities on the short- and long-term distribution of CO2. Sequestration capacity, the volume fraction of the subsurface available for CO2 storage, can be increased by geological heterogeneity. Numerical models demonstrate that in a homogeneous rock volume, CO2 flowpaths are dominated by buoyancy, bypassing much of the rock volume. Row through a more heterogeneous rock volume disperses the flow paths, contacting a larger percentage of the rock volume, and thereby increasing sequestration capacity. Sequestration effectiveness, how much CO 2 will be sequestered for how long in how much space, can also be enhanced by heterogeneity. A given volume of CO2 distributed over a larger rock volume may decrease leakage risk by shortening the continuous column of buoyant gas acting on a capillary seal and inhibiting seal failure. However, where structural heterogeneity predominates over stratigraphic heterogeneity, large columns of CO2 may accumulate below a sealing layer, increasing the risk of seal failure and leakage.
In order to reduce anthropogenic increases in atmospheric concentrations of CO2 significantly, an alternative to releasing very large volumes of this gas must be found. Simple volumetric calculations (for example Bergman & Winter 1995; Holloway, 1996; Sass et al 1998; Hovorka et al 2000) indicate that appropriately large volumes of storage exist in the subsurface in porous, brine filled formations below, and hydrologically separated from, potable water. These brine formations have an advantage as compared to sequestration in reservoirs of not requiring the cost or environmental impact of pipeline transport of CO2. For auspiciously located facilities, disposal could follow the deep-well disposal model of going straight down into the subsurface from within the producing plant. Basin-scale analysis of brine residence time and flow (Bachu et al 1994) indicates that storage for geological timeframes are feasible. Technologies for subsurface injection are well known because CO2 has been injected into hydrocarbon reservoirs as part of an enhanced oil recovery (EOR) process for decades. In addition, there is a long operational history for waste-fluid disposal by underground injection, and the technological and regulatory environments for this process are also well known. On average, US power plants release 2.7 billion metric tons of CO2 per year, with the largest plants
releasing more than 18 million metric tons per year (US Geological Survey 2001). For CO2 to be injected, we assume that there must be (1) extraction of the CO2 from a mixed waste stream and drying in order to avoid compression costs of the nitrogen and oxygen and equipment being corroded by water vapour; (2) compression to conditions above the critical point, and (3) injection through one or more well bores to the target horizon(s). A typical project's life span might be 20 years. During this time, the CO2 as a supercritical fluid will immiscibly displace brine in a large area around the injection well, and the pressure within the subsurface injection horizon will increase over a much larger area. Over time, the 'bubble' of supercritical CO2 is expected to migrate under the influence of gravity and pressure to decrease as CO2 dissolves into the brine and as brine or CO2 slowly leaks from the injection site into lower pressure areas. Depending on host formation mineralogy, some of the CO2 may also form solid carbonates over longer timescales (tens to hundreds of years).
Problem and approach The purpose of this paper is to begin sitescale assessment of the applicability of geological
From: BAINES, S. J. & WORDEN, R. H. (eds) 2004. Geological Storage of Carbon Dioxide. Geological Society, London, Special Publications, 233,147-163.0305-8719/047$ 15.00 © The Geological Society of London 2004.
148
S.D.HOVORKAErAL.
Fig. 1. The 7-county area (16700 km2) centred on Houston, Texas, showing the locations of the two case studies modelled.
sequestration in brine formations for greenhouse gas emission reduction and to determine if the regionalscale estimates of potential for storage of large volumes are replicated when the complexities expected in real geological settings are introduced. Our goal is to test the feasibility of high-volume sequestration in a realistic brine-bearing formation in a typical area where there are currently large CO2 emissions to the atmosphere. This study is based in part on geotechnical and modelling support for a proposed pilot injection project, as well as on a more general assessment of issues of geological sequestration in brine formations for long-term, large-scale greenhouse gas reduction.
Setting The upper Texas Gulf Coast around Houston, Texas, is one of the areas where US CO2 emissions to the atmosphere are concentrated because of a combination of electric power generation and industrial activity. Within a 7-county area (16700km2) centred
on Houston, Texas (Fig. 1), ten power plants released an estimated 32 million metric tons of CO2 in 1996 (US Geological Survey 2001). In addition, more than 100 chemical manufacturing plants and refineries in the same area continue to release an unknown additional volume of CO2. We selected this area as a representative region in which strategies for emissions reduction must be implemented if US emissions are to be reduced sufficiently to affect atmospheric concentrations. This area of concentrated emissions in the upper Texas Gulf Coast overlies a thick sedimentary section (up to 17000m) that provides an attractive target for geological sequestration. The geological and petrophysical characteristics of Gulf Coast sandstones are well known because of a long and intensive history of hydrocarbon exploration and production. Sandstones are thick, laterally extensive, and permeable and are separated by regionally extensive shales deposited during marine transgressions (Fig. 2). The entire thick section of Gulf Coast Tertiary sediments onshore as well as on the shallow continental shelf offshore are prospective sequestra-
IMPACT OF GEOLOGICAL HETEROGENEITY
149
Fig. 2. Regional stratigraphy of the Texas Gulf Coast. Modified from Galloway et al. (1982).
tion sites. The Frio Formation was selected for this study as a representative unit for additional study because the sandstone-rich, high quality, wellknown targets for storage are onshore, and because the interval is overlain by a thick, regionally extensive shale, the Anahuac Formation. A basin-scale estimate of the total pore volume in sandstone of the Frio Formation can be made from regional data. The sample area is the Oligocene Frio Formation as defined by Galloway (1982) along the entire Texas coast. Pore volume calculations were made by extracting the net sand thickness of the Frio at somewhat arbitrary target depths between 1000 and 3000 m below the ground surface and multiplying it by gridded porosity from regional data to calculate brine-filled porosity-height (Fig. 3). Porosity-height times cell area for the entire Frio yield volume of 5840km3 of brine-filled porosity in sand. However, not all this volume can be replaced by CO2 Van der Meer (1995) estimated that 1-6% of pore volume can be used for sequestration. In this paper we assess the variables that influence the storage efficiency and the storage that is useful for sequestration.
Evaluating geological heterogeneity Geological complexities in the Frio as well as other onshore and offshore prospective sequestration sites include: stratigraphic heterogeneity resulting from transitions between rocks deposited in sand- and mud-dominated depositional facies in fluvial, deltaic, and barrier island settings, leading to a heterogeneous permeability distribution; and structural heterogeneity from growth faults, folds, and salt diapirs that complicate regional fluid-flow paths and provide potential avenues for leakage of CO2 and displaced brine. Overprinted on these hetero-
geneities is a complex hydrological regime involving coastward movement of fresh water, pressure depletion related to hydrocarbon production, and geopressure at depth (Kreitler et al. 1988). These geological and hydrological complexities provide a challenge for modelling and predicting the suitability of the units for sequestration. We use a reservoir characterization and geological play approach to extend our knowledge from well-known areas of hydrocarbon reservoirs and outcrops to poorly-known, large volume, unproductive brine formations that may be targets for storage. Reservoir characterization and play approaches are standard techniques for hydrocarbon exploration and reservoir development, but require modification to determine what hydrogeological settings are acceptable for CO2 injection.
TOUGH2 simulator The numerical simulations presented in this paper were performed with TOUGH2, a general-purpose simulator for multiphase flows in porous and fractured media (Pruess et al. 1999). TOUGH2 solves mass balance equations (optionally also an energy balance) for multicomponent fluid mixtures in which the components can partition into several fluid and solid phases. Flow is represented with a multiphase version of Darcy's law that includes relative permeability and capillary pressure effects. The continuum equations are discretized by means of an integral finite difference method, which for systems of regularly shaped grid blocks is mathematically equivalent to conventional finite differences (Narasimhan & Witherspoon 1976). For numerical stability, time is discretized fully implicitly as a first-order backward finite difference. Time
150
S.D.HOVORKAE7ML.
Fig. 3. Porosity distribution for the Frio Formation (in porosity-metres).
steps are automatically adjusted (increased or reduced) during the course of a simulation, to cope with variable non-linearities and convergence rates, especially during appearance or disappearance of phases. Discretization results in a system of coupled non-linear algebraic equations that are solved by Newton-Raphson iteration. The linear equations arising at each iteration step are solved by means of preconditioned conjugate gradient methods or sparse direct solvers (Moridis & Pruess 1998). The fluid property description is based on correlations originally developed for geothermal applications (Battistelli et al 1997), and subsequently enhanced to represent phase partitioning and thermophysical properties of water/CO2/NaCl mixtures at near-ambient temperatures and supercritical CO2 pressures more accurately (Pruess & Garcia 2002). Densities, viscosities, and enthalpies of CO2 are cal-
culated from correlations developed by Altunin (1975), as implemented in a computer program kindly provided to us by V. Malkovsky. Dissolution of CO2 in NaCl brines is described with an extended version of Henry's law that accounts for effects of 2 CO fugacity, temperature, and salinity. Details of the fluid property model are given in Pruess and Garcia (2002).
Impacts of geological heterogeneity on CO storage We consider two interacting types of geological constraints on the volume of CO2 that can be sequestered in geological settings: the capacity of the formation to accept CO2; and the effectiveness of sequestration.
IMPACT OF GEOLOGICAL HETEROGENEITY
Capacity
151
such as partially penetrating injection wells, gravity segregation, and dipping formations with spill points.
Capacity is defined as the volume fraction of the subsurface within a defined stratigraphic interval available for CO2 sequestration (Doughty et al 2001). It is a function of the rock and injection geometries, and does not address the issue of how long the CO2 will stay within the interval in which it is placed nor the maximum volume that can be injected safely, factors that are considered below under sequestration effectiveness. We consider C to be the product of four factors:
(i) Ci is intrinsic capacity, which is controlled by multiphase flow and transport phenomena; C is geometric capacity factor, which is controlled by formation and injection geometry; Ch is heterogeneity capacity factor, which is controlled by geological variability; and
is porosity, the fraction of void space within the formation. We have found this formulation useful for investigating the different processes that influence C and comparing our work to that of other authors. Reservoir simulation can provide estimates for the entire capacity factor C, but it is not generally possible to identify the various factors given in Eqn (1) unambiguously. Intrinsic capacity. Ci Intrinsic capacity Ci is defined as the fraction of pore space occupied by CO2 assuming radial flow through a uniform medium. C can be divided into gas- and liquid-phase components: Ci = Cig + Cn. A Buckley-Leverett type analysis (Buckley & Leverett 1942; Pruess et al 2001) gives (2) where Sg is the average gas saturation behind the front (the small contribution of water vapour to S has been neglected). For CO2 dissolved in the aqueous phase, (3) C0
where S{ and X, 2 are the saturation and CO2 mass fraction, respectively, averaged over the liquid (aqueous) phase behind the front and p;/p is the liquid:gas density ratio. Inclusion of the p/p, term makes Cu the volume fraction that CO2 dissolved in the liquid phase would occupy if it were converted to the gas phase. This formulation ensures that intrinsic capacity is additive between gas and aqueous phases, regardless of the phase partitioning of CO2, which depends strongly on pressure-temperature conditions. Geometric capacity factor Cg. Geometric capacity factor Cg accounts for departures from the idealized radial flow geometry assumed for intrinsic capacity,
Heterogeneity capacity factor Ch. Heterogeneity capacity factor Ch accounts for bypass flow arising from geological heterogeneity. This factor has been referred to as horizontal sweep efficiency in the petroleum literature. Calculating capacity factors. For non-radial flow or a heterogeneous medium, there may not be a single CO2 front. We extend the definitions of Cig and Cn given in Equations (2) and (3) to <4>
(5) where < > identifies averages over regions of space. Examples of averaging regions include the volume of a natural CO2 trap (defined by the distance from the injection well to a spill point or caprock discontinuity), the volume of a targeted geological formation, or some relevant unit volume (e.g. the volume enclosed by a lease boundary and horizons corresponding to the injection depth and the base of potable water). Obviously, the choice of averaging region will have a large effect on the capacity value calculated, so it must be chosen carefully to ensure meaningful comparison of different scenarios. The stage of CO2 plume development at which capacity is calculated is also important. Averages may be taken when the boundary of the averaging region is first encountered or, alternatively, when quasi-steady flow conditions exist throughout the averaging region. Effectiveness Assessing the effectiveness of sequestration addresses issues such as: what is the short-, intermediate-, and long-term fate of the CO2? What is the residence time for the CO2 in the subsurface? How much CO2 will be sequestered for how long in what three-dimensional space? We expect that some leakage will occur, leading to non-zero flux of CO2 from the subsurface to the atmosphere. Any calculation of the effectiveness will involve the analysis of flux over time to determine whether sequestration will be able to achieve significant reduction of atmospheric concentrations over time frames of hundreds of years. Some areas are not effective for sequestration because of unacceptable risks of shortor long-term negative impacts. Initial analysis of the effectiveness of sequestration is based on knowledge of the migration and distribution of hydrocarbons as well known and widely applicable partial analogues for sequestered CO2.
152
S. D. HOVORKA ETAL.
Fig. 4. A representative field-scale cross-section demonstrates the three types of heterogeneity: (1) stratigraphic heterogeneity, formed by lateral and vertical variability in rock properties such as porosity, permeability, and mineralogy, (2) folding of heterogeneous strata; and (3) faulting.
Heterogeneities The distribution of permeability in the subsurface environment is complex, and conceptualizing important issues so that they can be modelled is a critical step. Three categories of heterogeneity characteristic of the Frio Formation in the Houston area are analysed in this paper (Fig. 4): stratigraphic heterogeneity, folding of heterogeneous strata, and faulting.
Stratigraphic heterogeneity Stratigraphic heterogeneity results from variation in sedimentary depositional processes through time and space, and is the cause of lateral and vertical variability in rock properties such as porosity, permeability, and mineralogy. In our study unit, the Frio Formation in the Houston area, the stratification is created by thin but laterally continuous sandstonerich high frequency cycles (3-20 m thick) separated by low permeability mudstones and shales deposited during regional marine flooding events Depositional facies geometry has created additional lateral and vertical heterogeneity within the sandstone-rich facies. Figure 4 shows stratigraphic representative volume of the subsurface from the Umbrella Point field, which was selected as a source of data for mod-
elling a less well-known analogue in brine-bearing strata at site 1, a typical site for sequestration at a depth of 1850m within the Frio Formation beneath a power plant and adjacent refinery. We used regional, dip-oriented, well log crosssections (Dodge & Posey 1981) and regional depositional system patterns (Galloway 1982) to determine that the selected interval was deposited in deltaic and barrier strandplain systems. Log shape and lateral variability of sands in cross-section from the nearby and stratigraphically similar Umbrella Point field (Vining 1997) provided information on facies and scale of sand-rich depositional features (barrier core, washover, splay, and channel) and muddy depositional features (bayfill offshore, and transgressive shales). Numerous past detailed studies of Frio sandstones in the Texas Gulf Coast (Galloway 1982; Galloway & Cheng 1985; Knox & Barton 1999) provide the background needed to interpret depositional features from limited site-specific information. Stochastic realizations using transition probability statistics (Carle & Fogg 1996) were then created from each idealized facies map (methods described by Hovorka et al 2001). Permeability for each facies (Table 1) was estimated using reservoir characteristics in similar facies in adjacent reservoirs (McConnell 1962; Pace 1962; Quinn 1962; Steer 1962; Surber 1962). Figure 5 shows the distribution of facies created for the model.
IMPACT OF GEOLOGICAL HETEROGENEITY
153
Table 1. Typical reservoir characteristics for each fades. Facies
Shale Splay 1 Channel Splay 2 Barrier core Washover
Porosity (%)
Horizontal permeability (mD)
Vertical permeability (mD)
10 28 30 30 32
0.001 150 400 250 700
0.0001 30 100 100 700
29
200
50
Structural heterogeneity -folds and faults Deformation of the bedded sedimentary rocks is ubiquitous in the Frio Formation in the Gulf Coast. Faulting, dominantly synsedimenatry growth faults and faulting associated with salt and mud diaperism, has offset the sandstones and compartmentalized the flow system. Folding is associated with faulting as well as uplift at diapers. Most strata between faults exhibit significant dip, and reversal of dip to form structural closure is common (Fig. 4). The scale of faulting and folding is variable, from small features with amplitudes of a few metres to large regional features with amplitudes of hundreds of metres. Large features can be mapped from well penetrations (e.g. Geomap 1981) and interpretation of seismic data sets; small features must be inferred from outcrop analogues. Our modelling approach is to break the complex structures into elements and model the elements in order to identify the significant parameters that control flow. We can then reduce model complexity whilst maintaining the significant parameters to scale-up to model CO2 flow path at basin scale. Our example for initial analysis of the impact of folding and faulting on sequestration is from site 2, our proposed pilot site on the south side of a salt dome (Fig. 6). Here the Frio Formation is strongly compartmentalized by a typical pattern of highangle radial faults and cross faults. The structure and fault boundaries used for modelling are based on upward extrapolation of structure and fault patterns mapped from 3D seismic on a producing interval about 1000m below the brine-bearing upper Frio Formation. This structural interpretation has a 440m-wide compartment with fault boundaries on the NW, NE and SE. A fault boundary in the SW side of the compartment was not imaged within the seismic volume, so the closure on this side is unknown and is used as a variable in the modelling experiment. Within the compartment, strata are tilted off the salt dome. At the injection well the top of the Frio Formation strikes N70°W and dips 15° toward the SW. Stratigraphy employed for the flow
Fig. 5. Distribution of facies for the TOUGH2 model.
modelling focuses on the selected injection interval, a 6m thick high porosity, high permeability sandstone of probable barrier island core facies overlying 9m of heterogeneous fluvial-deltaic sands with shale stringers. Locally extensive shales deposited within the Frio during cycle-bounding flooding events form the sealed boundaries at the top and bottom. The thick regionally extensive shale of the Anahuac Formation overlies the Frio Formation and is represented as an impermeable boundary in our current model.
Modelling Using the data outlined, models were constructed for two cases (Table 2). The first case, used for model runs 1,2 and 3, is designed to focus on the impacts of stratigraphic heterogeneity on capacity at a randomly selected power plant and refinery east of Houston, Texas (site 1). We modelled a 1km2 area with a 100m-thick model volume distributed into 10 layers (Fig. 5). Strata with stochastically derived permeability distribution are simplified to flat lying; the top and bottom boundaries are closed, representing unbroken regionally extensive major low permeability shale units. The lateral boundaries are open, and are used to conceptualize when CO2 leaves the lease, the only meaningful 'spill point' for this flatlying system. The CO2 was injected through a well completed either in the lower 50m or completed through the entire interval. The permeability is varied through a number of cases, using the stochastically generated permeability distributions. The second case, used for model runs 4 to 6 was constructed to investigate the impacts of dipping beds and faults on sequestration capacity and effectiveness, as well as to predict what concentrations might reach a monitor well. The porosity and
154
S.D.HOVORKA£7AL. 1984). The pressure gradient, averaged from surrounding fields, is 0.1 barm"1. In case 1 (flat lying), the initial temperature is 78 °C and pressure is 188 bars. In case 2, the tilted fault block is at slightly shallower depths, the initial pressure is about 150bars and the temperature is 64 °C at the injection well. Injection rate was varied in different numerical experiments.
Sequestration capacity
Fig. 6. Structure on top injection interval at site 2 in a brine-bearing compartment on the south flank of a salt dome. Interpretation is based on analysis of a 3D seismic volume and well-logs showing how geological complexity was simplifed to create the numerical model.
permeability distribution was generated stochastically from the Frio facies sequence interpreted from regional sandstone trends and local well log interpretation. The 6m thick injection sandstone is discretized into five layers and assigned a permeability distribution representative of a barrier-core dominated geological setting. The two units beneath the injection horizon, interpreted as interfingering sand and shale deposited in a fluvial setting are modelled as two grid-layers each. The upper and lower boundaries are closed, simulating the 5m-thick laterally extensive marine flooding shales that are expected to provide good vertical seals with subregional-scale lateral extent. The lateral boundaries to the NW, NE and SE, are modelled as closed to flow, creating a 440m wide compartment. The model extends far beyond the pilot site to the SW to allow for the possibility that no sealing fault exists nearby, but a provision is included to place a low-permeability vertical barrier in the model at various locations to investigate the impact of different hypothetical sealing fault locations. Lateral grid block spacing is finer around the injection and observation wells. The layers dip 15° with an average strike of N70W, which is simulated by rotating each layer 15° about the x- axis. Other inputs into the model are derived by interpolation of regional data to the two case studies. Reasonable values for Frio water chemistry at these depths are TDS lOOOOOppm, Na 35000ppm and Cl 45000ppm (Macpherson 1992). Both injection intervals are nonproductive of hydrocarbons. At depths above the top of geopressure, the regional geothermal gradient is 0.0326 °Grr' (Loucks et al
Three modelling experiments from a series derived from the first case are presented here to document relationships between stratification and capacity. The injection history modelled is a 20-year injection period during which 682000 (metric tonnes) of CO2 were injected per year, followed by a 60-year recovery period. Results are given as figures showing the spatial distributions of supercritical CO2 in an immiscible gas-like phase (S ) and CO2 dissolved in the aqueous phase (5/X/c°2p//pg) at three times: after one year of injection; at the end of the 20-year injection period; and after 40 years of subsequent recovery, during which no injection occurs. The capacity factor is also shown as a function of time for each case. The total capacity factor C represents the volumetric fraction of the subsurface containing CO2. C is divided into Cg, the fraction containing gas-phase CO2 and C/? the fraction containing CO2 dissolved in the aqueous phase (C=C + Ct). The injection well is completed in the lower naif the model in order to promote interaction of CO2 with the entire volume, except for case 3, which is modelled as a full penetration. Run 1: The model of uniform sandstone (Fig. 7) has high porosity of 0.30 and uniform high horizontal permeability of 400mD. The distribution of CO2 is dominated by buoyancy, reflecting the lower density of CO2 relative to brine. Most of the movement of CO2 is straight up from the injection point, then along the sealing shale at the top of the model. CO2 exits the run two years after the beginning of injection, at which time the flow within the model region becomes quasi-steady. The capacity declines slightly from a maximum of 0.04 during injection, and drops rapidly as CO2 escapes from the volume during the resting period. Concentrations of dissolved CO2 (Cj) indicate the regions that gas-phase CO2 has contacted, and increases gradually through time. Run 2: This heterogeneous model (Fig. 8) examines a complex flow system generated by assembling a series of plausible realizations in a facies-stacking pattern typical of the model area. Locally continuous maximum flooding shales between the layers are suppressed, so that the sand-rich layers are connected within the model. Discontinuous shale layers
Table 2. Parameters used for modelling Run
Parameter investigated
Variable
Injection rate
Model volume
Porosity
Horizontal permeability (mD)
Vertical permeability (mD)
Boundary conditions
Depth at injection well (m)
Initial Temperature pressure at at injection injection well (°C) well (bars)
1
Strati graphic heterogeneity
Uniform rock properties
50 ts/day
IkmXlkmX 100m
0.30
400
100
Top and bottom sealed, all sides open
1860
188
78
2
Stratigraphic heterogeneity
Heterogeneous interconnected sands and shales
50ts/day
IkmX IkmX 100m
Complex, created via geostatistical manipulation from Table 1
Complex, created via geostatistical manipulation from Table 1
Complex, created via geostatistical manipulation from Table 1
Top and bottom sealed, all sides open
1860
188
78
3
Stratigraphic heterogeneity
Heterogeneous sands separated by shale seals
50 ts/day
IkmXlkmX 100m
Complex, created via geostatistical manipulation from Table 1
Complex, created via geostatistical manipulation from Table 1
Complex, created via geostatistical manipulation from Table 1
Top and bottom sealed, all sides open
1860
188
78
4
Structure (dip and faults)
Sealed compartment
50 ts/day
440mX440mX 15m
0.30
400
100
Top and bottom sealed, all sides sealed
1524
150
148
5
Structure (dip and faults)
Open compartment
50 ts/day
440m X 440m X 15m
0.30
400
100
Top and bottom sealed, 3 sides sealed
1524
150
148
6
Structure (dip Open and faults) compartment, increased injection rate
150 ts/day
440m X 440m X 15m
0.30
400
100
Top and bottom sealed, 3 sides sealed
1524
150
148
156
S.D.HOVORKACTAL.
Fig. 7. Run 1: Uniform sandstone showing Sg (supercritical CO2 in an immiscible gas-like phase) and ,S/X/CC)2p/pg (CO2 dissolved in the aqueous phase). Injection lasts for 20 years followed by 40 years of recovery during which no injection occurs. The capacity factor is also shown as a function of time for each case. The total capacity factor C (volumetric fraction of the subsurface containing CO2) is divided into C , the fraction containing gas-phase CO2 and Cr the fraction containing CO2 dissolved in the aqueous phase.
interrupt vertical movement of CO2, and significant amounts of the CO2 are retained in the lower part of the model. Capacity is nearly 0.07, larger than in the uniform model, and increases slightly throughout the injection period. More CO2 is retained within the model during the recovery phase because increased complexity of flow paths causes a larger percentage of porosity to be accessed for storing CO2. However, more rapid spilling from the model volume in one year reflects the more direct flow paths to the model edge. Run 3: The common occurrence of stacked hydrocarbon reservoirs in the Frio demonstrates that maximum flooding shales can vertically isolate sandrich cycles on the reservoir scale. This is investigated in run 3 (Fig. 9) by removing the high and moderate permeability facies from shale-rich beds. Because the sands are strongly isolated, we distribute the injection over the entire model thickness. Injection volume is proportional to bed thickness. Capacity of
0.068 is similar to run 2 and also increases slightly during injection, but declines even more slowly than run 2 during the resting period. CO2 reaches the edge of the model spill point in two years. An additional model run identical to run 2, but with a full penetration injection, shows a decreased capacity of 0.06, because the lower sand-rich intervals are less completely utilized. In run 4 (Fig. 10) we investigate the impact of dip and structural compartmentalization on sequestration. The dip is steep (15°) and the fault-isolated compartment small (440m across) and is assumed to be completely sealed. Injection of 501 per day is simulated for 100 days. The simulated asymmetry of the plume shows a strong influence of gravity on the CO2 flowpaths. Interaction between gravity and facies-controlled heterogeneity in permeability is apparent in the irregularity of the plume. Pressure in the thin sand in the small closed compartment rises rapidly (Fig. 11).
IMPACT OF GEOLOGICAL HETEROGENEITY
157
Fig. 8. Run 2: Heterogeneous sandstone distribution created by stacking facies realization.
In run 5 (Fig. 12) we assess the impact of the compartment being hydrologically open. As a first approximation, the SW margin of the model is moved out to 9000m, and the injection rate maintained at 501 per day. Pressure does not increase as much in the extended model as in the closed model, so CO2 density is slightly lower, resulting in a somewhat bigger plume. Otherwise the effect of the SW boundary on CO2 plume development is minor. In fact, none of the lateral boundaries have much of an effect on the CO2 plume for this low injection rate and relatively short injection period, so buoyancy flow updip is the main influence on CO2 plume shape. In run 6, we triple the injection rate and volume injected from run 5-150t per day, for 100 days. In contrast with the lower injection rate, the lateral closed boundaries updip of the injection well counteract buoyancy flow to some extent, creating a CO2 plume that is more symmetrically spread around the injection well (Fig. 13). Additional simulation shows that continued injection for a year increases the pressure from 152 to 178bars (Fig. 11) and fills most of the compartment. Increased pressure allows CO2 to permeate imperfect seals and begin to
fill other compartments. Compartmentalization by faulting therefore appears to increase capacity, but at the expense of increased pressure.
Sequestration effectiveness CO2, as a buoyant, low viscosity, reactive phase, will not remain static in the subsurface over geological time periods. The residence time of the CO2 in the subsurface must be long enough to be effective in decreasing atmospheric concentrations. In order for CO2 disposal in brine formations to be effective in greenhouse gas abatement, the mechanisms and flows that change the distribution and concentration of the immiscible CO2 in the subsurface must be assessed. It is essential to assure the public and regulators that in implementing geological sequestration movement of CO2 in the subsurface will not have catastrophic or other negative impacts on the environment. A review of current information shows that geological heterogeneities have a strong impact on residence time and have provided the basis for future simulations.
158
S.D.HOVORKA£7AL.
Fig. 9. Run 3: Heterogeneous sandstone distribution created by stacking facies realizations, and adding regionally extensive shales to isolate the permeable strata within the model volume.
Several mechanisms that increase residence time have been documented by previous work and by the modelling presented in the preceding section of this paper. CO2 dissolved in brine remains behind as the immiscible plume moves under the influence of gravity, and may have had a long residence time within horizons below potable water because of slow flow typical of brines in the deeper part of basins (Bachu et al. 1994). Dissolved CO2 in the presence of favourable mineralogy will induce mineral precipitation, resulting in essentially permanent sequestration (Perkins & Gunter 1996). Comparison of the Ct in runs 1, 2, and 3 shows that the same stratigraphic heterogeneity that favours capacity increases dissolution of CO2 into water. Domes or fault traps where geometric relationships between permeable strata and low permeability seals form structural traps provide long residence times for accumulations of hydrocarbons in the subsurface. Analogy with hydrocarbon traps suggests that flux from these structures is low over geological periods of time. However, sequestration in closed structures is not the ideal solution to assuring effec-
tive sequestration of large volumes of CO2 for several reasons. The spill point of the structure limits the volume that can be trapped. A tall column of buoyant fluid within a large structure may create pressures at the high point on the structure that exceed the lowest capillary entry pressure of the seal causing CO2 to leak (Capuano 1993). Injection directly into the structure under pressures approaching lithostatic would also favour seal failure. For this reason, 85% of lithostatic pressure is used by the regulatory agency, the Texas Railroad Commission, as an upper limit to injection zone pressures. Likewise, analogy with the dynamics of sealing faults suggests that build-up of significant columns of CO2 under increased pressure might cause leakage along the faults until pressures decrease (Cartwright 1994; Finkbeiner et al 2001). Reliance on closed structures to assure adequately effective sequestration would be difficult to demonstrate during characterization, especially for those structures that have not demonstrated seal competence by trapping hydrocarbons. Structural closure along a fault or dome would
IMPACT OF GEOLOGICAL HETEROGENEITY
159
Fig. 10. Distribution of gas for run 4, a 440m2 fault-isolated compartment. In this run, the compartment is isolated on all four sides and CO2 is injected at 501 per day for 100 days.
effectively sequester a large volume of CO2 in a strongly layered situation such as run 3. The common occurrence in structures of hydrocarbon accumulation in multiple horizons within sandstoneshale cycles demonstrates the effective confinement of this setting. The cumulative storage in a structure that propagated through all of the layers would be large. The concept of using multiple storage layers and barriers would reduce some of the risk of relying on the seal on top of a single large storage reservoir, because structures in a series of nearly sealed beds could be quite effective in retarding flow through the system. In this concept, the CO2 would fill one small structure to its spill or failure point before leaking out of the structure or migrating vertically to fill the next overlying interval. Short CO2 columns resulting from dividing the CO2 injected among a number of traps would limit mechanisms for leakage. Stratigraphic traps or a combination of stratigraphic-structural traps retain hydrocarbons for geological periods of time, and analogous Stratigraphic trapping of CO2 could be significant in creating
Fig. 11. Comparison of pressure build-up versus time in the fault -bounded compartment. The variables are the rate of injection and the geometry of the compartment. The closed scenarios involve a sealed 440 m2 compartment; in the extended scenarios the model extends much further than shown, to 9000 m on the SW side.
160
S.D.HOVORKACTAL.
Fig. 12. Distribution of gas for run 5, a 440m faulted compartment into which CO2 is injected at 50t per day for 100 days. The compartment is open on the SW (bottom) side. The plume is slightly larger than in run 4 because pressures are lower; however, the plume distribution responds primarily to the interaction between gravity in the dipping system and heterogeneity in the facies.
effective sequestration. This mechanism has not been specifically modelled, but confinement of CO2 in poorly connected strata is apparent in run 2 (Fig. 8). Addition of dip along regional flowpaths would result in significant volumes of CO2 being trapped in numerous small to large stratigraphic traps. It is common for stratigraphic traps to be stacked, for example along tectonic features, and the same benefits that result from stacking of structural traps would result. Long-term retention of CO2 as an immiscible phase within the pore system has not been considered in the previous calculations. Processes similar to the retention of residual saturation of methane (Holtz 2002) would be expected to leave behind a significant amount of CO2. As pores are emptied of gas under gravity force or in response to a pressure differential, the displacement of gas by water is incomplete. When pore throats are filled with water, capillary pressure and relative permeability effects
stop the flow of gas and allow only water to move. Residual gas is left behind the water front. Concentration of residual gas varies widely in response to petrophysical properties of the rock. Rate and displacement process are also important. Experience during production of methane demonstrates that 10-70% of the initial saturation is typically left behind as residual saturation (Holtz 2002). Geological heterogeneity also may have significant negative consequences for sequestration effectiveness. Heterogeneous systems are difficult to characterize and to model, and predictions of transport and fate of CO2 are more expensive and less precise than they are in simple systems. Faults have a potential for providing a rapid and highly connected route for transport of CO2 or displaced brine toward potable water and the surface, and this risk is difficult to assess. An extensive literature exists detailing methods that are used to assess the reliability of fault seals (for example Yielding et al 1997); however
IMPACT OF GEOLOGICAL HETEROGENEITY
161
Fig. 13. Distribution of gas for run 6, a 440m faulted compartment into which CO2 is injected at an increased rate of 150t per day for 100 days. In this run, the compartment is open on the SW (bottom) side.
this remains an area of uncertainty. In addition, faults raise the additional risk of induced seismicity which has occurred during fluid injection in nonGulf Coast settings (Healy et al 1968; Nicholson & Wesson 1990). Additional data collection, modelling, and field experiments are needed to determine how to assess accurately the effectiveness of sequestration in heterogeneous settings, and the risk introduced by faults.
Conclusions The sequestration capacity of a rock volume is increased by the presence of multiple low permeability layers. In both the case of discontinuous low permeability layers and reservoir-scale continuous low permeability layers, modelling shows that volumes are more efficiently occupied than homogeneous permeable sand. This stratigraphic layering is the dominant deposition pattern in sedimentary
rocks that are the most viable high injectivity targets for large-scale sequestration. If a rock volume is defined as homogeneous, the flow path of CO2 is dominated by buoyancy, the CO2 moves upward from the injection point, and only a small volume of the injection horizon is used for transport. If the CO2 encounters a single low permeability layer on top of a high permeability injection horizon, it spreads out as a relatively thin sheet that does not fill most of the rock volume. If injection of CO2 is modelled within a realistic sedimentary rock volume where permeable rock is interbedded with multiple low permeability layers, the average effective porosity is decreased but the sequestration capacity is increased because baffling of flow results in a longer flow path and interaction of the CO2 with a larger fraction of the rock volume. Geological complexities such as variable permeability, dip, folds, fractures, and faults exist to a greater or lesser extent in all subsurface environments. Dip decreases capacity, because the volumes
162
S.D. HOVORKA £7AL.
higher than the injection point are used more than those down dip from the injection point. Compartmentalization by faulting appears to increase sequestration capacity by favouring filling of all pore space, at the cost of increased pressure and increased risk of leakage. Sequestration effectiveness involves absence of unacceptable short and long-term risks to humans and the environment and a long subsurface residence time for the CO2. Slow flux to the atmosphere without negative impact is expected because the CO2 is buoyant. The risk of upward flux must be determined first through characterization of flowpaths and then through modelling. Geological heterogeneity is seen to have a positive impact on residence time, because the same settings that trap hydrocarbons (fault traps, domal traps, and stratigraphic traps) are likely to trap CO9 for geologically long periods of time. A series of small laterally adjacent traps with spill points or vertically stacked intervals of somewhat leaky interbedded seal strata are likely to lead to the lowest risk of release, long residence times, and successful characterization because of the redundancy of multiple seals and the resulting long flow path. However, analogy with hydrocarbon reservoirs indicates that the increased pressure and CO2 column thickness resulting from increased capacity may result in seal failure, decreased CO2 residence time in the injection horizon, and increased risk of unacceptably high leakage rates. Additional data collection, modelling, and field experiments are needed to determine how to assess accurately the effectiveness of sequestration in heterogeneous settings, and the risk introduced by faults. This work was supported, in part, by NETL and by the Office of Basic Energy Sciences of the US Department of Energy under contracts No. DE-AC03-76SF000098 and DE-AC26-98FT40417.
References ALTUNIN, V. V. 1975. Thermophysical properties of carbon dioxide. Moscow, Publishing House of Standards (in Russian). BACHU, S., GUNTER, W. D. & PERKINS, E. H. 1994. Aquifer disposal of CO2. Hydrodynamic and mineral trapping. Energy Conversion and Management, 35, 269-279. BATTISTELLI, A., GALORE, C. & PRUESS, K. 1997. The simulator TOUGH2/EWASG for modeling geothermal reservoirs with brines and noncondensable gas. Geothermics, 26,437-464. BERGMAN, P. D. & WINTER, E. M. 1995. Disposal of carbon dioxide in aquifers in the US. Energy Conversion and Management, 36, 523-526. BUCKLEY, S. E. & LEVERETT, M. C. 1942. Mechnism of fluid displacement in sands. Transactions of the American Institute of Mining and Metallurgical Engineers, 146, 107-116.
CAPUANO, R. M. 1993. Evidence of fluid flow in microfractures in geopressured shales. American Association of Petroleum Geologists Bulletin, 77,1303-1314. CARLE, S. F. & FOGG, G. E. 1996. Transition probability based on indicator geostatistics. Mathematical Geology, 28,4,453-477. CARTWRIGHT, J. A. 1994. Episodic basin-wide fluid expulsion from geopressured shale sequences in the North Sea basin. Geology, 22,447-450. DODGE, M. M. & POSEY, J. S. 1981. Structural cross sections, Tertiary formations, Texas Gulf Coast. The University of Texas at Austin, Bureau of Economic Geology. DOUGHTY, C. A., PRUESS, K., BENSON, S. M., HOVORKA, S. D. & GREEN, C. T. 2001. Capacity investigation of brine-bearing sands of the Frio Formation for geologic sequestration of CO2. In: Proceedings of the 1st National Conference on Carbon Sequestration, May 14-17, Washington, DC, sponsored by National Energy Technology Laboratory, on CD. FlNKBElNER, T, ZOBACK, M., FLEMINGS, P. & STUMP, B.
