DEVELOPMENTS IN WATER SCIENCE 52
UNDERGROUND INJECTION SCIENCE AND TECHNOLOGY
i
OTHER TITLES AVAILABLE IN DEVELOPMENTS IN WATER SCIENCE: 41. D. STEPHENSON AND M.S. PETERSON WATER RESOURCES DEVELOPMENT IN DEVELOPING COUNTRIES 43. J. ZÁRUBA WATER HAMMER IN PIPE-LINE SYSTEMS 44. W.E. KELLY AND S. MARES (EDITORS) APPLIED GEOPHYSICS IN HYDROGEOLOGICAL AND ENGINEERING PRACTICE 46. G.A. BRUGGEMAN ANALYTICAL SOLUTIONS OF GEOHYDROLOGICAL PROBLEMS 47. S.M. HASSANIZADEH, R.J. SCHOTTING, W.G. GRAY AND G.F. PINDER COMPUTATIONAL METHODS IN WATER RESOURCES 48. LENA M. TALLAKSEN AND HENNY A.J. VAN LANEN HYDROLOGICAL DROUGHT 49. P. WILDERER AND S.WUERTZ MODERN SCIENTIFIC TOOLS IN BIOPROCESSING 50. A.S. ALSHARHAN AND W.W. WOOD WATER RESOURCES PERSPECTIVES: EVALUATION, MANAGEMENT AND POLICY 51. S.K. JAIN AND V.P. SINGH WATER RESOURCES SYSTEMS: PLANNING AND MANAGEMENT 54. S.E. JØRGENSEN, H. LOFFLER, W. RAST AND M. STRASKRABA LAKE AND RESERVOIR MANAGEMENT 55. C.T. MILLER, M.W. FARTHING, W.G. GRAY AND G.F. PINDER COMPUTATIONAL METHODS IN WATER RESOURCES (Volumes 1 and 2)
ii
DEVELOPMENTS IN WATER SCIENCE 52
UNDERGROUND INJECTION SCIENCE AND TECHNOLOGY EDITED BY CHIN-FU TSANG and JOHN A. APPS EARTH SCIENCES DIVISION LAWRENCE BERKELEY NATIONAL LABORATORY BERKELEY, CALIFORNIA USA
2005
Amsterdam – Boston – Heidelberg – London – New York – Oxford Paris – San Diego – San Francisco – Singapore – Sydney – Tokyo iii
ELSEVIER B.V. Radarweg 29 P.O. Box 211, 1000 AE Amsterdam The Netherlands
ELSEVIER Inc. 525 B Street, Suite 1900 San Diego, CA 92101-4495 USA
ELSEVIER Ltd The Boulevard, Langford Lane Kidlington, Oxford OX5 1GB UK
ELSEVIER Ltd 84 Theobalds Road London WC1X 8RR UK
© 2005 Elsevier B.V. All rights reserved. This work is protected under copyright by Elsevier B.V., and the following terms and conditions apply to its use: Photocopying Single photocopies of single chapters may be made for personal use as allowed by national copyright laws. Permission of the Publisher and payment of a fee is required for all other photocopying, including multiple or systematic copying, copying for advertising or promotional purposes, resale, and all forms of document delivery. Special rates are available for educational institutions that wish to make photocopies for non-profit educational classroom use. Permissions may be sought directly from Elsevier's Rights Department in Oxford, UK: phone (+44) 1865 843830, fax (+44) 1865 853333, e-mail:
[email protected]. Requests may also be completed on-line via the Elsevier homepage (http://www.elsevier. com/locate/permissions). In the USA, users may clear permissions and make payments through the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA; phone: (+1) (978) 7508400, fax: (+1) (978) 7504744, and in the UK through the Copyright Licensing Agency Rapid Clearance Service (CLARCS), 90 Tottenham Court Road, London W1P 0LP, UK; phone: (+44) 20 7631 5555; fax: (+44) 20 7631 5500. Other countries may have a local reprographic rights agency for payments. Derivative Works Tables of contents may be reproduced for internal circulation, but permission of the Publisher is required for external resale or distribution of such material. Permission of the Publisher is required for all other derivative works, including compilations and translations. Electronic Storage or Usage Permission of the Publisher is required to store or use electronically any material contained in this work, including any chapter or part of a chapter. Except as outlined above, no part of this work may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without prior written permission of the Publisher. Address permissions requests to: Elsevier's Rights Department, at the fax and e-mail addresses noted above. Notice No responsibility is assumed by the Publisher for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions or ideas contained in the material herein. Because of rapid advances in the medical sciences, in particular, independent verification of diagnoses and drug dosages should be made.