2001. Stress, pore pressure, and dynamically constrained hydrocarbon columns in the South Eugene Island 330 field, northern Gulf of Mexico. American Association of Petroleum Geologists Bulletin, 85, 1007-1031. GALLOWAY, W. E. 1982. Depositional architecture of Cenozoic gulf coastal plain fluvial systems. The University of Texas at Austin, Bureau of Economic Geology Geological Circular 82-5. GALLOWAY, W. E. & CHENG, E. S. 1985. Reservoir fades architecture in a microtidal barrier system; Frio Formation, Texas Gulf Coast. The University of Texas at Austin, Bureau of Economic Geology Report of Investigations 144. GALLOWAY, W. H., HOBDAY, D. K. & MAGARA, K. 1982. Frio Formation of the Texas Gulf Coast Basin—depositional systems, structural framework, hydrocarbon origin, migration, distribution, and exploration potential. The University of Texas at Austin, Bureau of Economic Geology Report of Investigations 122. GEOMAP, 1981. Geomap of the Gulf Coast, Inc. 1981. Regional Map No. 12, Upper Texas Gulf Coast, Horizon A (scale 1 inch = 4000 ft), Dallas, Texas. HEALY, J. H., RUBEY, W. W, GRIGGS, D. T. & RALEIGH, C. B. 1968. The Denver earthquakes. Science, 161, 1301-1310. HOLTZ, M. H. 2002. Residual gas saturation it aquifer influx: a calculation method for 3D computer reservoir model construction. Society of Petroleum Engineers, SPE 75502. HOLLOWAY, S. 1996. An overview of the Joule II Project 'The underground disposal of carbon dioxide'. Energy Conversion and Management, 37,1149-1154. HOVORKA, S. D., ROMERO, M. L., WARNE, A. G., AMBROSE, W. A., TREMBLEY, T. A., TREVINO, R. H. & SASSON, D. 2000. Sequestration of greenhouse gases in brine formations. Word Wide Web Address: http://www/beg. utexas.edu/co2/. HOVORKA, S. D., DOUGHTY, C., KNOX, P. R., GREEN, C. T, PRUESS, K. & BENSON, S. M. 2001. Evaluation of brine-bearing sands of the Frio Formation, Upper Texas Gulf Coast for geologic sequestration of CO2, In: Proceedings of the 1st National Conference on
IMPACT OF GEOLOGICAL HETEROGENEITY Carbon Sequestration, May 14-17, Washington, DC, sponsored by National Energy Technology Laboratory, on CD. KNOX, P. R. & BARTON, M. D. 1999. Predicting interwell heterogeneity in fluvial-deltaic reservoirs: effects of progressive architecture variation through a depositional cycle from outcrop and subsurface observations. In: SCHATZINGER, R. & JORDAN, J. F. (eds) Recent advances in reservoir characterization. American Association of Petroleum Geologists, Memoir 71, 57-72. KREITLER, C. W, AKHTER, M. S., DONNELLY, A. & SOFER, Z. 1988. Hydrologic/ hydrochemical characterization of Frio Formation used for deep-well injection of chemical wastes, Texas Gulf Coast. Eos, Transactions of the American Geophysical Union, 69, 348. LOUCKS, R. G., DODGE, M. M. & GALLOWAY, W. E. 1984. Regional controls on diagenesis and reservoir quality in lower Tertiary sandstones along the lower Texas Gulf Coast. In: MCDONALD, D. A. & SURDAM, R. C. (eds) Clastic diagenesis. American Association of Petroleum Geologists, Memoir 37,15^6. MACPHERSON, G. L. 1992. Regional variation in formation water chemistry; major and minor elements, Frio Formation fluids, Texas. American Association of Petroleum Geologists Bulletin, 76,740-757. MCCONNELL, R. R. 1962. Umbrella Point field, Chambers County. In: DENHAM, R. L. (ed.) Typical oil and gas fields of Southeast Texas, 1. Houston Geological Society 225-228. MORIDIS, G. & PRUESS, K. 1998. T2SOLV: an enhanced package of solvers for the TOUGH2 family of reservoir simulation codes. Geothermics, 27,415-444. NARASIMHAN, T. N. & WITHERSPOON, P. A. 1976. An integrated finite difference method for analyzing fluid flow in porous media. Water Resources Research, 12, 57-64. NICHOLSON, C. & WESSON, R. L. 1990. Earthquake hazard associated with deep well injection; a report to the U.S. Environmental Protection Agency. US Geological Survey Bulletin 1951,74. PACE, R. C. 1962. Cedar Point field; Galveston County. In: BURNS, G. K. (ed.) Typical oil and gas fields of Southeast Texas, 2. Houston Geological Society, 75-79. PERKINS, E. H. & GUNTER, W. D. 1996. Mineral traps for carbon dioxide. In: HITCHON, B. (ed.) Aquifer disposal of carbon dioxide: hydrodynamic and mineral trapping- proof of concept. Geoscience Publishing Ltd,
163
Alberta, Canada. PRUESS, K. & GARCIA, J. 2002. Multiphase flow dynamics during CO2 injection into saline aquifers. Environmental Geology, 42,282-295. PRUESS, K., OLDENBURG, C. & MORIDIS, G. 1999. TOUGH2 user's guide, version 2.0. Berkeley, CA, Lawrence Berkeley National Laboratory Report LBNL-43134, November. PRUESS, K., Xu, T., APPS, J. & GARCIA, J. 2001. Numerical modeling of aquifer disposal of CO2. Society of Petroleum Engineers, SPE/EPA.DOE Exploration and Production Conference, San Antonio, TX, 26-28 February 2001. QUINN, M. J. 1962. Goose Creek field; Harris County, Texas. In: BURNS, G. K. (ed.) Typical oil and gas fields of Southeast Texas, 2. Houston Geological Society, 242-247. SASS, B., GUPTA, N., SMINCHAK, J. & BERGMAN, P. 1998. Geochemical modeling to assess the capacity of a Midwestern United States geologic formation for CO2 sequestration. In: REIMER, P., ELIASSON, B. & WOKAUN, A. (eds) Proceedings of the Fourth International Conference on Greenhouse Gas Control Technologies. Interlaken, Switzerland, GHGT-4, 1079-1086. STEER, R. K. 1962.Trinity Bay field, Chambers County. In: DENHAM, R. L. (ed.) Typical oil and gas fields of Southeast Texas, L Houston Geological Society, 221-224. SURBER, N. C. 1962. South Cotton Lake field, Chambers County. In: DENHAM, R. L. (ed.) Typical oil and gas fields of Southeast Texas, L Houston Geological Society, 198-199. US Geological Survey 2001. Geode, United States Gulf Coast region. World Wide Web Address: http://geode.usgs.gov/. VAN DER MEER, L. G. H. 1995. The CO2 storage efficiency of aquifers. Energy Conversion Management, 36, 513-581. VINING, M. R. 1997. Reserve growth in a mixed sequence of deltaic and barrier-island Frio sandstones; Umbrella Point Field, Chambers County, Texas. In: CRAIG, W. W. & KOHL, B. (eds) Transactions - Gulf Coast Association of Geological Societies, 47,611-619. YIELDING, G., FREEMAN, B. & NEEDHAM, D. T. 1997. Quantitative fault seal prediction. American Association of Petroleum Geologists Bulletin, 81, 897-917.
This page intentionally left blank
Reservoir geology of the Utsira Formation at the first industrial-scale underground CO2 storage site (Sleipner area, North Sea) PETER ZWEIGEL1, ROB ARTS2, ANE E. LOTHE1 & ERIK B. G. LINDEBERG1 1
SINTEF Petroleum Research, S.P. Andersens vei 15b, NO-7052 Trondheim, Norway (e-mail: [email protected]) 2 TNO-NITG, PO Box 80015, 3508 TA Utrecht, The Netherlands Abstract: At the Sleipner fields in the North Sea, CO2 is being injected into sands of the Miocene-Pliocene Utsira Formation, which is overlain by thick Pliocene shales. The highly porous (35%^K)%) and extremely permeable (approximately 2 D) Utsira sands are organized into approximately 30m thick packages. These packages are separated by thin (predominantly 1 m thick), lowpermeability shale layers, which are assumed to contain potential fluid pathways of erosive or deformational origin. A 6.5m thick shale layer close to the top of the sands separates an eastward thickening sand wedge from the main sand package below. Migration simulations indicate that the migration pattern of CO2 below the shale layer would differ strongly from that within the sand wedge above. Time-lapse seismic data acquired prior to the start, and after three years, of injection confirmed a reservoir model based on these findings and showed that the thin shale layers act as temporary barriers and that the 6.5 m thick shale layer does not fully inhibit upward migration of CO2.
Long-term underground storage of CO2 generated by point sources has been suggested as a means of reducing the emission of CO2 into the atmosphere (e.g. Holloway 1996; Herzog et al 2000). Injection of CO2 into the Miocene-Pliocene Utsira Formation at the Sleipner fields in the North Sea (Fig. 1) is the first attempt to store large quantities of CO2 underground with the sole objective of preventing it reaching the atmosphere (Baklid et al 1996). Injection started in 1996 at an annual rate of approximately 0.85 Mt. The planned injection rate is 1 Mt per year during the life of the Sleipner Vest field, which is estimated at about 20 years. Under reservoir conditions, the CO2 is expected to have a lower density than water. Consequently, it should rise to the top of the formation and accumulate beneath the overlying shale package, which is several hundred metres thick. The CO2 injection activities are accompanied by the multinational Saline Aquifer CO2 Storage (SACS) research project which aims to monitor the CO2 in the subsurface and to predict its long-term fate. Since the Utsira Formation is not a hydrocarbon reservoir unit, it had not been well studied prior to its emergence as a candidate for underground CO2 storage. Initial reservoir models used during the planning stage of the injection project were accordingly relatively schematic. For example the definition of the reservoir top geometry was based on a network of 2D seismic lines and was therefore approximate. Similarly, the reservoir was taken to be internally homogeneous and assumed not to contain internal barriers that might inhibit or affect CO2 migration (Baklid et al 1996). Here, we present results of a detailed study of the
reservoir formation based on 3D seismic, wire-line log, and rock sample data. The reservoir model which resulted from this study formed the basis for reservoir simulations (Lindeberg et al 2000; Zweigel et al 2000), for feasibility studies on geophysical monitoring techniques, and for the interpretation of seismic time-lapse data (Arts et al 2000, 2004; Chadwick et al 2004a). The latter data, including a seismic survey over the storage site shot after 3 years of injection, confirmed several aspects of the new reservoir model presented here. Analysis of the seal formation has not yet been completed and its results will therefore be documented separately.
Geological framework The Utsira Formation has been defined by Degan & Scull (1977) as the first thick sand unit below the Pliocene to Recent deposits in the central parts of the northern North Sea. Further definitions and subdivisions were made by Isaksen & Tonstad (1989) and Gregersen et al (1997). Recently, the age of the Utsira Formation has been determined biostratigraphically in a well a few kilometres to the south of the CO2 injection site (Eidvin et al 1999) to range from late middle Miocene (c. 11 Ma) to earliest late Pliocene (c. 3 Ma). Regional mapping of the Utsira Formation revealed that it occupies a north-south elongated area in the centre of the northern North Sea with two distinct depocentres (Gregersen et al 1997; Chadwick et al 2000,2004£). The Sleipner area is at the centre of the southern depocentre, in which the Utsira Formation reaches up to 300m in thickness (Figs 2
From: BAINES, S. J. & WORDEN, R. H. (eds) 2004. Geological Storage of Carbon Dioxide. Geological Society, London, Special Publications, 233,165-180.0305-8719/047$ 15.00 © The Geological Society of London 2004.
166
P.ZWEIGELCTAZ,
Fig. 1. Location of the Sleipner area in the northern North Sea. Positions of the 3D survey ST98M11 and of semiregional seismic and well cross sections are indicated. Data from the wells shown were used in reservoir interpretation.
Fig. 2. Regional profiles based on interpreted seismic lines. East-west profile based on line CNST82-06, north-south profile based on line CNST82-18B. The area of 3D seismic survey ST98M11 is indicated. Location of profiles is given in Fig. 1.
SLEIPNER C07 INJECTION: UTSIRA RESERVOIR
& 3). The lithostratigraphic definition of the Utsira Formation in its type well 16/1-1 (Degan & Scull 1977; Isaksen & Tonstad 1989) includes a shalecontaining package at its top. This study concerns mainly the lower package comprising almost exclusively sands which form a hydrological aquifer unit. This lower package is termed the Utsira Sand. The depth to its top ranges regionally from approximately 550-1500m (Chadwick et al 2000). The Utsira Sand is underlain by predominantly shaly sediments of the Hordaland Formation which exhibit severe deformation by soft sediment mobilization (e.g. Heggland 1997) and polygonal faulting (similar to that described by Cartwright & Lonergan 1996 in other North Sea areas). Presence of mud diapirs and mud volcanoes at the base of the Utsira Formation results in significant local thickness variations. The direct overburden of the Utsira Sand consists of clay-rich sediments of the Nordland Group which are approximately 250m thick in the Sleipner area (Fig. 2). These 'Nordland Shales' are the distal parts of sediment wedges that prograded from the western and eastern basin margins. They are overlain by several hundreds of metres of coarser-grained Quaternary sediments. The Utsira Sand sediments are basin-restricted marine lowstand deposits. However, their depositional environment is a matter of dispute. Interpretations range from tidal sand ridge complexes (Rundberg 1989) through geostrophic-induced contourites (Galloway et al. 1993), turbidites (Gregersen et al 1997) to linked strandplain and sandy shelf shoal deposits (Galloway 2002). Recently, Eidvin et al. (1999) interpreted the deposition to have taken place under middle to outer neritic conditions, in a water depth of c. 30-200 m. Recent biostratigraphic data of Eidvin et al. (1999) suggest sedimentation rates for the Utsira Sand that are lower than those of the overlying shales and considerably lower than sedimentation rates of shaly deposits in the neighbouring Central North Sea which are correlative to the Utsira Sand (Fig. 4). Depositional models that include intensive sediment reworking in a broad channel linking the Central North Sea with the North Atlantic and contemporaneous deposition of finer grained or mixed material in the Central North Sea therefore seem favourable.
Geology of the Utsira Sand in the Sleipner area Wire-line log and seismic characteristics The sand-dominated package of the Utsira Formation, i.e. the Utsira Sand, is identified on wire-line
167
Fig. 3. Two-way travel time in ms to (a) the base and (b) the top of the Utsira Formation in the Sleipner area and its surroundings. Based on interpretations of a network of regional 2D seismic lines, linked to the 3D seismic survey ST98M11 (outline indicated by rectangle) and calibrated by wells shown in Fig. 1.
logs (Fig. 5) from its relatively low gamma-ray, sonic velocity, and neutron density values. Sharp downward decreases and increases at the unit top and base, respectively, of sonic velocity and density constitute large impedance contrasts and cause strong, clearly identifiable reflections (Fig. 6) that can be traced with confidence. The base of the Utsira Sand has a northeastwardtrending depression in the greater Sleipner area (Fig. 3a). The 3D survey ST98M11 (Fig. 7a) is situated largely in the depression axis but includes part of the northwestern flank. Overprinted on this general trend are elongate, circular- and ring-shaped positive
168
RZWEIGELCTAL.
topographic features of typically 1-2 km in width, up to more than 10km in length, and approximately 100m in height. These are mud edifices (mud diapirs and mud volcanoes) caused by localized mobilization of the underlying Hordaland Shale (Fig. 6a). In areas away from such mud edifices, eastward toplap of reflections against the base of the Utsira Sand has been observed; this indicates some erosion prior to Utsira Sand deposition and signifies that the base of the Utsira Sand constitutes an unconformity. The base of the Utsira Sand in the Sleipner area occurs at depths of 900-1100m. The top of the Utsira Sand has a general southward dip on semi-regional (Fig. 3b) and local (survey ST98M11, Fig. 7b) scale. The northeastwardtrending depression at base Utsira level (Fig. 3a) is not visible at top Utsira level (Fig. 3b). This implies the presence of a depocentre that was filled during Utsira Sand deposition. On the 3D seismic data, local variations of the general trend are evident at top Utsira level, but these are broader and of smaller amplitude than at the base Utsira level. Local depressions at top Utsira level are situated above mounds at base Utsira level (Fig. 6). The depressions can be attributed to preferential compaction of the mud edifices during burial. Connected depressions enclose complex sets of domal structures that serve as structural traps for injected CO2. The top of the Utsira Sand in the Sleipner area occurs at depths of 750-900 m. On seismic traces, the interior of the Utsira Sand is characterized by relatively weak, but generally continuous reflections. They are usually parallel to
Fig. 4. Sedimentation rates in the Miocene to Recent of (a) the Sleipner area and (b) the Ekofisk area, based on age determinations of Eidvin et al. (1999). Note that in the Ekofisk area the unit corresponding to the Utsira Sand consists of shales and has a hiatus at its base.
Fig. 5. Well cross section across the injection site area (location see Fig. 1). Gamma-ray and resistivity/induction logs clearly show the Utsira Sand, which is characterized by generally low gamma-ray and low resistivity values. An eastward thickening sand wedge in the uppermost part of the Utsira Sand is separated from the main sand package by t 6.5 m thick shale layer. Note shale peaks within the Utsira Sand.
SLEIPNER CO, INJECTION: UTSIRA RESERVOIR
169
Fig. 6. Seismic sections from 3D seismic cube ST98M11; locations see Fig. 7. Top and base Utsira Sand are characterized by strong reflections, (a) Typical appearance of features linked to mud edifices at the base of the Utsira Sand. Mud edifice is a pop-up structure, limited by reverse faults. Horizons in lower part of Utsira Sand downlap on top of edifice, but this may be rotated due to pronounced compaction of mud edifice. Compaction is evidenced by depression above the edifice. Occurrence of laterally restricted reflections above the edifices indicate existence of a depression during deposition of the upper part of the Utsira Sand, (b) Seismic section through the injection site. Domal trap shape results passively from pronounced subsidence at the margins of the structure, where strong compaction of mud edifices at the base of the Utsira Sand caused depression of the overlying layers (see also Fig. 7). Note that intraUtsira reflections which possibly indicate shale layers, have domal shape similar to top Utsira Sand, (c) The sand wedge at the top of the Utisra Sand in seismic. In the west, the single contact shale over sand causes a strong, single amplitude trough (black reflection). In the east, the three contacts shale-sand, sand-shale and again shale-sand produce a reflection set characterized by two strong amplitude troughs. Note that the lower trough correlates to the trough in the west, indicating continuity of the shale layer at the base of the wedge. Weak reflections in the central area are due to thin-bed tuning effects.
Fig. 7. Depth maps in total vertical depth below sea level, based on 3D seismic survey ST98M11 and depth converted (see Appendix), (a) Depth of the base of the Utsira Sand. Note positive topography features, interpreted to be mud volcanoes and mud diapirs. (b) Depth of the top of the main sand package of the Utsira Sand, which illustrates trap shape for at least parts of injected CO2. Depressions at this level are above 'hills' at the base Utsira level. General shallowing direction is northwestward, implying potential long term migration in that direction, (c) Depth of the top of the main sand package of the Utsira Sand in the close surrounding of the CO2 injection site. The domal trap above the injection site is linked to other traps by saddles towards north, west and south. Depth to these saddles differs by approximately 1 m.
SLEIPNER CO2 INJECTION: UTSIRA RESERVOIR
171
Fig. 8. Photomicrographs of Utsira Sand samples from the core in well 15/9-A23. (a) and (b) from 1080.04m MD, (c) and (d) from 1080.70mMD. (a) and (c) with parallel and (b) and (d) with crossed Nichols. Arrow in (c) points to scarce broken mineral. Note scarcity of grain contacts and of fractures in grains. Quartz grains are only slightly undulous with crossed Nichols.
the base and the top but converge above mud edifices (Fig. 6a). The lower reflections show downlap or onlap onto the mud edifices. Above some mud mounds, additional reflections with limited areal extent are present, with onlap onto underlying reflections at their margins. These restricted reflections document the presence of local depressions during Utsira Sand deposition at sites which previously constituted topographic highs. There is no evidence for faults in the interior of the Utsira Sand, with the exception of the margins of some of the mud edifices (Fig. 6a). In rare cases, faults from the polygonal fault pattern of the underlying Hordaland Formation penetrate into the lowermost part of the Utsira Sand. However, at the most they reach up to the first reflection above the base Utsira Sand reflection.
Petrography Rock samples from the Utsira sand are almost exclusively ditch cuttings. This study also had access to
three 1 m long segments of a 9m long core from the inclined well 15/9-A23 which covers the depth range from approximately 905-910m below mean sea level. The core samples consist of loose, unconsolidated sand and disintegrate when shaken in unfrozen conditions. Macroscopic inspection did not reveal any layering or other sedimentary structures. Shell fragments of several millimetres in length are common and do not show any preferred orientation. Thin sections reveal a homogeneous, fine to medium grained, moderately to well sorted sand (Fig. 8). The grains are predominantly angular to sub-angular; large grains are often rounded. The quartz grain content is very high, and feldspar (predominantly K-feldspar, S. J. Kemp, pers comm. 1999) is second in abundance (Table 1). Only minor amounts of cement (quartz or chlorite) were observed. Microscopy reveals that in two-dimensional thin sections grain contacts are scarce and that grains are usually not deformed (Fig. 8), attesting to the unconsolidated state of the sediment.
P.ZWEIGEL£TAL.
172
Table 1. Modal analysis results of Utsira Sand samples. Well Sample depth (mMD)
15/9-A23 1080.04
15/9-A23 1080.80
15/9-A23 1084.19
15/9-16 850-860
15/9-16 890-900
15/9-16 950-960
15/9-16 15/9-15 1000-1010 880-890
Detrital grains Quartz Feldspar Muscovite Biotite
72.3 8.5 0.5 0.5
71.4 14.1 0.0 0.0
76.1 4.9 0.5 0.0
50.7 7.3 0.0 0.0
66.7 3.7 1.3 0.0
58.7 4.0 0.3 0.0
76.7 2.7 0.0 0.0
65.3 2.0 0.7 0.0
Calcite (shell fragments) Clay pellets Glauconite Organic matter Matrix Heavy minerals
11.2 3.7 0.0 0.0 0.5 0.0
7.3 2.1 0.0 0.0 1.0 1.6
5.4 3.3 0.0 0.5 0.0 0.0
18.0 1.3 1.3 1.0 0.0 0.0
17.0 1.0 1.0 0.3 0.0 0.0
8.0 0.0 0.7 0.0 0.0 0.0
7.7 0.3 0.7 0.0 0.0 0.0
4.7 0.3 0.7 0.0 0.0 0.0
Detrital rock fragments Chert Quartzite Gneiss Sandstone Shale/schist
0.0 0.0 0.5 0.0 0.0
0.0 0.0 0.0 0.0 0.0
0.0 0.5 0.0 0.0 0.0
0.0 0.3 0.3 0.0 4.3
0.0 0.7 1.0 0.0 1.0
1.3 2.3 0.0 0.3 16.3
0.3 3.7 0.7 0.3 4.7
0.0 0.7 0.0 0.3 23.3
1.1 0.0 0.0 0.0 0.5 0.0 0.0 0.5 4.8 100.0
0.0 0.5 0.0 0.0 0.5 0.5 0.5 0.0 2.6 99.5
1.6 1.1 5.4 2.7 0.5 0.0 0.0 0.0 14.1 102.7
0.0 1.3 0.0 0.0 0.0 0.0 14.0 0.0 4.0 100.0
1.0 0.3 0.0 0.0 0.0 0.7 4.3 0.0 3.3 100.0
0.0 0.0 0.0 0.0 0.0 0.0 8.0 0.0 0.7 100.0
0.7 0.0 0.0 0.0 0.0 0.3 1.3 0.0 1.7 100.0
0.3 0.0 0.0 0.0 0.0 0.0 1.7 0.0 1.3 100.0
Diagenetic (cements etc) Illite cement Illite pseudomorph Chlorite rim Chlorite cement Calcite cement Quartz cement Pyrite Siderite Sum clays Sum components
Samples from well 15/9-A23 are from the Utsira Sand core. The other samples are from ditch cuttings. All numbers in Vol. % of the solid portion of the sample. Porosity was counted separately.
Intra-Utsira shale layers Wire-line logs from the Utsira Sand show primarily low gamma-ray, velocity, and density values (indicative for sands) with thin spikes of higher gamma-ray, velocity, and density values in between (Fig. 5). The log values of these spikes are similar to those of the overlying Nordland shale, and they are accordingly interpreted as shale layers. An alternative interpretation would be that the gamma-ray spikes are due to the presence of glauconite (having been reported from the Utsira Formation, e.g. Isaksen & Tonstad 1989; Rundberg & Smalley 1989), but we would not then expect commensurately increased velocity and density at these depths. Most shale layers are relatively thin, with an average thickness of 1-1.5m, but a single 6.5 m thick shale layer exists close to the top of the Utsira Sand (Fig. 5). This 6.5 m thick shale layer separates an eastward thickening sand body from the underlying main part of the Utsira Sand. Detailed interpretation of 3D and
2D seismic data revealed that the shale package in the east corresponds to the lowermost part of the Nordland Shale in the west, where there is no separate sand body at the top of the Utsira sand (Fig. 6c). The reflection at the base of the shale layer is fairly continuous throughout the area of the 3D seismic survey ST98M11 and suggests continuity of the shale layer. Accordingly, the wedge-shaped sand body was expected to constitute a hydrological entity separated from the main part of the Utsira Sand. The topography of the top of the sand wedge in the uppermost Utsira Sand (Fig. 9a) is similar to that of the top of the main, lower part of the Utsira Sand (Fig. 7b). The major disparity is a subtle difference in regional dip; due to the eastward thickening sand wedge between the two horizons, the top sand wedge dips 0.3° steeper to the west than the deeper top of the main sand package. Correlation of individual thin shale layers from well to well is possible over distances of up to
SLEIPNER CO, INJECTION: UTSIRA RESERVOIR
173
Fig. 9. (a) Topography of the top sand wedge (in m total vertical depth below sea level) in survey ST98M11. (b) Thickness map of the sand wedge (in mTVDss) showing sinuous features interpreted as channels, (c) Seismic sections through interpreted channels. Note that channel corresponds in inline 4054 to a deeper base of the sand wedge but in crossline 3045 to an elevated top. The latter may be due to less compaction of the sands as compared to the adjacent shales. Location of seismic sections indicated in Fig. 9b.
approximately 1km in the case of closely spaced wells. Over distances of several kilometres, e.g. between exploration wells, unambiguous correlation was not possible. On average, the sand packages between individual thin shale layers are approximately 30m thick. In several wells of the Sleipner A platform, a thick shale package close to the base of the Utsira Sand was identified. Its maximum thickness is 46m in well 15/9-A2T2. This is interpreted as outflows of a mud edifice (mud volcano) which occurs in the immediate neighbourhood of the well. The thin intra-Utsira shale layers may constitute hydrological barriers, inhibiting or retarding vertical
flow in the reservoir. We do not expect these shale layers to be completely sealing for three reasons: (i) relatively high transport energy active during deposition of the overlying sand may have locally eroded the shale layers; (ii) lateral subsidence variation caused by higher compaction intensity of the base Utsira mud edifices than the Utsira Sand may have induced local displacements (faults) of the shales; and (iii) the shales may have been dissected by injected sand dykes, which could have been caused by geological loading and overpressuring due to reduced drainage through the shales.
174
RZWEIGELETAL.
Fig. 10. Histogram of porosity values for the Utsira Sand.
The injection site area in detail The CO2-injection site is positioned below a dome structure in the top Utsira Sand (Fig. 7c) with a diameter of about 1.2km and a height above spillpoints of approximately 12m. The top of the dome is at approximately 820m TVDss. The spillpoints are saddles or 'channels' that link the dome with other domal structures in the north, south and west. According to the depth map derived from 3D seismic data, the depth of these 'channels' varies only by approximately 1m from one to the other. This is just above 0.1 % of the depth to the top Utsira Sand, and is much below any possible accuracy for depth determination. In addition, seismic signals of the top Utsira Sand at parts of the channels are disturbed by the presence of seismic anomalies (possibly shallow gas) in the overburden. It is thus impossible to predict the actual spill point with confidence.
Reservoir properties Porosity The porosity of the Utsira Sand has been determined by modal analysis of thin sections, by liquid invasion measurements on core samples, and from density logs. Thin sections were prepared with utmost care from the Utsira Sand core in well 15/9-A23. Boxshaped samples were cut from the core in frozen condition, thawed, and slowly penetrated at one side with blue epoxy in vacuum, avoiding any movement
or shaking in unfrozen condition. Modal analysis of four samples from the core segments 1080-1081m MD (i.e. c. 906.0-906.6m below sea level) and 1084-1085m MD (i.e. c. 908.4-909.0m below sea level) yielded porosity values ranging from 36.0%^0.1% (Fig. 10) with an average of 38.0%. Liquid invasion using water and/or helium ('core analysis') yielded effective porosities ranging from 35%-42.5%. These latter values are in accordance with laboratory data from GEUS (Springer, pers. comm. 2002), based on both core samples and reconsolidated loose Utsira sand samples, measured at reservoir pressure (Fig. 10). Wire-line log density data were used to calculate porosities by the formula where <1> is porosity (in %), p rain is density of the grains (2.66gem"3 for Utsira Sand samples from Sleipner; Springer, pers. comm. 2002), pbulk is the bulk density as determined from density logs, and Pfluid *s me density of the pore fluid (1.022gem"3 in the Sleipner case; Statoil data). Analysis of data from six wells in block 15/9 with suitable logs (density log available, few cavings) yielded porosities ranging from 35%-39% with an average of37% (Fig. 10). The porosities determined by three different methods range from 35%-42.5%. However, all but one of the porosity values above 40% are from measurements at atmospheric conditions (modal analysis, core analysis at atmospheric pressure) and are therefore likely to be overestimated. In particular, modal analysis is thought to provide artificially high
SLEIPNER CO2 INJECTION: UTSIRA RESERVOIR
values because the preparation procedure (including freezing of the sample and intrusion of the unfrozen, loose sample by epoxy) might loosen the fabric and separate the grains from each other. We therefore favour porosities in the range between 35% and 40% as representative for the Utsira Sand at reservoir conditions. This porosity range may seem high for a sand buried below 800m overburden, but is supported by other observations: (1)
(2)
(3)
The sand is very friable, indicating that it is almost unconsolidated. We would expect a substantial porosity reduction to be accompanied by consolidation. Microscopic inspection of the grains shows only minor indications for deformation or shape changes of the grains during compaction (Fig. 8). The amount of compaction should accordingly be low. Compaction experiments carried out by GEUS show that loose Utsira Sand grains reduce porosity from approximately 41.5% at surface conditions to approximately 38%-39% at reservoir conditions.
Values in the range 35%-40% are, therefore, preferable for reservoir simulations aiming at a simulation of the present CO2 distribution. However, risk assessments, e.g. determinations of maximum migration distances, should include a conservative ('worst case') porosity estimate of c. 30%, which would then also include non-perfect sweep.
Permeability The permeability of Utsira Sand to water was measured in four 1.5 inch cores of different length. Two vertical cores and two horizontal cores were cut from two different brine-saturated, frozen 4 inch X 1 m core samples. The cores were kept frozen during mounting in the core holders. Water permeability was measured with a confining pressure corresponding to the lithostatic pressure in the Utsira Sand. The measurements of the short cores have a much larger uncertainty than the longer core and this was taken into account when the average and standard deviation was calculated (Table 2). The average permeability from core analysis is approximately 2D. A significant difference in vertical and horizontal permeability cannot be verified from the results. The scatter in the results represents the physical variation between the samples more than uncertainty in the measurements. Additional permeability measurements by GEUS (Springer, pers. comm. 2002) are in accordance with the values presented here. Well-test data do not exist for the Utsira Sand in the Sleipner area. For the Grane area (90km from
175
Table 2. Measured permeability from horizontal and vertical Utsira core samples. Core
Length (m)
Permeability (mD)
Weight factor
Vertical Vertical Horizontal Horizontal Weighted average and standard deviation:
1.15 0.10 0.10 0.10
1845 3252 2550 1621 1975±418
0.7930 0.0690 0.0690 0.0690
Sleipner) well tests yielded 5.8D, and for the Oseberg area (250km from Sleipner) they yielded permeabilities ranging from 1.1-8.14 D (Norsk Hydro data). The permeability values derived from core samples appear low when compared to the very high porosity. A possible explanation may be the grain size distribution of the Utsira Sand (Fig. 11). The lack of sufficient quantities of intermediate size grains may cause the pore spaces between large grains to remain open, resulting in high porosity, whereas the fine grain fraction tends to block pore throats, reducing permeability.
Net/gross ratio Net/gross ratio (N/G) is defined as the ratio between the thickness of potential reservoir rocks and the thickness of the whole reservoir unit. The N/G for the main part of the Utsira Sand beneath the base of the 6.5 m thick shale layer in the Sleipner area is in the order of 0.90 to 0.97 (data from five wells with suitable logs in block 15/9). For calculations of CO2 storage capacity and migration distances a conservative ('worst case') N/G of 0.85 was used to include residual water saturation and non-perfect sweep efficiency (see next chapter).
Lateral reservoir heterogeneity Thickness maps of the sand wedge in the upper part of the Utsira Sand reveal narrow, elongate, sinuous features of increased thickness compared to their surrounding (Fig. 9b). The detailed cross-section geometry of the sinuous features is not resolved on seismic sections in survey ST98M11. Their shape in map view and the fact that they are thicker than the surrounding (Fig. 9c) lead to their interpretation as channels. These observed channels illustrate that the sand wedge probably possesses a lateral heterogeneity which is not directly evident from wireline-log data.
176
RZWEIGELCTAL.
Fig. 12. Temperature, fluid pressure, and CO2 density as functions of depth for the Sleipner case. Fig. 11. Grain-size spectrum of an Utsira Sand sample from the core in well 15/9-A23, derived by laser light scattering particle size distribution measurements (apparatus: Malvern Master Size S).
For example, the channel sands might exhibit higher (or lower) permeability than their surroundings, therefore providing preferred pathways (or baffles, respectively) for fluid migration. This heterogeneity could reduce sweep efficiency, especially when a strong viscosity contrast exists between the two pore-filling fluids (as is the case for CO2 and water), which will cause viscous fingering. Seismic analysis of the Utsira Sand below the thick shale layer did not reveal any structures that might imply lateral heterogeneity. This lack of indicators is, however, largely due to the lack of suitable intra-Utsira reflections that could be traced over longer distances with any degree of confidence. Since the main part of the Utsira Sand and the sand wedge at its top are lithologically similar, it is likely that the Utsira Sand contains similar lateral heterogeneity, as is evident for the sand wedge. Reduced sweep efficiency in both reservoir units was tentatively included in CO2 storage capacity and migration distance calculations using the net/gross ratio of 0.85, which is lower than observed values (see above). For these calculations, permeability and porosity values determined for the main part of the Utsira Sand were also used for the sand wedge because log properties are similar for both units.
Reservoir pressure and temperature, CO2 density One single temperature measurement for the shallow subsurface in the Sleipner area exists, giving 37 °C at 1038m below sea level (Statoil data). For a water depth of 80m and assuming 4.8 °C for the sea floor (from regional experience), this yields a subsurface temperature gradient of 33°Ckm~ 1 (Fig.
12). Temperatures in the Utsira formation in the Sleipner area are therefore expected to range from 28 °C at the reservoir top to 41°C at the reservoir base. There are no indications for overpressure in the Utsira Formation in the Sleipner area. Hydrostatic pressure (Fig. 12) was accordingly used for fluid pressure and was calculated based on tabulated pressure- and temperature-dependent density data for brine containg 3.5% salt (salinity data from Statoil). The density of CO2 depends on temperature, pressure and the amount of impurities. Relevant impurities are in the Sleipner case mainly methane and butenes, toluenes, and xylenes (BTX). Methane tends to reduce the density of the CO2-mixture, whereas BTX increases it. These effects cancel each other out in the Sleipner case. Therefore, the thermodynamic properties of pure CO2 can be used for density calculations which yield approximately 700kgm~3 at reservoir conditions (Fig. 12). Details about the calculation principle and the thermodynamic database are available from the author E. Lindeberg.
Discussion The present CO2 reservoir model in summary The detailed reservoir analysis presented above resulted in an updated reservoir model, which is illustrated in Fig. 13. The important elements of this model are: The main part of the Utsira Sand consists of highly permeable, high porosity sand packages of approximately 30m thickness each. CO2 is expected to rise vertically upwards within each of the packages, driven by buoyancy and facilitated by low capillary resistance. Thin shale layers of approximately 1 m thickness each separate the sand packages. The shale layers are expected to provide barriers to upward flow resulting in localized, laterally
SLEIPNER CO9 INJECTION: UTSIRA RESERVOIR
177
Fig. 13. Scheme of the reservoir and of the predicted migration behaviour of CO2 at the injection site.
extensive CO2 accumulations in the upper parts of the sand packages. Where erosion or faulting caused local absence of shales, CO2 will escape from the accumulation and will migrate vertically upwards until it reaches the next barrier under which it will again accumulate. A thick shale layer of approximately 6.5m thickness close to the top of the Utsira Sand separates the main part of the Utsira Sand hydrologically from an overlying, eastward thickening sand wedge. A dome-shaped trap exists above the CO2 injection site with a maximum height above spillpoint of approximately 12m both below the thick shale layer and below the top of the Utsira Sand. The intra Utsira shale layers seem to have a varying geometry from roughly parallel to the top Utsira Sand horizon in the upper parts to more or less flat (horizontal) in the lower parts of the Utsira Sand. CO2 accumulations within the main part of the Utsira Sand and (if at all) in the sand wedge are therefore expected to be roughly circular in map view.
Predictions based on the reservoir model Reservoir simulations (Lindeberg et al 2000) based on a simplified version of this reservoir model predicted that CO2 should reach the base of the thick shale layer close to the top of the Utsira Sand approximately three years after injection start. In contrast, initial simulations based on a homogeneous Utsira Sand without internal shale layers predicted arrival of CO2 at the top after approximately 2 months. The new model predicted the presence of
intra-Utsira CO2 accumulations below shale layers. These accumulations are transient features. After injection ends, they would shrink, much of the CO2 would migrate upwards through holes or fractures and accumulate below the final migration barrier, either the top Utsira Sand or the top of the sand wedge, depending on the sealing capacity of the thick shale layer. The detailed reservoir model presented here has positive effects on storage safety compared to the original, homogeneous reservoir model: (a) upward migration would be delayed, and (b) both the intraUtsira accumulations and the distributed migration paths through 'holes' in the shales to the next barrier would provide a larger contact surface to the formation water, which would enhance solution of CO2 into formation water. Dissolved CO2 will no longer rise, in fact, the density of CO2-containing formation water is slightly higher than normal formation water.
Confirmation by the 1999 time-lapse survey A 3D seismic time-lapse survey over the injection site was acquired in October 1999, i.e. approximately 3 years after injection start and after injection of some 2.28 Mt of CO2. Analysis of this survey (Arts et al 2000, 2004; Chadwick et al 20040) revealed the presence of several intra-Utsira CO2 accumulations. The accumulations have roughly elliptical shape with long axes (trending NE) ranging from 190-1900m (average 725m) and short axes ranging from 150-750m (average 400m). A small accumulation exists at the base of the thick shale layer close to the top of the Utsira sand and two even smaller accumulations were observed in the
178
P. ZWEIGEL £7AL.
Fig. 14. Predicted migration and trap filling of CO2 injected into the Utsira Sand, from Zweigel et al. 2000. (a) Migration exclusively below the thick shale layer close to the top of the Utsira Sand assumed, (b) Migration within sand wedge only assumed.
sand wedge. The small size of these accumulations indicates that the top of the reservoir had been reached by CO2 just prior to the acquisition of the 1999 seismic survey. In essence, the time-lapse survey confirmed the main predictions of the reservoir simulations, and consequently supports the geological reservoir model. It also revealed that the 6.5m thick shale layer at the base of the sand wedge was not totally sealing.
Predicted long-term migration paths A simplified reservoir model, neglecting the intraUtsira shales, was used as input for first-order simulations of long-term migration pathways (Zweigel et al. 2000). Assuming migration only within the main part of the Utsira Sand below the 6.5m thick shale layer, it was predicted that migration would primarily occur towards the west or NW (Fig. 14a). The maximum distance in the case of a total injected quantity of 20Mt CO2 would be 12km, i.e. the CO2 would stay within the Norwegian sector of the North Sea and would not be expected to reach the western margin of the Utsira Sand where potential pathways to shallower levels might exist. Assuming migration into and within the sand wedge, simulations produced a significantly different result. In the various realizations of this case migration was predicted to occur in a northeastward direction (Fig. 14b). The eastern margin of the existing 3D seismic survey provided a limit for the possible simulations. Within the survey area, 2.8-5.2MI of CO2 could be stored in structural traps. The indications from time-lapse seismic data that CO2 has
migrated into the sand wedge, require the wedge to be mapped in the area east of the studied seismic 3D survey, an activity that has been carried out recently in the SACS project.