First edition 2005 Library of Congress Cataloging in Publication Data A catalog record is available from the Library of Congress. British Library Cataloguing in Publication Data A catalogue record is available from the British Library
ISBN-13: 978-0-444-52068-5 ISBN-10: 0-444-52068-6 ISBN (CD-ROM): 0-444-52069-4 ISSN: 0167-5648 The paper used in this publication meets the requirements of ANSI/NISO Z39.48-1992 (Permanence of Paper).
Printed in The Netherlands.
Working together to grow libraries in developing countries www.elsevier.com | www.bookaid.org | www.sabre.org
iv
CONTENTS
Contributing Authors. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xvii Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxv
I. HISTORY, REGULATION, AND RISK ASSESSMENT Chapter 1. AN OVERVIEW OF INJECTION WELL HISTORY IN THE UNITED STATES OF AMERICA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . J.E. Clark, D.K. Bonura, and R.F. Van Voorhees 1.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.2. Prior to EPA UIC Regulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.3. EPA UIC Regulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.4. Class I Hazardous Well Regulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.5. Risk Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.6. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Chapter 2. DEEP INJECTION DISPOSAL OF LIQUID RADIOACTIVE WASTE IN RUSSIA, 1963–2002: RESULTS AND CONSEQUENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A.I. Rybalchenko, M.K. Pimenov, V.M. Kurochkin, E.N. Kamnev, V.M. Korotkevich, A.A. Zubkov, and R.R. Khafizov 2.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2. Characteristics of Deep-Well Injection Sites and Preliminary Investigation . . . . . . . . . . . . . . . . . . . . . 2.3. Investigations of Waste Behavior and New Data Received While Operating Deep-Well Injection Sites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.4. Safety Requirements and Criteria While Performing Site Injection for Liquid Radioactive Waste at the Present Time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.5. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Chapter 3. APPLICATIONS OF DEEP-WELL INJECTION OF INDUSTRIAL AND MUNICIPAL WASTEWATER IN TEXAS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. Knape 3.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2. Regulatory Jurisdiction and Federal Program Authorization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3. Definition of Terms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4. Well Numbers and Locations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.5. Technical Requirements of Rules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.6. General Characteristics of Injected Waste Streams . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.7. Industries with Current or Historical Use of Class I Injection Wells . . . . . . . . . . . . . . . . . . . . . . . . . . 3.8. Proposed Uses of Class I Injection Wells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.9. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Chapter 4. WHY CURRENT REGULATIONS PROTECT FLORIDA’S SUBSURFACE ENVIRONMENT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Muniz, M. Tobon, and F. Bloetscher 4.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2. Current Raw Water Supply . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
v
3 3 4 5 7 8 10 11
13
13 14 16 17 18 19
21 21 21 22 22 22 23 24 26 27 28
29 29 30
vi 4.3. 4.4. 4.5. 4.6. 4.7.
Contents Effluent Disposal Alternatives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Risk Issues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Environmental Regulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Injection Well Construction and Testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Chapter 5. AN INTERPRETATION OF THE SAFE DRINKING WATER ACT’S “NON-ENDANGERMENT” STANDARD FOR THE UNDERGROUND INJECTION CONTROL (UIC) PROGRAM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B.J. Kobelski, R.E. Smith, and A.L. Whitehurst 5.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2. Statutory and Regulatory Authority . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.3. Approaches for Preventing Endangerment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.4. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgments and Disclaimer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Chapter 6. THE APPROPRIATE METHODOLOGY FOR DETERMINING THE USE OF A FIXED-RADIUS AREA OF REVIEW OR ZONE OF ENDANGERING INFLUENCE, WHEN CONDUCTING AN AREA-OF-REVIEW ANALYSIS FOR UNDERGROUND INJECTION CONTROL OPERATIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . S. Stephen Platt and D. Rectenwald 6.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.2. Methodologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.3. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Chapter 7. ANALYSIS OF INJECTATE LOCATION AT DUPONT BEAUMONT WORKS . . . . . . . . . . . J.W. Mercer, C.R. Faust, C. Brown, and J.E. Clark 7.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.2. Geology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.3. Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.4. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Chapter 8. AQUIFER STORAGE AND RECOVERY WELLS IN FLORIDA: HOW AND WHEN WILL IMPACT BE REGULATED? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Muniz, M. Tobon, and F. Bloetscher 8.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.2. The Concept . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.3. Floridan Aquifer System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.4. ASR Development in Florida . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.5. Everglades Restoration—ASR Component . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.6. Concerns about Aquifer Storage and Recovery on a Large Scale . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.7. Governing Regulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.8. Conclusions and Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Chapter 9. ALASKA-SAKHALIN 2002 SYMPOSIUM DISCUSSION OF UNDERGROUND INJECTION CONTROL IN ARCTIC OILFIELDS . . . . . . . . . . . . . . . . . . . T. Cutler and D. Thurston 9.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.2. Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.3. Well Drilling and Operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
31 34 34 35 38 38
39 39 39 40 42 43 43
45 45 46 49 51 51 54 55 64 64
65 65 66 67 69 70 72 75 76 77
79 79 80 82
Contents
vii
9.4. 9.5. 9.6.