Storage capacity The total pore volume in structural traps at the top of the Utsira Sand in the area of survey ST98M11 is 0.135km3. However, only approximately half of this volume would be accessible in practice, i.e. with a small number of injection wells. Migration simulation showed that 0.05km3 could be stored in traps at the top Utsira Sand along the migration path from at the present injection site until it would leave the survey area at its northern margin. Compared to the total pore volume of the Utsira Sand in the area of survey ST98M11 which is approximately 44.5km3 (all numbers conservatively assuming 30% porosity and 0.85 N/G ratio), these potential storage volumes correspond to a storage efficiency of 0.12%-0.15%, i.e. only 0.12%-0.15% of the available pore space would be used if CO2 is stored only in structural traps. These numbers are consistent with the value (0.12%) used by Holloway et al (1996) for calculation of the total CO2 storage potential for storage in structural traps of the Utsira Sand in the North Sea. We thank our colleagues in the SACS project group for valuable discussions and the SACS funders for financial support. SACS was funded by BP, ExxonMobil, Norsk Hydro, Statoil, Total, Vattenfall AB, the European Commission and national programmes such as the KLIMATEK program of the Norwegian Research Council. Statoil is thanked for providing samples and seismic and
SLEIPNER CCL INJECTION: UTSIRA RESERVOIR
179
Central North Sea - an assessment of its potential for regional CO2disposal. In: WILLIAMS, D. J., DURIE, B., MCMULLAN, P., PAULSON, C. & SMITH, A. (eds) Greenhouse Gas Control Technologies, CSIRO Publishing, Collingwood, Australia, 349-354. CHADWICK, R. A., ARTS, R., EIKEN, O, KIRBY, G. A., LINDEBERG, E. & ZWEIGEL, P. 2004<2. 4D seismic imaging of an injected CO2 plume at the Seipner Field, central North Sea. In: DAVIES, R., CARTWRIGHT, Appendix: J., STEWART, S., LAPPIN, M. & UNDERBILL, J. (eds) 3D Seismic Technology: Application to the Exploration of Time-depth conversion of interpreted seismic horiSedimentary Basins. Geological Society, London, zons was carried out employing a layer cake velocity Memoirs, 29, 311-320. model, using the layers and properties as listed in CHADWICK, R. A., HOLLOWAY, S., BROOK, M. S. & KIRBY, Table Al. Acoustic velocities for each layer were G. A. 2004b. The case for underground CO2 sequestration in Northern Europe. In: BAINES, S., GALE, J. & derived by minimization of the square-sum of WORDEN, R. (eds) Geological Storage of Carbon misfits between calculated and measured horizon Dioxide. Geological Society, London, Special depths at nine wells in the survey area. Publications, 233,17-28. DEEGAN, C. E. & SCULL, B. J. 1977. A standard lithostratiTable Al. Layers, horizons delimiting them and acoustic graphic nomenclature for the Central and Northern velocities used during seismic time-depth conversion. North Sea, Report 77/25, 1. Institute of Geological Science, London, Oljedirektoratet, Stavanger. EIDVIN, T, Rus, F. & RUNDBERG, Y. 1999. Upper Cainozoic Layer Layer top Layer base Velocity stratigraphy in the central North Sea (Ekofisk and [ms"1] Sleipner fields). Norsk Geologisk Tidsskrift, 79, 97-128. Water Sea level Sea floor 1659* GALLOWAY, W. E. 2002. Paleogeographic setting and depoQuaternary Sea floor Top Pliocene 1785 sitional architecture of a sand-dominated shelf depoUpper Pliocene Top Pliocene Intra Pliocene 2208 sitional system, Miocene Utsira Formation, North Sea Lower Pliocene Intra Pliocene TopUtsira/ Basin. Journal of Sedimentary Research, 72, Top sand wedge 2077 476^90. UtsiraSand TopUtsira Base Utsira 2056 GALLOWAY, W. E., CAREER, J. L., Liu, X. & SLOAN, B. J. 1993. Sequence stratigraphy and depositional frame* Water velocity accounts for choice of reflection parallel work of the Cenozoic fill, Central and Northern North to sea floor reflection. Sea Basin. In: PARKER, J. R. (ed.) Petroleum Geology of Northwest Europe: Proceedings of the 4th Conference, 33-43, Geological Society, London. References GREGERSEN, U., MICHELSEN, O. & SORENSEN, J. C. 1997. Stratigraphy and facies distribution of the Utsira Formation and the Pliocene sequences in the northern ARTS, R., BREVIK, I., EIKEN, O., SOLLIE, R., CAUSSE, E. & North Sea. Marine and Petroleum Geology, 14, VAN DER MEER, B. 2000. Geophysical methods for 893-914. monitoring marine aquifer CO2 storage - Sleipner HEGGLAND, R. 1997. Detection of gas migration from a experiences. In: WILLIAMS, D. J., DURIE, B., deep source by use of exploration 3D seismic data. MCMULLAN, P., PAULSON, C. & SMITH, A. (eds) Marine Geology, 137,41—41. Greenhouse Gas Control Technologies, CSIRO HERZOG, H., ELIASSON, B. &KAARSTAD, O. 2000. Publishing, Collingwood, Australia, 366-371. Capturing greenhouse gases. Scientific American, ARTS, R., EIKEN, O., CHADWICK, A., ZWEIGEL, P., VAN DER February 2000,54-61. MEER, B. & KIRBY, G. 2004. Seismic monitoring at the Sleipner underground CO2 storage site (North HOLLOWAY, S. (ed.) 1996. The underground disposal of carbon dioxide. Final Report of JOULE II Project No. Sea). In: BAINES, S. & WORDEN, R. (eds) Geological CT92-0031. British Geological Survey. 1-355. Storage of Carbon Dioxide. Geological Society, ISAKSEN, D. & TONSTAD, K. 1989. A revised Cretaceous London, Special Publications, 233,181-191. and Tertiary lithostratigraphic nomenclature for the BAKLID, A., KORB0L, R. & OWREN, G. 1996. Sleipner Vest Norwegian North Sea. Norwegian Petroleum CO2 disposal, CO2 injection into a shallow underDirectorate Bulletin, 5, Oljedirektoratet, Stavanger. ground aquifer. 1996 Society of Petroleum Engineers LINDEBERG, E., ZWEIGEL, P., BERGMO, P., GHADERI, A. & Annual Technical Conference and Exhibition, Denver, Colorado, USA, SPE paper 36600,1-9. LOTHE, A. 2000. Prediction of CO2 dispersal pattern CARTWRIGHT, J. A. & LONERGAN, L. 1996. Volumetric conimproved by geology and reservoir simulation and verified by time lapse seismic. In: WILLIAMS, D. J., traction during the compaction of mudrocks: a mechaDURIE, B., MCMULLAN, P., PAULSON, C. & SMITH, A. nism for the development of regional-scale polygonal fault systems. Basin Research, 8,183-193. (eds) Greenhouse Gas Control Technologies, CSIRO Publishing, Collingwood, Australia, 372-377. CHADWICK, R. A., HOLLOWAY, S., KIRBY, G. A., GREGERSEN, RUNDBERG, Y. 1989. Tertiary sedimentary history and basin U. & JOHANNESSEN, P. N. 2000. The Utsira Sand, wireline data. Reidar B0e carried out the modal thin section analysis. GEUS and BGS are thanked for their consent to present their porosity and grain density data. Editor Richard Worden and reviewer Paul Nadeau are thanked for very helpful comments improving clarity of presentation.
180
P.ZWEIGELCTAL.
evolution of Norwegian North Sea between 60°N-62°N, an integrated approach. Dr. ing. thesis, Universitetet i Trondheim, Norges Tekniske H0gskole, Institutt for Geologi og Bergteknikk, Norway, 292 pp. RUNDBERG, Y. & SMALLEY, P. C. 1989. High resolution dating of Cenozoic sediments from northern North Sea using 87Sr/86Sr Stratigraphy. American Association of Petroleum Geologists Bulletin, 73,298-308.
ZWEIGEL, P., HAMBORG, M., ARTS, R., LOTHE A. & T0MMERAs, A. 2000. Prediction of migration of CO2 injected into an underground depository: Reservoir geology and migration modelling in the Sleipner case (North Sea). In: WILLIAMS, D. J., DURIE, B., MCMULLAN, P., PAULSON, C. & SMITH, A. (eds) Greenhouse Gas Control Technologies, CSIRO Publishing, Collingwood, Australia pp. 360-365.
Seismic monitoring at the Sleipner underground CO2 storage site (North Sea) ROB ARTS1, OLA EIKEN2, ANDY CHADWICK3, PETER ZWEIGEL4, BERT VAN DER MEER1 & GARY KIRBY3 1
TNO-NITG, PO Box 80015, 3508 TA Utrecht, The Netherlands (e-mail: [email protected]) 2 Statoil Research Centre, Rotvoll, N-7005 Trondheim, Norway 3 BGS, Kingsley Dunham Centre, Keyworth, Nottingham, NG12 5GG, UK 4 SINTEFPetroleum Research, S.R Andersens vei 15b, NO-7052 Trondheim, Norway Abstract: The growing emissions of greenhouse gases, especially CO2, are seen worldwide as one of the major causes of climate change. International treaties like the Kyoto Protocol are supposed to contribute to reducing the emission of greenhouse gases. Underground sequestration has the potential to play an important role in keeping large volumes of CO2 from escaping into the atmosphere in the short term. The first case of industrial scale CO2 storage in the world (close to one million tonnes per year since 1996) is taking place at the Sleipner underground CO2 storage site in the North Sea offshore Norway. Careful monitoring of the behaviour of the storage facility is required to establish its safety. To this end, two time-lapse seismic surveys have been acquired; the first repeat survey was completed in October 1999 and the second in October 2001. The presence of CO2 beneath thin intra-shale layers within the reservoir has caused significant changes both in reflection amplitudes (up to a factor 10) and in travel time (more than 40ms) through the CO2 plume (the velocity pushdown effect). Some aspects of the interpretation of these time-lapse seismic surveys will be presented here.
Since October 1996, Statoil and its Sleipner partners have injected CO2 into a saline aquifer, the Utsira Sand, at a depth of 1012m below sea level. The CO2 is separated on the platform from natural gas produced at the Sleipner field in the central North Sea (Norwegian block 15/9) and injected into the aquifer through a horizontal well at a lateral distance of about 2.3 km from the platform (Fig. 1). A multi-institutional research project SACS (Saline Aquifer CO2 Storage) was formed to predict and monitor the migration of the injected CO2. To this end two time-lapse seismic surveys over the injection area have been acquired so far, one in October 1999 after 2.35 Mt CO2 had been injected and the second in October 2001 after 4.26Mt CO2 had been injected. Comparison with the baseline seismic survey of 1994 prior to injection provides detailed insights in the subsurface distribution and migration of the CO2. The effects of the injected CO2 on the seismic data are significant (Eiken et al. 2000; Brevik et al 2000), the presence of CO2 inducing a pronounced drop in the compressional wave velocity even for moderate CO2 saturations, leading to a clear change in seismic response. This is expressed as marked changes, both in reflection amplitudes and in travel time through the CO2 plume (the velocity pushdown effect). Some aspects of the ongoing interpretation of the two time-lapse seismic surveys are presented here,
together with a brief description of the reservoir geology; a more detailed one can be found in Zweigel^a/. (2004). Geology of the Utsira Sand in the Sleipner area Reservoir geology of the injection site area The Utsira Sand forms part of the Mio-Pliocene Utsira formation (Gregersen et al 1997, Chadwick et al. 2000), and is overlain by a thick, dominantly shaly, overburden. Four key reflectors based on well information have been interpreted on the 1994 baseline survey (Arts et al 2000a, b), top Pliocene prograding unit, intra-Pliocene prograding unit, top Utsira Sand and base Utsira Sand (Fig. 2). The Utsira Sand has a thickness of more than 200m near the injection site but shows marked thickness variation laterally. Mounds at its base are interpreted as due to mud diapirism that was active during deposition of the lower part of the Utsira Sand. The presence of these shale mounds induced differential compaction, which led to depressions forming in the Utsira Sand and overlying units above the mud volcanoes (Figs 2 & 3). These depressions constitute local modifications of the general southward dip of the top Utsira Sand, and include local
From: BAINES, S. J. & WORDEN, R. H. (eds) 2004. Geological Storage of Carbon Dioxide. Geological Society, London, Special Publications, 233,181-191. 0305-8719/047$ 15.00 © The Geological Society of London 2004.
182
R.ARTSETAL
Fig. 1. Schematic overview of the CO2 injection scheme at Sleipner (Arts et al. 2000a).
domal and anticlinal structures, which act as potential traps and/or channels for CO2 flow. Within the Utsira Sand, several thin shale layers with an average thickness of 1-1.5m have been identified from well log data. Although they are not resolved on the baseline (pre-injection) seismic data, it is possible to establish a wireline correlation between the 'shale peaks' in closely-spaced wells over distances up to 1 km (Arts et al. 20000; Zweigel et al 2004). About 20m below the top of the reservoir a thicker shale unit of 5-7 m thick is present. In previous publications the sand unit above this thicker shale layer has been referred to informally as the 'sand wedge'. However because the upper sand is more correctly considered as part of the main Utsira reservoir, the term 'sand wedge' will no longer be used. The Pliocene shales of the caprock can be subdivided into two units separated by the intra-Pliocene horizon. The lower unit, directly overlying the Utsira Sand, includes at its base a shale drape that can be distinguished on a regional scale. Locally, this lower unit exhibits anomalously high amplitudes. The upper Pliocene prograding unit is characterized by irregular internal reflectors and some very high amplitudes which might be due to the presence of shallow gas. A more detailed discussion of these anomalies can be found in Arts et al. (2000/?).
Injection of the CO2 CO2 is injected near the base of the Utsira Sand at depths of 1010-1013 m below sea level, within a 38m long well perforation interval. The main mechanism driving flow of the CO2 is gravity, the CO2 rising buoyantly until it reaches an intra-reservoir shale (Lindeberg et al. 2000). These thin shale layers form temporary flow-barriers for the CO2, but are not expected to be fully tight (Zweigel et al. 2004). Reservoir simulations (van der Meer et al 2000) indicate that most of the CO2 remains trapped in relatively thin, high saturation accumulations beneath the shale layers and follows their topography. An example of a reservoir simulation at the time of the 1999 time-lapse seismic survey is shown in Figure 4.
Analysis of the seismic response Petrophysteal properties of the Utsira Sand reservoir The petrophysical properties of the Utsira Sand pertinent to its seismic response have been determined from well log data (Fig. 5). The Utsira Sand is a very weakly cemented sandstone with measured porosities in the range 30-42% (mean value 37%). The
SEISMIC MONITORING OF CO7 IN AN AQUIFER
183
Fig. 2. Key horizons interpreted on the 1994 baseline survey. The arrow at the bottom indicates the extent of the 1999 and the 2001 time-lapse seismic surveys (Arts et al. 2000£).
acoustic velocity in the water-saturated sandstone is 2050ms"1 with an uncertainty range of 1950-2100 ms"1. Bulk densities of water-saturated Utsira Sand have been estimated in the range of 1960-2080kgm"3. Norwegian well 15/9-A23 is the only well with shear wave velocity information through a DSI-log. The average value of the shear wave velocity is 643 ms~ ! , varying in the range 600-680ms"l.
Petrophysteal properties of the caprock and of the intra-shale layers For seismic modelling, estimations of the P- and Swave velocities and of the bulk density of the caprock and of the intra-reservoir shale layers are required. Values for the caprock have been estimated from well-log data which indicate a P-wave velocity of Vp = 2270ms"1, an S-wave velocity of Vs = 850ms"1 and a bulk density of p = 2100kgm"3. Uncertainties are estimated in the order of 4%. The well log expression (including the sonic and density) of the intra-shale 'spikes' is similar to the logresponse of the Pliocene caprock shales, so the same values have been used (Zweigel et al 2004).
Fig. 3. Images of the top Utsira Sand horizon and the base Utsira Sand horizon in TWT (ms) around the injection site.
a fluid, but the compressibility of a gas. The most likely values for the density of the CO2 are between 600-700kgm-3. The bulk modulus of the CO2 is considered likely to have a value not greater than 0.0675 GPa.
Petrophy steal properties of the injected CO2 Gassmann modelling The natural pressure and temperature at the injection point are estimated at lOMPa and 36 °C respectively. Under these conditions CO2 is in a supercritical state. In practice this means that it has the density of
Seismic velocities were modelled as a function of CO2 saturation from the Gassmann relationships (Gassmann 1951) which enable the elastic properties
184
R.ARTSErAZ,
Fig. 4. Result of a reservoir simulation at the time of the first time-lapse seismic survey in October 1999 (after 3 years of injection) after 2.28 Mt of CO2 injected (van der Meer et al 2000).
of a porous medium saturated with a fluid to be derived from the known properties of the same medium saturated with a different fluid. As a basic assumption in this analysis homogeneous mixtures of brine and CO2 with respect to the seismic wavelength are assumed. In reality this condition is likely to be fulfilled only approximately. The densities and compressibilities of the saturating fluids, the rock matrix and the porosity of the rock are assumed to be known. In the Sleipner case the 100% water-saturated Pand S-wave velocities are known from well logs. The main constituent of the rock matrix is quartz with a known density and compressibility. The sand porosity, and the densities and compressibilities of the water and of the CO2 are also known. From this information, the Gassman relationships can be used to calculate the elastic velocities (P- and S-wave) and the density of the rock, saturated with a CO2-water mixture, for a range of saturation states. Because of the uncertainty on the bulk modulus of the CO2 under reservoir conditions (P—T), velocities have been calculated for three values within extreme limits (Fig. 6). For values of bulk modulus ^T co < 0.0675 GPa velocities are fairly constant for CO2-saturations over 20%. For higher bulk moduli, the elastic properties of CO2 more closely resemble those of the replaced water and the effect on seismic velocity decreases. In general, even for small CO2 saturations, the drop in P-wave velocity is considerable (about
30%). This can be explained by the fact that only a few small 'bubbles' of CO2 have a dramatic effect on the overall compressibility of the saturated rock. Therefore a sharp decrease in the P-wave velocity can be observed for small CO2 concentrations. The influence of the CO2 on the S-wave velocities is minimal because shear waves are not sensitive to the saturating fluids. The minor variation that is observed is due to the variation in the bulk density (CO2 is less dense than water).
Pressure effect During the injection process at Sleipner, until 2001 no significant increase in well head pressure has been observed at the injection well (not exceeding the unsystematic data scatter of about 0.2MPa), the CO2 flowing easily into the very high permeability reservoir. The pressure-temperature conditions of the reservoir around the CO2 plume are such that the CO2 is expected to remain in a supercritical state. Taking this into consideration, the pressure effect on the seismic velocities is expected to be marginal.
Wavelet determination and synthetic modelling In order to perform seismic modelling, a wavelet was estimated from the seismic data. The frequency
SEISMIC MONITORING OF CO2 IN AN AQUIFER
185
Fig. 5. Crossplots of the compressional wave velocity (V ) versus the bulk density (p) and the porosity (cj>) of the Utsira Sand derived from 14 wells in the Sleipner area
Fig. 6. Elastic velocities of the Utsira Sand as a function of water-CO2-saturation using Gassmann's model. The required input parameters are reported including three extreme values for the bulk modulus of the CO2.
content of the baseline seismic data has been determined and shows a bandwidth from approximately 10-70 Hz with a central peak frequency of around 40 Hz. Assuming zero phase data, a wavelet was estimated from the spectra at various inlines to check the spatial variability of the estimated wavelet. The results show no significant variation, leading to an optimum mean wavelet (Fig. 7). Using the estimated elastic parameters for the shale layers, for the 100% water-saturated sandstone and for the 100% CO2-saturated sandstone, a simplified acoustic impedance model was created to illustrate the seismic response of the injected CO2. Figure 8 shows an example of CO2 accumulating at the top of the reservoir and beneath two illustrative intra-reservoir shale layers. A range of CO2 thicknesses was modelled to investigate the influence on the seismic signal. Two dominant effects determine the seismic response: The negative seismic impedance contrast between shale and underlying sand becomes more negative (larger in absolute value) when CO2 is present in the sand. The seismic response is a composite wavelet caused by interference from sequences of water-
saturated sand, shale, CO2-saturated sand, and water-saturated sand again. The first effect leads to stronger negative seismic amplitudes as for a classical 'bright spot'. The second effect (tuning) can lead to destructive or constructive interference depending on the thickness of the CO2 layer, evident from the seismic modelling (Fig. 8). As the thickness of the CO2 column increases from 0-8 m a gradual increase of the (negative) amplitude is observed. Maximum constructive interference corresponds to a CO2 thickness of about 8m, the so-called 'tuning thickness'
Tuning effect at the top of the reservoir At the top of the reservoir an interesting phenomenon is observed. Due to the tuning effect of CO2 trapped at the top of the Utsira Sand, the main reflection trough appears to be aligned slightly above the reservoir top for thin CO2 accumulations. Modelling shows (Fig. 8) that the tuning effect can force the peak upward by as much as 3 ms two-way time.
186
R.ARTS£TAL.
Fig. 7. Zero-phase wavelet estimated from the baseline 1994 seismic survey from different inlines (3750-3950 with steps of 50) within the time window of 500-1200 ms.
Velocity 'pushdown' effect Below the CO2 plume a prominent velocity 'pushdown' is expected due to the lower velocities through the CO2-saturated sand with respect to the water-saturated sand (Arts et al. 2002). This effect can be quantified by a so-called 'gross pushdown factor'. The 'gross pushdown factor' has units of sm"1 and expresses the amount of pushdown in seconds (or milliseconds) per metre thickness of fully CO2-saturated rock. The mathematical expression of the 'gross pushdown factor' is:
To illustrate, using the calculated Gassman velocities, if a 10m column of fully water-saturated sand were replaced by a 10m column of fully CO2-saturated sand, the resulting TWT pushdown or timedelay, would be about 4ms.
Interpretation of the time-lapse seismic data Observed seismic reflectivity As predicted from seismic modelling, introducing CO2 into the Utsira Sand has a dramatic effect on the reflectivity. Strong negative reflections are observed at nine stratigraphical levels both on the 1999 and the 2001 time-lapse surveys. Figure 9 shows an inline through the injection area for the 1994, 1999 and the 2001 surveys including the difference between the 1999-1994 data and the 2001-1994 data. The interpreted CO2 levels are shown in white as interpreted on the 1999 data and in black as interpreted on the 2001 data. The consistency between the 1999 and the 2001 CO2 levels is obvious. In general the 2001 CO2 levels have a larger lateral extent and have been 'pushed
down' slightly more with respect to the 1999 CO0 levels. The latter can be easily understood considering that more injected CO2 causes more pushdown. The two highest CO2 reflections observed correspond respectively to trapping at the top of the Utsira Sand reservoir (level 9) and trapping beneath the 5 m thick shale layer (level 8). In 1999 the CO2 had just reached the top of the reservoir. By 2001 the CO2 had migrated laterally beneath the caprock, but also appears to have penetrated slightly above the top Utsira Sand. This is likely to be an artefact, in part to the tuning effect described above and possibly also due to unresolved stratigraphical complexity at the reservoir top. No evidence has been found of CO2 migrating significantly into the caprock. The seven deeper interpreted reflections are caused by CO2 accumulated beneath the thin intrareservoir shale layers. Any reflections from these thin shales on the 1994 baseline data are too weak to identify. Only the presence of trapped CO2 illuminates them sufficiently to make interpretation possible. In general these CO2 reflections do not show the gentle antiformal geometry of the Utsira stratigraphy as imaged on the 1994 data, but rather show a downward pointing V-profile, which becomes more pronounced down through the reservoir. This is attributed to the velocity 'pushdown' effect caused by the CO2 present above the reflectors. Similarly, this explains the chaotic structure observed in the difference sections - the difference signal arises both from changes in reflectivity and also from changes in two-way travel-time due to pushdown. To derive proper difference sections the seismic datasets would have to be time-depth converted very precisely. So far this has not proved feasible in such detail due to the complicated interference patterns occurring within the 'bubble' and the highly accurate velocity model required. An additional complication to imaging the lower CO2 layers properly is the presence of sea-bottom multiple energy from the upper reflections. These multiples can be recognized by their polarity reversal with respect to CO2 reflections, but at some locations they interfere with the primary reflections making interpretation very difficult. A prominent vertical feature, characterized by localized pushdown and much decreased reflection amplitudes (Fig. 9), is interpreted as a 'chimney' of CO2. This is situated approximately above the injection point and forms an important vertical migration path which conducts CO2 almost directly to the top of the reservoir. The total velocity pushdown below the CO2 chimney seems very large (probably 50ms or more), consistent with the presence of a column of CO2. The total velocity pushdown observed below the CO2 for the 1999 survey and for the 2001 survey (Fig. 10) gives a good overview of the lateral extent of the CO2 plume present. These maps have been
SEISMIC MONITORING OF CO2 IN AN AQUIFER
187
Fig. 8. Illustrative model of injected CO2 trapped in a simplified reservoir scenario representative of the Sleipner case. The thickness of the CO2 layers increases to the right and the corresponding synthetic seismic response is shown below. At the bottom the amplitudes of the synthetic seismic signals caused by the three CO2 accumulations are plotted as a function of the CO2 layer thickness. The so-called tuning thickness corresponds to a CO2 column of 8 m. Note the increase in seismic energy above the actual top of the reservoir caused by tuning in the thickening layer of CO2.
obtained by cross-correlating the seismic signals below the CO2 plume of the time-lapse surveys with those of the baseline survey. Seismic modelling has shown that within the tuning thickness of approximately 8m the reflection amplitude of the seismic data is directly related to
the local thickness of the CO2 column. A more quantitative analysis of this relation is given in Arts et al. (2002). Here only the amplitude maps of the different accumulations are given (Fig. 11). This shows the lateral extent of the individual CO2 accumulations. The largest accumulation can be observed in
188
R.ARTSCTAL.
Fig. 9. Inline 3838 through the injection area for the 1994, 1999 and the 2001 surveys including the difference between the 1999-1994 data and the 2001-1994 data. The CO2 levels interpreted in 1999 (white) and in 2001 (black) are shown on both difference sections.
the middle of the Utsira Sand (level 5). For all maps, the seismic amplitudes decrease markedly at the edges indicating very thin CO2 layers pinching out to zero thickness. The shape of the total 'pushdown' map is very similar to the cumulative amplitude contours of the interpreted CO2 accumulations (Fig. 10). This increases our confidence that amplitude and layer thickness are related by the tuning effect and that even very thin CO2 layers of the order of 1 m thick can be detected. The ways in which the CO2 plume has changed between 1999 and 2001 are quite interesting, in that major growth of the CO2 accumulations is restricted to the upper part of the plume. Thus, small accumulations in 1999 at top Utsira Sand and level 8 had grown considerably by 2001. Also, the largest accumulation in the upper part of the reservoir (level 5)
has grown. In contrast, the deeper accumulations (levels 1 and 2) have not changed very much and appear to be in a state approaching equilibrium. One explanation for this is that the major CO2 chimney is now effectively by-passing the deeper accumulations, with the implication that it is becoming a more effective conduit with time. It is also possible to relate the amount of pushdown observed to the amount of CO2 in the plume (Arts et al 2002; Chadwick et al 20Q4a). The calculated amounts of pushdown for various models of CO2 saturation are broadly consistent with the observed pushdown. However the fact that CO2 present in low saturations produces similar pushdown to CO2 in high saturation (Fig. 6) means that modelling is intrinsically non-unique. Further work is continuing to try to decrease uncertainties in the saturation models.
SEISMIC MONITORING OF CO2 IN AN AQUIFER
189
Fig. 10. Velocity 'pushdown' in time (ms), observed below the CO2 plume in 1999 and in 2001. The 'pushdown' has been determined by cross-correlating the seismic signals below the CO2 of respectively the 1999 and the 2001 surveys with the 1994 survey. The isolines indicate the shape of the cumulative (total) seismic amplitude anomaly caused by the injected CO2.
Discussion and conclusions The most important conclusion to be taken from the seismic monitoring at the Sleipner underground CO2 storage site is probably that time-lapse seismic data enables us to map the CO2 distribution clearly and to image pathways in time. The effective detection limit for local CO2 accumulations is of the order of a metre or less. Such resolution makes us confident that any significant leakage into the Pliocene shales would have been picked up by the data. The velocity pushdown effect provides a powerful means of estimating the volume of CO2 in situ, albeit with uncertainties related to non-uniqueness. The tuning phenomenon enables effective thin-bed resolutions down to about a metre, providing a quantitative tool for volumetric estimation independent of the pushdown analysis. Effective imaging of the reservoir geology, however, particularly the thin intra-shale layers, has proven very difficult in the Sleipner case. This is due to the fact that the topography of the shales in the
Utsira Sand cannot be interpreted on the 1994 baseline seismic data. Only when they are 'illuminated' by trapped CO2 is seismic imaging of these layers possible. Therefore a perfect reservoir simulation model with a short-term predictive power that exactly honours the seismic data is difficult to obtain. The latter is not valid for the longer-term predictions (more than 10 years). As is already observed for the lower intra-reservoir shale layers, a kind of steady-state flow towards the top of the reservoir may eventually be established. Then the main lateral migration of CO2 will be governed by the topography of the top Utsira Sand, which has been mapped accurately. We thank our colleagues in the SACS project group for valuable discussions and the SACS funders for financial support. SACS is funded by BP, ExxonMobil, Norsk Hydro, Statoil, Total-Fina-Elf, Vattenfall AB, the European Commission and national programmes. Statoil is thanked for providing samples, seismic data and geophysical log data.
190
R.ARTSETAL.
Fig. 11. Amplitude maps of the individual CO2 accumulations in the 1999 seismic data (upper image) and in the 2001 data (lower image). The levels go down from the shallowest accumulation at the top of the reservoir at level 9 to the deepest at level 1.
SEISMIC MONITORING OF CO2 IN AN AQUIFER
References ARTS, R., BREVIK, I., EIKEN, O., SOLLIE, R., CAUSSE, E. & VAN DER MEER, L. 20000. Geophysical methods for monitoring marine aquifer CO2 storage - Sleipner experiences. In: WILLIAMS, D. J., DURIE, R. A., MCMULLAN, P., PAULSON, C. A. J. & SMITH, A. Y. (eds) Greenhouse Gas Control Technologies. CSIRO Publishing, Collingwood, Australia, 366-371. ARTS, R. J., ZWEIGEL, P. & LOTHE, A. E. 2000£: Reservoir geology of the Utsira Sand in the Southern Viking Graben area - a site for potential CO2 storage. 62nd European Association of Geoscientists and Engineers meeting, Glasgow 2000, paper B-20. ARTS, R., ELSAYED, R., VAN DER MEER, L., EIKEN, O., OSTMO, S., CHADWICK, A., KIRBY, G., ZWEIGEL, P. & ZINSZNER, B. 2002. Estimation of the mass of injected CO2 at Sleipner using time-lapse seismic data. 64th European Association of Geoscientists and Engineers meeting, Florence 2002, paper HO 16. BREVIK, L, EIKEN, O., ARTS, R. J., LINDEBERG, E. & CAUSSE, E. 2000. Expectations and results from seismic monitoring of CO2 injection into a marine acquifer. 62nd European Association of Geoscientists and Engineers meeting, Glasgow 2000, paper B-21. CHADWICK, R. A., HOLLOWAY, S., KIRBY, G. A., GREGERSEN, U. & JOHANNESSEN, P. N. 2000. The Utsira Sand, Central North Sea - an assessment of its potential for regional CO2 disposal. In: WILLIAMS, D.J., DURIE, R.A., MCMULLAN, P., PAULSON, C. A. J. & SMITH, A. Y. (eds). Greenhouse Gas Control Technologies. CSIRO Publishing, Collingwood, Australia, 349-354. CHADWICK, R. A., ARTS, R., EIKEN, O, KIRBY, G. A., LINDEBERG, E. & ZWEIGEL, P. (2004). 4D seismic imaging of an injected CO2 plume at the Seipner Field, central North Sea. In: DAVIES, R., CARTWRIGHT, J., STEWART, S., UNDERBILL, J. & LAPPIN, M. (eds) 3D
191
Seismic Technology: Application to the Exploration of Sedimentary Basins. Geological Society, London, Memoirs, 29, 311-320. EIKEN, O., BREVIK, I., ARTS. R., LINDEBERG, E. & FAGERVIK, K. 2000. Seismic monitoring of CO2 injected into a marine aquifer. SEG Calgary 2000 International conference and 70th Annual meeting, Calgary 2000, paper RC-8.2. GASSMANN, F. 1951. Uber die Elastizitat poroser Medien. Vier, derNatur, Gesellschaft in Zurich, 96, 1-23. GREGERSEN, U., MICHELSEN, O. & SORENSEN, J. C. 1997. Stratigraphy and facies distribution of the Utsira Formation and the Pliocene sequences in the northern North Sea. Marine and Petroleum Geology, 14, 893-914. LINDEBERG, E., ZWEIGEL, P., BERGMO, P., GHADERI, A. & LOTHE, A. 2000. Prediction of CO2 dispersal pattern improved by geology and reservoir simulation and verified by time-lapse seismic. In: WILLIAMS, D. J., DURIE, R. A., MCMULLAN, P., PAULSON, C. A. J. & SMITH, A. Y. (eds) Greenhouse Gas Control Technologies. CSIRO Publishing, Collingwood, Australia, 372-377. VAN DER MEER, L. G. H., ARTS, R. J. & PETERSON, L. 2000. Prediction of migration of CO2 injected into a saline aquifer: Reservoir history matching to a 4D seismic image with a compositional gas/water model. In: WILLIAMS, D. J., DURIE, R. A., MCMULLAN, P., PAULSON, C. A. J. & SMITH, A. Y. (eds) Greenhouse Gas Control Technologies, CSIRO Publishing, Collingwood, Australia, 378-384. ZWEIGEL, P., ARTS, R., LOTHE, A. E. & LINDEBERG, E. (2004). Reservoir geology of the Utsira Formation at the first industrial-scale underground CO2 storage site (Sleipner area, North Sea). In: BAINES, S. J. & WORDEN, R. H. (eds) Geological storage of CO2. Geological Society, London, Special Publications, 233,165-180.
This page intentionally left blank
Carbon dioxide sequestration in the Campine Basin and the adjacent Roer Valley Graben (North Belgium): an inventory B. LAENEN1, P. VAN TONGEREN1, R. DREESEN1 & M. DUSAR2 1
Energy Department, Vito, Boeretang 200, 2400 Mol, Belgium, (e-mail: [email protected]) 2 Belgian Geological Survey, Jennerstraat 13, 1000 Brussels, Belgium Abstract: The Campine Basin offers a variety of potential CO2 storage sites. Within the Roer Valley Graben area Triassic Buntsandstein rocks have an estimated CO2 storage capacity of several hundred million tons. Mesozoic shales and younger sediments provide adequate sealing. Westphalian D sandstones may possess storage opportunities in the southern graben area, and the karstified/dolomitized parts of the Dinantian carbonates in the western sub-basin have an estimated CO2 storage capacity of 130 X 106tons. In places, the Dinantian reservoirs may be combined with carbonate reservoirs in the underlying Devonian strata. Sealing is provided by Namurian shales and Westphalian coals and shales. Along the western and southern edge of the basin the Dinantian reservoir is sealed by Cretaceous chalks and marls. Besides aquifers, the available coal qualifies for CO2 storage. Six coalbed methane target areas contain a producible enhanced coalbed methane (ECBM), volume of 53-79 X 109 m3. This provides a minimum CO2 sequestration potential of 400 X 106tons. ECBM development factors include the sedimentological setting of the coal sequence as well as changes in porosity, permeability and stress conditions induced by former mining activities. The CO2 storage capacity in abandoned coalmines is limited. Nevertheless, CO2 storage in the abandoned mines may be an option when it is combined with coalmine methane extraction or ECBM production in neighbouring areas.
Together with the other member states of the European Community, the federal and regional governments of Belgium have endorsed the 6th Environmental Action Programme. The first priority point of this programme is the fight against the increase in the atmospheric concentrations of greenhouse gases. The goal is to stabilize the concentrations of greenhouse gases at a level that will not cause unnatural variations of the earth's climate. The key priority for the climate part is the ratification of the Kyoto protocol, which imposes a cut in greenhouse gas emissions by the European Community of 8% over 1990 levels by 2008-2012. Belgium's contribution as part of the joint effort of the EC, is put at a reduction of 7.5% over 1990 levels. The commitment levels can be reached by reducing anthropogenic emissions from sources, enhancing anthropogenic removal by sinks, or trading of emission reduction units. Under removal by sinks, the protocol also mentions carbon dioxide sequestration techniques. In an attempt to reach Belgium's international commitments, the federal and regional governments formulated a series of climate plans that are based on the principle of sustainable development and a rational use of energy resources. However, these measures alone taken will not be sufficient to reach the imposed reduction level. In this context, geological sequestration of CO2 is a viable and safe technology to cut greenhouse gas emissions in the short term. With this in mind, Vito and the Belgian
Geological Survey made an inventory of the potential CO2 storage sites in the Flemish subsurface. The study focused on storage in deep saline aquifers and in coals. The selection of potential storage sites was based on the knowledge that geological sequestration of carbon dioxide is efficient under supercritical conditions and that for safety reasons natural traps and adequate sealing have to be present. With respect to the average actual geothermal gradient of 30-33 °Ckm-1 (Vandenberghe & Fock 1989) and given the subsurface geology in Flanders, storage in aquifers shallower than 800m were not deemed suitable. For economic reasons, the lower limit for potential reservoirs was placed at 2000m. The depth range rules out most of the Mesozoic and Tertiary aquifers and limits the target area for CO2storage to the subsurface of the Campine Basin and the adjacent southwestern edge of the Roer Valley Graben. This paper summarizes the principal results of the inventory study. It consists of four parts. The first sketches the general geological setting of the target area. The second deals with storage in deep aquifers. The third describes the possibilities for CO2 sequestration in combination with enhanced coalbed methane production (ECBM) in Westphalian coals and the former mine infrastructure. The inventory ends with a brief discussion of the economic conditions in which CO2 storage and ECBM production may take place.
From: BAINES, S. J. & WORDEN, R. H. (eds) 2004. Geological Storage of Carbon Dioxide. Geological Society, London, Special Publications, 233,193-210.0305-8719/047$ 15.00 © The Geological Society of London 2004.
194
B.LAENENCTAL.
Geological setting The Campine Basin covers most of the Flemish provinces of Antwerpen and Limburg (Fig. 1). It is part of the extensive Carboniferous basin of northwestern Europe. The northern and northeastern border is formed by the Hoogstraten fault and the NW-SE striking boundary faults of the Roer Valley Graben. Eastward the basin extends into Dutch Limburg, where the NE-SW striking Variscan Anticlinaal fault/Oranje fault system forms the boundary with the German Carboniferous Wurm Basin. To the west and south, the basin is bounded by the subcropping early Palaeozoic rocks of the Caledonian London-Brabant Massif. At the southern edge of the basin, clastic Givetian rocks disconformably overlie the Caledonian basement (Fig. 2). The succession continues with carbonates and siliciclastic sediments the Banjaard Group, which is of Frasnian to Famennian age. The predominantly siliciclastic Banjaard Group is covered by the Lower Carboniferous dolostones and limestones of the Carboniferous Limestone Group. The Tournaisian and lowermost Visean strata are generally developed as dolostones. The upper Visean succession consists primarily of limestones. In a large part of the basin, the carbonates of the Carboniferous Limestone Group are intensely karstified. It is in one of these karst structures that Distrigas successfully developed the underground gas-storage facility of Heibaart. The Visean / Namurian transition is marked by a shift from a carbonate to a siliciclastic setting, which is characteristic for the Upper Carboniferous paralic coal basin of northwestern Europe. The Silesian sequence, known as the Belgian Coal Measure Group, starts with the deposition of open marine shales. The facies gradually becomes more proximal: this results in the development of coastal marshes during early Westphalian times and continental settings (back-swamps, fluvial plains and 'hinterland' facies) during the late Westphalian. The gradual shift towards continental facies culminated in the deposition of the thick, porous, multi-storey, fluvial sandstone bodies during the Westphalian D (Paproth et al 19830, b\ Langenaeker2000). In the northeastern part of the basin, the Belgian Coal Measure Group is disconformably covered by a wedge of late Palaeozoic and Mesozoic sediments (Langenaeker 2000) (Fig. 3). The base of this sequence is formed by impure, marly limestones and shales of the Helchteren Formation. These Permian deposits are overlain by clastic, carbonate and mixed Triassic sediments of the Buntsandstein, Muschelkalk and Keuper Formations, and thick, Lower Jurassic claystone deposits of the Sleen and Aalburg Formations (Fig. 2). Finally, the Palaeozoic and early Mesozoic successions are disconformably
covered by 300-1000m of gently dipping Upper Cretaceous carbonates and predominantly clastic, Tertiary deposits (Fig. 2). The Campine Basin is transected by a set of predominantly NNW-SSE-striking normal faults, which locally display a shear component (Fig. 1). Most of these faults already existed during the Carboniferous. The more important ones were reactivated during the Jurassic. Some of the faults, e.g. the Heerlerheide Fault and the Feldbiss Fault, are still active today. A sub-Hercynian tectonic inversion of these reactivated faults during the Late Cretaceous and Early Tertiary was followed by the subsidence of the Roer Valley Graben in the midTertiary. As a result, the thickness of the Cretaceous sequence is strongly reduced in the graben whereas the Tertiary deposits are much thicker than in the adjacent Campine Basin (Van Wijhe 1987; Langenaeker 1999). Locally the NNW-striking faults intersect with subordinate N-S to NE-SW striking thrust faults that are relicts of the compressional regime related to the Variscan uplift of the basin. The resulting pattern is one of a series of elongated, NW-SE striking fault blocks, which are generally tilted towards the NNE. The tilting was caused by the uplift of the London-Brabant Massif during the Kimmerian erogenic phases (Langenaeker 2000). It caused the Carboniferous subcrop to deepen quickly towards the north and NE and resulted in the preservation of the most complete Silesian sequence in the NE of the province of Limburg (Fig. 1). On each of the fault blocks, the Westphalian strata generally display a gentle monoclinal dip of 5°-10° towards the NNE. The same trends in deepening of the subcrop and preservation of strata is seen in the remains of Permian to Jurassic sediments (Fig. 3). Another striking tectonic feature is the Donderslag Fault Zone. It is a roughly north-south running lineament that divides the Palaeozoic basin into a western and an eastern part (Fig. 1). Evidence for synsedimentary tectonics during the Carboniferous is found in the burial histories and the contrasting sedimentological styles on both sides of the Donderslag Fault Zone. These differences warrant the subdivision of the Palaeozoic Campine Basin into eastern and western sub-basins (Dreesen et al. 1995; Van Keer et al 1998; Helsen & Langenaeker 1999). There is no clear evidence for post-Palaeozoic activity of the Donderslag Fault Zone (Langenaeker 2000). Other important, roughly north-south striking, fault zones are the Leut Fault and the parallel running Dilsen Fault System, both crossing the Belgian-Dutch border. The Donderslag Fault Zone is thought to have a significant dextral strike-slip component (Dusar & Langenaeker 1992). The Leut Fault also appears to have a small dextral horizontal component. Between the Leut Fault and the Dilsen Fault System either a
Fig. 1. Palaeozoic subcrop map of the Campine Basin (compiled after Langenaeker 2000 and Patijn & Kimpe 1961). Important faults and structures mentioned in the text are indicated.
196
B. LAENEN ETAL.
Fig. 2. Stratigraphic position and lithological sketch of the potential storage sites.
GEOLOGICAL CO, STORAGE IN NORTH BELGIUM
197
Fig. 3. Distribution of Permian (Helchteren Formation), Triassic (Buntsandstein, Rot, Muchelkalk and Keuper Formations) and Jurassic (Sleen and Aalburg Formations) rocks in the north-eastern area of the Campine Basin (after Langenaeker 2000)
small Carboniferous graben has developed, or it has a larger than average wrenching component.
Storage in aquifers Within the geological setting of the Campine Basin and the adjacent Roer Valley Graben, four aquifers were identified that comply with the depth and safety restrictions mentioned above, and hence qualify for geological CO2 storage (Table 1). These are the karstified reservoirs within Tournaisian and Visean carbonates of the Carboniferous Limestone Group, the upper Westphalian Neeroeteren Sandstones, the thick Buntsandstein deposits in the Roer Valley Graben and poorly explored structures in the upper Cretaceous and lower Tertiary strata in the north and NW of the Campine Basin. Each of these aquifers is discussed briefly below.
Tournaisian and Visean carbonates Two seismic surveys acquired in the early 1980s enabled the preliminary mapping of the Dinantian subcrop in the northwestern part of the Campine
Basin (Dreesen et al. 1987). An apparent feature of the seismic profiles is the sagging of the overlying Silesian, Cretaceous and in places even Tertiary strata above collapse structures within the Visean limestones. The collapse structures are the result of supposed dissolution of anhydrites within the Visean sequence and karstification that occurred during the various geological periods of uplift in the Campine Basin. The percolating meteoric waters caused preferential dissolution along faulted zones. In several places sagging is associated by a clear change in seismic facies within the overlaying strata. This is indicative of a sudden changes in limological character over the collapsed zones and so illustrates the influence of dissolution on the overlying sedimentation (Dreesen et al 1987). The published isochron map clearly displays the morphological and structural complexity of the area (Fig. 4). The overall pattern of the Visean subcrop in the mapped area shows an alternation of flattened and rather wide antiformal highs surrounded by relatively narrow depression belts. In the west, the antiforms have relatively steep slopes and are separated by well-developed depression belts that reach depths of 250-300m and range in width from 250-lOOOm. Many of these depressions appear to be
B.LAENENETAZ,
198
Table 1. Reservoir properties and CO2 storage potential for the selected geological storage sites in the Campine Basin and in the adjacent Roer Valley Graben. Karstified carbonates Geological characteristics Lithology limestones and dolomites Age Dinantian (extending into Tournasian and Devonian) Location West Campine Sub-basin Type aquifer Reservoir characteristics Depth (m) 800-2300 Temperature (°C) 40-150 Porosity (%) 1-20 Permeability (mD) 2-3000 CO
Westphalian Coal Measures
Neeroeteren sandstones
Buntsandstein
Chalks
coal
sandstones
sandstones
calcarenites
Westphalian
Westphalian D
Early Triassic
Late Cretaceous Early Tertiary
Campine Basin
Roer Valley Graben aquifer
Roer Valley Graben aquifer
northern Campine Basin aquifer
800-2000
800-2000
750-800
adsorption on coal (ECBM) 500-1500 30-65
6-20 0.1-120
5-20 0.3^00
<30 <500
>880
Limited
CC2 storage potential
Total volume2 ) (10 6 tonCO remarks
130
400
Several small reservoirs small scale storage facilities
Small ECBM fields - limited storage facilities in mine
Small scale storage facilities - need further investigation
Large storage volume - needs exploration
Limited storage facilities
related to NNW-SSE striking fault systems, often form covered the entire West Campine Sub-basin. displaying en-echelon fault steps. The antiformal Wells within this part of the basin reveal a general block pattern is further fragmented by a series of trend from restricted marine facies in the west, over east-west oriented cross-faults. In the east, the reefs towards fore-reef facies in the east (Bless et al. east-west structures are less frequent and the 1976;Muchez^a/. 1987;Muchez^<2/. 1991). In all NNW-SSE striking depressions are shallower and wells from the western half of the basin, the transiwider. As a result the antiformal blocks are larger, tion between the Lower and the Upper Carbonibut generally also flatter, than in the west. The differ- ferous is abrupt and the top of the limestone ences in morphology have been explained by differ- sequence is karstified. A totally different picture ences in the extent of uplift between both areas emerges from the East Campine Sub-basin. Here, the early and middle Visean platform carbonates are (Dreeseneffl/. 1987). A number of cored wells in the area convincingly replaced by thick turbiditic limestone sequences prove that dissolution / karstification and tectonics characteristic of slope-settings during the late are not the whole story behind the antiformal blocks. Visean. Moreover, the Visean-Silesian transition The morphology is also partly controlled by the sed- appears to be gradual and uninterrupted without imentology of the Visean carbonates. The best docu- signs of late Visean palaeo-karstification (Bless et al. mented structures are the Heibaart dome between 1976). The differences in sedimentology and karstificathe villages Loenhout and Rijkevorsel, and the antiform at Poederlee (Fig. 4). The former structure tion are explained by a difference in stress regime for is being used by the gas-distribution company both parts of the basin during the late Visean: transDistrigas to store natural gas. The actual stored pressive in the west, transtensive in the east. This volume exceeds 1 X 109Nm3 (Nm3 is the volume of caused slower subsidence and intermittent uplift in gas under normal atmospheric conditions: 20 °C and the West Campine Sub-basin and uninterrupted 1 atm). The top of the porous limestone reservoir lies gradual subsidence in the East Campine Sub-basin. at 1080m below sea-level. Its final spill-point is situ- Besides the contrast in sedimentology, this also ated at 1295 m. Both explored domes are late Visean resulted in the deposition of a thicker late Visean algal to crypto-algal reef mounds that developed on sequence in the east (more than 500m thick) than in a shallow carbonate platform. The carbonate plat- the west (less than 375 m thick) (Bless et al 1976).
GEOLOGICAL CO? STORAGE IN NORTH BELGIUM
199
Fig. 4. Isochron map for the top of the karstified top Visean limestones in the northwestern part of the Campine Basin (after Dreesen et al. 1987)
From this geological framework it is deduced that reef mounds similar to those found at Heibaart and Poederlee may be present in the unexplored areas of the West Campine Sub-basin. The search should initially focus on uplifted structural blocks, for example the Donderslag structure and the GenendijkHulst structure, for these are expected to be the most intensely karstified (Fig. 1). Adequate sealing would be provided by thick shale accumulations in the overlying Namurian and Westphalian strata, which due to their ductile behaviour guarantee good reservoir closures even in the surrounding fault zones. Additional sealing will be provided by the overlying coal seams, which would capture any escaping CO2. The thickness of the karstified reservoir in the West Campine Sub-basin is a function of carbonate facies and burial history. Well observations indicate that the thickness of the karst interval varies from 5-160m. Within this range, various karstified horizons may be superposed. At Heibaart for example, several levels of open fractures occur in the upper 60-80 m of the Dinantian and other, underlying karstified horizons are present as well, including two intervals with increased porosity (7.5% average matrix porosity) in the Frasnian dolostones (Heibaart 1/lbis well). The combined thickness of these two zones amounts to 10m. In the Turnhout
well, a 46m thick karstified zone has been encountered starting at a depth of 2200m. This is important as it proves that karstified carbonates retain useful porosity and permeability conditions at deeper levels. Although the lateral extent of the karstified zone is well known, its true depth extent remains uncertain. Depending on the palaeo-topography and the shift in sedimentary facies, the thickness of the karstified zone(s) is likely to diminish towards the east (Dreesen et al 1987; Vandenberghe et al 2000). Moreover, questions remain about the homogeneity of and the connectivity between the Dinantian reservoirs. However, hydraulic pressure measurements in the Merksplas-Beerse well show connectivity with wells in the Heibaart gas-storage site 10km towards the west (Vandenberghe et al 2000). The interference occurs in the karstified reservoir(s) in the uppermost Dinantian. It thus seems likely that the karst reservoirs in large parts of the West Campine Subbasin area are in some way connected. The hydraulic communication between wells 10km apart also suggests that most fault zones in the Dinantian act as open conduits through the carbonate rocks. Based on the structures mapped during the seismic campaigns of 1980 and 1981 (Fig. 4) and reservoir data from the Heibaart gas-storage and the Merkplas-Beerse well (Vandenberghe et al. 2000),
200
B.LAENENEFAL.
Fig. 5. Possible storage locations for CO2 in karstified Palaeozoic carbonate rocks below the Cretaceous unconformity. The upper Cretaceous limestones form both a chemical and conventional seal to the injected CO2. The dashed lines mark the lower boundary of the hypothetical reservoirs.
an estimate of the total cumulative CO2 storage capacity of the karstified strata has been made (Table 1). However, except for Heibaart and Poederlee, the mapped structures are only tentative and generally not well defined. This is due to inadequate and insufficient seismic information. Therefore, any new structure that might be considered for CO2 storage must be redefined by new seismic data before any drilling. Using a seismic velocity of 3400ms"1 for the Namurian sediments and an average reservoir porosity of 2.5%, the storage capacity of the mapped structural closures in the top Dinantian strata is estimated at 46 X 106 of supercritical CO2. This figure includes the Heibaart structure. The explored area covers roughly one fifth of the West Campine Sub-basin. However, about half of the area is judged unsuitable for CO2 storage as the potential reservoirs are situated too deep in the NNE area or lack appropriate sealing by shales or coalbearing strata in the west and SW. Extrapolation of the estimated storage volume and reservoir characteristics of the explored area towards the remaining unexplored regions of the West Campine Sub-basin leads to an estimated total storage volume of 115 X 106 of supercritical CO2. In view of the observations made in well Heibaart 1/lbis, local additional storage potential at lower stratigraphic levels in the Dinantian or Frasnian, may add to this figure. Therefore, the minimal CO2-storage potential of the western part of the basin in Lower Carboniferous and Devonian strata is tentatively estimated at 130 X 106 of supercritical CO2. It may be possible to inject and store CO2 in Dinantian strata that have been karstified during the erosion period preceding the deposition of the Upper
Cretaceous chalks. This situation occurs along the western and southern edge of the basin (Fig 1). The CO, storage trap would then be formed by the dipping karstified limestones of the Dinantian and the onlapping Cretaceous strata (Fig. 5). Sealing would be provided by the Cretaceous shales or tight chalk. The sealing quality of the chalks may be improved by chemical reactions between the CO2, the formation water and the surrounding carbonates. This possibility appears promising. However, to date, the potential for effective chemical sealing has not been investigated thoroughly. Therefore this option has not been taken into consideration during calculation of the CO2-storage capacity of the Dinantian. As the explored antiformal structures are relatively small, the Dinantian reservoirs identified are judged suitable for CO2 storage by local CO2 producers. Moreover, close to this western part of the Campine Basin, there are many large CO2 producers that could use some of the reservoirs to limit their CO2 emissions on the short term.
Neeroeteren sandstone In the NE of the Campine Basin, as well as in the Roer Valley Graben, Upper Westphalian D sandstones of the Neeroeteren Formation have been preserved (Fig. 1). The wedge of Neeroeteren sandstones thickens rapidly towards the east. It reaches an estimated thickness of 500m along the Feldbiss Fault (Dusar 1989). The thickness distribution within the graben is not known. The Neeroeteren Formation is characterized by thick, multi-storey, coarse- to fine-grained sub-arkose
GEOLOGICAL CO, STORAGE IN NORTH BELGIUM
sandstone complexes that have been deposited by large, fan-shaped braided rivers in a 'hinterland' setting. At varying levels, overbank intercalations occur that are composed of silty shale, shale and/or coals (Dreesen et al. 1995). The sandstones generally have good reservoir characteristics (Table 1). Porosities of 12% or more are frequently measured in the coarser sediments. The finer overbank deposits usually contain porosities up to 6%. The permeability of the coarser sandstones is good, with average values around 120mD. Depending on facies and lithology, the overbank fines still have permeabilities of 0.1-2 mD. In the area directly south of the Roer Valley Graben, the subcrop-depth of the Neeroeteren sandstones lies between 650-1200m. Large parts of the Neeroeteren Formation in this area are therefore considered too shallow to guarantee stable supercritical CO2 storage conditions. Another point of concern is the seal integrity. In the Campine Basin, the sandstones are partly overlain by thin marly sediments of the Helchteren Formation and Lower to Middle Triassic deposits, and partly by Upper Cretaceous chalks and Tertiary deposits (Fig. 2). As no Jurassic shales have been preserved in this area and as potentially sealing Cretaceous chalks and Tertiary shales occur at shallow depths, a conventional seal appears to be absent. Therefore, good CO2 storage opportunities in the Neeroeteren sandstones can only be provided by local intraCarboniferous traps and seals (shales and coals). In the Belgian part of the Roer Valley Graben, the top of the Neeroeteren Formation deepens quickly below the arbitrarily chosen depth limit of 2000m. Only in the southeastern part of the graben are the sandstones expected to be present at shallower depths. Moreover, the transition zone between the Campine Basin and the northeastern part of graben is formed by a relatively narrow zone that consists of a series of wrench faults. In this area, the Neeroeteren sandstones locally occur at depths of 800-2000 m. Moreover, both potentially sealing Jurassic shales as well as potential structural traps may occur here (Langenaeker 2000). Nevertheless, the CO2 storage potential of the Westphalian-D sandstones is considered limited. The best opportunities are likely to exist in the centre of the transition zone between the Campine Basin and the Roer Valley Graben and in the most southeastern part of the graben itself. This needs to be confirmed by adequate exploration of these target areas.
Buntsandstein The Triassic Buntsandstein Formation forms an important reservoir within the subsurface of the northeastern Campine Basin and the adjacent Roer
201
Valley Graben (Fig. 3). The formation comprises a stacked series of red fluvial conglomeratic to finegrained sandstones, siltstones and shales. It is overlain by shaly limestones of the Muschelkalk Formation and shales of the Keuper Formation, which locally contains thin anhydrite intercalations. In the graben, the Triassic rocks are covered by a thick sequence of Jurassic shales (Sleen and Aalburg Formations), which have good sealing capacity, a thin sequence of Cretaceous chalks and a thick package of clastic Tertiary sediments, many of which are tightly-packed shales and have good sealing properties. The combined thickness of the sandstone members of the Buntsandstein commonly exceeds 450m (Dusar et al. 1998; Langenaeker 2000). Porosities measured in the shallower fault block areas south of the graben, show an average of 7.1%, but the coarse sandstone sequences commonly have higher values. The depth of the top of the Buntsandstein in the Campine Basin is too shallow and the sealing capacity of the overlying formation is insufficient to guarantee safe CO2 storage. Inside the Roer Valley Graben, the formation occurs at sufficient depth to allow CO2 storage under supercritical conditions. Moreover, the thickness of sealing overburden (Muschelkalk, Keuper Sleen and Aalburg Formations) increases rapidly towards the NNW. In conjunction with an increasing burial depth of the Buntsandstein reservoir in the same direction, which is caused by the dip of the southern graben blocks in a northwesterly direction, and sealing by Cretaceous and Tertiary sediments in the SE, a potential reservoir with an area of about 186km2 occurs in the Belgian part of the Roer Valley Graben (Fig. 3). Using an average thickness of 450m of the sandstone sequences and an average porosity of 5%, the reservoir volume is estimated at 4.2km3. Even if only 2km3 of this volume is suitable and/or available for CO2 storage, this would still allow storage of 880Xl0 6 t of CO2 under supercritical conditions. Although this figure is only a rough estimate, it clearly demonstrates the necessity for a careful investigation of the storage potential of the graben.
Late Cretaceous and Early Tertiary chalks In the Campine Basin, a 300-1000m thick sequence of the Upper Cretaceous and Tertiary sediments discordantly overlie all the older deposits. This sequence dips gently and thickens towards the north. The basal layers of the Cretaceous are formed by sands and marls of the Aachen and Vaals Formations. These sandy facies are resticted to the eastern and central parts of the Campine Basin. Towards the west, the sediments become more clay-rich (lignitecontaining clays in the Aachen Formation, marly
202
B.LAENENErAZ,
Fig. 6. Isochron map for the Base Cretaceous in the northwestern part of the Campine Basin (after Dreesen et al. 1987).
clays in the Vaals Formation). The change in lithology is the result of the inversion of the Roer Valley Graben during the Upper Cretaceous. This led to erosion of pre-Cretaceous sediments in the graben area and supplied sands to the adjacent part of the Campine Basin. Moreover, the thickness of the Upper Cretaceous deposits in the graben was significantly reduced (Bless et al. 1986; Langenaeker, 1999). The siliciclastic and marly deposits of the Aachen and Vaals Formations are capped by fine, often flint-rich chalks and marls of the Gulpen Formation. Along the western edge of the basin, the Gulpen Chalks directly overlie Palaeozoic strata. The top of the Cretaceous sequence and the basal layers of the Tertiary are formed by calcarenites of the Maastricht and Houthem Formations. The chalk deposits are covered by a predominantly siliciclastic Tertiary sequence, which displays a cyclic alternation of sandy and clay-rich formations (Fig. 2). Most of the sand packages are important sources for drinking water (Bless et al. 1986). Moreover, two intervals with good aquifer properties occur within the Cretaceous/Early Tertiary sequence: the calcarenites of the Maastricht and Houthem Formations and the coarse sand bodies within the Aachen Formation (Table 1). The former extend over the entire study area. The calcarenites have an effective porosity of up to 30% and a permeability approaching 0.5 D.
The reservoir units are up to 50m thick. They are sealed by Tertiary shales of variable sealing quality. At present, the calcarenites are an important source for drinking water in the southern parts of the basin. In contrast, the geographical extension of the basal sand bodies is restricted to the eastern part of the Campine Basin adjacent to the graben area. Both aquifer intervals are separated by the chalks and marls of the Gulpen Formation. These deposits are compacted, have low permeability and may even act as a chemical trap when brought into contact with CO2. For the past 80 years, the tight chalks have protected the overlying drinking-water aquifer in the Maastricht Formation from being drained into the underlying collieries. In the largest part of the Campine Basin, the base of the Cretaceous lies at depths of less than 800m (Fig. 6). Therefore, the Cretaceous aquifers of the central and southern parts of the basin are largely excluded as targets for supercritical CO2 storage. In the north and NW of the Campine Basin the calcarenites of the Maastricht and Houthem Formations exceed the 800m depth. This also holds for the Roer Valley Graben. However, the graben has a low potential for suitable Upper Cretaceous reservoirs due to changes in sedimentary facies, more intense diagenetic cementation, and a strong thickness reduction of the Cretaceous strata (Bless et al 1986). Although the
GEOLOGICAL CO, STORAGE IN NORTH BELGIUM
quality of the reservoirs deteriorates with depth, the calcarenites are still suitable for the production of drinking water at intermediate depths. In the north, the water becomes too mineralized. Here, the porous chalks are used locally for the extraction of low enthalpy geothermal energy. The potentially interesting regions for CO2 storage have been only partly explored. Mapping of the area did not reveal any large closed structures that would be suitable for CO2 storage. The only suitable structure identified within the calcarenites is located just south of the village Westmalle (Fig. 6). The structure extents over an area of 6.5 km2 and its height is limited to 20-30 m. Because of lower geothermal gradients in the overlying Tertiary strata, its depth of 750-800m may just be enough to warrant supercritical CO2 storage. Overlying Tertiary clays should provide ample sealing. In conjunction with likely additional, underlying Dinantian reservoirs nearby, it may well complement any desired local storage capacity and/or may be regarded as an additional safety buffer. Similar structures may exist on fault-bounded horst blocks in the north and northeastern parts of the basin along the Roer Valley Graben boundary faults, but this needs to be verified by further exploration.
CO2 sequestration in Westphalian coals Geological background Besides the aquifers, the Westphalian Coal Measures may serve as a large potential CO2 sink in the northern part of Belgium. In a large part of the provinces of Antwerpen and Limburg, coal-bearing strata occur at depths of 500-3500m. The Coal Measures belong to the extensive West European paralic coal belt that came into existence during the late Carboniferous. Sedimentological evidence indicates a diachronous shift in these Westphalian sediments from a marineinfluenced depositional setting in the late Namurianearly Westphalian to more fluvial-influenced continental settings during the late Westphalian. In addition, a gradually shifting palaeoclimate and synsedimentary tectonics were responsible for the observed changes of the fluvial environment and architecture and consequently, in the geometry and properties of the peat deposits. The latter eventually affected the maceral composition, geochemistry and hence the methane generation potential of the resulting coal beds (Dreesen et al. 1995; Langenaeker 2000). The burial history of the Campine Basin shows two episodes of major subsidence followed by two stages of erosion. Most parts of the basin reached their maximum burial by the end of the Westphalian (Helsen & Langenaeker 1999). The subsequent
203
Fig. 7. Evolution of the methane content of coal seams in the CBM test well at Peer with increasing depth below the Permian / Carboniferous unconformity.
Variscan orogeny halted the first major burial stage and caused uplift and erosion during Stephanian and early Permian times. During the erosion event, the coals became partially degassed, as shown by data from the single CBM-test well at Peer (Fig. 7). An important second burial phase took place during the Lower to Middle Jurassic. However, only in the northeastern part of the basin was a new burial maximum reached, giving rise to new gas generation (Van Keer et al 1998; Helsen & Langenaeker 1999).
Coal and CBM-reserves To calculate the amount of CO2 that may be sequestrated in the Campine coals, the geological coal resources of the basin were calculated down to a depth of 1500m. To this end, geological data from more than 140 wells, as well as over 6000 coal rank analyses (volatile matter content and vitrinite reflectance values) have been processed. Problems due to regional structural and stratigraphical variability were overcome by calculating the coal resources per stratigraphic interval and for each identified tectonic block. For each block, a coal resource map was drawn. The summation of the results for all blocks gave the estimate of the total coal resources. The results were verified by comparison with stratigraphic maps, data from underground boreholes and information from coal exploration campaigns during the 1980s. Where necessary, the derived data were corrected and subsequently stored
204
B.LAENENCTAL.
Fig. 8. Relation between coal rank and remaining gas content in the area of the Dutch State Mines in the SouthLimburg province of the Netherlands (after Stuffken 1957). The labelled dots represent the average volatile matter and methane content of individual coal layers (numbers refer to the coal classification scheme used by the Dutch 'Geologisch Bureau voor het Mijngebied'), Dots connected with a line represent couples of adjacent coal layers of which the methane content is mutually influenced due to mining. The crosses stand for the average gas content of these coal layer couples. This average value more accurately represents the gas content of the coal couples prior to mining. The hatched area corresponds to the envelope of the individual measurements on single coal layers (not influenced due to mining of an adjacent coal seam) or coal layer couples.
into a database. During this step, a coal rank was assigned to each coal package. In the next step, the gas content of the coals was calculated using the relationship between methane content and coal rank given by Stuffken (1957, 1960). The Stuffken relationship was experimentally derived from collieries in the Dutch part of Campine Basin (Fig. 8). It was established for coalgas emission purposes in the former Dutch State Mines in Dutch Limburg. It also appeared to be applicable in the Belgian part of the basin (Vandeloise 1971). The Stuffken relationship describes the gas content of coal seams that fall within the degassed zone below the top of the Carboniferous, as a function of coal rank and, to a minor extent, dip of the strata. At low to moderate dips and in an undisturbed Westphalian succession, the degassed zone roughly extends to a maximum depth of 600m below the top of the Westphalian (Stuffken 1957,1960; Teichmuller et al 1970; Von Treskow 1985). The gas content of coals in the degassed zone appears to be largely a function of the overall coal porosity. This parameter changes with rank (Stuffken 1957). The maximum gas content is found in coals of medium volatile to low volatile rank. At greater depths the low volatile to anthracite flank of the Stuffken-curve is likely to level off to a constant value of approximately 15Nm3/ton coal, due to larger remaining overburden pressures and hence less gas loss during uplift. The empirical Stuffken relationship gives a reasonable first approximation of the coalbed methane
reserves of the Campine Basin. The method is likely to underestimate the gas content in the northern and northeastern parts of the basin due to the occurrence of a second coalification stage and deeper burial of the most gaseous Westphalian strata. In contrast, comparison with coal-gas emissions from the former coal-mines proves that the calculated gas-content is fairly reliable in the southern parts. Besides coal rank, differences in maceral composition are also known to influence porosity and cleat formation, as well as the adsorption capacities of coals. In the light of lateral constancy of the average maceral composition of individual coal packages throughout the basin, those effects are judged to be averaged out by the number of measurements. Any changes in gas contents due to temperature differences and/or water saturation conditions have not been taken into account. The calculated total geological coal resource for the Campine Basin corrected for mining amounts to 38.8 X 109t (Fig. 9a; Table 2). As it is assumed that only coal seams with a minimum thickness of 90 cm can be mined conventionally, the total reserve is estimated at 8.5 X 109 tons. Of this reserve, 5 X 109 occur in the West Campine Subbasin and 3.5 X 109 in the East Campine Sub-basin. From this estimate, the corresponding amount of gas-in-place was calculated using the coal rank assigned to each coal in conjunction with the Stuffken relationship (Fig. 9b; Table 2). In the resource calculations the gas concentration contour of 50Xl0 6 Nm 3 CH4km-2 has been taken as a threshold value. So the relatively low gas containing
GEOLOGICAL CO2 STORAGE IN NORTH BELGIUM
205
Fig. 9. (a) Coal-concentration map for the Campine Basin down to a depth of 1500 m. Coal concentration is given in million tons per km2. Contour interval is 5 X 106 km"2, (b) Coalbed methane concentrations in the Campine Basin down to a depth of 1500 m calculated using the Stuffken-relation between coal rank and gas content. Gas-concentration contours are in million Nm3 kmr2. Contour interval every 50 X 106 Nm3 km~ 2 . The target areas are hatched.
B.LAENENCTAL.
206
Table 2. Coal and coalbed methane resources in the Campine Basin. Coal resources 9
Campine Basin Total West Campine Sub-basin Total Target areas East-Campine Sub-basin Total Target areas
Coalbed methane resources Concentration (Nm3 CH4/ton coal)
Total (10 tons)
Mineable (109tons)
Gas in place (109Nm3)
38.8
8.5
291
22.4 4.3
5.0 0.8
174 44.9
8.2 10.9
16.4 9.5
3.5 1.9
117 87.2
7.2 9.1
7.8
areas between the 0-50Xl0 6 Nm 3 CH4krrr2 contour lines have not been taken into account, as this gas is assumed not producible. For the remaining part of the basin an overall gas resource of 291 X 109Nm3 has been calculated. This figure is corrected for loss of gas due to former coal mining activity. The calculated average CBM concentration amounts to 7.8Nm3 CH4 per ton coal (Table 2). Six target areas for CBM-production were marked (Fig. 9b). They have a gas content over five times the threshold value. Their cumulative methane resource totals 132 X 109Nm3 CH4. One third of this volume is situated in coals that do not warrant any gas production because they are isolated, too thin, or located too close to faults or permeable overburden to be of any gas-prospective value. Of the remaining coals we estimate that 60-90% of the gas could be produced successfully, provided this happens by enhancing production techniques with injections of CO2, N2 or flue gases. For the six target areas, this leaves a producible methane volume of 53 X 109Nm3 to 79 X 109Nm3. The lower estimate represents about five times the yearly gas consumption in Flanders.
Using a conservative 2:1 exchange ratio for methane replacement by CO2 (Gunter et al 1997) and a 60% methane recovery only, the minimum geological CO2 sequestration capacity within the six target zones is estimated at 106X109Nm3 or 208 X 106t under standard conditions. If by the use of CO2 stripping, an enhanced CBM recovery of 90% could be obtained, the CO2 sequestration capacity by single exchange would increase to 312 X 106. This figure does not take into account the fact that the coal seams in the Campine Basin are on average 20% undersaturated in methane. Moreover, at depths below 800m, CO2 will be injected under supercritical conditions. Experimental work performed at the Technical University of Delft proves that when both supercritical and multi-layered sorption conditions for CO2 are considered, the maximum amount of CO2 sequestration within the coals rises by a factor of 1.5 or more (Wolf et al 2000). Taking the undersaturation of the coals and the high-pressure injection conditions into account, the total CO2 sequestration potential within the six target zones is cautiously expected to be in the order of 400 X 106t CO2 under standard conditions.
CO2 storage potential
Use of abandoned coal mines
In general, storage of CO2 in coal seams is independent of their methane saturation. However, for both economical and practical reasons it is preferable to limit the subsurface CO2 sequestration potential of the Westphalian coals to those areas where sequestration is likely to be associated with an important CBM production. This restricts any future CO2 sequestration within the coals to the six target areas in which the CBM concentration exceeds 250X106Nm3 CH4kmr2. Consequently, this also means that large areas of the basin are likely to stay excluded from any viable CO2 sequestration scenarios in coals.
The main problems for the development of an ECBM-production site with simultaneous CO2 sequestration are the geometry and low permeability of the coal seams (Table 1). Because of the low permeability, ECBM production would require a dense grid of injection and production wells. This would be difficult within the densely populated area of northern Belgium. In this context, the abandoned collieries may be interesting sites for small scale ECBM production and CO2 injection. In the former mining areas, tensional release and related porosity increase may facilitate the injection and migration of propellants and enhance methane expulsion. Antici-
GEOLOGICAL CO, STORAGE IN NORTH BELGIUM
pated advantages in production time and the possible reduction of production costs are likely to render any future development of the many ECBM targets in these areas more economic. Economic development of ECBM production sites in the northern target areas will be more difficult. The greater depth at which the coal layers occur will result in higher compressive stresses, and so reduced permeability. The greater depth and lower permeability will increase the development costs of a ECBM field (through higher drilling and well-stimulation costs). Moreover, a closer well-spacing will be required here as well. Therefore, any economic target development in the northern target areas must start with a search for geological conditions that would optimize ECBM production. This may involve looking for certain structural settings (anticlines, intraformational cut-off up-dip slopes, etc.) and/or for areas with enhanced cleat/fracture formation and stress shielding strata (coals in or under competent rock layers). In this way, enhanced fracture development may be expected in differential compaction settings involving channel sands and coal seams. Sandstone bodies overlain by coal also provide potential permeability pathways for CO2 injection in ECBM production schemes. Besides the release in tension on remaining coal seams within the range of influence of the coal mines, the former mining activities created an extensive man-induced reservoir. After closure of the mines, the shafts were filled and sealed, but the rest of the infrastructure was left intact. Although part of it will now have collapsed, an important reservoir, consisting of a extensive network of high-permeability conduits (more than 1000km of stone-drifts and blind shafts), mined out strata and fractured interbedded Westphalian rocks, still exists at depths of 500-lOOOm. Based on literature-derived data (Berding 1952; Labasse 1965a, b, c\ Stassen 1970; Malolepszy & Ostaficzuk 1999; Kunz 2000), we calculated an average effective porosity (including both man-induced and intrinsic porosity volumes) of about 8.5% for the caved areas and about 5.5% for the back-filled zones of the former mining panels. Of the former mine infrastructure we have estimated that only 25% have remained open and intact. Based on these data and a detailed geometrical model, the residual porosity volume for the Beringen coal mine was calculated to be 5X10 6 m 3 . Extrapolation of this figure to all former collieries in the Campine Basin, results in a total residual pore-volume of at least 35.5 X 106m3. Previous calculations have suggested a total residual pore-volume of approximately 60X10 6 m 3 for all former Campine coal mines (Vansteelandt 1993). The artificial reservoir can be used for various purposes, for example disposal of chemical and other types of industrial waste, storage of methane under low to medium pressure conditions or sequestration
207
of CO2. Adequate sealing of the reservoir is provided by impermeable clay layers within the 270-600 m thick overburden. This should guarantee safe sequestration of CO2, methane or waste products.
Economic conditions Driven by a sound economy, the energy demand in Belgium grew steadily during the last decade. Moreover, the relatively high oil prices in combination with a growing concern about the environmental impacts of increased fossil energy consumption (ratification of the Kyoto protocol) have greatly increased the demand for cleaner fuels, notably gas. The rise is expected to accelerate in the light of the anticipated closure of the Belgian nuclear power plants by 2025. Nevertheless, Belgium will have to meet its Kyoto commitment. This means that it will have to cut its CO2 emissions by 7.5% over 1990 levels by 2008-2012. The presence of subsurface CO2 storage potential in various geological formations, including the opportunity to start a local ECBM-industry, is both of strategic and economic importance. The volumetrieally small reservoirs in the Visean limestones offer companies within or near the western part of the Campine Basin an opportunity to reduce their CO2 emissions in the short term. The opportunities increase substantially if the estimates of the Triassic storage volumes (approximately 880 million tons) and the possibilities of sequestration in the Neeroeteren Formation along the Roer Valley Graben prove to be valid. However, subsurface storage should be accompanied by long-term efforts to develop sustainable production techniques and a more rational use of energy. Besides the deep saline aquifers, storage in coals and abandoned mines is another short-term measure that may contribute to Belgium's efforts to cut its CO9 emissions. The amount of CO2 that a starting ECBM-industry can handle yearly is small. Based on data from the Black Warrior Basin and taking a 90% methane recovery and a 2:1 CO0/CH4 exchange ratio, it is estimated that an ECBM gasfield of about 7 km2 and comprising 22 wells requires the injection of about 129000 tons of CO2 per year (Mostade 1999). Thus the large CO2 sequestration capacity of the Campine coals cannot be used immediately for the storage of large quantities of CO2. Nevertheless, the development of a few small ECBM fields in the selected target areas or the former mining area provide ample opportunities to store Kyoto-required amounts of CO2. The quantities of CO9 needed by the ECBM fields can be supplied by the numerous small CO2 producers present in or close to the target areas (CO2 emissions in the range of 100000-200000 tons/year). Except for a few small petrochemical
208
B.LAENENCTAL.
Fig. 10. Location of some important CO2-producers and gas pipeline infrastructure in relation to the CBM concentrations in the Campine Basin.
plants, CO2 is produced as flue gases. Large quantities of pure process CO2 are only available at two petrochemical complexes at the edges of the basin. The first is situated near the town Geleen, just across the Belgian-Dutch border. This complex lies exactly within the cross-border continuation of an anomalous target zone (Fig. 10). Another Belgian producer of fertilizers is present close to the western edge of the basin at a distance of 50-80 km of the prime ECBM-target zones. Nitrogen, another interesting propellant in ECBM production, may be produced easily on site from air (cryogen separation) or at certain production localities. In addition to the presence of a perfect road system, important waterways and railway connections, an efficient natural gas transportation system of pipelines crosses the Campine Basin and Belgium (Fig. 10). These gas transportation pipelines ran from the Netherlands to France, Germany and Luxemburg and beyond. Major branches of this system, both within and just around the basin area, lead to industrial facilities and cities. Subsidiary pipelines even go to the Dutch petrochemical industry at Geleen. Thus in the Campine Basin, CBM reserves, producers of propellants (notably CO2), a technical infrastructure and a market for the produced methane are all available.
Conclusions In Flanders, the possibilities for geological sequestration in aquifers are limited. A series of potential storage sites occur in the Campine Basin and in the adjacent Roer Valley Graben: the karstified reservoirs within Tournaisian and Visean carbonates, the Upper Westphalian Neeroeteren Sandstones, the thick Buntsandstein deposits in the Roer Valley Graben and structures in the Upper Cretaceous and Lower Tertiary in the north and NW of the Campine Basin. Except for the Bundsandstein, the storage capacity of each of these deep, saline aquifers is limited. Therefore, they only qualify for small-scale sequestration projects. The Campine Basin still possesses large coal reserves and at least six target areas with an increased CBM concentration. In combination with CO2 sequestration and/or the use of other propellant gases, the estimated producible CBM reserves in these areas vary between 53-79 X 109Nm3. Initial development of these reserves will vary from target to target. In the southern regions of the basin it will certainly involve the use of the induced changes in the general stress pattern by the former mining activities. In the north an initial development will necessitate the search for geological traps and/or particular coal settings.