83 84 90 90 90
Water Discharge Disposal Routes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The Underground Injection Control Presentation to Sakhalin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Chapter 10. A PROBABILISTIC RISK ASSESSMENT OF CLASS I HAZARDOUS WASTE INJECTION WELLS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . W.R. Rish 10.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.2. Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.3. Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.4. Class IH Injection Well System Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.5. Failure Modes and Effects Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.6. Event and Fault Tree Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.7. Event-Frequency-Distribution Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.8. Quantitative Analysis of Event Trees . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.9. Probabilistic Risk Assessment (PRA) Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.10. Overall Loss of Waste Isolation Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.11. Conclusions and Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
93 93 94 96 97 101 101 102 102 102 118 120 122
II. WELL TESTING AND HYDROLOGIC STUDIES Chapter 11. REPLACING ANNUAL SHUT-IN WELL TESTS BY ANALYSIS OF REGULAR INJECTION DATA: FIELD-CASE FEASIBILITY STUDY . . . . . . . . . . . . . . . D. Silin, C.-F. Tsang, and H. Gerrish 11.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.2. Description of the Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.3. Analysis of Field Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.4. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Chapter 12. EXPERIMENTAL STUDY OF INJECTION-INTERVAL HYDRAULIC ISOLATION FROM OVERLYING FORMATION AT THE DISPOSAL SITE OF THE SIBERIAN CHEMICAL COMPLEX, USING HIGH-ACCURACY HYDRAULIC HEAD MEASUREMENTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A.A. Zubkov, V.A. Sukhorukov, A.I. Zykov, E.A. Redkin, V.M. Shestsakov, S.P. Pozdniakov, V.A. Bakshevskay, and V.M. Kurockin 12.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.2. Monitoring Equipment and Measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.3. Monitoring Data Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.4. Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Chapter 13. GULF COAST BOREHOLE-CLOSURE-TEST WELL NEAR ORANGE, TEXAS . . . . . . . J.E. Clark, D.K. Bonura, P.W. Papadeas, and R.R. McGowen 13.1. Introduction and Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.2. Test Interval Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.3. Borehole-Closure Testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.4. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
139 139 140 141 148 149 149
151
151 152 153 154 155 155 157 157 158 159 166 166
viii
Contents
Chapter 14. INTERPRETATION OF TRANSIENT PERMEABILITY TESTS TO ANALYZE THE EVOLUTION OF A BRINE-FILLED SALT CAVERN . . . . . . . . . . . . . . Aron Behr 14.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14.2. Experimental Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14.3. Code PaTe for Test Evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14.4. Test Evaluation Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14.5. Evaluation Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14.6. Relationship between Permeability and Stress State . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14.7. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Chapter 15. POTENTIAL CORROSION AND MICROBIOLOGICAL MECHANISMS AND DETECTION TECHNIQUES IN SOLUTION MINING AND HYDROCARBON STORAGE WELLS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ken E. Davis and Larry K. McDonald 15.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15.2. The Corrosion Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15.3. Microbiologically Influenced Corrosion (MIC) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15.4. Conclusions and Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Chapter 16. CHARACTERIZATION OF SUBSURFACE HETEROGENEITY: INTEGRATION OF SOFT AND HARD INFORMATION USING MULTIDIMENSIONAL COUPLED MARKOV CHAIN APPROACH . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Eungyu Park, Amro Elfeki, and Michel Dekking 16.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16.2. Theoretical Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16.3. THE 3-D Coupled Markov Chain Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16.4. Application of the 2-D and 3-D CMC Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16.5. Summary and Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Chapter 17. MODELING OF WASTE INJECTION IN HETEROGENEOUS SANDY CLAY FORMATIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . S.P. Pozdniakov, V.A. Bakshevskay, A.A. Zubkov, V.V. Danilov, A.I. Rybalchenko, and C.-F. Tsang 17.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17.2. Development of a 3-D Model for Heterogeneity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17.3. Flow and Transport Model Calibration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17.4. Analysis of Effective Hydraulic and Transport Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17.5. Modeling of the Injection History . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17.6. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Chapter 18. NON-DARCY FLOW BEHAVIOR NEAR HIGH-FLUX INJECTION WELLS IN POROUS AND FRACTURED FORMATIONS . . . . . . . . . . . . . . . . . . . . . . . . . Y.-S. Wu 18.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18.2. Mathematical Model and Numerical Formulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18.3. Dimensionless Variables and Analytical Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18.4. Type Curves of Non-Darcy Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18.5. Summary and Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