GEOLOGICAL CO, STORAGE IN NORTH BELGIUM
CO2 sequestration possibilities in conjunction with the CBM development are important; under certain conditions even up to 400 X 106 tons of CO2 might be sequestrated. Within the Campine basin, CBM-reserves, producers of propellants (notably CO2), a technical infrastructure as well as an expanding market for the produced methane are all present or very near. This warrants further ECBM development activities. The inventory study was commissioned by the Flemish Government (Afdeling Natuurlijke Rijkdommen en Energie). Part of the data used has been collected within the context of the European project on the 'European potential for geological storage of carbon dioxide from fossil fuel combustion (GESTCO)'. The authors are grateful to K. Van Baelen for making the drawings and to the colleagues of the environmental department of Vito for supplying CO2production data for the Campine Basin.
References BERDING, C. J. A. 1952. Le controle des coups d'eau dans les mines de houille. In: VAN AELST, E. (ed.) Compte Rendu du Troisieme Congres pour Vavancement des etudes de Stratigraphie et de Geologic du Carbonifere. Part 1. Netherlands Institute for Applied Geosciences - Geological Survey of the Netherlands (NITG-TNO), Utrecht, The Netherlands, 15-38. BLESS, M. J. M., BOUCKAERT, J., BOUZET, P. & 12 OTHERS. 1976. Dinantian rocks in the subsurface north of the Brabant and Ardenno-Rhenish massifs in Belgium the Netherlands and the Federal Republic of Germany. Mededelingen Rijks Geologische Dienst, 27,81-195. BLESS, M. J. M., FELDER, P. J. & MEESSEN, J. P. M. T. 1986. Late Cretaceous sea level rise and inversion: their influence on the depositional environment between Aachen and Antwerp. Annales de la Societe Geologiquede Belgique, 109, 333-355. DREESEN, R., BOUCKAERT, J., DUSAR, M. SOILLE, J. & VANDENBERGHE, N. 1987. Subsurface structural analyses of the late-Dinantian carbonate shelf at the northern flank of the Brabant Massif (Campine Basin, N. Belgium). Toelichtende Verhandelingen voor de Geologische en Mijnkaarten van Belgie, 21, 1-37. DREESEN, R., BOSSIROY, D., DUSAR, M., FLORES, R. M. & VERKAEREN, P. 1995. Overview of the influence of syn-sedimentary tectonics and palaeo-fluviatile systems on coal seam and sand body characteristics in the Westphalian C strata, Campine Basin, Belgium. Geological Society, London, Special Publications, 82, 215-232. DUSAR, M. 1989. Non-marine lamellibranchs in the Westphalian C/D of the Campine coalfield. Bulletin van de Belgische Vereniging voor Geologie, 98, 483-493. DUSAR, M. & LANGENEAKER V. 1992. De Oostrand van net Brabant Massief, met beschrijving van de geologische verkenningsboring Martenslinde. Professional Paper of the Belgian Geological Survey, 225,1-22.
209
DUSAR, M., BLESS, M. J. M., BURGER, K. & 9 OTHERS. 1998. De steenkoolverkennings-boring Hechtel-Hoef. Professional Paper of the Belgian Geological Survey, 286,1-129. GUNTER, W. D., GENTIS, T, ROTENFUSSER, B. A. & RIDCHERDSON, R. J. H. 1997. Deep coalbed methane in Alberta, Canada: a fuel resource with the potential of zero greenhouse gas emissions. Energy Conversion Management, 38 Supplement, 217-222. HELSEN, S. & LANGENAEKER, V. 1999. Burial history and coalification modelling of Westphalian strata in the eastern Campine basin (Northern Belgium). Professional Paper of the Belgian Geological Survey, 289,1-22. KUNZ, E. 2000. Entsehung und Vorkommen von Grubengas im Steinkohlenbergbau. In: KUNZ, E. & MEINERS, H. (eds) Proceedings of the Oberhausener GrubengasTage 2000. Deutsche Montan Technologic GmbH (DMT), Essen, Germany, 1-19. LABASSE, H. I965a. Les pressions des terrains dans les mines de houille: les mouvements de terrain. Annales des Mines de Belgique, 64,643-668. LABASSE, H. 1965&, Les pressions des terrains dans les mines de houille: les mouvements de terrain. Annales des Mines de Belgique, 64, 837-859. LABASSE, H. 1965c. Les pressions des terrains dans les mines de houille: les mouvements de terrain. Annales des Mines de Belgique, 64,1081-1090. LANGENAEKER, V. 1999. Indications for a transpressional Late Cretaceous inversion in the Belgian part of the Roer Valley Graben. Aardkundige Mededelingen, 9, 139-142. * LANGENAEKER, V. 2000. The Campine Basin: Stratigraphy, structural geology, coalification and hydrocarbon potential for the Devonian to Jurassic. Aardkundige Mededelingen, 10,1-142. MALOLEPSZY, Z. & OSTAFICZUK, S. 1999. Geothermal potential of the Upper Silesian Coal Basin, Poland. Bulletin d'Hydrogeologie, Editions Peter Lang, Universite de Neuchdtel, 17, 67-76. MOSTADE, M. 1999. Coalbed methande potential of the southern coal basin of Belgium. Proceedings of the 1999 International Coalbed Methane Symposium, University of Alabama, Tuscaloosa, USA, 35-45. MUCHEZ, P., CONIL, R., VlAENE, W, BOUCKAERT, J, & POTY,
E. 1987. Sedimentology and biostratigraphy of the Visean Carbonates of the Heibaart (DzHl) borehole (Northern Belgium). Annales de la Societe Geologiquede Belgique, 110, 199-208. MUCHEZ, P., VIAENE, W, KEPPENS, E., MARSHALL, J. D. & VANDENBERGHE, N. 1991. Vein cements and geochemical evolution of the subsurface fluids in the Visean of the Campine Basin (Poederlee borehole, Belgium). Journal of the Geological Society, London, 148, 1005-1017. PAPROTH, E., CONIL, R., BLESS, M. J. M. & 30 OTHERS. 1983. Bio- and lithostratigraphic subdivision of the Dinantian in Belgium: a review. Annales de la Societe Geologiquede Belgique, 106,185-239. PAPROTH, E., DUSAR M., BLESS, M. J. M. & 10 OTHERS. 1983. Bio- and lithostratigraphic subdivisions of the Silesian in Belgium, a review. Annales de la Societe Geologiquede Belgique, 106, 241-283. PATIJN, R. J. H. & KIMPE, W. F. M. 1961. De kaart van het
210
B.LAENEN£FAL.
Carboon-oppervlak, de profielen en de kaart van het dekterrein van het zuid-Limburgse mijngebied en Staatsmijn Beatrix en omgeving. Mededelingen Geologische Stichting, Serie, 44,5-12. STASSEN, P. 1970. Gestion des voies. Annales des Mines de Belgique, 69,161-199. STUFFKEN, J. 1957. De Mijngasafgifte van kolenlagen. PhD thesis, Technical University Delft. STUFFKEN, J. 1960. Bin Berechnungsverfahren zur Bestimmung der Ausgasung von Steinkohlenflozen. Bergbau-Archiv, 1 (Sonderabdruck), 40-48. TEICHMULLER, R., TEICHMULLER, M., COLOMBO, U., GAZZARINI, R, GONFIANTINI, R. & KNEUPER, G. 1970. Das Kohlenstoff-Isotopen -Verhaltnis im Methan von Grubengas und Flotzgas und seine Abhangigkeit von den geologischen Verhaltnissen. Geologische Mitteilungen, 9,181-206. VAN KEER, I, ONDRAK, R., MUCHEZ, P., BAYER, U., DUSAR, M. & VIAENE, W. 1998. Burial history and thermal evolution of Westphalian coal-bearing strata in the Campine basin (NE Belgium). Geologic en Mijnbouw, 76,301-310. VAN WIJHE, D. H. 1987. Structural evolution of inverted basins in the Dutch offshore. Tectonophysics, 137, 171-219. VANDELOISE, R. 1971. Methoden voor het opstellen van voorcalculaties van de specifieke mijngasontwikkeling in pijlers met een vlakke ligging in Belgie.
Annales des Mines de Belgique, 9, 867-898. VANDENBERGHE, N. & FOCK, W. 1989. Temperature data in the subsurface of Belgium. Tectonophysics, 164, 237-250. VANDENBERGHE, N., DUSAR, M., BOONEN, P., FAN, S. L., VOETS, R. & BOUCKAERT, J. 2000. The MerksplasBeerse geothermal well (17W265) and the Dinantian reservoir. Geologica Belgica, 3,349-367. VANSTEELANDT, P. 1993. De gevolgen van de mijnverzakkingen in Limburg. In: VANSTEELANDT, P., MOSTAERT, F., VAN Roo, J. & BROOTHAERS, L. (eds) Studiedag Geologische Kartering en Geologisch Onderzoek in het Vlaamse Gewest. Departement Economic, Werkgelegenheid en Binnenlandse Aangelegenheden, Ministerie van de Vlaamse Gemeenschap, Bestuur Natuurlijke Rijkdommen en Energie, Brussels, Belgium, 135-156. VON TRESKOW, A. 1985. Die Zusammenhange zwischen dem Gasinhalt und der Geologic im Ruhrrevier. Gluckauf, 121/23,1747-1754. WOLF, K-H., BARZANDJI, O. H. BERTHEUX, W. & BRUINING, J. 2000. CO2-sequestration in the Netherlands. CO2injection and CH4-production as related to the Dutch situation: laboratory experiments and field simulations. In: WILLIAMS, D. & HUBER, G. (eds) Proceedings of the 5th International Conference on Greenhouse Gas Control Technologies, Cairns, Australia, D5 3.
Geological sequestration of CO2 in the subsurface: lessons from CO2 injection enhanced oil recovery projects in oilfields RICHARD H. WORDEN1 & LETA K. SMITH2 Department of Earth Sciences, University of Liverpool, 4 Brownlow Street, Liverpool L69 3GP, UK (e-mail: [email protected]) 2 IHS Energy Group, 5333 Westheimer Road, Suite 100, Houston, Texas 77056, USA Abstract: In this paper production geochemical data from oil fields where CO2 has been injected to enhance oil recovery (CO2-EOR) and experimental simulations of this process are reviewed. These data show that over the timescale of days to many years, CCX, injected into the subsurface typically results in the bulk dissolution of carbonate minerals. There is little evidence for the sequestration of the injected greenhouse gas as a solid phase carbonate mineral on the timescale of the CO2-EOR projects or experiments. There is extensive aqueous geochemical, petrographic and core analysis evidence that supports the conclusion that CO2, injected into oil fields to enhance secondary recovery, leads to the bulk dissolution of calcite, dolomite and siderite. Although carbonate dissolution leads to enhanced porosity, the expected commensurate increase in permeability may be offset by the migration of clays, liberated by the action of the acidic water on the rock, with consequent blocking of pore throats. Additionally, injection of CO2 into oil fields can result in asphaltene deposition on mineral surfaces. Such a bitumen coat could ultimately isolate the mineral matrix from injected fluids and insulate the rock to the injected greenhouse gas. Localized precipitation of calcite scale has been reported in the topside facilities of CO2-EOR projects and in the low-pressure region of experimental simulations.
Subsurface storage of CO2 is seen as a key mechanism for reducing the emission of greenhouse gases to the atmosphere. It is thought that CO2 taken directly from a petroleum stream or from the exhaust gases of electricity generating stations (or industrial plants) can be stored safely in the subsurface rather than being vented to atmosphere and contributing to greenhouse gas emissions (Gale 2004). Simple logic suggests that since non-aqueous fluids can remain naturally in the subsurface (e.g. petroleum accumulations), so CO2 could also remain in the subsurface. There are several options under investigation including injection into deep aquifers, injection of CO2 into former or current oil and gas accumulations and injection into coal beds. There are numerous concerns about the injection of large quantities of CO2 into the subsurface. Petroleum gases are relatively inert when accumulated in geological reservoirs. However, it is naive to model the accumulation of CO2 in the subsurface assuming that CO2 is also inert. It is plausible that simply injecting CO2 into a saline aquifer in the subsurface might result in carbonate mineral precipitation. The reason for this process is that saline waters typically contain aqueous calcium, so that adding CO2 might lead to calcite precipitation by a reaction of the type: Calling upon such reactions to sequester a continuing stream of CO2 would rely upon an inexhaustible
supply of aqueous calcium that is clearly unreasonable, i.e. once the calcium in the water has been used up where does more calcium come from? Injected CO2 might react further with calcium bearing minerals in the subsurface by reactions of the type:
However, the dissolution of CO2 into the aquifer water can lead to the formation of an acidic solution:
Such an acidic solution is capable of causing carbonate mineral (or any mineral) dissolution by reactions of the type:
Thus, depending on conditions, CO2 injection could theoretically lead to either carbonate mineral precipitation or dissolution. Injecting CO2 into the subsurface does not have a guaranteed outcome; it need not always lead to carbonate mineral precipitation. Several attempts have been made to understand the impact of CO2 on subsurface rocks and fluids. Some of these have involved geochemical modelling whereas others have employed laboratory-based experiments intended to replicate the anticipated
From: BAINES, S. J. & WORDEN, R. H. (eds) 2004. Geological Storage of Carbon Dioxide. Geological Society, London, Special Publications, 233,211-224.0305-8719/04/$ 15.00 © The Geological Society of London 2004.
212
R. H. WORDEN & L. K. SMITH
conditions during CO2 disposal (e.g. Czernichowski et al. 1996; Gunter et al 1997; Gupta& Sass 1999). These approaches suffer from a number of problems including the globally poor knowledge of geochemical kinetics in rocks under subsurface conditions and the short timescale of laboratory experiments in comparison to the length of time that CO2 will be stored in the subsurface. One alternative approach to understanding the consequences of injecting CO2 into subsurface aquifers has been to study rocks that have experienced a natural influx of CO2 over geological timescales (e.g. Baines & Worden 2004). Such an approach is useful since it gives an insight into the very long-term consequences of CO2 injection. Another approach, adopted here, is to review the fate of CO2 used for enhanced oil recovery (EOR). CO2EOR is used to improve the extraction of liquid phase petroleum, and can have an operations history of several decades. It is thus a useful adjunct to the very long timescales revealed by geological case studies.
hances oil recovery are somewhat different in the two injection methods described above. In the WAG process, the CO2 is injected at temperature and pressure conditions that make it miscible with the oil. This does two things. First and foremost, it decreases the viscosity of the oil allowing it to flow more easily through the reservoir (Holm & Josendal 1974; Archer & Wall 1992). Second, the injection of CO2 (and water) into an oil field increases the pressure gradient between the injector and producer wells so that the petroleum will be pushed out faster (Archer & Wall 1992). This is not unique to CO2 injection; it is the basis behind most water-injection EOR projects as well. In the cyclic CO2 injection process, the CO2 is injected at conditions where it is immiscible with the oil. Although immiscible, a finite amount of CO2 still dissolves into the oil, so this process also reduces the viscosity of the oil. However, the primary mechanism of cyclic CO2 EOR is believed to be a volume increase or 'swelling' of the oil that causes it to be forced from the pores (Monger et al 1991). Some of the earliest models of CO2 injection into Why is CO2 injection used for enhanced oil the subsurface have involved the assumption that the CO2 reacts only with the oil and that the water-rock recovery? system is unaffected (Archer & Wall 1992). HowCarbon dioxide has been used to enhance oil recov- ever, production rates from an early CO^-EOR injecery for over fifty years. Both experimental and field tion scheme suffered from corrosion and scaling of data have shown the process to work, with incremen- the pumps and other facilities (Patterson 1979), It tal recovery being as high as 22% of original oil in was concluded that injection of CO2 had caused carplace (Brock & Bryan 1989; Goodrich 1980). There bonate mineral dissolution in the subsurface and that are essentially two types of CO2 injection methods. precipitation of calcite occurred as the pressure In one type, CO2 is injected at the periphery of a field dropped during production. The CO2-oil-rockwhere production has largely been depleted by water system is clearly not inert. primary means, and the oil and CO9 are swept along as a front toward production wells (Fig. la). In this process, water is usually injected alternately with the Effect of CO2 injection on water-rock CO2 (a process referred to WAG: water-alternating- interaction: field evidence gas), so as to avoid two common problems associated with CO2 injection: viscous fingering of the There have been relatively few published studies of CO2 through the reservoir and/or gravity over-ride of CO9-EOR that include data on water geochemistry the oil. Both of these reduce the sweep efficiency of following CO2 injection. Usually, an enhanced oil the CO2 through the reservoir. recovery project using CO2 is put into place when oil The other injection method is a cyclic injection- yields have fallen and the estimated recovery rates production process. In this process, the CO0 is are deemed to be unacceptably low. Sophisticated injected into a well and then the well is shut-in for CO2 injection programmes commonly follow water 2-4 weeks. Afterwards, the CO2 and oil are pro- injection (typically in a pattern flood). The result is duced back from the same well (Fig. Ib), with the oil that even when produced water samples are colproduction from the well increasing over the pre- lected for analysis following CO2 injection, such CO2 injection levels by as much as eight-fold (Deans water samples will exhibit a complex geochemistry & King 1991). The injection-production cycle is reflecting a variety of induced processes. However, generally repeated two to three times. The amount of the primary interest is in the immediate conseincreased oil recovery from each successive treat- quences of CO2 injection and assessing whether ment generally declines from the previous treatment water geochemistry reflects mineral precipitation or until it is no longer economically viable to inject mineral dissolution. Thus, although it is likely that more CO2. the results do not reveal much about natural The mechanisms by which CO2 injection en- processes (since the systems will have been cor-
CO2 STORAGE: LESSONS FROM EOR-INJECTION
213
Fig. 1. Two types of CO2 enhanced oil recovery (EOR) scheme, (a) Injection of water alternated with gas (CO2) into dedicated injection wells. The CO2-bearing fluid passes through the reservoir towards the production well, (b) Oil production from a single well is in decline. CO2 is injected into the well. The well is then shut in to allow the oil to soak up the CO2. After several weeks, the petroleum is produced again from the same well, hopefully with improved oil production rates.
rupted by pre-CO2 injection recovery practices), they may reveal data that should be of immediate relevance to CO2 sequestration. Case study 1: Hungary Early CO2 injection programs were introduced in Hungary in the late 1960s and 1970s. Data reported from these schemes are incomplete and there seem to be no tabulated data published (Table 1). However, Meneth (1983) published a review of local CO2 injection schemes and published some timeaqueous concentration variation diagrams from the Budafa Field, Pannonian Basin. It is not possible to check the quality of the data (using for example charge balance calculations) since major element aqueous geochemistry data were not fully reported. Before CO2 injection, this carbonate-cemented sandstone reservoir (Table 1) was flooded with fresh water to maintain field pressures and sweep remaining reserves to the producers. The data show that following CO9 injection, the aqueous calcium concentration increased year on year for as long as data were collected (up to seven years; Fig. 2). Conversely, aqueous bicarbonate either increased but
then decreased in later years or decreased slowly but progressively. Ede concluded that the calcium concentration changes may have been due to clay mineral exchange reactions, such as ion exchange in smectite minerals but proffered no supporting evidence for this. It is rather more likely that the changes are due to CO2-induced calcite dissolution if only because Ca-smectite is much less common than calcite (McKinley et al 2003) in sandstones. It is not possible to determine accurately the saturation state of the formation water with respect to calcite before and after CO2 injection since insufficient water geochemical data were reported. However, since the reservoir was reported to be carbonate cemented and since only aqueous calcium data were reported (Table 1), it is reasonable to assume that the water was saturated with respect to calcite prior to CO2 injection. The pattern of increasing aqueous calcium and bicarbonate indicate that calcite underwent dissolution as the acidity increased. Unfortunately, Meneth (1983) did not report the historical pattern of changes of other aqueous species so it is impossible to assess the dissolution or precipitation of other minerals. Similarly, it is not possible to deduce how much calcite dissolved as a function of volume of injected CO2 since insufficient data were reported.
Table 1, Summary of case studies and examples Case Study Authors
Location
Petroleum Reservoir Field or Porosity Temperature Cement minerals core-flood type or core-rock (%) (°C) data? lithology
Brine type
Initial Ca2+ (nigT 1 )
Post CO2flood Ca2+ (mgl- 1 )
Initial HCO3" (mgr 1 )
Post C02flood HCO, (mgr 1 )
moderate salinity, Na-Cl dominated, approx. 50000 mgl- 1 ) low to moderate salinity (approx. 30000 mgr 1 ) (l)low salinity, Na-Cl dominated, 5000 mg r1 (2) v low salinity, Ca-HCO3-SO4 dominated, lOOOmgr 1 low salinity
-30
-130
-6000
-2000
868
1036
548
1250
14
184
1182
1879
258
828
588
2361
not reported
not reported not reported not reported
not reported but very
not reported not reported not reported -800
Hungary, Europe
Meneth 1983 Budafa Field
field
38° API
sandstone
21.50
68
'carbonate cement'
Colorado,
Bowker & Shuler 1991
Rangely Field
field
oil (not defined)
sandstone
not
70
USA Wyoming,
Smith 1998
(1) Crooks Gap Field
field
35° API
sandstone
(1)23
(1)64
(2) Bonanza Field
field
35° API
sandstone
(2)20
(2)29
n/a
core flood n/a
dolostone
25
n/a
Fe-calcite, Fedolomite, illite, ± chlorite calcite, dolomite, ± quartz, kaolinite, chlorite calcite, dolomite, ± quartz, kaolinite, chlorite dolomite
core flood n/a (l)Mid. Jurassic outcrop (2)L. core flood n/a Permian core
sandstone
(1) ferroan calcite ( 1 ) brine of unspecified salinity (2)80 (2) dolomite (2) brine of (2) not reported unspecified salinity moderate salinity, 12 45 siderite and calcite Na-Cl dominated, approx. 50000 mgl-') 12 not reported Fe-calcite, Femoderate salinity dolomite, illite, ± chlorite 10-14 80 low salinity dolomite, anhydrite 10-14 80 dolomite, low-moderate anhydrite salinity
USA
Dolomite North Sea, UKCS
Omole & Osoba 1983 Ross et al. 1982
reported
sandstone
Western Canada Basin
Sayegh et al. Pembina 1990 Cardium Formation
core flood n/a
sandstone
Colorado,
Bowker & Shuler 1991
core flood n/a
sandstone
core flood n/a
sandstone
core flood n/a
sandstone
USA Wyoming,
USA
Shiraki & Dunn 2000
Weber Sandstone Formation Tensleep Formation
(1)16
(1)20
low 230
not reported not reported
low 0
-250
not reported not reported
920
1500
60
1970
-630
-750
-520
-1
-920
none detected none detected
not reported but very
-1070
CO2 STORAGE: LESSONS FROM EOR-INJECTION
215
Case study 2: Colorado, USA A rather more detailed water geochemistry dataset from a CO2-EOR scheme at the Carboniferous Rangely Field, Colorado, USA, was published by Shuler et al (1989) and Bowker & Shuler (1991). CO2 was injected as part of a water-alternating gas pattern flood into Carboniferous sandstone reservoirs in Colorado, USA (Table 1). Their related experimental data will be discussed later but the field data show a range of significant features. Following CO2 injection, not only did the aqueous bicarbonate concentration increase (as would be expected by reference to reaction R2a) but also the aqueous calcium concentration increased steadily (Fig. 3). The aqueous calcium concentration was nearly twice the pre-CO2 injection concentration after 22 months of CO2 injection into the field. It is possible to force a number of trends through the data in Figure 3 although a linear increase with time results in a statistically acceptable correlation coefficient. These authors also assessed which other aqueous species changed following CO2 injection. These data are here plotted as a function of aqueous bicarbonate concentration and confirm that calcium, magnesium, iron and, somewhat oddly, bromine, increased following CO2 injection (Fig. 4). According to data presented by Bowker & Shuler (1991) geochemical variations in other parameters such as aqueous sodium, potassium, chloride and sulphate were not altered by CO2 injection. Concentration of aluminium and silica were not reported so it is not possible to assess any consequences of CO2 injection on these species. Bowker & Shuler (1991) concluded that CO2 injection resulted in dissolution of carbonate minerals, such as ferroan calcite and ferroan dolomite, in the reservoir.
Case study 3: Wyoming, USA
Fig. 2. Downhole CO2-EOR data from two wells in Hungarian oil fields (after Meneth 1983). The data show that up to seven years of CO2 injection led to everincreasing aqueous calcium concentrations. Aqueous bicarbonate concentration data are more ambiguous since these seem to decrease with time. The increasing calcium concentrations suggest that calcite (or dolomite) is being dissolved as a result of CO2 injection.
A detailed analysis of water geochemistry and production rates following cyclic CO2 injection (Fig. 1) into Cretaceous and Carboniferous reservoir sandstones in Wyoming, USA (Table 1), were published by Smith et al (1991) and Smith (1993, 1998). Similar to the Bowker & Shuler (1991) case study, following CO2 injection the aqueous bicarbonate concentration initially increased significantly relative to the pre-CO2 injection (Figs 5 & 6). After the initial increase in bicarbonate concentration, there was then a slow progressive decrease with time as the produced fluids progressively reflected fluids that had not been affected by the CO2 injection and soak process. Aqueous calcium and magnesium concentrations followed the same pattern of initial increase and then progressive decrease as the bicarbonate and thus typically correlate very well with the
216
R. H. WORDEN & L. K. SMITH
water-rock interaction but have also simulated CO2 injection into two phase (oil and water) reservoirs.
Case study 1: Middle Jurassic and Lower Permian carbonate-cemented sandstones
Fig. 3. Downhole CO2-EOR data from Upper Carboniferous-Permian reservoirs in the USA (after Bowker & Shuler 1991). The data show that up to two years of CO9 injection have led to constantly rising aqueous calcium concentrations even after the CO2 reached saturation and formed a separate phase (indicated by the time when gas-phase CO2 breaks through).
bicarbonate (Figs 5 & 6). The pH of the water was somewhat variable and did not correlate well with the other aqueous geochemical parameters. Thus although dissolution and dissociation of CO2 should lead to the generation of both aqueous bicarbonate and acidity, the hydrogen ions must be involved (used) in other geochemical processes (for example R2b). Smith et al (1991) and Smith (1993, 1998) observed similar patterns of increased calcium and magnesium concentrations in water produced in four other cyclic CO2 injection projects, although geochemical analyses were insufficiently detailed to examine correlations with other parameters. The dataset in Smith (1998) is more thorough than others published since it includes aqueous silica and aluminium. Aqueous silica increased following CO2 injection and then proceeded to decrease in the same way as aqueous calcium and magnesium (Fig. 7a, b). Like calcium, aqueous silica shows a good correlation with bicarbonate (Fig. 7c, d). Conversely, aluminium had a concentration that peaked much later than the other aqueous species and showed no simple correlation with other aqueous species, including bicarbonate (Fig. 7).
Effect of CO2 injection on water-rock interaction: experimental evidence To assist with trying to understand the consequences of CO2 injection on reservoir rocks, a number of experimental studies have been undertaken. These have attempted to simulate simple CO2-rich
Ross et al. (1982) simulated subsurface conditions during CO9 injection into a range of carbonate cemented North Sea sandstones (Table 1). They compared the effects of flooding with brine and the effects of flooding with brine plus CO2. They showed that significant carbonate cement dissolution occurred directly as a result of CO2 addition to the system whereas injection of brine alone caused negligible dissolution. Core analysis measurement revealed that the rocks had higher permeability after CO2 injection although porosity was little changed. They studied the core material using optical thin section petrography before and after the CO2 injection and observed that carbonate cement had been dissolved. Mercury-injection porosimetry showed that the biggest pores had been enlarged although the smallest pores seemed to be unchanged in size. In some cases the cores developed high permeability streaks due to CO2-induced carbonate dissolution that increased the flow rate of the CO2-rich brine and thus reduced the contact time between fluid and rock. Indeed, the experiments revealed decreasing calcium and magnesium concentrations with increasing cumulative core flood volume due to the CO2-brine mixture passing through the dissolution-induced high permeability streaks and avoiding contact with the bulk of the rock. Flowing CO2 and brine through cores containing residual oil slowed, but did not stop, the rate of carbonate dissolution in comparison to core samples without oil. These seminal results show that CO2 injected into a brine-bearing rock will induce dissolution of any carbonate cements present in the rock. The development of a dual porosity system has potentially great significance for the way that the CO2 would move in the subsurface.
Case study 2: dolostone core Omole & Osoba (1983) used dolostone cores of unspecified origin, previously saturated with low salinity brine, to show that injected CO2 led to dissolution of the core material (Table 1). Their focus was on the increase in permeability although petrography of the core samples after the experiment showed that the rock had undergone dissolution. Injection of CO2 led to dolomite dissolution. They also showed that dolomite was reprecipitated if the pressure dropped although they did not state whether the precipitation was a function of total pressure or the partial pressure of CO2.
CO, STORAGE: LESSONS FROM EOR-INJECTION
217
Fig. 4. Downhole CO2-EOR data from Carboniferous reservoirs in the USA (after Shuler et al. 1989, Bowker & Shuler 1991). Aqueous concentration data for (a) calcium, (b) magnesium, (c) iron and (d) bromine have been plotted against bicarbonate concentrations. Bicarbonate increased with time since CO2 injection commenced (as does Ca, see Fig. 3) so that the injection led to increases in common divalent metals, as well as the enigmatic rise in the halide, bromine.
Case study 3: Pembina Cardium Formation core Sayegh et al. (1990) circulated CO2-saturated brines through Cardium Formation core samples (Table 1) and observed that calcium increased from the initial Oppm to an ultimate value of 250 ppm. The increase was assigned to the dissolution of carbonate minerals. They used petrographic (SEM) and XRD analysis to confirm that calcite and siderite had dissolved during the experiment. They also recorded that, after an initial permeability decrease (thought to be due to fines-migration), the permeability increased as the experiment progressed.
Case study 4: Weber Formation ferroan carbonate-cemented sandstone Shuler et al. (1989) and Bowker & Shuler (1991) reported the results of experimental simulation of CO2
injection (Table 1) as well as the empirical field data discussed previously. The CO2 injection was initially miscible (one phase) but they increased the CO2 concentration progressively and simulated CO2 breakthrough as a separate phase. They injected CO^-brine mixtures in various ratios and at various rates into dolomite-cemented Upper Carboniferous sandstones from the same reservoir unit that was tested for the effects of CO2 injection on water geochemistry (see earlier). The aqueous bicarbonate content was strongly affected by the percentage of CO2 in the brine-CO2 mixture (Fig. 8). Some aqueous species, for example sodium, were unaffected by the process. Other aqueous species including calcium, magnesium, iron and silica were strongly affected by the CO2 content of the fluid (Fig. 8), SEM analysis showed that dolomite cement had been partially dissolved as a result of CO2 injection. In some experiments, following CO2 addition, CO2-free brine was flushed through the cores. The outflow aqueous concentrations of Ca2+, Mg2+, HCO3~ and aqueous SiO2 were higher
218
R. H. WORDEN & L. K. SMITH
Fig. 5. Downhole CO2-EOR data from the Crooks Gap Carboniferous reservoir in the USA (after Smith 1998). Aqueous concentration data have been plotted as a function of time since CO2 production started for (a) magnesium, (b) calcium, and (c) bicarbonate. Data have also been plotted as a function of bicarbonate concentration for (d) pH, (e) calcium and (f) magnesium. After an initial increase due to CO2 injection and carbonate mineral dissolution, magnesium, calcium and bicarbonate fall progressively as the produced water increasingly derived from a part of the reservoir unaffected by the injected CO2.
CO7 STORAGE: LESSONS FROM EOR-INJECTION
219
Fig. 6. Downhole CO7-EOR data from the Bonanza Carboniferous reservoir in the USA (after Smith 1998). Aqueous concentration data have been plotted as a function of time since CO2 production started for (a) magnesium, (b) calcium, and (c) bicarbonate. Data have also been plotted as a function of bicarbonate concentration for (d) pH, (e) calcium and (f) magnesium. Note that the Bonanza Field had more saline water than the Crooks Gap field in Figure 5. After an initial increase due to CO2 injection and carbonate mineral dissolution, magnesium, calcium and bicarbonate fall progressively as the produced water was increasingly derived from a part of the reservoir unaffected by the injected and soaked CO2.
220
R. H. WORDEN & L. K. SMITH
Fig. 7. Downhole CO2-EOR data from two Carboniferous reservoirs in the USA (after Smith 1998). Aqueous concentration data have been plotted as a function of time since CO2 production started for aluminium and silica from the (a) Crooks Gap and (b) Bonanza oil fields. Silica displays a similar pattern to the calcium, magnesium and bicarbonate suggesting that a silica-bearing mineral underwent dissolution in a similar way to carbonate. Aluminium concentration evolved differently with a punctuated yet progressive increase for the Bonanza field and a late peak and then decrease for the Crooks Gap field, (c and d) Silica and aluminium concentration data plotted as a function of bicarbonate concentration Crooks Gap and Bonanza oil fields.
than inflow showing that CO2-induced dissolution continued to have an effect even after the CO2 injection had ceased. This presumably shows that micropores contained raised levels of dissolved CO2 that continued to cause carbonate cement dissolution even after the primary addition of CO2 had ceased. Bowker & Shuler (1991) and Shuler el al (1989) showed that elevated calcium and magnesium occurred irrespective of the flow rates and flow volumes.
Case study 5: Tensleep Formation dolomite and anhydrite-cemented sandstones Shiraki & Dunn (2000) injected CO2-brine mixtures into dolomite and anhydrite cemented Carbonifer-
ous sandstones at 80 °C and reservoir pressure and analysed a wide range of solutes (Table 1). From the effluent water geochemical data, they showed that dolomite dissolved and K-feldspar was altered to kaolinite during the experiments. Some species, such as sodium and chloride were unaffected by the treatment. SEM micrographs unequivocally show dissolution fabrics. Dolomite dissolution occurred at a constant rate during the experiments whereas Kfeldspar dissolution was interpreted to have slowed during the experiment. Aqueous sulphate concentration in those experiments that used brine that was undersaturated with respect to anhydrite. The addition of CO2 during the injection of high sulphate brines had no effect on anhydrite since the brines were already saturated with respect to anhydrite.
CO, STORAGE: LESSONS FROM EOR-INJECTION
221
Fig. 8. Experimental CO2-EOR data (after Shuler et al. 1989; Bowker & Shuler 1991). Aqueous concentration data for (a) sodium, (b) iron, (c) bicarbonate, (d) magnesium, (e) calcium, (f) silica, have been plotted as a function of the amount of CO2 added to the brine during the core flood. Sodium is totally unaffected by the treatment whereas all the other species increased. These data imply that carbonate and silicate minerals have dissolved due to flooding the core with CO2. The data show that previous CO2-induced dissolution processes have affected the final water geochemistry.
222
R. H. WORDEN & L. K. SMITH
Core analysis before and after the CO2-brine injections revealed that porosity tended to increase slightly (up to 9% relative increase). Permeability, though, tended to decrease despite the increased porosity although one sample showed a significant relative increase in permeability. The predominant decrease in permeability was thought to be due to fines-migration and blocking of pore throats with clay minerals released by the dissolution of other minerals. Shiraki & Dunn (2000) modelled their results to simulate the effects of the injection of CO2 on the rock around an EOR injector well and concluded that the injected brine-CO2 mixture would be saturated with dolomite (thus dissolving no more dolomite) within about a 3m radius. However, they ignored the dynamic nature of their system since, as dolomite would be progressively dissolved and removed from the near-well bore region, the dissolution would take place in progressively more distant locations from the well bore.
Evidence of effect of CO2 injection on petroleum-rock interaction from core-flood data Injecting CO2 into oil-bearing reservoirs can lead to changes to the solubility of asphaltenes in the oil. Increasing the gas:oil ratio (GOR) by increasing the methane content of a petroleum system routinely causes asphaltene fall-out. Monger & Fu (1987), Wolcott et al (1989) and Monger & Trujillo (1991) injected oil-CO2 mixtures into brine-bearing cores and studied the effect of CO2 on asphaltene precipitation on mineral surfaces. They observed that the amount of asphaltene deposited on grain surfaces was a function of the amount of asphaltene dissolved in the primary oil. Asphaltene precipitation occurred when CO2 was added to the fluid but the amount depended on the quantity of asphaltene originally dissolved in the oil. The injection of oil-CO2 mixtures led to the deposition of higher molecular weight compounds than the injection of oil alone. The site of asphaltene deposition was mineral-specific with clay minerals favouring the formation of the organic film over most other minerals although organic-bearing calcite ooids also experienced preferential asphaltene deposition. One of the major consequences of CO2 injection into oil-bearing rocks is that grains will become coated with bituminous coatings. These may serve to isolate mineral grains from the reactive fluid as well resulting in the rock becoming increasingly oil-wet as CO2 injection proceeds.
Discussion of the CO2-EOR data The elementary reactions listed earlier (Rla & b) suggest that adding CO2 to calcium-rich brine or a
rock rich in calcium-bearing aluminosilicate minerals will lead to calcite precipitation: the ultimate benefit being the locking up of CO2 as a solid mineral in the subsurface. However, adding CO2 to subsurface brines will lead to dissolution and then dissociation to form an acidic solution (R2a & b). The acidic solution is capable of dissolving calcite or other carbonate minerals. So what happens, calcite (or dolomite) precipitation or dissolution? An answer comes clearly from the CO2-EOR results discussed above. From both the months to year timescales of the downhole data and shorter timescale of the experimental data, adding CO2 to carbonate mineral-bearing sandstones that contain saline brine causes dissolution of the carbonate. Neither the downhole data nor the core flood experimental data lead to the expectation that injecting CO2 for sequestration purposes will result in solid phase precipitation of the CO2 as a carbonate mineral. Thermodynamic analysis previously led to the conclusion that CO2 might be sequestered as carbonate minerals. Gunter et al. (1993) modelled CO2 sequestration by quartz-anorthite sandstones. Gunter et al (1997, 2000) simulated CO2 injection into quartz-feldspar-kaolinite-biotite-(Fe, Mg, Ca) carbonates. Wu et al (2001) simulated the reaction of CO2 with wollastonite (CaSiO3 pyroxene). Gunter et al (1997) concluded that timescales of 10000-100000 years would be necessary to allow solid phase sequestration of CO2 to occur if it were injected into reactive sandstone. The downhole CO2-EOR data (Figs 2-7 after Ede 1983; Shuler et al 1989; Bowker & Shuler 1991; Smith 1998) reveal the clearest picture of what will happen when CO2 is injected into the subsurface as a means of disposal or storage. All three field-based case studies explored above lead to the conclusion that carbonate mineral dissolution is the predominant result of injecting CO2 into rocks that contain carbonate minerals. The large increase in calcium (and magnesium) concentrations with increasing bicarbonate shows that the dissolution of minerals occurs much faster than the reaction of CO2 with any alkaline earth silicates or even with Fe-rich clay minerals. The much smaller absolute increase in silica concentration relative to calcium, magnesium and bicarbonate are most likely to be due to the slower dissolution rate of silicates relative to carbonates. The evidence from the downhole CO2-EOR projects is corroborated by experimental data. The data from core-flood experiments have also shown that dissolution of carbonate and other minerals occur. One of the main physical consequences of the dissolution is a minor increase in porosity. However, perhaps contrary to simple notions of the relationship of permeability to porosity, the dissolution does not always lead to increased permeability. In some cases, permeability decreased in the experiment,
CO, STORAGE: LESSONS FROM EOR-INJECTION
probably due to liberated clay minerals and other fine grains migrating under the fluid pressure gradient and blocking pore throats (e.g. see Stonecipher 2000). The analysis of the downhole CO2-EOR data also shows that carbonate dissolution occurs much faster than silicate mineral dissolution. Thus the pair of carbonate dissolution reactions R2a and b occurs much faster than silicate dissolution and reaction processes Rla and b. Not one of the field data sets discussed above is detailed enough to permit an analysis of CO2-rock ratios and their influence on water geochemistry. There are not enough data presented to reveal a quantitative picture of the variable effects of a given amount of CO2 on the extent of carbonate mineral dissolution. Moreover, to gain a quantitative understanding of the rate of carbonate dissolution (e.g. in typical kinetic terms such as mol.cm~ 2 .s~') it would be necessary to have data on the CO2-rock ratio, the porosity of the rock, the proportions of different minerals, and the specific surface area of the carbonate minerals that underwent dissolution. It would also be necessary to have an appreciation of the presence of high permeability streaks in the rock and information on the natural variability of all the controls on mineral dissolution. If all these data were available, they would be the best way to gain kinetic data for use in forward predictive models of the effects of CO2 injection for greenhouse gas storage projects.
Scale formation in topside facilities at CO2EOR facilities During the early stages of the CO2-EOR project at SACROC excess calcite scale was an unfortunate byproduct of production following CO2 injection (Patterson 1979). Exacerbated development of scale on downhole submersible pumps, by design at a much lower pressure than the formation, was a significant problem. This was also observed during a cyclic-CO2 treatment conducted at Cole Creek field, Wyoming USA. In that example, the downhole pump scaled so badly that it was completely destroyed within a matter of a few weeks (Smith 1993, 1998; Smith et al 1991). It is likely that the injected CO2 initially dissolved carbonate minerals in the reservoir thus increasing both the aqueous calcium and bicarbonate concentrations. As the CO2-affected fluid flowed into the production well, the pressure would have fallen, causing a reduction in the solubility of CO2 and inducing precipitation of calcite. Some experimental simulations of CO2 injection have recorded precipitation of calcite scale where fluid pressures have fallen (Omole & Osoba 1983). The quantity of calcite precipitation was found to be a function of the magnitude of the pressure drop.
223
Implications for geological storage of CO2 by injection into the subsurface The CO2-EOR projects allow us to make some general comments of relevance to CO2 sequestration. (1)
(2) (3)
(4) (5)
(6)
Downhole EOR-CO2 data, collected over a period of months to years, show that dissolution of calcite, dolomite and siderite should be expected when CO2 is injected into rocks that contain these ubiquitous carbonate minerals. Dissolution of feldspar minerals may occur, but typically at a slower rate than the dissolution of carbonate minerals. Dissolution is likely to be most extensive nearest to injector wells where pH values are most affected and where cumulative fluid-rock ratios are highest. Solid-phase sequestration of CO2 as carbonate minerals within the reservoir is not evinced from any of the EOR-CO2 data. Core-flood experiments have shown that minor enhancement of the porosity and permeability of the host rock may occur due to carbonate mineral dissolution, especially in the near well-bore region but spread throughout the rock volume in long term injection projects. Mineral dissolution may be heterogeneous and lead to very high permeability streaks. Permeability may be degraded by fines migration from clays that are released by the action of the CO2 on the rock matrix. This may occur several metres from the CO2 injector well and may eventually slow or prevent further CO2 injection.
We would like to thank the reviewers Craig Smalley and Beverly Saylor and editor Shelagh Baines for thoughtful comments that led to an improved manuscript.
References ARCHER, J. S. & WALL, P. G. 1992. Petroleum Engineering. Principles and Practice. Graham and Trotman. 362 p. BAINES, S.J. & WORDEN, R. H. 2004. The long-term fate of CO2 in the subsurface: natural analogues for CO2 storage. In: BAINES, S. J. & WORDEN, R. H. (eds) Geological Storage of Carbon Dioxide. Geological Society, London, Special Publications, 233,59-85. BOWKER K. A. & SHULER P. J. 1991. Carbon dioxide injection and resultant alteration of the Weber sandstone, Rangely Field Colorado. American Association of Petroleum Geologists Bulletin, 75,1489-1499. BROCK, W. R. & BRYAN, L. A., 1989, Summary results of CO2 EOR Field Tests, 1972-1987. SPE 18977. Proceedings of the Society of Petroleum Engineers Joint Rocky Mountain Regional/Low Permeability Reservoirs Symposium and Exhibition. Denver, Colorado, March 6-8,1989.
224
R. H. WORDEN & L. K. SMITH
CZERNICHOWSKI-LAURIOL, L, SANJUAN, B., ROCHELLE, C., BATEMAN, K., PEARCE, J. & BLACKWELL, P. 1996. Analysis of the geochemical aspects of the underground disposal of CO2. In: APPS, J.A. & TSANG, C.F. (eds) Deep injection disposal of hazardous and industrial waste. Academic Press Inc, 565-583. DEANS, H. A. & KING, R. A. 1990. The Crooks Gap huff'n'puff project. Proceedings of the 6th Wyoming Enhanced Oil Recovery Symposium. Casper, Wyoming, 1 May 1991. GALE, J. 2004. Why do we need to consider geological storage of CO2? In: BAINES, S. J. & WORDEN, R. H. (eds) Geological Storage of Carbon Dioxide. Geological Society, London, Special Publications, 233,7-15. GOODRICH, J. H. 1980. Review and analysis of past and ongoing carbon dioxide injection field tests: SPE 8832, 6th Society of Petroleum Engineers-DOE Symposium on Improved Oil Recovery, Tulsa, OK, April 20-23,1980. GUNTER, W. D., PERKINS, E. H. & MCCANN, T. J. 1993. Aquifer disposal of CO2-rich gases: reaction design for added capacity. Energy Conversion Management, 34,941-948. GUNTER, W. D., WIWCHAR, B. & PERKINS, E. H. 1997. Aquifer disposal of CO2-rich greenhouse gases: extension of the time scale of experiment for CO2sequestering reactions by geochemical modelling. Mineralogy and Petrology, 59,121-140. GUNTER, W. D., PERKINS, E. H. & HUTCHEON, I. 2000. Aquifer disposal of acid gases: modelling of water-rock reactions for trapping acid wastes. Applied Geochemistry, 15,1085-1095. GUPTA, N. & SASS, B. 1999. Hydrodynamic and geochemical modeling for carbon dioxide sequestration in deep saline formations. US DOE report for project DEAF26-98FT00874. HOLM, L. W. & JOSENDAL, V. A. 1974. Mechanisms of oil displacement by carbon dioxide. Journal of Petroleum Technology, December 1974. MCKINLEY J. M., WORDEN R. H. & RUFFELL, A. H. 2003. Smectite in sandstones: a review of the controls on occurrence and behaviour during diagenesis. In: WORDEN, R.H. & MORAD, S. (eds) Clay mineral cement in sandstones. International Association of Sedimentologists, Special Publication, 34,109-128. MENETH, E. 1983. Main results of field experiment establishing exploitation with carbon dioxide in Hungary. International symposium on CO2 enhanced oil recovery. Budapest, Hungary, March 8-11, 327-352. MONGER, T. G. & Fu, J. C. 1987. The nature of CO2induced organic deposition. Proceedings of the 1987 Society of Petroleum Engineers Annual Technical
Conference, 147-159 (SPE 16713). MONGER, T. G., RAMOS, J. C. & THOMAS, J. 1991. Light oil recovery from cyclic CO2 injection: influence of low pressures, impure CO2 and reservoir gas. SPE Reservoir Engineering, 6,25-32. MONGER, T. G. & TRUJILLO, D. E. 1991. Organic deposition during CO2 and gas rich flooding. SPE Reservoir Engineering, 6,17-24. OMOLE, O. & OSOBA, J. S. 1983. Carbon dioxide-dolomite rock interaction during CO2 flooding process. Paper 83-34-17 Petroleum Society Canadian Institute for Mining and Metallurgy, 1-13, Banff, Alberta. PATTERSON, K. W. 1979. Fighting downhole corrosion and scale in flood CO2 at SACROC. Petroleum Engineering International, 50, 36—44. Ross, G. D., TODD, A. C., TWEEDIE, J. A. & WILLS, A. G. S. 1982. The dissolution effects of CO2-brine systems on the permeability of UK and North Sea calcareous sandstones. SPE 10685. SAYEGH, S. G., KRAUSE, F. F, GIRARD, M. & DABREE, C. 1990. Rock/fluid interactions of carbonated brines in a sandstone reservoir, Alberta, Canada. SPE Formation Evaluation, 5, 399-405. SHIRAKI, R. & DUNN, T. L. 2000. Experimental study on water-rock interactions during CO2 flooding in the Tensleep Formation, Wyoming, USA. Applied Geochemistry, 15, 265-279. SHULER, R J., FREITAS E. A. & BOWKER, K. A. 1989. Selection and application of barium sulfate scale inhibitors for a carbon dioxide flood, Rangely Weber sand unit, Rangely Colorado. SPE 18973. SMITH, L. K. 1993. Aspects of oilfield water chemistry, PhD Dissertation, University of Wyoming, 290pp. SMITH, L. K. 1998. Carbonate cement dissolution during a cyclic CO2-enhanced oil recovery treatment. In: MORAD, S. (ed.) Carbonate cements in sandstones. International Association of Sedimentologists Special Publication, 26,483-499. SMITH, L. K. MACGOWAN D. B. & SURDAM R. C. 1991. Scale prediction during CO2 huff 'n 'puff enhanced oil recovery, Crooks Gap Field, Wyoming, SPE 21838. STONECIPHER, S. A. 2000. Applied sandstone diagenesis practical petrographic solutions for a variety of common exploration, development and production problems. Society of Economic Mineralogists and Paleontologists Short Course Notes, 50. WOLCOTT J. M., MONGER T. G., SASSEN, R. & CHINN, E. W. 1989. The effects of CO2 flooding on reservoir mineral properties. SPE 18467. Wu, J. C. S., SHEEN, J. D., CHEN, S. Y. & FAN, Y. C. 2001. Feasibility of CO2 fixation via artificial rock weathering. Industrial Engineering Chemistry Research, 40, 3902-3905.
Acid-gas injection in the Alberta basin, Canada: a CO2-storage experience STEFAN BACHU1 & WILLIAM D. GUNTER2 1
Alberta Geological Survey, 4999-98 Avenue, Edmonton, AB, T6B 2X3, Canada (e-mail: [email protected]) 2 Alberta Research Council, 250 Karl Clark Road, Edmonton, AB, T6N1E4, Canada Abstract: Over the past decade, oil and gas producers in the Alberta basin have been faced with a growing challenge to reduce atmospheric emissions of hydrogen sulphide (H2S) that is produced from 'sour' hydrocarbon pools. Since surface desulphurization is uneconomic, increasingly operators are turning to acid-gas disposal by injection into deep geological formations. Acid gas, a mixture of hydrogen sulphide and carbon dioxide (H2S and CO2), is the by-product of 'sweetening' sour hydrocarbons. Although the purpose of the acid-gas injection operations is to dispose of H2S, significant quantities of CO2 are also being injected because it is uneconomical to separate the two gases. The acid-gas injection operations in the Alberta basin represent an analogue to geological sequestration of CO2. Large-scale injection of CO2 into depleted oil and gas reservoirs and into deep saline aquifers is one of the most promising methods of geological sequestration of CO2, and in this respect it is no different from acid-gas disposal operations. However, before implementation of greenhouse-gas geological sequestration, a series of questions need to be addressed; the most important ones relate to the short- and long-term fate of the injected CO2. Thus, the study of the acid-gas injection operations in Alberta provides the opportunity to learn about the safety of these operations and about the fate of the injected gases, and represents a unique opportunity to investigate the feasibility of CO2 geological sequestration.
A significant portion of raw natural gas production in the Alberta basin, Canada, contains varying percentages of acid-gas components, specifically hydrogen sulphide (H2S) and carbon dioxide (CO2). To meet pipeline and sales gas specifications, this raw sour natural gas must pass through gas processing facilities to separate out the acid-gas components. Furthermore, a number of sour-oil pools produce solution gas that contains significant amounts of H2S and CO2 (e.g. 4% and 17% respectively at the Wayne -Rosedale site in Alberta; Ho et al 1996). Prior to 1988, if sulphur-recovery technology could not economically remove the sulphur from the acid gas, the alternative was to burn the acid gas in flare stacks or incinerators. Each flaring source was permitted to emit up to lOt/d of sulphur. To reduce the waste and pollution resulting from sour-gas flaring, since 1989 regulatory agencies in Alberta and British Columbia have required that gas plants with a sulphur throughput of more than 1 t/d recover the sulphur from the gas stream (Longworth et al. 1996). The amount of mandated recovered sulphur increases from 70% for plants producing 1-5 t/d, to 96% for plants producing more than 10t/d. Nevertheless, even with sulphur recovery in place, significant volumes of tail, or acid gas, are generated, and this is often flared from stacks at the well sites. In the early 1990s, the yearly average emission from gas flaring in the Alberta basin was in the order of 0.3 Mt SO2 and 6Mt CO2 (Long worth etal. 1996).
Until the late 1980s and early 1990s, sulphur recovery and acid-gas flaring were considered the most economic ways of dealing with the sulphur in sour and acid gases. As a result of public concern about flaring, including environment degradation, human and animal health, and waste, the regulatory agencies increasingly require that operators reduce emissions, particularly in regions where the public concern is high. Because the world supply of sulphur is dominated by recovered sulphur and environmental pressure has driven desulphurization to the point that the supply has exceeded demand, large tonnages of sulphur throughout the world are being poured into block storage above or immediately below the ground (Connock 2001), including the Alberta basin. Since desulphurization is uneconomical in a weak sulphur market, and flaring is publicly unacceptable, operators in the Alberta basin are increasingly turning to acid-gas disposal through injection into deep subsurface formations, mostly in saline aquifers, but also in depleted oil and gas reservoirs. The on-site geological disposal is particularly attractive if, alternatively, operators have to build their own sulphur recovery plant or a pipeline to the nearest plant with available capacity - each alternative having its own difficulties or undesirable effects, such as cost, economic risk, pipeline right of way, and plant proliferation (Keushnig 1995). The number of acid gas disposal sites in the Alberta basin grew from one site in 1989, to four
From: BAINES, S. J. & WORDEN, R. H. (eds) 2004. Geological Storage of Carbon Dioxide. Geological Society, London, Special Publications, 233,225-234.0305-8719/047$ 15.00 © The Geological Society of London 2004.
226
S. BACHU & W. D. GUNTER
Fig. 1. Location of acid-gas injection sites in the Alberta basin at the end of 2002.
sites in 1994, to 25 sites in 1998, and 39 active sites at the end of 2002. Of the 42 approved operations at the end of 2002, one was never implemented by the operator. Two other operations have been rescinded after an active period, either because the injection volumes reached the approved limit, and the operator moved to another site or injection formation, or because the gas plant producing the acid gas has been decommissioned, with the sour gas being processed at another, nearby plant. At an operation with multiple sites, at Zama in northern Alberta, injection into two reefal carbonate reservoirs was shut down in 1998 and 2000 by the regulatory agency because the operator had greatly exceeded the approved injection volume or pressure. Figure 1 presents the location and status of acid-gas injection operations in western Canada at the end of 2002. The composition of the injected acid gas varies from site to site, from 85% H2S and 15% CO2 to 2% H2S and 98% CO2. These sites are located mostly in the region close to the Rocky Mountain Thrust and Fold Belt, which corresponds to sour-gas production from deep reservoirs (see Fig. 1). At 24 sites the acid gas is injected into a deep saline aquifer, at 15 sites it is injected into a depleted oil or gas reservoir, and at four sites in the water leg underlying a reservoir. Because acid gas is miscible with oil, it may be used in conjunction with a miscible-flood scheme to
enhance the oil recovery (Davison et al. 1999). Also, acid gas may increase oil or gas recovery if it is injected to maintain reservoir pressure (Longworth et al 1996; Connock 2001). Site selection and approval must meet a set of requirements imposed by regulatory agencies (Keushnig 1995; Longworth et al. 1996). As a result, the sites also meet the criteria for and are located in areas suitable for CO2 sequestration in geological media in the basin (Bachu 2000; Bachu & Stewart 2002). At the end of 2000, the last year with complete records at the time of writing, the cumulative injection rate for acid-gas injection in the Alberta basin reached 0.26Mt/a CO2 and 0.20Mt/a H2S, with a total of 1.1 Mt CO2 and 0.6 Mt H2S injected, driven by the need to dispose of H 2 S(seeFig.2). In the context of current debates regarding global climate change as a result of anthropogenic greenhouse gas emissions, of which CO2 is the most important, the acid-gas injection operations in the Alberta basin represent an analogue to geological sequestration of CO2. Large-scale injection of CO2 into depleted oil and gas reservoirs and into deep saline aquifers is one of the most promising methods of geological sequestration of CO2, and in this respect it is no different from acid-gas disposal operations. However, before implementation of greenhouse gas geological sequestration, a series of questions need addressing, the most important
ACID-GAS INJECTION
227
ones relating to the short- and long-term fate of the injected CO2. Thus, the acid-gas injection operations in the Alberta basin provide the opportunity to learn about the safety of these operations and about the fate of the injected gases, representing a unique opportunity to investigate the feasibility of CO2 geological disposal. In the 13 years since the first operation in the world started injecting acid gas into a depleted reservoir on the outskirts of the city of Edmonton in Alberta, no safety incidents have been reported. Given the nature of H2S, which is more toxic and corrosive than CO2, the success of these acid-gas injection operations indicate that the technology for CO2 geological storage is in a mature stage and ready for large-scale deployment.
Review of operations In Alberta, applications for acid gas disposal need to conform to the specific requirements listed in Chapter 4.2 of Guide 65 that deals with applications for conventional oil and gas reservoirs (AEUB 2000). Requirements in BC are modelled after those in Alberta. The selection of an acid-gas injection site needs to address various considerations which relate to: proximity to sour oil and gas production that is the source of acid gas; confinement of the injected gas; effect of acid gas on the rock matrix; protection of energy, mineral and groundwater resources; equity interests; wellbore integrity and public safety (Keushnig 1995; Longworth et al 1996). The surface operations and the subsurface aspects of acid-gas injection depend on the properties of the H2S and CO2 mixture, which include, but are not limited to non-aqueous phase behaviour, water content, hydrate formation and the density and viscosity of the acid gas (Carroll & Lui, 1997; Ng et al. 1999). Acid gas properties The acid gas obtained after the removal of H2S and CO2 from the sour gas may also contain 1-3% hydrocarbon gases, and is saturated with water vapour in the range of 2-6%. Unlike the case of hydrocarbon gases, for which water solubility decreases with increasing pressure, the solubility of water in both H2S and CO2, hence in acid gas, decreases as pressure increases up to 3-8 MPa, depending on temperature, after which it increases dramatically (Fig. 3). The solubility minimum reflects the pressure at which the acid gas mixture passes into the dense liquid phase form, where the solubility of water can increase substantially with increasing pressure due to the molecular attraction
Fig. 2. Cumulative amount (all sites) of CO2 and H2S injected in deep aquifers and reservoirs in the Alberta basin, Canada, to the end of 2000: (a) injection rate; (b) mass.
between these polar compounds (Wichert & Royan 1996, 1997). The ability of acid-gas to hold water increases with temperature and decreases with the addition of small amounts of methane (Fig. 3). This property of the acid gas mixture is used in dewatering the acid-gas to avoid pipe and well corrosion. The acid gas is usually compressed from about lOOkPa to around 8-10MPa for injection (Wichert & Royan 1996,1997) and the water content is generally reduced to less than half a mole %. Although there are not many published properties of the acid-gas mixture, the properties of pure CO2 and H2S have been thoroughly examined and reported. In their pure state, CO2 and H2S have similar phase equilibria, but at different pressures and temperatures (Carroll 1998a). They exhibit the normal vapour/liquid behaviour with pressure and temperature (Fig. 4), with CO2 condensing at lower temperatures than H2S. Methane (CH4) also exhibits this behaviour, but at much lower temperatures. The phase behaviour of the acid-gas binary
228
S. BACHU & W. D. GUNTER
mize storage and minimize risk, the acid gas needs to be injected: (1) in a dense-fluid phase, to increase storage capacity and decrease buoyancy; (2) at bottom-hole pressures greater than the formation pressure, for injectivity; (3) at temperatures in the system generally greater than 35 °C to avoid hydrate forming, which could plug the pipelines and well; and (4) with water content lower than the saturation limit, to avoid corrosion.
Surface operations
Fig. 3. Solubility of water in acid gas as a function of pressure for: (a) different acid-gas composition (CO2 and H2S) at 30 °C; (b) different temperatures for an acid gas with a composition of 49% CO2, 49% H2S and 2% CH4 (see also Lock 1997; Wichert & Royan 1996, 1997).
system is represented by a continuous series of twophase envelopes (separating the liquid and gas phases) located between the unary bounding systems in the pressure-temperature space (Fig. 4). If water is present, both CO2 and H2S form hydrates at temperatures up to 10°C for CO2 and more than 30°C for H2S (Carroll & Lui 1997). If there is too little water, the water is dissolved in the acid gas and hydrates will generally not form. However, phase diagrams show that hydrates can form without free water being present (Carroll 1998a,&), thus operating above the hydrate-forming temperature is desirable. The properties of the acid-gas mixture are important in facility design and operation because, to opti-
The sweetening of acid gas takes place at gasprocessing plants where the natural gas is processed to reach pipeline and sales specifications. After separation of any gas liquids, the sour gas is flowed through a contactor or absorber tower in a one- or two-stage process where the sour gas typically comes in contact with an amine solution. The amine reacts with the acid gas, dissolving it, trapping it in the water phase and allowing the sweet components of the natural gas to pass on through. By the time the natural gas reaches the top of the separation tower, all of the CO2 and H2S should have been absorbed by the amine solution. The gas is now sweet, meeting pipeline and market specifications for allowable acid-gas content. The gas-saturated amine solvent is collected from the bottom of the tower and is regenerated by heating the aqueous solution to approximately 170 °C. The water-saturated acid-gas stream leaves the regeneration unit at 35-70kPa and must be cooled and then compressed in excess of the subsurface storage formation pressure for injection. Typically, four stages of compression are required to provide the required discharge pressure (Fig. 4). During compression, the acid gas mixture heats up and, therefore, cooling is required after each stage of compression. Dehydration may be necessary depending on the final discharge pressure. Once the acid gas is compressed, it is transported through a pipeline to the injection wellhead usually at a short distance from the gas plant. The well consists of a central steel tubing string with an outer annulus bounded by a steel casing that is cemented to the subsurface formations. The acid gas flows down the well tubing and into the subsurface storage formation through perforations in the well casing. The discharge pressure from the compressor must exceed the sum of: formation pressure minus the pressure-head in the well tubing (weight of the fluid column), the pressure loss or gain due to elevation change from the compressor to the well head, and all the pressure losses along the way from the compressor into the formation (due to friction in the pipeline to the wellhead and in the tubing string, and across perforations). Ultimately, in
ACID-GAS INJECTION
229
Fig. 4. Phase diagrams for methane (CH4), carbon dioxide (CO2), hydrogen sulphide (H2S) and a 50%-50% acid-gas mixture; hydrate conditions for CO2 and H2S; and a four-stage acid-gas compression cycle (after Wichert & Roy an 1996, 1997).
establishing the discharge pressure one must consider the desired injectivity into the subsurface storage formation, which depends on the thickness and permeability of the formation and on the viscosity and density of the acid gas at in-situ conditions. The high pressures after the fourth compression stage stabilize, upon cooling, the high-density liquid-phase of the acid gas, which can have a density of up to 75% of the density of water, providing the hydrocarbon content is not greater than approximately 2%. Considering that in most cases the pressure of the subsurface disposal formation is near hydrostatic, the hydraulic head provided by the acid gas provides a significant amount of the total pressure required to inject it into the subsurface formation. Several safety features are incorporated in the injection well to prevent leakage. The casing is isolated from the tubing string and the acid gas by installing a packer in the annulus between the casing and the tubing string just above the subsurface storage formation, which is pressure tested for integrity once a year. A downhole safety valve or a check valve can be incorporated in the tubing string so that if equipment fails at surface, the back-flow of acid gas from the formation to the surface will be prevented. The wellhead of the injection well is similarly protected with valves. Such precautions have
been successful to date as proven by the absence of safety incidents. Phase behaviour is one of the keys to acid-gas injection (Carroll & Lui 1997). It is desirable to keep the compression cycle at a higher temperature than the stability field for hydrates to prevent compressor breakdown and plugging. The hydrate-forming temperature at any pressure of an acid gas increases with increasing content of H2S. Compression temperatures maintained above 35°C will avoid hydrates forming (see Fig. 4). Even if hydrate forming is avoided, excess water accelerates corrosion of the steel in contact with the acid gas, and ultimately this is the main reason why dehydration of the acid gas is needed. However, by the fourth stage in a cycle (Fig. 4), compression will tend to dewater the acid gas up to a maximum pressure between 3 and 5 MPa (Fig. 3) if there are no hydrocarbon impurities present. Further compressing the acid gas to higher pressures increases the solubility of water in the acid gas, such that any residual excess water dissolves into the acid gas, and more than counteracts the decrease in solubility due to interstage cooling. Hence, a separate dehydration step may not be needed because no freewater will be present in the system (Clark et al 1998). Thus, by collecting the condensed water during each compression stage and by keeping the temperature a few degrees above the hydrate stability
230
S. BACHU & W. D. GUNTER
temperature, the probability of hydrate formation and corrosion are minimized, and lower-quality, less expensive steels can be used for transporting the acid gas mixture to the subsurface storage formation.
Subsurface The general location for an acid-gas injection well is often influenced by the proximity to sour gas oil or gas production facilities that are the source of acid gas. The specific location is based on a general assessment of the regional geology and hydrogeology, which is designed to evaluate the potential for leakage (Longworth et al 1996) and which includes: (1) size of the disposal zone, to confirm that the disposal zone is large enough to hold volumetrically all of the injected acid gas over the project lifetime; (2) thickness and extent of the overlying confining layer (caprock), and any stratigraphic traps or fractures that may affect its ability to contain the acid gas; (3) location and extent of the underlying or lateral bounding formations; (4) folding or faulting in the area, and an assessment of seismic (neotectonic) risk; (5) rate and direction of the natural flow system, to assess the potential for migration of the injected acid gas; (6) permeability and heterogeneity of the injection zone; (7) chemical composition of the formation fluids (water for aquifers, oil or gas for reservoirs); (8) formation temperature and pressure; (9) core analyses of the injection formation and caprock (if available); and (10) a complete and accurate drilling history of offsetting wells within several kilometres of the disposal well, to identify any wells or zones that may be impacted by the injected acid gas. In addition to the above technical requirements, the regulatory agencies require that environmental concerns must be addressed, such as disposal-formation suitability, wellbore integrity, operating parameters (to ensure formation and well integrity), and optimization of the disposal space, considered to be a limited resource. Disposal wells in the Alberta basin are classified into four classes, depending on the nature of the injected fluid (AEUB 1994). Class la wells are used for the disposal of oilfield or industrial waste fluids. Class Ib wells are used for the disposal of produced water and common oilfield waste streams. Class II wells are used for the disposal of brine and brine-equivalent fluids. Class III wells are used for the injection of hydrocarbons, or inert or
other gases. Class IV wells are used for injection of potable water or steam. The classification serves to define well completion requirements. Because of the nature of acid gas injection operations, the wells are considered as Class III disposal wells, unless the acid-gas is dissolved in produced water prior to injection, in which case the well is designated as either Class Ib or Class II, depending on the produced-water designation (AEUB 1994). Completion and logging requirements are similar for Class II and III wells: (1) (2) (3) (4) (5)
all geological zones must be identified using logs and/or core; all potential hydrocarbon-bearing zones and the shallow potable groundwater aquifers must be hydraulically isolated by cement; injection must be through tubing, and the annulus must be with a corrosion-inhibiting fluid; hydraulic isolation and cement integrity must be confirmed by a full-length casing log; and safety devices must be installed both above the ground and in the wellbore to ensure that failure of any component in the system does not result in environmental damage.
If the injection pressure drops for any reason, the well is automatically shut-in, to prevent acid-gas backflow. Knowledge of the geological setting and characteristics is critical to assess the integrity of the host formation or reservoir, and the short- and long-term fate of the injected acid gas. Of particular importance are potential migration pathways from the injection zone to other formations, shallow groundwater and/or the surface. These potential pathways are of three types: the caprock pore space ('membrane'), natural and/or induced fractures ('cracks') through the confining strata, and improperly completed and/or abandoned wells ('punctures'). To avoid diffuse gas migration through the caprock pore space, the difference between the pressure at the top of the injection aquifer or reservoir and that in the confining layer must be less than the caprock threshold displacement pressure, which is the pressure needed for the acid gas to overcome the capillarity barrier and displace the water that saturates the caprock pore space. To avoid acid gas migration through fractures, the injection zone must be free of natural fractures, and the injection pressure must be below a certain threshold to ensure that fracturing is not induced. The maximum bottomhole injection pressure is set by regulatory agencies at 90% of the fracturing pressure. In the absence of site-specific tests, the pressures are limited by pressure-depth correlations, based on basin-wide statistical data for the Alberta basin. From this point of view, injection
ACID-GAS INJECTION
into a depleted oil or gas reservoir has the advantage of injection pressures being low and of wells and pipelines being already in place (Keushnig 1995). To ensure that environmental and subsurface matters are adequately addressed, regulatory agencies in Alberta and British Columbia require that, in their applications, operators submit information about the geology, lithology and rock properties of the acid-gas injection zone and confining strata, about the formation or reservoir fluids, about operating characteristics such as pressure and temperature, and about acid-gas composition. After approval, operators have to submit semi-annual progress reports detailing the actual composition, rate, pressure and temperature of the injected acid gas.
Issues Critical issues are for the most part environmental and safety-related and they directly affect the economics of acid-gas injection. Either a slow or rapid acid gas leak can result in loss of life or contamination of the biosphere and atmosphere leading to degradation of the environment. Surface issues are addressed through engineering, installation of safety valves and monitoring systems, and emergency procedures in case of FLS leaks.
Surface issues The cost of the separation step is an important part of the acid-gas injection operation. Better chemical amine forms and derivatives are being developed to increase the efficiency of the process, and, in some cases, other acid-gas solvents are being considered (e.g. K2CO3). In cases where the acid-gas stream is predominantly CO2 (e.g. up to 98%), consideration has been given to further processing the gas to remove the CO2, which then would be vented. Cost savings would arise from the lower compression costs. Obviously this is undesirable from a greenhouse gas-emissions reduction viewpoint. As discussed previously, the phase behaviour, water content, and physical properties of the acid gas are important in facility design. To avoid pump cavitation, the acid gas must not enter the two-phase region during compression. Higher temperatures, although required in the process to avoid hydrate formation and plugging of the lines, lead to accelerated corrosion. Both plugging and corrosion can lead to containment failure. Additional safety precautions require addition of chemical inhibitors to the acidgas stream (e.g. methanol to lower hydrate formation temperatures, and glycol to extract the excess water from the acid-gas stream) or use of special corrosionresistant steels. Both add to the cost of the operation.
231
Corrosion monitoring is normally done to assess the effectiveness of the treatments. Although a number of safety valves are always installed, both in the well and in the surface facilities to be able to isolate the containment lines for the acid-gas injection system into small segments, the release of even small volumes of acid gas can be harmful. Consequently, operators are required to have a detailed emergency response plan (ERP) if a leak occurs that may impact humans. An emergency planning zone, the EPZ (i.e. area of land which may be impacted by the release of H2S), is defined around the sour gas facility. To have an effective EPZ, the area surrounding the injection site should be equipped with H2S detection and alarm systems, windsocks, self-contained breathing apparatus, and remote unit and plant shutdown stations.
Subsurface issues Subsurface issues are of two inter-related categories: the effect of the acid gas on the rock matrix and well cements, and plume containment. When the acid gas contacts the subsurface formation, it will readily dissolve in the formation water in an aquifer, or connate water in a reservoir, and create weak carbonic and sulphuric acids. This leads to a significant reduction in pH and accelerates water-rock reactions. Depending on mineralogy, dissolution or precipitation may occur, affecting the porosity and permeability of the host rock. The fact that both CO2 and H2S dissolve in the formation water leads to some complex reaction paths where carbonates precipitate and dissolve, and pyrite/pyrrhotite precipitates (Gunter et al. 2000; Hitchon et al. 2001). If it can be demonstrated that the H2S is immobilized quickly in a sulphide mineral, then concerns over the escape of H2S from the subsurface formation are negligible. In a companion paper in this issue, Gunter et al. (2004) discuss the issue of security of geological storage of acid gases. Dissolution of some of the rock matrix in carbonate strata, or of the carbonates surrounding the sand grains in sandstone units results in lower injection pressures in the short term. A major concern with the injection process is the potential for formation damage and reduced injectivity in the vicinity of the acid gas scheme. The reduction in injectivity could possibly be the result of fines migration, precipitation and scale potential, oil or condensate banking and plugging, asphaltene and elemental sulphur deposition, and hydrate plugging (Bennion et al. 1996). Cement compatibility with the acid gas, primarily in the injection well, but also in neighbouring wells, is crucial for safety and containment. For example a non-carbonate and calcium cement blend shattered
232
S. BACHU & W. D. GUNTER
when tested in an acid-gas stream for several weeks (Whatley 2000). Thus, the compatibility of the acid gas with the cement that bonds the casing to the formation must be tested at a minimum. Although the cement for the newly implemented acid-gas operation can be tested and properly selected prior to drilling, the cements in nearby wells are already in place and their condition is largely unknown. Some of these wells may be quite old, with the cement already in some stage of degradation as a result of brine composition. The acid gas, when reaching these wells, may enhance and speed up the cement degradation, leading to possible leaks through the well annulus and/or along casing. If the acid gas is injected into the originating or other oil or gas pool, the main concern is the impact on further hydrocarbon recovery from the pool and acid-gas production at the pump, although the operation may prove successful, like the Slaughter field in Texas (Adams & Rowe 1981) and the Zama X2X pool in Alberta (Davison et al. 1999). If the gas mixture is injected into an aquifer, the degree to which it forms a plume and migrates from the injection well depends on various factors, including pressure and temperature, solubility, interplay between driving forces like buoyancy and aquifer hydrodynamics, and aquifer heterogeneity, which controls gravity override and viscous fingering.
Case histories For three years, between 1976 and 1979, Amoco Production Co. experimentally used acid gas feedstock from a local sulphur recovery plant in a CO2 miscible flood for tertiary enhanced oil recovery (EOR) in west Texas (Adams & Rowe 1981). Problems were encountered, particularly with acid gas dehydration and hydrate blockages in the approximately 11 km pipeline from the plant to the site, because the gas stream contained too much water moisture, but they were resolved by replacing the molecular sieve being used as a dessicant with a silica gel dessicant (Adams & Rowe 1981). The first true acid-gas injection operation, designed for this purpose, became operational in 1989 on the outskirts of Edmonton in Alberta, Canada. By the time Union Pacific Resources installed a very small-scale acidgas injection operation at Ozona, also in west Texas, USA, later shut down because the acid-gas source watered out, there were already 17 acid gas disposal operations in Alberta (Longworth et al. 1996). Applications and progress reports submitted by operators to regulatory agencies in Alberta and British Columbia become part of the public domain and can be reviewed by any interested party. In addition, several reports of acid-gas operations have been presented previously in the literature. In 1989, Chevron
Canada Ltd started the first acid-gas injection operation at Acheson on the outskirts of Edmonton, Alberta, designed to handle a mixture of 15% H2S and 85% CO2 by injecting it at a depth of 1100 m into a depleted sandstone oil reservoir (Lock 1997). The operation experienced elemental sulphur deposition, which plugged the injection well (Longworth et al. 1996). The next operation, by a different company, started in 1994 by injecting 20% H2S and 80% CO2 into a watersaturated carbonate reef in central Alberta. Also in 1994, Chevron Canada Ltd started acid-gas injection at SOMPa pressure into a carbonate aquifer 2800m deep at Pembina west of Edmonton. The approved acid-gas composition was 74% H2S, 21% CO2, 1% hydrocarbon gases and 4% water (Lock 1997). In 1995, PanCanadian Petroleum Ltd started mixing at surface acid gas with produced water at elevated pressure, and injected the solution ('sour water') into a water-saturated sandstone zone underlying an oil pool at Hansman Lake (operation rescinded since then) and into a deep, dolomitized carbonate-shelf aquifer at Thompson Lake in eastern Alberta (Kopperson et al. I998a,b). Both injection zones are overlain by shales. The sandstone aquifer has high porosity (30% on average) and permeability (up to several darcies, or 10^12 m2), whereas the carbonate zone has 22% porosity. Also in 1995, PanCanadian started an acid-gas injection operation at Wayne-Rosedale, NE of Calgary, Alberta. The acid gas mixture (20% H2S and 80% CO2), with an estimated daily production rate of 21000 m3/d, is the result of sweetening a 34°API oil with an associated solution gas containing 4% H2S and 17% CO2 (Ho et al. 1996). Injection takes place in a 26 m thick watersaturated carbonate-reef unit, characterized by 6% porosity, 150X10-15m2 (150mD) permeability, 65 °C temperature and 15.5MPa formation pressure, confined by tight dolomite units. At in-situ conditions, the acid gas density is about 543 kg m3 (Ho et al. 1996). In 1995, Pennzoil Canada Inc. started injecting a mix of 20% H2S and 80% CO2 at 120000m3/d into a depleted oil reservoir in a pinnacle carbonate reef at a depth of about 1510m in the Zama area in northern Alberta (Davison et al. 1999). The c. 115m thick dolomitized reef, with an average porosity of 7.6%, is underlain by approximately 45 m of tight lime mudstone and is overlain by 60-90 m of anhydrite. The critical point of the acid-gas mixture is 42 °C and 7.64MPa. At target operating conditions of 14.5MPa and 76 °C (initial reservoir pressure and temperature), the injected acid gas is supercritical, with a density of 500kgm~3 and a viscosity of 0.04cp (Davison et al. 1999). The caprock threshold displacement pressure was measured in the laboratory to be 22.3MPa. Considering that the initial differential pressure across the caprock was about 7MPa, the maximum operating
ACID-GAS INJECTION
pressure was established at 30MPa. Because the formation water naturally contains significant amounts of both CO2 and H2S, acid-gas injection was not expected to cause dissolution of reservoir rock and/or precipitation. The West Stoddart acid-gas injection operation in NE British Columbia, Canada, started in 1998, is the largest to date in the Alberta basin, with a licensed capacity of 390000m3/d, but operating only at 280000 m3/d. The acid gas stream contains 85% H2S and 15% CO2 (340 and 78t/d H2S and CO2, respectively). Currently, acid-gas injection in the Alberta basin occurs over a wide range of aquifer and reservoir characteristics, acid-gas composition, and operating conditions. Three operations have been rescinded after an active period, either because the injection volumes reached the approved limit, and the operator moved to another site or injection formation, or because the gas plant producing the acid gas has been decommissioned. Two injection operations have been shut in by the regulatory agency because the operator greatly exceeded the approved injection volume or pressure. Also, at five sites the acid gas is or was mixed with water at the surface prior to injection, of which two are actually water disposal sites with minor amounts of dissolved acid gas ('sour water' disposal). The fate of the injected acid gas in the subsurface is not known, because subsurface monitoring is not required, and it is difficult and expensive. Only the well-head gas composition, pressure, temperature and rate have to be reported to regulatory agencies. There is only one recently reported case (2002) where the operator believed, when it applied for the injection license, that the disposal zone was isolated by a fault from an adjacent pool in the same formation producing oil with a high H2S content, only to have acid-gas breakthrough at two offset producing wells some 900m away from the injection well several months after the start of injection. However, the operator, owner of both the oil pool and of the acid-gas disposal facility, decided to continue both operations. Since the mid 1990s, acid-gas injection operations were implemented also in the United States, mostly in Wyoming and west Texas, but also one each in New Mexico, Oklahoma and Michigan. At several sites, the acid gas is mixed with water at surface prior to injection. The largest acid-gas project operating in the United States was installed by Union Pacific Resources in SW Wyoming in 1997 and injects 86% CO2 and 14% H2S at approximately 250000 mYd (Miller et al 1999). This was eclipsed in 2003, when the Exxon-Mobil LaBarge project in Wyoming came on stream. It was designed to inject 1700000 m3/d of acid gas having a composition of 65% H2S and 35% CO2, equivalent to
233
approximately 0.4Mt/a storage of CO2 in deep geological formations.
Conclusions All of the acid-gas injection sites in North America (Canada and the United States), except for the Michigan one, are in the string of sedimentary basins that extend from the Canadian north to the Gulf of Mexico in the foreland of the Rocky Mountains. More deep, and increasingly sour, gas will be produced from these basins in the future, which will lead to more injection sites for the disposal of the acid gas produced as a by-product of the gas sweetening process. Acid-gas injection has been developed into a safe and environment-friendly technology. It can be injected as a dry gas or a solution gas depending on the wastewater available. Compared to other options, acid-gas injection has fewer environmental consequences than sulphur recovery (where leaching of the sulphur piles can lead to groundwater contamination) or flaring (which essentially substitutes SO2 for H2S in the atmosphere, as well as releasing CO2). Knowledge of the phase-equilibria and properties of the acid gas is extremely important in designing a safe injection operation that minimizes corrosion, hydrate formation and cavitation, leading to a dense liquid phase for injection. Both monitoring for gas leaks and corrosion, and plans for mitigation if a leak occurs, are an essential part of the technology. Equally important for a safe acid-gas injection operation is the proper characterization and selection of the subsurface disposal zone (reservoir or aquifer). Critical elements are containment and prevention of leakage and/or migration through natural or man-made conduits, such as fractures ('cracks') and abandoned wells ('punctures'). The integrity of the confining aquitard (or caprock), which is subjected to physical and chemical stresses, is essential. Acid-gas injection is more than an industrial, scaled-up analogue for geological storage of CO2. It is a mature technology and applications of this process are growing in number and size. The experience gained so far in North America, particularly in Canada, can be applied elsewhere in the world. As gas resources become more depleted, even more sour gas will be produced; and therefore we predict that this technology will result in the reduction of significant CO2 emissions even without the incentive of CO2 emission trading. With the advent of trading, acid-gas injection should become even more appealing to oil and gas producers. By the end of 2002, close to 1.5Mt CO2 and 1 Mt H2S had been successfully injected into deep hydrocarbon reservoirs and saline aquifers in Canada
234
S. BACHU & W. D. GUNTER
alone. In addition, considering operations in the United States, experience shows that CO2 storage in geological media is a mature technology that can successfully be expanded to and applied in largescale operations that will reduce CO2 emissions into the atmosphere from large CO2 point sources. Discussions with Ed Wichert, Sogapro Engineering, have been very helpful to the authors.
References ADAMS, G. H. & ROWE, H. G. 1981. Slaughter Estate Unit CO2 pilot - surface and downhole equipment construction and operation in the presence of H2S. Journal of Petroleum Technology, 33,1065-1074. AEUB (Alberta Energy and Utilities Board) 1994. Guide 51: Injection and Disposal Wells. Alberta Energy and Utilities Board, Calgary, AB. DOI eub.gov.ab.ca/bbs/ products/guides/g5 l-1994.pdf. AEUB (Alberta Energy and Utilities Board) 2000. Guide 65: Resources Applications for Conventional & Gas Reservoirs. Alberta Energy and Utilities Board, Calgary, AB, 113-136. DOI eub.gov. ab.ca/bbs/products/guides/g65 .pdf. BACHU, S. 2000. Sequestration of carbon dioxide in geological media: criteria and approach for site selection. Energy Conversion and Management, 41,953-970. BACHU, S. & STEWART, S. 2002. Geological sequestration of anthropogenic carbon dioxide in the Western Canada Sedimentary Basin. Journal of Canadian Petroleum Technology, 41, 32-40. BENNION, D. B., THOMAS, F. B., BENNION, D. W. & BIETZ, R. F. 1996. Formation screening to minimize permeability impairment associated with acid gas or sour gas injection/disposal. CIM Paper 96-93 Presented at the 47th Annual Technical Meeting of the CIM Petroleum Society, Calgary, AB, June 10-12,1996. CARROLL, J. J. 1998a. Phase diagrams reveal acid-gas injection subtleties. Oil & Gas Journal, 96, 92-96. CARROLL, J. J. 1998&. Acid-gas injection encounters diverse H2S, water phase changes. Oil & Gas Journal, 96,57-59. CARROLL, J. J. & Lui, D. W. 1997. Density, phase behavior keys to acid gas injection. Oil & Gas Journal, 95, 63-72.
CLARK, M. A., SYREK, W. Y, MONNERY, W. D. & 6 OTHERS 1998. Designing an optimized injection strategy for acid gas disposal without dehydration. In: Proceedings of the 77th Gas Processors Association Annual Convention, Dallas, Tx, May 16-18 1998. Gas Processors Association, Tulsa, OK, 49-56. CONNOCK, L. 2001. Acid gas injection reduces sulphur burden. Sulphur, 272, 35-41. DAVTSON, R. J., MAYDER, A., HLADIUK, D. W. & JARRELL, J. 1999. Zama acid gas disposal/miscible flood implementation and results. Journal of Canadian Petroleum Technology, 38,45-54. GUNTER, W. D., PERKINS, E. H. & HUTCHEON, I. 2000. Aquifer disposal of acid gases: modeling of waterrock reactions for trapping acid wastes. Applied
Geochemistry, 15,1085-1095. GUNTER, W. D., BACHU, S. & BENSON, S. M. 2004. The role of hydrogeological and geochemical trapping in sedimentary basins for secure geological storage of carbon dioxide. In: BAINES, S. J. & WORDEN, R. H. (eds) Geological Storage of Carbon Dioxide. Geological Society, London, Special Publications, 233,129-145. HITCHON, B., PERKINS, E. H. & GUNTER, W. D. 2001. Recovery of trace metals in formation waters using acid gases from natural gas. Applied Geochemistry, 16,1481-1497. Ho, K. T., MCMULLEN, J., BOYLE, P., ROJEK, O., FORGO, M. & BEATTY, T. 1996. Subsurface acid gas disposal scheme in Wayne-Rosedale, Alberta. SPE Paper 35848, In: Proceedings of the Gas Technology Symposium, Calgary, AB, Canada, 28 April - 1 May 1996, Society of Petroleum Engineers, Richardson, Tx, 699-704. KEUSHNIG, H. 1995. Hydrogen sulphide - if you don't like it, put it back. Journal of Canadian Petroleum Technology, 34,18-20. KOPPERSON, D., HORNE, S., KOHN, G., ROMANSKY, D.,
CHAN, C. & DUGWORTH, G. L. 19980. Injecting acid gas with water creates new disposal option. Oil & Gas Journal, 96, 33-37. KOPPERSON, D., HORNE, S., KOHN, G., ROMANSKY, D., CHAN, C. & DUGWORTH, G. L. 1998&. Two cases illustrate acid gas/water injection scheme. Oil & Gas Journal, 96, 64-70. LOCK, B. W. 1997. Acid gas disposal - A field perspective. In: Proceedings of the 76th Gas Processors Association Annual Convention, San Antonio, TX, March 10-12, 1999. Gas Processors Association, Tulsa, OK, 161-170. LONGWORTH, H. L., DUNN, G. C. & SEMCHUK, M. 1996. Underground disposal of acid gas in Alberta, Canada: regulatory concerns and case histories. SPE Paper 35584, In: Proceedings of the Gas Technology Symposium, Calgary, AB, Canada, 28 April - 1 May 1996, Society of Petroleum Engineers, Richardson, Tx, 181-192. MILLER, E. W., SOYCHAK, S. J., REED, A. E., BARTOO, R. K. & ACKMAN, R. 1999. Brady plant treating project. In: Proceedings of the 78th Gas Processors Association Annual Convention, Nashville, TN, March 1-3, 1999, Tulsa, OK, 76-81. No, H-J., CARROLL J. J. & MADDOCKS, J. R. 1999. Impact of thermophysical properties research on acid gas injection process design. In: Proceedings of the 78th Gas Processors Association Annual Convention, Nashville, TN, March 1-3, 1999. Gas Processors Association, Tulsa, OK, 114-120. WHATLEY, L. 2000. Acid-gas injection proves economic for west-Texas plant. Oil & Gas Journal, 98,58-61. WICHERT, E. & ROYAN, T. 1996. Sulfur disposal by acid gas injection. SPE Paper 35585, In: Proceedings of the Gas Technology Symposium, Calgary, AB, Canada, 28 April-1 May 1996. Society of Petroleum Engineers, Richardson, TX, 193-200. WICHERT, E. & ROYAN, T. 1997. Acid gas injection eliminates sulfur recovery expense. Oil & Gas Journal, 95, 67-72.
Monitoring experience associated with nuclear waste disposal and its application to CO2 sequestration projects MICHAEL J. STENHOUSE1 & DAVID SAVAGE2 1
Monitor Scientific, 3900 S. Wadsworth Blvd, Suite #555, Denver Colorado 80235, USA, (e-mail: mstenhouse @ monitorsci. com) 2 Quintessa Limited, 24 Trevor Road, West Bridgford, Nottingham, NG2 6FS, UK Abstract: Monitoring is a key component of the development and operation of nuclear waste repositories, and some of the underlying considerations and lessons learned can benefit projects involving the geological storage of CO2. Before reviewing the general monitoring objectives for nuclear waste repositories, the key differences between nuclear waste disposal and CO2 storage projects are emphasized. The philosophy underlying monitoring after closing/sealing a repository is discussed. Important aspects of this philosophy include the need to collect adequate baseline data representative of the unperturbed site, and the desire to engender public confidence, but not at the expense of compromising the protective barriers of the repository itself. Pre-operational and operational monitoring provide important data that feed into safety assessment calculations, either as input data or as information that can be used to confirm, and/or refine, predictions. Using a specific example of a deep (geological) repository, monitoring experience at the WIPP site in New Mexico is discussed, focusing on methods and techniques that are relevant to CO2 sequestration projects. Such monitoring includes geotechnical (characterization of the evolving behaviour of underground facilities), groundwater (quality and quantity), environmental (impacts on ecosystems), and subsidence (to support subsidence predictions for the WIPP site located in bedded salt formations).
Monitoring is an important aspect of the development and operation of a nuclear waste repository starting from the initial site characterization and continuing through to closure and sealing of the repository and possibly longer. Although the type of monitoring carried out at a repository is not always directly applicable to that considered for a long-term CO2 storage reservoir, there are some general lessons to be learned from the broad experience of nuclear waste disposal, which should be useful for future CO2 storage projects. This paper provides a general overview of monitoring objectives for radioactive waste repositories, as well as providing a specific example of such monitoring. As an example of an active monitoring programme in support of a radioactive waste repository, - the Waste Isolation Project Plant (WIPP) in New Mexico, has been selected, this being the only licensed deep (geological) site for long-lived radioactive waste. 'Deep' in the context of radioactive waste disposal means generally at depths of more than 500m, below which the likelihood of human intrusion is considered to be minimal. In fact, several monitoring programmes are ongoing at WIPP (e.g. Department of Energy, DOE 1996, 1999) but the focus here is on those programmes that directly support assessments of the long-term safety of the site. Before discussing specific aspects of monitoring at radioactive waste disposal sites, it is worthwhile acknowledging key differences between radioactive waste disposal and long-term CO2 storage in the
context of monitoring and prediction of long-term performance. First, radioactive waste comprises hazardous materials that must be isolated from the environment for periods of hundreds to thousands of years to assure public safety. On the other hand, CO2 is not normally considered a hazardous gas, although there are safety concerns associated with an atmosphere having CO2 concentrations greater than 4% (i.e. about 100 times normal atmospheric concentrations) (Material Safety Data Sheet). Indeed, the possibility exists that future generations may consider this gas a valuable resource. The second key difference lies in the nature of the storage volume itself and how it is characterized. The area in which radioactive waste is to be stored (near-field rock) is well characterized, even if uncertainties exist about the surrounding geosphere. Detailed near-field rock characterization is possible because this volume has been excavated and is therefore accessible in situ during the construction and operational phases of the project. Furthermore, additional barriers are added to the excavated volume to provide a multi-barrier engineered system for waste containment. Figure 1 shows the various components of an engineered barrier system associated with high-level waste disposal (Posiva Oy 1999). The barriers of such a system are carefully engineered and, therefore, have been characterized in detail. In contrast, the characterization of a potential CO2 storage reservoir is carried out remotely, albeit with state-of-the-art techniques.
From: BAINES, S. J. & WORDEN, R. H. (eds) 2004. Geological Storage of Carbon Dioxide. Geological Society, London, Special Publications, 233, 235-247.0305-8719/047$ 15.00 © The Geological Society of London 2004.
236
M. J. STENHOUSE & D. SAVAGE
Fig. 1. Components of an engineered barrier system - spent nuclear fuel disposal, Finland. The planned disposal facilities, at a depth of around 400-700 m, comprise disposal tunnels at roughly 25 m intervals, 100-150m in length, linked by a central tunnel. The total length of the disposal tunnels is about 13km. The total estimated quantity of spent nuclear fuel to be disposed of, is estimated at about 4000 tonnes by 2040 (Posiva Oy 1999).
Lastly, the underground storage of CO2 relies on the intrinsic storage capacity of the host rock (a reservoir rock) within its natural porosity and permeability rather than in an excavated cavern. Despite these obvious differences, the similarities in temporal and spatial scales for such projects are sufficient to justify an examination of the monitoring carried out at nuclear waste facilities. The isolation requirements for radioactive waste and CO2 storage cover periods of at least hundreds of years in each case (and possibly thousands), and the areal extent of underground sites for both radioactive waste disposal and CO2 storage, is in the order of tens of square kilometres.
Monitoring objectives for nuclear waste disposal: general The main objectives of monitoring at nuclear waste disposal facilities depend to a large extent on the stage of development of the disposal site. Broadly, these objectives are: (1)
(2)
To provide information towards a good understanding of geology, hydrogeology and geochemistry of the site (site characterization and construction phases). To ensure that operations are conducted in a safe and environmentally acceptable manner (operational phase).
(3) (4) (5) (6)
(7)
(8)
To maintain a detailed record of waste emplacement (and ensure its memory is retained). To enable the possibility of retrieval at some future date (the concept of retrievability or reversibility). To boost public confidence and thereby to enhance public acceptance of geological disposal. To assist in the societal decision-making process by, for example, monitoring system performance and providing data on conditions relevant to the retrievability of the waste. To provide information and data that can be used, either directly or indirectly, in safety assessment calculations (all phases of the repository) for submission to regulators; either as input data or for validation purposes; To provide information and data once the repository has been closed and sealed (postclosure phase).
The emphasis of this paper is on the last two objectives since they relate to the long-term performance of the repository system. However, there is overlap in the monitoring carried out to meet the above objectives. During the characterization of a site which has been selected for a potential repository, the monitoring techniques to be used are expected to be similar to those used for a CO2 storage reservoir, e.g. a variety of
MONITORING STRATEGIES FOR NUCLEAR WASTE
seismic survey techniques, borehole studies for geological and hydrogeological data, and geochemical analyses. Thus, these are not discussed in detail in this paper. However, one important lesson learned in this area over the past ten years is the need to collect 'good' baseline data (representative of unperturbed conditions) and the need to carry out geochemical analyses on groundwaters as early as possible during site characterization. These considerations have important implications for the site characterization methods used, and in particular for the design and completion of investigative boreholes (UK Nirex 1996). Inappropriate use of conventional drilling fluids and completion materials can contaminate and ruin the groundwater geochemical information. Appropriate use of geochemical 'tracers' can be used to establish the effects of drilling-induced contamination (UK Nirex 1996). Commonly, in the past, emphasis has been placed on obtaining geological and hydrogeological information first, at the expense of groundwater chemistry data representative of unperturbed conditions. In the absence of unperturbed data, interpretation of changes in water chemistry is more complex. With regard to deep disposal facilities, only one site is at the operational stage (WIPP site) and the monitoring conducted at this site is discussed in a later section. However, an interim stage in most national deep disposal programmes for nuclear waste, is the construction of one or more underground research laboratories (URLs) to enable in-situ investigations to be carried out. For example, such facilities have been constructed in Sweden (Aspo), France (Aisne, Paris Basin), and Canada (Manitoba). Under these circumstances, the monitoring and investigations carried out are relevant to an improved understanding of the site and to safety assessment calculations, i.e. objective 7 above (see next subsection). For shallow disposal facilities, (surface or near-surface, at a depth of less than 150m; e.g. Centre de la Manche, surface repository in France, and SFR1 about 50m below the Baltic Sea, at Forsmark in Sweden), much of the monitoring conducted during the operational phase is typical of all nuclear facilities. Such monitoring is designed to demonstrate that there are no significant releases of contaminants away from the site, which would impact on humans or the environment. If there are significant releases, the monitoring programme offers an early indication of the need for remedial action.
Support for performance or safety assessment purposes Performance assessment (PA) involves the analysis of possible ways in which a repository site might evolve and prediction of the associated consequences. PA is 'an analysis to predict the perfor-
237
mance of a system or subsystem, followed by comparison of the results of such analysis with appropriate standards or criteria' (NEA 1991). For most countries, a base or reference scenario exists, which is the expected evolution of the site, i.e. in the absence of perturbations. Alternative scenarios are then explored by examining the possibility of one or more perturbations acting on the repository system to affect its evolution. Results from PA calculations are then compared with existing standards or regulatory limits (typically a radiation dose limit or the risk of a radiological health impact to members of the population) to see if the site performance will be acceptable under its current design, or whether modifications in the design are necessary to ensure longterm safety. Since assessment calculations typically focus on the transport of radioactive contaminants to the biosphere (or accessible environment), the type of information and data used as input to PA involves transport pathways, specifically via gas and groundwater, as well as processes that can affect transport. Hence, the monitoring carried out in support of PA activities concerns geological, hydrogeological, geomechanical and geochemical data. Initial information should be available at the site characterization stage, and be updated as site-specific data are provided during the site construction and operational phases. In those countries without a deep disposal facility in operation, a URL allows in-situ investigations and monitoring. Examples of the types of monitoring conducted in different facilities include (NEA 1988): Geomechanical characteristics of deep clay formation and long-term behaviour using various types of dilatometers, convergence/confinement testing, and pressure measurements (URL in clay, Belgium). Hydrogeological: determination of hydraulic and hydrogeological configuration around URL using pressure meters and piezometers, both in individual and array modes (URL in clay, Belgium). Thermo-mechanical: monitoring of rock stress with and without a heat load (URL in crystalline rock, Switzerland). Hydro-thermo-mechanical: monitoring of strain, stress and hydraulic conductivity variations in rock (URL mine in crystalline rock, France). Hydrogeological: hydraulic measurements in boreholes (various URLs: granite/crystalline rock, clay). Hydro geochemical: groundwater characterization (various URLs: granite/crystalline rock, clay).
M. J. STENHOUSE & D. SAVAGE
238
Table 1. Examples of geochemical and hydrogeological monitoring in underground research laboratories (URLs). Location of URL
Monitoring
Purpose
Description / Comment
Reference
Aspo, Sweden
Periodic long-term groundwater sampling and monitoring of boreholes
Changes in chemical characterization of groundwaters (EH, pH, redox-sensitive components, salinity, isotopic)
Downhole and surface chemical probes; on-site chemical analysis laboratory; conducted during site characterization, construction and operational phases
Almen & Zellman 1991 SKB 1992
Aspo, Sweden
Groundwater level and Study changes in pressure monitoring groundwater pressure due to excavation activities in boreholes
Integrated monitoring system; SKB 1993 borehole transducers
Stripa Mine, Sweden
Hydrogeological and hydrochemical monitoring
Long-term sampling and Davis & analysis programme involving Nordstrom minerals from drill cores and 1992 rock fractures, and groundwaters; hydraulic tests in boreholes
Stripa Mine, Sweden
Tracer monitoring tests Evaluate groundwater flow and solute transport in fractured rock; including dispersion, matrix diffusion and sorption
TVO Research Hydraulic and tracer Tunnel, Finland monitoring between boreholes
Hydraulic, chemical and isotopic characterization of Stripa granite and groundwaters; understanding of geochemical origin and evolution of groundwaters, including mixing
Investigate transport in single fractures prior to boring of full-scale boreholes Study the mass outflow pattern from sides and base of borehole; monitor breakthrough curves for tracers for main outflow
2D and 3D migration tracer Birgersson et tests; monitoring water inflow al. 1992 rates and distributions into a Olsson 1992 fracture zone; comparison with modelling predictions Water flow rates measured Hautjarvi et al. using dilution probe 1994 technique; flowpaths could not be deduced unambiguously, but discharge points were identified clearly
areas
Grimsel Test Site, Switzerland
Hydrochemical monitoring of groundwaters in vicinity of large shear zone
Change in chemical characterization of groundwaters (Ew pH, redox-sensitive components, salinity, isotopic); determination of groundwater mixing and residence times
In addition, monitoring of man-made vault and shaft seals is considered important in waste disposal programmes. Atomic Energy of Canada Limited, responsible for the disposal of Canada's nuclear fuel waste, considered a component test area as the optimum means of monitoring seal quality, to avoid compromising the integrity of a disposal vault (Johnson etal 1994). A large portion of monitoring associated with national waste disposal site programmes is focused on geochemical monitoring, to establish existing conditions at and around the site as well as providing data for geochemical modelling to determine the
Long-term sampling and analysis; results did not indicate spatial variability; fracture mineral sampling and analysis
Alexander 1991 Frick et al. 1992
extent of mixing of different groundwaters. Examples of this type of monitoring are shown in Table 1, including, where appropriate, supporting hydrogeological monitoring.
Post-closure considerations An important premise applicable to all national radioactive waste disposal programmes is that nuclear waste generated from past and present years should not place any onus on future generations. Thus, there should be no requirement on future gen-
MONITORING STRATEGIES FOR NUCLEAR WASTE
erations to assure the safety of nuclear waste repositories. It follows from this premise that no deep repository should be sealed and closed unless longterm safety is assured. Given this requirement, the need for post-closure monitoring appears not to be justified technically. However, the two main reasons for conducting post-closure monitoring using in-situ techniques are: to establish that integrity is maintained by confirming the predicted behaviour of a repository, predicted by safety assessment calculations; to provide public confidence. In confirming the predictions of safety assessment calculations, at least for a short time after closure, those parameters that might be monitored may not necessarily relate directly to events which are predicted in a safety assessment. For example, a safety assessment might include some assumption about resaturation of a facility, whereas monitoring might detect changes in the near-field stress which could be indicative of resaturation. Thus, it is likely that a major interpretive effort would be required if a monitoring programme were established. In the postclosure phase, monitoring would continue as long as society considered it beneficial. In this context, consideration needs to be given to who will see and use the information that will be obtained, and what specific parameters are to be monitored. If post-closure monitoring is required for performance confirmation, the licensing authorities will require technical data to indicate that the behaviour of the storage system is in accordance with the quantitative predictions made before closure. If the general public requires reassurance they will want to see simple, easily understandable, illustrative data that demonstrate that the repository remains safe. Whatever the reason for monitoring, if the monitoring data were to indicate a failure to reach the required standards, or if the data were to indicate that the storage systems were behaving in an unacceptable way, this would precipitate a need for corrective action. The ultimate corrective action would be retrieval of the waste. It should also be appreciated that any changes in repository behaviour would be small or extremely delayed relative to any monitoring period being considered. Thus, if the monitoring confirms such a slow (predicted) degradation of the repository, given that this is expected behaviour, such monitoring might only raise concerns or unnecessary questions about repair and improvements. Factors that influence the monitoring period include expectations on the lifetime of in situ monitoring equipment combined with the idea of no onus being placed on future generations. For shallow or near-surface nuclear waste repositories, a period of
239
institutional control is factored into long-term safety considerations. For those surface disposal sites for which such an institutional control period is planned, timescales have been proposed ranging from 50 years (low-level waste site at Drigg, NW England) to up to 300 years (various disposal sites including low-level waste sites in France). Given such institutional control periods, some period of post-closure monitoring appears reasonable for near-surface facilities, although expected lifetimes for monitoring equipment attaining the upper end of the range in control period seem unlikely. The situation is different for deep disposal, however. For example, with regard to a deep disposal facility in France, monitoring for even a few hundred years was not considered to be of significant use in ensuring repository performance owing to the long timescales over which safety should ideally be studied. On the other hand, post-closure monitoring is a regulatory requirement for the WIPP site in the USA (discussed in the subsequent section). Possibly more important than any technical confirmation of assessment predictions, post-closure monitoring can have a non-technical objective, to provide comfort to anyone with an interest in the safety of the disposal, in particular the general public. Whatever the basis for carrying out post-closure monitoring, it seems advisable, where some form of monitoring programme is contemplated, to have complementary plans for responding to unfavourable monitoring results. For example if there were indications that a repository was degrading faster than predicted, there should be some provision for reversing the process or at least addressing any safety implications of the faster-than-predicted degradation. Given such circumstances, post-closure monitoring should be based on a clear, preconceived description of what constitutes a failure of the repository system and what action should be taken in the event of a failure arising. Thus, in the case of nuclear waste disposal, post-closure monitoring seems to be appropriate only for the identification of gross failure of the repository which can be linked directly to specific remedial measures. For most national deep disposal programmes for nuclear waste, there is likely to be some period of monitoring after nuclear waste has been emplaced but prior to the repository being closed, possibly focusing on measurements of rock deformation, hydrogeology and groundwater chemistry. Currently, there are no specific regulatory requirements covering the period between emplacement of waste and closure. However, once the repository is sealed and closed, there are significant differences in how monitoring is approached, or even whether it should be addressed at all.
M. J. STENHOUSE & D. SAVAGE
240
Monitoring considerations for deep geological CO2 storage Pre-operational monitoring The pre-operational phase of any monitoring programme would include surface exploration for site selection and site characterization, and design of CO2 injection patterns. The objectives of monitoring in this phase should be to collect baseline data to understand the future evolution of the storage system, for use in long-term performance and safety assessment, and to develop confidence in the conceptual and numerical models of system behaviour. Such baseline data in the form of soil-gas analysis, were collected as part of the International Energy Agency's (lEAs) Weyburn CO2 monitoring and storage project (Strutt et al 2003). Baseline monitoring activities should begin at the earliest possible stage of a CO2 sequestration development programme, before the perturbations caused by CO2 injection start to accumulate. In practice, the monitoring programme will, therefore, commence during the pre-operational phase, with some measurements also forming part of the site characterization programme (plus incorporation of historical records: seismic activity, river water levels, climate etc.). Key geoscience data requiring collection over long periods, extending into the operational and post-closure phases, might include: seismicity - national and local seismic networks, primarily to improve estimates of earthquake probabilities and magnitudes; tectonic deformation - precise levelling surveys for disruptive events and to confirm rates of tectonic deformation; hydrological data to improve estimates of recharge and infiltration; meteorological and climatic data; hydrogeological data (hydraulic heads and conductivities to increase confidence in the estimates of groundwater travel time and the site hydrogeological model; temperature; groundwater flow rates and directions); and hydrogeochemical data (groundwater chemistry -Ew pH, dissolved ions, salinity). The important parameters for defining the state of the surface environment are likely to include: natural (background) levels of CO2 in water and soils; meteorological conditions including precipitation levels; surface hydrology, including infiltration rates; characterization of natural habitats and eco-systems. The establishment of baseline values for surface environmental indicators is relatively straightforward, because the process of measurement will generally not affect the parameters being measured e.g. mea-
surements relating to climatic factors and surface hydrology. Where parameter values are expected to vary within a natural range, an extended monitoring is advisable to enable the pattern of natural variation (e.g. seasonality), to be established. Where parameter values are found to follow a trend of increasing or decreasing values, baseline monitoring will need to be continued until that trend is established with confidence. The assessment of when sufficient baseline data have been collected and interpreted is an important part of the planning process. Some general guidance on this topic relating to site characterization, can be found in UK Nirex (1996). Drilling from surface will cause perturbations to the natural system. Baseline monitoring must therefore be continued for sufficient time for conditions to approach steady-state, allowing the implications of the perturbation to be understood. This monitoring cannot be regarded as measurement of the original conditions, but merely the first set of measurements that could be made. In summary, the parameters of primary interest in the context of establishing baseline conditions in the geological environment are: groundwater occurrence and flux in the target formation and in the surrounding aquifers; and geochemical characteristics of the water (including pH, EH, rock-water interaction products). Baseline values for the above parameters should be established as part of the site characterization activity, e.g. from measurements of a pattern of regional boreholes and from the preliminary investigations. These measurements must be continued for sufficient time to enable the implications of natural variations, and of any perturbations caused by the drilling and monitoring activities themselves, to be understood.
Operational phase monitoring The parameters to be measured might include: temperature; pressure; water inflows; pH; CO2 levels in groundwater etc. Borehole geophysical methods for rock structure stability and disturbance monitoring include seismic tomography, vertical seismic profiling (VSP) and complex VSP, cross-hole tomography, radar and acoustic emission/microseismic monitoring. For example a range of seismic techniques is being used in the IEA Weyburn CO2 Monitoring and Storage Project to evaluate the ability to track CO2 movement in, as well as away from, the oil reservoir. In particular, 4D seismic monitoring offers encouraging prospects in being able to characterize CO2 movement, although the technology used in this way is still in its infancy (Kendall^al 2003).
MONITORING STRATEGIES FOR NUCLEAR WASTE
In the case of a borehole intruding directly into a storage reservoir, it would be necessary to ensure that its construction (and use) had minimum impact on the overall performance of the disposal system. This assurance is likely to extend into the period after which the site is closed, to ensure that wellbore and casing seals provide adequate long-term (at least several decades, but ideally, hundreds of years) performance.
Post-closure monitoring In developing a strategy for monitoring after a subsurface site has been closed (sealed), the limitations and the potential detriments of the monitoring process need to be understood and balanced against the benefits from improved understanding of the behaviour of the system components. Many of the lessons learned from such considerations applied to nuclear waste disposal are also relevant here. The durability and reliability of the instrumentation will severely limit the temporal extent of some types of monitoring. The ability to ensure continued calibration of equipment is also cause for concern. Monitoring from the surface could, however, be carried out for as long as society wished it. The extent of monitoring (i.e. the amount of instrumentation) will also be limited by the practical difficulty of taking measurements, by the fact that replaceable instruments will need to be accessible, and by the need to minimize the number of penetrations into the storage system. The need to preserve passive, longterm safety will broadly determine the type of monitoring to be applied. Additionally, the monitoring data should adequately reflect the repository behaviour and be capable of being interpreted with the least possible ambiguity. Potential detrimental aspects of monitoring include the formation of pathways by the installation of monitoring equipment, leading to increased potential for CO2 migration and an increased likelihood of human intrusion or adverse impacts by natural or induced processes if monitoring is continued into the long term. For each monitoring objective it will be necessary to: (1) (2) (3) (4) (5) (6)
list the parameter(s) to be measured and define its relationship to the environmental or performance monitoring objectives; define the monitoring method/measurement technique; analyse the current instrumental technology and define the measurement resolution; estimate the cost of monitoring/measurement; define the expected range of observations; define the bounds of tolerable discrepancy, justified by some safety-related rationale; and
(7)
241
define the action to be taken on identification of intolerable discrepancy.
The significance of determining the bounds of tolerable discrepancy and the action to be taken should the results fall outside that range should not be underestimated. For performance monitoring, a relatively detailed modelling capacity is required, to determine what would be intolerable and what to do about it. For monitoring which is only likely to produce significant results in the event of gross malfunction, it is hard to devise in advance what action would be appropriate. To the extent that failures can be anticipated, the design should be changed to mitigate either the probability or consequences of occurrence. If only 'highly unlikely' events are left, then it seems counter-intuitive to plan for their eventuality, beyond saying that all that needs to be done will be done. However, this may not be seen as sufficient in the eyes of the general public. The overall implication is that monitoring should be limited to those targets for which a sensible response to results can be demonstrated. An obvious target for monitoring programme will be wellbores drilled into, and in the vicinity of, a geological storage site for CO2. Currently, the material used for wellbore seals is cement that degrades with time, even in the absence of aggressive groundwater species. In addition, corrosion of metal casings will occur over time. Thus, CO2 migration via abandoned wells is a key issue concerning future release of CO2 back to the atmosphere and/or surface environment and, as such, is a focus of the long-term assessment being conducted for the IEA Weyburn CO2 monitoring and storage project (Stenhouse et al. 2003).
Monitoring experience at WIPP Background and development The Waste Isolation Pilot Plant (WIPP) is the first geological repository in the United States that has been submitted for regulatory approval. WIPP, located in a semi-arid southeastern part of New Mexico, is intended for the permanent disposal of defence-related transuranic (TRU) waste and TRU mixed waste. The underground facility is being excavated in bedded salt approximately 658 m below ground surface in the Salado Formation. The application for certification was submitted in 1996 (DOE 1996) and, after another period of formal comment and public hearings, the US Environmental Protection Agency (EPA) provided certification for the site in 1998. The operational period, during which waste will be received and emplaced, is set at 25 years, until 2023. Towards the end of the operational period, a Closure Review
242
M. J. STENHOUSE & D. SAVAGE
Fig. 2. Schematic diagram of repository system components relevant to environmental standards (from National Research Council 1996). The EPAs 'accessible environment' is shown as the region beyond a vertical cylinder of 5km radius, centred on the WIPP repository. DOE calculations are aimed at demonstrating compliance within a smaller region, a volume generated by a square of cross section 6.44km2 (4 miles X 4 miles), extending downwards without limit.
Study will be submitted. Assuming a favourable decision based on this study, a period of 8-10 years has been allocated to decontaminate and decommission surface facilities. Closure occurs when the shafts have been backfilled and sealed. Thus, the expected time between the operational phase (when waste is emplaced) and closure is no more than ten years. There is, however, no specific time requirement in the federal regulations. The WIPP repository system, as required by US environmental federal regulations (40 Code of Federal Regulations, CFR Part 191) (EPA 1993), comprises a combination of geological and engineered barriers.
With regard to compliance requirements, the US EPA has defined the 'accessible environment' as the region outside a vertical cylinder, of 5km radius, around the centre of the WIPP site (see Fig. 2). 'Accessible environment' comprises '....(1) the atmosphere; (2) land surfaces; (3) surface waters; (4) oceans: and (5) all of the lithosphere that is beyond the controlled area' (EPA 1993 section 191.12(k)). The DOE defines the boundary used in compliance calculations as a smaller, 6.4km by 6.4km, square region within the 5km radius circle, also extending downwards without limit from the surface (see Fig. 2). Any migration of radionuclides across this boundary in excess of the
MONITORING STRATEGIES FOR NUCLEAR WASTE
243
Table 2. Pre- and post-closure monitoring parameters at WIPP (DOE 1996). Monitoring parameters
Pre-
Post-
Basis for measurement
Salado1 physical parameters Creep closure Extent of deformation Initiation of brittle deformation Displacement of deformation features
Yes Yes Yes Yes
No No No No
Direct measurement in open areas of repository Direct measurement in open areas of repository Analysis of monitored data Direct observation and measurement in open areas of repository
Non-Salado Hydrological Properties Culebra2 water (brine) composition
Yes
Yes
Yes Yes Yes
Yes Yes Yes
Analysis of brine samples collected from water quality sampling programme wells Direct measurement from WIPP wells Analysis of well water levels, water level changes over time Based on drilling activity in the Delaware Basin
Yes
Yes
Based on drilling activity in the Delaware Basin
Waste-related parameters Waste radioactivity
Yes
No
Waste characterization information
Subsidence Subsidence
Yes
Yes
Direct measurements at benchmark locations
Yes
Can be measured and evaluated against predictions and baseline database
2
Culebra well water level Culebra2 groundwater flow direction Castile formation3 brine reservoir location and pressure Drilling practices (including drilling intensity, borehole plugging)
Repository performance parameter Subsidence
Compiled from Tables MON-1 and MON-5 of Appendix MON [DOE 1996] 1 Salado Formation, in which WIPP is located 2 Culebra Dolomite lies above WIPP and provides a potential path to the accessible environment 3 Castile Formation lies below WIPP and contains pockets of pressurized brine
limits specified in the federal regulations, constitutes a violation of this standard.
Role of monitoring at WIPP In general, monitoring activities at WIPP site are aimed at one or more of the following objectives: fulfilling governing requirements and compliance with federal and state regulations, as well as health and safety considerations; providing characterization data, used to support site selection; providing baseline data; and providing data for performance assessment needs and for testing predictions and reducing uncertainties. In the USA, PA forms the basis of the evaluation of compliance with federal regulations (40 CFR Part 191, Subparts B and C). The focus in the following sections is on the first of the above objectives. Provision for monitoring associated with long-term compliance with EPAs
requirements is stated in federal regulation 40 CFR Part 191 (14b): Disposal systems shall be monitored after disposal to detect substantial and detrimental deviations from expected performance. This monitoring shall be done with techniques that do not jeopardize the isolation of the wastes and shall be conducted until there are no significant concerns to be addressed by further monitoring.
The compliance criteria require pre-closure monitoring of significant parameters affecting the disposal system's performance, e.g. brine inflow, composition and spatial distribution, stress measurements, and subsidence. Such monitoring must also be in place prior to the emplacement of any waste. Once the shafts have been backfilled and sealed, the site is deemed to be closed and post-closure monitoring must be conducted. Examples of post-closure monitoring provided in 40 CFR Part 194 include geophysical methods (surface subsidence, groundwater monitoring, geo-electrical properties and seismic, gravity and magnetic properties). Table 2 contains a list of the parameters identified for preand post-closure monitoring at WIPP.
244
M. J. STENHOUSE & D. SAVAGE
DOE's Compliance Monitoring Programme (CMP) was developed to support the DOE 5-year WIPP site re-certification process (DOE 1999). (The WIPP site must be re-certified every 5 years, EPA 1996). The intent of the CMP is to be able to recognize anomalous conditions and report specific performance conditions to EPA. Ten Compliance Monitoring Parameters (COMPs) are based on safety assessment work and sensitivity analyses and are monitored under the CMP. They are (cf. Table 2 parameters): groundwater composition of the Culebra Dolomite (Table 2, footnote 2); change in groundwater flow through the Culebra Dolomite; probability of encountering a Castile Brine reservoir (Table 2, footnote 3); drilling rate; subsidence measurement; waste activity (radioactivity); creep closure and stresses; extent of deformation; initiation of brittle deformation; displacement of deformation features. A detailed process exists for evaluating complianceparameter-related monitoring data and observations against independently-derived trigger values (Sandia National Laboratories 2000). A trigger value is a measure or limiting value for a particular set of compliance-parameter-related monitoring data, that when exceeded, indicates the data may represent a significant deviation from the Compliance Baseline, and merit further evaluation/analysis. Pre-closure monitoring Data for the ten compliance-related monitoring parameters are generated by five separate monitoring programmes: geotechnical, groundwater, environmental, Delaware Basin drilling, and Subsidence. These programmes are discussed briefly below. Geotechnical monitoring programme. In accordance with the WIPP Hazardous Waste Permit (New Mexico Environmental Department 1998), the purpose of this programme is to obtain in-situ data to support the continuous assessment of the design of the underground facilities. The programme provides: early detection of conditions that could affect operational safety; guidance for design modifications and remedial actions; and data for interpreting the behaviour of underground openings, in comparison with established design criteria. Groundwater monitoring programme. The original purpose of the groundwater monitoring programme (GMP), initiated in 1985 (under a different name),
was to define background groundwater quality and quantity near the WIPP site. The continuing objective of this programme is to collect representative and reproducible groundwater samples from waterbearing zones in the area of the WIPP site, to provide defensible data for PA, as well as for regulatory compliance and permitting. Water-level monitoring is included in this programme, the objective being to extend the documented record of water-level fluctuations in the Rustler Formation (above WIPP). Some overlap exists between the GMP and the environmental monitoring programme. Environmental monitoring programme. The goal of the environmental monitoring programme is to determine if the local ecosystem has been affected during the pre-disposal and disposal phases of the WIPP and, if so, to evaluate the severity, geographical extent, and environmental significance. The programme has both radiological and non-radiological components and involves the collection of environmental media for subsequent analysis, normally using routine laboratory chemical and radiochemical techniques. Prior to operations, the programme involved a compilation of radiological and non-radiological environmental data, thereby providing essential baseline data as well as enabling an evaluation of seasonal changes. This programme is similar to the programmes conducted routinely at all nuclear facilities. Non-radiological portions of the programme focus on the immediate area surrounding the site, whereas radiological surveillance generally covers a broader geographical area including nearby ranches, villages and cities. Meteorological monitoring is also carried out. Environmental monitoring is planned to continue for 2-5 years after decommissioning of the site, in accordance with a formal agreement between DOE and New Mexico. Delaware Basin drilling monitoring programme. The EPA provides criteria in 40 CFR Part 194 (section 33) that addresses human intrusion by taking into account future and shallow drilling in performance assessment. Accordingly, DOE has used the historical rate of drilling for resources in the Delaware Basin (in which the WIPP site is located) as a basis for calculating a future drilling rate. The purpose of the Delaware basin drilling monitoring programme (DOE 1997) is to provide for active monitoring of drilling activities within the Delaware Basin, with specific emphasis on the ninetownship area (about 2000km2) that includes the WIPP site. The monitoring programme is one of recording and compiling information and data rather than technical measurements. It addresses a number of parameters regarding drilling, which are incorporated in PA calculations addressing the possibility of
MONITORING STRATEGIES FOR NUCLEAR WASTE
human intrusion from drilling activities at the site at some time in the future. Subsidence monitoring programme. Within this programme, subsidence data are compiled for comparison with subsidence predictions. The goal of the study is to determine the most reliable subsidence predictions for the repository and to define the bounding limits within which acceptable subsidence repository performance may be defined. Subsidence monitoring carried out during the operational period of the facility generates data that help to establish a baseline against which long-term subsidence monitoring data and information may be evaluated. Note that subsidence was selected by DOE as the parameter indicative of repository performance (see next section). Post-closure monitoring. The post-closure programme represents a significantly reduced form of the pre-closure monitoring, its main components being: geophysical survey; radiological environmental monitoring; and subsidence monitoring. The planned post-closure geophysical survey will comprise seismic, resistivity, environmental monitoring, gravitational and radiological aerial surveys. The radiological environmental monitoring programme (currently a component of the environmental monitoring programme) will be conducted in two phases; the existing (operational) programme will be extended for two years after site closure. Thereafter, a reduced version of this programme will be conducted. Note, as shown in Table 2, that post-closure monitoring of the repository will use subsidence monitoring as the repository's primary performance indicator. Subsidence monitoring is discussed below.
Monitoring techniques Only the geotechnical and subsidence monitoring programmes at WIPP are discussed here, as being of some relevance to long-term CO2 storage. Geotechnical monitoring. Geotechnical monitoring incorporates geomechanical, geological and excavation effects monitoring. The instrumentation for Geomechanical, or rock mechanics monitoring includes tape extensometer stations, convergence meters, borehole extensometers, piezometers, embedment strain gauges, stress gauges, inclinometers, load cells and crackmeters. Instrument sensitivity is such that small changes in rock displacements (resolution 0.001 in. or 0.0025cm) and rock stresses
245
(resolution 1 |min./in. or 10 6) can be detected. This resolution is not a specific requirement, but was determined as the minimum necessary to establish significant changes in stress. The geomechanical datalogging system consists of a surface computer and a network of computers at various locations underground. The underground data acquisition system consists of instruments, polling devices, and a communicating network. Polling devices are connected by datalink cables and modems to a surface computer. Geomechanical data are subsequently entered into database files of a geographical information systems (GIS) data processing system. The principal activities of geological monitoring include seismic monitoring and evaluation, and geological and fracture mapping. The fracture maps, based primarily on visual examination, document the propagation of fractures through time, thereby enabling the analysis of strain in the immediate roofbeam. Fracture surveys, typically performed on an annual basis, are just one of the basic methods of monitoring for excavation effects. Subsidence monitoring. Subsidence is defined as vertical movement of the land surface anywhere in the subsidence basin. This movement can be up (uplift) or down (subsidence) and is relative to an assumed fixed reference. It is assumed that the reference is fixed, even though it may be subject to the same factors that cause the surface movement and may also be moving. Subsidence can be caused by a variety of factors including mining, hydrocarbon extraction, water injection and extraction, geological deformation, and formation dissolution. Typically, a levelling survey is used to monitor subsidence by measuring the vertical height difference between an array of markers on the surface. Usually, one reference benchmark (ideally outside the potential subsidence basin) is used as the standard and the relative movement of other stations or benchmarks is measured to detect vertical movement over time. A first-order survey has an error of one part in 100000; a second-order survey has an error of one part in 20000. With current technology, several thousandths of an inch vertical movement can be measured to this precision. State-of-the-art levelling equipment is specified for subsidence monitoring, the technical specifications being: Standard deviation: electronically 1.5mm; optically 2.0mm; Range: electronically 1.8-100m; optically 0.6m up wards; Distance measurements: electronically 3-5 mm; optically 0.2-0.5 m; Temperature range: measurement -20-50°C; storage -40-70°C.
246
M. J. STENHOUSE & D. SAVAGE
Conclusions The previous sections give some indication of the breadth and scope of monitoring conducted in support of site programmes for nuclear waste disposal. In discussing such monitoring, it is useful to recognize the different stages in development of the disposal site. Monitoring techniques used at the surface or borehole stage of site characterization for nuclear waste disposal, for example, are expected to be similar to those used to characterize potential CO2 storage reservoirs. The main difference in the nature of the characterization information or data in the former case occurs once the decision is made to excavate down to a formation that is a potential host for the repository, thereby allowing the application of in-situ monitoring and investigation techniques. Geochemical data are important for the characterization of both potential waste disposal sites and CO2 storage reservoirs. Experience from site characterization work in support of national nuclear waste disposal programmes emphasizes the importance of collecting geochemical data as early as possible so that perturbation of the natural system is minimal and that interpretation of subsequent perturbations is achievable. In-situ monitoring can continue through repository construction and operation, allowing the compilation of data important to supporting assessment calculations of long-term safety, while also assuring safety and environmental protection during the operational phase. This situation can be compared with the use of sophisticated seismic surveys, for example, which might be used to help identify CO2 flooding patterns in a CO2 storage reservoir. Post-closure monitoring at deep nuclear waste disposal sites, if carried out at all (not planned for many national waste disposal programmes in Europe but a regulatory requirement in the USA, as confirmation of performance) will almost certainly involve non-intrusive monitoring concepts, e.g. monitoring of groundwater beyond the repository area, of surface soil gas, surface uplift or subsidence, and seismicity. In this situation, similar types of monitoring could be anticipated for remote monitoring of a CO2 storage reservoir. The difficulty in such cases will be an accurate interpretation of the resultant data to avoid unnecessary remedial action and associated safety concerns. Post-closure monitoring should not be necessary to demonstrate post-disposal safety. Its main role will be one of generating public confidence, in providing confirmation to the public that radioactive material is safely contained in the repository and will remain so for the foreseeable future. Surface monitoring of a closed geological repository may be carried out, to provide assurance that the quality of the environment is maintained. However, long-term
monitoring for thousands of years is impractical and, in any case, is contrary to the generally-accepted philosophy that nuclear waste disposal should not be a burden on future generations. The authors are grateful to the reviewers for constructive comments which helped to guide substantial improvements in the clarity of the text.
References ALEXANDER, W. R. 1991. Redox state of the Grimsel Test Site (GTS) Radionuclide Migration Experiment groundwater/rock system. Nagra Technical Note AN 90-216. Nagra, Wettingen, Switzerland. ALMEN, K-E. & ZELLMAN, O. 1991. Aspo Hard Rock Laboratory. Field investigation methodology and instruments used in the Pre-Investigation Phase, 1986-1990. SKB Technical Report TR 91-21. Swedish Nuclear Waste Management Co, (SKB), Stockholm, Sweden. BIRGERSSON, B., WIDEN, H., AGREN, T & NERETNIEKS, I. 1992. Tracer migration experiments in the Stripa Mine. Stripa Project Technical Report TR 92-25. Swedish Nuclear Waste Management Co, (SKB), Stockholm, Sweden. DAVIS, S.N. & NORDSTROM, K. 1992. Hydrogeochemical investigations in boreholes at the Stripa Mine. Stripa Project Technical Report TR 92-19. Swedish Nuclear Waste Management Co, (SKB), Stockholm, Sweden. DOE 1996. Preclosure and Postclosure (Long-Term) Monitoring Plan, Appendix MON. 40 CFR Part 191 Compliance Certification Application for the Waste Isolation Pilot Plant. Document DOE/CAO1996-2184, 21 volumes. DOE Carlsbad Area Office, New Mexico. DOE 1997. Delaware Basin Drilling Surveillance Plan. DOE WIPP Document WP 02-PC.02, Revision 0, March 1997. DOE, Carlsbad Area Office, Carlsbad, New Mexico. DOE 1999. 40 CFR Parts 191 and 194 Compliance Monitoring Implementation Plan. Document DOE/WIPP-99-3119. DOE, Carlsbad Area Office, Carlsbad, New Mexico. EPA 1993. Environmental Radiation Protection Standards for the Management and Disposal of Spent Nuclear Fuel, High-Level and Transuranic Radioactive Wastes, 40 CFR Part 191. Final Rule, Federal Register 58, 66398 (December 20, 1993). U.S. Environmental Protection Agency, Radiation Protection Division, Washington, D.C. EPA 1996. Criteria for Certification and Re-certification of the Waste Isolation Pilot Plant's Compliance with 40 CFR 191 Disposal Regulations, Final Rule, 40 CFR 194, Federal Register 63, 27353-27406 (February 9, 1996), U.S. Environmental Protection Agency, Radiation Protection Division, Washington, D.C. FRICK, U., ALEXANDER, W. R., BAEYENS, B. & 9 OTHERS. 1992. The Radionuclide Migration Experiment Overview of Investigations 1985-1990. Nagra Technical Report 91-04. Nagra, Wettingen, Switzerland
HAUTJARVI, A., VIENO, T, AUTIO, J., JOHANSSON, E.,
MONITORING STRATEGIES FOR NUCLEAR WASTE OHBERG, A. & SALO, J-P. 1994. Characterisation and tracer tests in the full-scale deposition holes in the TVO Research Tunnel. In: GEOVAL '94: Validation through model testing. Proceedings NEA/SKI symposium, Paris, France, October 11-14,1994. JOHNSON, L. H., TAIT, J. C, SHOESMITH, D. W., CROSTHWAITE J. L. & GRAY, M. N. 1994. The disposal of Canada's nuclear fuel waste: engineered barriers alternatives. Atomic Energy of Canada Limited Report AECL-10718, COG-93-8. AECL, Pinawa, Manitoba, Canada. KENDALL, R. R., WINARSKY, R., DAVIS, T. L. & BENSON, R. D. 2003. 9C, 4D seismic processing for the Weyburn CO2 flood in the Weyburn Field, Saskatchewan, Canada. Proceedings of the 2003 Canadian Society of Exploration Geophysicists Conference, Calgary, Alberta, June 2-4, 2003. CSEG Conference Abstracts, Calgary, Alberta, Canada.. Materials Safety Data Sheet, Occupational Safety and Health Act format. Carbon Dioxide. NEA 1988. Geological Disposal of Radioactive Waste: In Situ Research and Investigations in OECD Countries. Status Report, Nuclear Energy Agency Advisory Group. NEA/OECD, Paris, France. NEA 1991. Review of Safety Assessment Methods Disposal of Radioactive Waste. Report of the Nuclear Energy Agency Performance Assessment Advisory Group. NEA/OECD, Paris. NATIONAL RESEARCH COUNCIL 1996. The Waste Isolation Pilot Plant: A Potential Solution for the Disposal of Transuranic Waste. National Academy Press, Washington D.C. NEW MEXICO ENVIRONMENTAL DEPARTMENT 1998. WIPP Hazardous Waste Facility Permit, NM 4890139088TDSF. NMED, Santa Fe, New Mexico. OLSSON, O.J. 1992. Site characterisation and validation project - Final Report. Stripa Project Technical Report TR 92-22. Swedish Nuclear Waste Management Co, (SKB), Stockholm, Sweden. POSIVA OY 1999. The final disposal facility for spent nuclear fuel. Environmental Impact Assessment
247
Report. Posiva Oy, Helsinki, Finland. SANDIA NATIONAL LABORATORIES 2000. An analysis plan for annually deriving compliance monitoring parameters and their assessment against performance expectations to meet the requirements of 40 CFR 194.42. Sandia National Laboratories Document AP-069. Sandia National Laboratories, Albuquerque, New Mexico. SKB 1992. Treatment and final disposal of nuclear waste. R&D Programme, Aspo Hard Rock Laboratory Background Report. Swedish Nuclear Waste Management Co, (SKB), Stockholm, Sweden. SKB 1993. Aspo Hard Rock Laboratory Annual Report 1992. SKB Technical Report TR 93-08. Swedish Nuclear Waste Management Co, (SKB), Stockholm, Sweden. STENHOUSE, M. J., ZHOU, W, CHALATURNYK, R., MORENO, F. & JAZRAWI, W. 2003. IE A CO2 Monitoring and Storage Project: Long-term assessment of fate of CO2: Treatment of abandoned wells. Proceedings of 2nd Annual Conference on Carbon Sequestration, Alexandria, May 5-8, 2003. Exchange Monitor Publications and Forums, Washington D.C. USA. CD ROM. STRUTT, M. H., BEAUBIEN, S. E., BEAUBRON, J. C. & 8 OTHERS. 2003. Soil gas as a monitoring tool of deep geological sequestration of carbon dioxide: preliminary results from the Encana EOR Project in Weyburn, Saskatchewan, Canada. In: GALE, J. & KAYA, Y. (eds) Greenhouse Gas Control Technologies. Proceedings of the 6th International Conference on Greenhouse Gas Control Technologies, October 1^, 2002, Kyoto, Japan. Pergamon, Oxford. Volume 1, 39-396. UK NIREX 1996. Sellafield geological and hydrogeological investigations: Rock Characterisation Facility, Longlands Farm, West Cumbria, Report on baseline groundwater pressures and hydrochemistry. UK Nirex Limited Science Report SA/96/006. UK Nirex Ltd., Harwell, U.K.
This page intentionally left blank
Index Page numbers in italic refer to figures, in bold refer to tables Aachen Formation 201-202 Aalburg Formation 201 acid-gas dewatering 227 injection 2, 11, 137 Alberta Basin, Canada 225-234, 226, 227 case histories 232-233 disposal wells 230 choice of location 230 environmental issues 230-231, 233 formation damage 231 phase behaviour 229 safety 227, 229, 231-232 subsurface issues 230, 231-232 surface operations 228-230, 231 interaction with subsurface formation 231 migration pathways 230 properties 227 sweetening 228 Alberta basin acid-gas injection 225-234 C02 138, 139, 141,142 CO2 injection 93, 94 CO2 solubility 90 flow rates 132, 134,735 Alpine Fold Belt 30 Anahuac Formation 149, 152, 153 andesite, Eocene 37 anhydrite-CO2 interaction 97, 98 aquiclude 87, 89, see also caprock aquifers drinking water 202-203 saline CO2 sequestration 1, 2, 11, 21-26, 134-138 Campine Basin 197,198 reactive transport simulation 109-126 screening criteria for CO2 storage 124-125 see also Saline Aquifer CO2 Storage project (SACS) aquitard 132 asphaltene precipitation, effect of CO2 injection 222 asphyxiation 38 bacteria, source of CO2 62 Badoit, mineral water 37 barriers, engineered 235, 236, 242 basins sedimentary 130 characteristics 132-133 CO2 source 129, 131 CO2 storage capacity 131 formation 132-133 geothermal gradient 138 geothermal regime 132-133 Belgian Coal Measure Group 194 Belgium, greenhouse gas reduction 193, 207-208 Black Warrior Basin 207
bleaching 55 blow outs, CO2 34 Blue Whale gas field, Vietnam 10 Blue Whale high CO2 accumulation 64-65, 66, 67, 78-79, 80 Bonanza Carboniferous reservoir, CO2-EOR 219, 220 borehole completions, reactions with CO2 88, 99-102 Bowen-Gunnedah-Sydney Basin system, CO2-mineral reactions 96, 119 Brae Fault, North Sea 66 Bravo Dome, New Mexico, high CO2 reservoir 10, 67, 71-73, 74, 76, 78, 79, 82-83, 95-96 brine formations, CO2 storage 147-162 brine injection experiments 216-222 Buckley-Leverett theory 140 buffering 73, 76, 79, 90, 92 Bunter sandstone 2,19, 21, 26, 33 reservoir properties 24, 25, 26 Buntsandstein Campine Basin 194, 201 Vorderrhon 33 calcarenite, Campine Basin 202-203 calcite precipitation 46, 47-48, 88, 119, 120-122, 125, 211 scale formation at CO2-EOR facilities 223 calcium silicate hydrate (CSH), interaction with CO2 101, 102 Campine Basin, Northern Belgium 2 carbonates 197-200 CO2 sequestration 193-209 coal reserves 204, 205, 206 coalbed methane reserves 203-206 geology 194, 795,196, 197, 203 karstification 194, 197-200 reservoir properties 198 capacity 151 Frio Formation 154, 156-157 Utsira Formation 178 caprock carbonate cementation 123-124 optimal performance 124 pore space ('membrane') 230 porosity-permeability 123-124 reaction with CO2 96-99 Utsira Formation 12-13, 23, 183 carbon emission trading schemes 13, 193 uptake by terrestrial ecosystems 17-18 carbon dioxide see CO2 carbonate mineral dissolution 76, 78-79 carbonate mineral precipitation 47^8, 76, 87-88, 96, 136-138,211 carbonate rind 120-122, 123, 125 ray-crystal crusts 47, 48, 52 see also cementation, carbonate carbonate rock, addition of CO2 80, 81 carbonate veins, isotope analysis 51-52, 53
250
carbonates Campine Basin 197-200 decarbonation 54 injection of CO2 1,95 modelling 91-94 Castile Formation brine 243, 244 CBM see coalbed methane cement carbonation in borehole completions 88, 99-102 compatibility with acid-gas 231-232 cementation carbonate, in shales 122-124 dawsonite 119-120 dolomite, dissolution 217, 220 Cenozoic, rift system 29, 30, 31, 39 CFCs (chlorofluorocarbons) 8 CH4 see methane Chaffin Ranch Geyser 55 chalk, Campine Basin 201-203 Chaunoy Formation 72-73, 77 climate change 1, 7 C02 acid-gas 225-228, 229 atmospheric 61, 62 levels 17-18 biogenic 32, 62 'bubble' 87, 89, 147 buoyancy 20, 141,142 capture 8-10, 18-19,59 chemical fixing 21 chimney 186 contribution to global warming 8 crusta!61,62, 63 density 20, 138-141,759 dissolution 3, 21, 65, 87-89, 135, 211 effect on seismic reflectivity 186-188 effect on subsurface 61 emission reduction 1, 107 flow paths 54-55, 88 fluid properties 138-141, 739 gas cap 32 gas pools 30, 31,34, 36 gas vents 31, 38 geochemical fate 3, 21 injection 20 ECBM production 206-207 effect on host rock experiments 94-95 simulation 91-94 effect on seismic data 181-190 effect on water-rock interaction field evidence 212-216, 217-220 simulation 216-217, 220 EOR projects 212-223 methods 212, 213 see also Utsira Sand Formation, CO2 injection leakage 37-39, 43-56 mantle 34, 37 methane production 2-3 migration 3, 21 immiscible 108, 111, 114 natural as analogues 3, 29^0, 43-56, 61-63, 63
INDEX reservoirs 10, 60 sources 43, 53-54, 61-63, 62, 65, 71 northern Europe, point sources 79 oil and gas recovery 2, 11, 21, 22 partial pressure 73 phase change 20, 32, 88,138 reactive transport simulation 109-126 in rock geochemistry 61, 73, 76-83 solubility 21, 89, 90 stripping 2, 109,11 supercritical 37, 87, 88-90, 97, 99, 138-139, 147 tuning effect 185,187, 188 viscosity 138,139, 140 volcanic 32, 62 CO2-brine injection 216-222 CO2-EOR programs 2, 11, 212-223 Colorado, USA 214, 215 experimental data 227 Hungary 213, 214, 275 transport 19 Wyoming, USA 214, 215-216 CO2-water reactions 87-90 CO2-water-rock interaction 3, 4, 21, 94, 96-97, 212-223 CO2ROCK simulator 91 coal mines, CO2 storage 206-207 coal rank-methane relationship 204 coal seams unmineable CO2 storage 1, 11 methane extraction 2-3 Westphalian 203-207 CO2 storage potential 206-207 coal reserves 204, 205 gas content 204 coalbed methane (CBM) 203-207, 205, 206 Colorado Plateau geology 43^4 natural CO2 reservoirs 43-56 Colorado, USA, CO2-EOR programs 214, 215 Compliance Monitoring Programme (CMP) 244 containment 87-88 acid-gas 231-232, 233 caprock-CO2 reaction 96-99 core flooding experiments 95, 216-222 corrosion, acid-gas 227, 228, 231 'cracks' 230, 233 Crooks Gap Carboniferous reservoir, CO2-EOR 218, 220 Crystal Geyser 44, 46, 47, 48, 54 water chemistry 50-51, 51, 52, 53 Culebra Dolomite 243, 244 Da Nang Basin, offshore Vietnam 64-65 dawsonite 80, 90, 93, 96, 98 intra-plume cementation 119-120, 125 decarbonation 54 decomposition, organic 62 degassing mantle 53, 62 natural 43-56 desulphurization 2, 225 diagenesis 62, 63, 64-65, 66, 67, 71, 78 Dilsen Fault System 194, 795
INDEX disequilibrium 108, 709 dissolution 65 CO2 3, 21,87-89, 135,211 dolomite 217, 220 K-feldspar 37, 79, 119-120 Visean 197 dolostone, CO2-water-rock interaction 214, 216 Dolphin low CO2 accumulation 64-65, 66, 67, 78-79, 80 Donderslag Fault Zone 194, 795, 199 Drax coal-fired power-station 18 ECBM see enhanced coalbed methane ecosystems, uptake of CO2 17-18 emission anthropogenic 7-8, 8, 17, 19, 193 trading schemes 13, 193 energy alternative 5, 13, 17 renewable 8, 13, 142 enhanced coalbed methane (ECBM) production 193, 206-208 CO2 injection 206-208 enhanced oil recovery (EOR) 2, 11, 21, 22, 212-223 CO2 injection methods 212, 213 transport 19 Entrada Sandstone 50, 55 environment, accessible 242 Europe natural CO2 accumulation 30 underground CO2 sequestration 17-27 evaporites Crystal Geyser system 51, 52 Triassic 37 faults 153 CO2 fluid flow 54-55 'low permeability', CO2 leakage 43-56 feldspar, dissolution 37, 79, 80, 119-120 fingering, viscous 89, 108, 232 fixing, chemical 21 Flanders, CO2 sequestration 193-209 flaring, gas 225 Fiorina CO2 field 38 flow fluid 20, 54 down-dip 134, 141 in sedimentary basins 132-133,133, 134 up-dip 3, 55 folds 153 forests, CO2 sequestration 9 fractures 230 Montmiral reservoir 37 Vorderrhon reservoir 31-34 France, CO2 accumulations 36 Frio Formation 149-150, 152-154 fuel, fossil combustion 1, 17 in sedimentary basins 129 gas accumulations 31 acid see acid-gas
251
CO2 stripping 2, 11, 109 flaring 225 greenhouse 8 anthropogenic 7-8, 8, 13, 107, 193 reduction 1, 4-5, 7-9, 13, 59, 193 natural sour-gas 225, 228 storage 10, 198, 199 non-condensable (NCG) 39 recovery, CO2 enhanced 2 gas cap 32 gas fields, CO2 sequestration 21 gas pools, CO2-rich 30, 31, 34, 36 gas vents (moffettes) 31, 38, 39 Gassmann modelling 183-184, 186 GEMBOCHS database/software 108, 109 Genendijk-Hulst structure 795, 199 geochemistry CO2 3, 21 rock 73, 76-83 geysers 44, 46, 47, 48, 49-51, 55 Glauconite Sandstone aquifer, CO2 injection simulation 93,94 Glen Ruby 1-X well 44, 54 global warming 1, 7-8, 18 Global Warming Potential (GWP) 8 Gorgon gas field, Australia, CO2 stripping 2 gravimetry 34 Green River 45 gas seeps 46 gross pushdown factor 186-188 Gulf Coast sandstones 148 Gulpen Formation 202 Heibaart dome 198-200 Helchteren Formation 194, 201 heterogeneity effect on sequestration effectiveness 151 effect on storage capacity 151 geological 141,142, 147-162 modelling 149-150 Frio Formation 153-161 stratigraphic 152 structural 153 Utsira Sand Formation 175-176 HFCs (hydrofluorocarbons) 8 Hoogstraten fault 194, 795 Hordaland Formation 167, 168, 171 Houston, Texas CO2 emissions 148 geological heterogeneity 148-153 Houthem Formation 202 H2S in acid-gas 225-228, 229 storage 2, 137 Hungary CO2 gas fields 34, 35 leakage 37-38 C02-EOR programs 213, 214, 275 hydrate, formation in acid-gas 228, 229-230, 232 hydrocarbon migration 3 as source of CO2 53-54 hydrogen sulphide see H2S
252
INDEX
injection see CO2, injection injectivity 26 International Panel on Climate Change (IPCC) 1, 7 isotope analysis Da Nang Basin CO2 accumulation 64 travertine 51-52, 53 K-feldspar, dissolution 119 karstification Campine Basin 194, 197-200 Matraderecske 37-38 kerogen, decarboxylation 62, 63 Keuper Formation 194, 201 Kupferschiefer 31 Kyoto Protocol 8, 13, 193, 207 Latera geothermal field 38-39 leakage 37-39, 43-56, 133 acid gas 229, 230-232, 233 Leut Fault 194, 795 Little Grand Wash fault, Utah CO2 leakage 43, 44-19 geology 44, 45, 46 mineral deposits 47^9 Maastricht Formation 202 magnesite precipitation 98, 119, 120-122, 123 Magnus oil field, North Sea, low CO2 reservoir 65, 67, 68,69,70,71,79,80 Massif Central 36, 37 Matraderecske, CO2 seepage 37-38 Mercia Mudstone Group 97 methane (CH4) in acid-gas 227-228, 229 effect on density of CO2 140 Latera geothermal field 39 production 1 reduction 8 Westphalian Coal Measures, Campine Basin 203, 204, 206 see also coalbed methane methanogenesis 62 migration acid-gas 230 CO2 3, 12-13, 21 immiscible 108, 114-116 in natural accumulations 29, 34, 37-39 Utsira Sand Formation 12-13, 176-178 Mihalyi-Repcelak reservoir 34, 35 Miller oil field, North Sea, high CO2 reservoir 65-66, 67, 68,69,70,71,79,80 minerals, aquifer, reaction with CO2 90-96 mines, coal, CO2 storage 206-207 Moab fault 54 modelling CO2-mineral interaction 91-94 Gassmann 183-184 geochemical 39, 141 CO2 addition to geological systems 79-83, 91-94 impact of heterogeneity on sequestration 149, 153-161 reactive transport 108-109 hydrological and compositional data 110-112 thermodynamic and kinetic data 113-114 Utsira Sand reservoir 176-179, 183-189
mofetta see gas vents moffettes see gas vents monitoring compliance-related 243-244 drilling, Delaware Basin 244-245 environmental 244 geophysical 23-24 geotechnical 244, 245 groundwater 244 injected CO2 4, 11-13, 23-24 nuclear waste disposal 235-239 operational phase 240-241 post-closure 238-239, 241, 243, 245 pre-closure 243, 244 pre-operational 240 rock mechanics 245 seismic 12, 181-190 subsidence 245 WIPP 241-245 Montmiral reservoir 34, 36, 37 Montreal Protocol 8 mud edifices, Utsira Sand Formation 167, 168, 769, 770, 171,173, 181 Muschelkalk Formation 194, 201 Natuna gas field, Indonesia 10 Natural Analogues for the Storage of CO2 in the Geological Environment (NASCENT) project 29,31 NCG see gas, non-condensable Neeroeteren Formation 200-201, 207 N2O (nitrous oxide) 8 Nordland Shale 167, 172 North Sea Basin 19 oil fields, CO2 content 65-71, 68 sequestration reservoirs 21-26 Norway, CO2 emission 2 nuclear power, as alternative to fossil fuel 8 nuclear waste disposal see waste disposal, nuclear NUFT reactive transport simulator 108-109 oceans, storage of CO2 9-10, 13, 19-20 oil fields, CO2 sequestration 21-22 oil recovery, CO2 enhanced see enhanced oil recovery (EOR) Pannonian Basin 30, 34, 35 Paradox Basin, Utah geology 43-44 oil and gas deposits 44, 53, 54 Paradox Salt 54 Paris Basin 72-73, 77,91 Pembina Cardium Formation, CO2-water-rock interaction 214, 217 performance assessment (PA), nuclear repository sites 237, 243, 244 permeability Montmiral reservoir 37 sedimentary basins 132 Utsira Sand Formation 175 Perrier, mineral water 37 petroleum, North Sea reservoirs 65-71 PFCs (perfluorocarbons) 8 pH evolution 116-119
INDEX pipelines 13, 19, 208 plume, immiscible interaction with formation water 116-119 interaction with shale 123-124 migration 114-122, 124, 138 monitoring 23-24 Poederlee antiform 198-199 pore-water 20 porosity Frio Formation 150 monitoring 23-24 Montmiral reservoir 37 Rotliegend31 secondary 37, 39 Utsira Sand Formation 23, 174-175 porosity-permeability evolution 122, 123-124 power plants CO2 capture 1, 18-19,211 CO2 emissions 1,18, 147, 148 precipitation calcite 46, 47-48, 88, 119, 120-122, 125, 211 carbonate 3, 76, 87-88, 211 carbonate rind 120-122, 123, 125 dawsonite 90, 98, 119-120 magnesite98, 119, 120-122, 123 pressure fluid 73 reservoir 176 well head 184 'punctures' 230-233 'pushdown' effect 181, 186-188,189 radon 38 reactions in cement 99-102 dissolved CO2-aquifer minerals 90-96 dissolved CO2-caprock 97-99 in rock, CO2 addition 73, 76-83, 87-103, 89 supercritical CO2-borehole completions 99-102 supercritical CO2-caprock 97 supercritical Conformation water 88-90 within borehole completion 99-102 within caprock 96-99 within host formation 88-96 reef mounds 198-199 reflectivity, seismic 186-188 reforestation 9, 17-18 regulation 4, 13 Repcelak see Mihalyi-Repcelak reservoir reservoir simulation 13 reservoirs carbonate 64-65 degassing 43-54 fractures 31-34, 37 geochemical models 39 geological barriers to implementation 11-13 risk assessment 13 storage of CO210-11, 13-14, 20-27, 197, 198 natural CO2 31^0 oil and gas CO 2 1,2 CO2 sequestration 21, 140-141, 197
253
quartzose sandstone 65-66, 69, 71 red-bed sandstone 71-73 Utsira Formation 18, 22-24, 165-179 CO2 injection 182 geology 165, 167-168, 171 heterogeneity 175-176 model 176-179, 183-186,184 net/gross ratio 175 permeability 175 porosity 23, 174-175 pressure and temperature 176 Rhenish Massif 32 rind, carbonate 120-122, 123, 125 Rio Grande rift system 72 risk assessment, geological storage 13 Roer Valley Graben 194, 197, 201-202, 207 Rotliegend CO2 reservoir 31, 33 SACS see Saline Aquifer CO2 Storage project Salado Formation 241, 243 Saline Aquifer CO2 Storage project (SACS) 12-13, 165, 181 salt deformation 44, 54 structures 24, 25, 153,154 Salt Norm 51,52 Salt Wash faults, Utah CO2 leakage 43-56 geology 45, 46 salting-out 89, 117 San Juan Basin, USA, coal seam storage of CO2 11 sandstone bleached 44, 50, 55 carbonate cemented, CO2-water-rock interaction 214, 216 sandstone, addition of CO2 80, 81, 82 laboratory experiments 94-95 modelling 91-94 saturation, immiscible 114-116 scale at CO2-EOR facilities 223 Fiorina CO2 field 38 screening criteria, storage sites 124-125, 240 seals borehole completions, reaction with CO2 99-102 caprock 23, 96, 124 sedimentary basins see basins, sedimentary segregation, gravity 108, 115 sequestration 17-27, 108, 157-161, 223 CO2 8-9, 13, 17-27 deep ocean 1, 9-10, 13, 19-20 effectiveness 151, 157-161 impact of folding and faulting 153 impact of heterogeneity, modelling 153-157 natural analogues 59-83 Colorado Plateau 55-56 options 8-9, 13 underground 20-27 chemical reactions 87-102 environmental issues 26-27 Westphalian Coal Measures 203-207 serpentine, reaction with CO2138 SF6 (sulphur hexafluoride) 8
254
INDEX
shale intra-aquifer 114, 115-117, 118, 122 carbonate cementation 122-124 Frio Formation 153-154 Utsira Sand 12, 24, 111, 172-173 mineral trapping 123-124 porosity 98 siderite 119, 137 simulation see modelling sinks, natural 9, 13 Sleen Formation 201 Sleipner gas field, North Sea 19,166 borehole completion 102 CO2 injection project 11, 72, 22-24, 25, 165, 182, 182 reactive transport simulation 108-126,110, 111, 112 seismic monitoring 72, 181-190 CO2 injection site 174,182 CO2 migration 4, 178 CO2 stripping 2, 11 see also Utsira Sand Formation Sleipner-like settings, site screening criteria 124-125 soil, CO2 sequestration 9 sour-gas sulphur recovery 225 sweetening 228 Southeast Basin, France 34, 36, 37 Southern Permian Basin 31 springs CO2-rich 29, 30, 31, 36, 37, 38 Little Grand Wash fault, Utah 44, 46^7, 52, 55 Salt Wash faults, Torrey's Spring 50, 52, 55 steel, corrosion by CO2 99-100 storage barriers to implementation 11-13 Campine Basin 193-209 capacity, Utsira Sand Formation 178 in coal seams 206-207 effect of geological heterogeneity 4, 141,142, 147-162 effectiveness 11, 151 efficiency 138-141 geological options 10-11 barriers to implementation 11-13 monitoring CO2 240-241 nuclear waste 235-239 operation 87-88, 89 permanence 9 regulation 4, 13 requirements 131 security 133-138, 141, 142,143 in sedimentary basins 129-142 site monitoring 4, 235-246 site screening criteria 124-125, 240 subsurface 1-5, 59 modelling 61 see also sequestration Stuffken relationship 204 sulphur recovery 225 survey, seismic, Utsira Sand Formation 166,169,170, 173, 177-178 sweetening, gas 2, 11, 228 Tenmile Geyser 49-50, 55 water chemistry 50-51, 51, 52, 53
Tenmile Graben see Salt Wash faults Tensleep Formation, CO2-water-rock interaction 94, 220, 222 Texas Gulf Coast, stratigraphy 148-149,149 tobermorite, interaction with CO2 101 Torrey's Spring 50 water chemistry 50-51, 51, 52 TOUGH2 simulator 149,153 Tournaisian carbonates 197-200 trading schemes see carbon, emission trading schemes trapping geochemical 134-138 hydrodynamic 3, 21, 88, 89, 108, 114-116, 134, 135 hydrogeological 133 hydrostratigraphic 134 ionic 135-137 mineral 88, 90-91, 108, 119-124, 135, 136-137 in saline aquifers 114-124 solubility 108, 116-119,135 stratigraphic 3, 4, 39, 133,134 structural 108, 133 traps Bunter Sandstone 24 closed 133,134, see also reservoirs; trapping, stratigraphic open 134, see also trapping, hydrodynamic travertine deposits 38, 44, 46, 47-48, 49-50, 54 fluid flow 54-55 isotope analysis 51, 52 Tri Ton Horst, offshore Vietnam 64 Tubb Formation 72, 95-96 Tumbleweed Geyser 55 tuning effect 185,187, 188 United Kingdom, greenhouse gas reduction 4 Umbrella Point field 152 underground research laboratory (URL), monitoring 237, 238 United Nations Framework Convention on Climate Change (UNFCC) 7-8 URL see underground research laboratory United States Enhanced Oil Recovery 2 fossil fuel combustion 1 Global Climate Change Initiative 4-5 Utsira Sand Formation 19 caprock properties 23, 183 CO2 injection 11, 12, 21, 22-24, 109-110, 165, 777, 181-182 injection site 174, 782,183,184 reactive transport simulation 110-126 CO2 migration model 4, 12-13, 176-179, 777 geology 165, 167-168, 181-182 mineral precipitation simulation 93 mud edifices 167, 168,169,170, 111, 173, 181 petrography 171, 772 petrophysical properties 182-183 reservoir geology 165-179, 181-182 reservoir model 176-179, 183-189 reservoir properties 22-23, 174-176 sedimentation rates 161,168 seismic monitoring 23-24, 181-190
INDEX seismic survey 22,166,167,169,170,173, 177-178, 181 seismic velocity 183-189 shale 172-173, 183 storage capacity 178 Vaals Formation 201-202 Variscan uplift 194 velocity pushdown effect 24, 181, 186-188,189 seismic, Utsira Sand Formation 183-189 Vert le Grand, Paris Basin, low CO2 reservoir 67, 71-73, 74, 75, 78, 79, 82 Vichy, mineral water 37 Vietnam, Blue Whale gas field 10, 64-5 Viking Graben 65, 66, 69 Visean carbonates 197-200, 207 volcanism roleofCO 2 61 Tertiary 29, 30, 31,37, 39 Vorderrhon CO2 deposits 30, 31-34, 33 WAG see water-alternating-gas wanning, global see global warming waste disposal nuclear monitoring 235-239 performance assessment 237 Waste Isolation Project Plant (WIPP), New Mexico 235, 237, 241-245 monitoring 243-245
water CO2-rich 29, 31,39 flow 54-55 mineral water industry 37 drinking, Campine Basin 202-203 formation displacement by injected CO2 108 reaction with acid-gas 231 reaction with CO2 73, 77-78, 87-90, 116-119 in sedimentary basins 132 geyser, chemistry 50-51 solubility in acid-gas 227, 225, 229 water-alternating-gas (WAG) 212 wavelet determination 184-185 Weber Formation, 220, CO2-water-rock interaction 214, 217, 220 Weber Sandstone, CO2 injection simulation 95 Werra potash mine 31, 32 Werra Rock Salt 31, 32, 34 Westphalian Coal Measures CO2 sequestration 203-207 geology 203 Weyburn oil field, Canada EOR/CO2 storage 2, 11 Monitoring Project 12, 13 Woodside Geyser 55 Wyoming, USA, CO2-EOR programs 214, 215-216 Zechstein Limestone 31, 32, 33 Zechstein Salt 24, 25, 34
255