167 167 167 168 170 171 174 176 176
177 177 178 186 191 192
193 193 194 194 196 199 201 202
203
203 204 210 212 215 217 218 218
221 221 222 224 225 232 233 233
Contents Chapter 19. MODELING DENSITY CHANGES IN HAZARDOUS DISPOSAL WELL PLUMES . . . . R.G. Larkin and J.E. Clark 19.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19.2. Generic Model Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19.3. Generic Model Inputs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19.4. Modeling Results—Variation of Rate Schedule, Runs 1 and 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19.5. Modeling Results—Variation of Injectate Density . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19.6. Results—Model Run 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19.7. Results—Model Run 4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19.8. Results—Model Run 5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19.9. Results—Model Run 6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Chapter 20. LEAKOFF MODELING OF FLUID INJECTED IN GAS RESERVOIR AT FRACTURE STIMULATION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Behr and G. Mtchedlishvili 20.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20.2. Reconstruction of Fracture Propagation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20.3. Estimation of the Leakoff Coefficient . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20.4. Identification of Exponent γ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20.5. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
ix 235 235 236 237 237 238 239 239 240 240 241 242
243 243 245 246 246 251 253 253
III. GEOCHEMISTRY Chapter 21. PREDICTING TRACE METAL FATE IN AQUEOUS SYSTEMS USING A COUPLED EQUILIBRIUM-SURFACE-COMPLEXATION DYNAMIC-SIMULATION MODEL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . J.A. Dyer, N.C. Scrivner, B.C. Fritzler, D.L. Sparks, S.J. Sanders, and P. Trivedi 21.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21.2. OLI Software . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21.3. DynaChem Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21.4. Surface Complexation Model Calibration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21.5. Definition of Case Studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21.6. Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Chapter 22. REVIEW OF THE STUDIES OF RADIONUCLIDE ADSORPTION/DESORPTION WITH APPLICATION TO RADIOACTIVE WASTE DISPOSAL SITES IN THE RUSSIAN FEDERATION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V.G. Rumynin, L.N. Sindalovskiy, P.K. Konosavsky, A.V. Mironova, E.V. Zakharova, E.P. Kaimin, E.B. Pankina, and A.A. Zubkov 22.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22.2. Radon Site . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22.3. EUR Site . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22.4. Tomsk-7 Site . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22.5. Krasnoyarsk-26 Site . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22.6. Lake Karachai Site . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22.7. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix A. A Non-equilibrium Model of Dual-Site, One-Component Adsorption . . . . . . . . . . . .
257 257 258 258 260 260 262 269
271
271 273 288 290 297 301 304 306 306 308
x
Contents Appendix B. Sorption Coefficients (KFS, NS, and KD) from Batch Sorption Experiments with Lomonosovsky Sand (the Radon site) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix C. A Kinetic Model of Adsorption with Concomitant Mineral Dissolution . . . . . . . . . . .
Chapter 23. CHEMICAL INTERACTIONS BETWEEN WASTE FLUID, FORMATION WATER, AND HOST ROCK DURING DEEP-WELL INJECTION . . . . . . . . . . . . . . . . . . . . . . . . . . N.F. Spycher and R.G. Larkin 23.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23.2. Chemical Characterization of Formation Waters Prior to Waste Injection . . . . . . . . . . . . . . . . . . . . 23.3. Simulation of Chemical Interaction between Native Fluid, Waste, and Host Rock . . . . . . . . . . . . . . 23.4. Discussion and Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Chapter 24. WATER–ROCK GEOCHEMICAL CONSIDERATIONS FOR AQUIFER STORAGE AND RECOVERY: FLORIDA CASE STUDIES . . . . . . . . . . . . . . . . . . . . . . . . J.D. Arthur, A.A. Dabous, and J.B. Cowart 24.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24.2. Historical Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24.3. Research Goals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24.4. Hydrogeologic Setting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24.5. Water-Quality Changes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24.6. The Aquifer System Matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24.7. Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24.8. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Chapter 25. GROUTING WITH MINERAL-FORMING SOLUTIONS—A NEW TECHNIQUE FOR SEALING POROUS AND FRACTURED ROCK BY DIRECTED CRYSTALLIZATION PROCESSES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . G. Ziegenbalg 25.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25.2. Fundamentals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25.3. Sealing of Porous or Fractured Rock Formation by Induced Crystallization . . . . . . . . . . . . . . . . . . 25.4. In Situ Immobilization by Crystallization Processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25.5. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
309 310
313 313 313 316 324 325 325
327 327 328 328 329 329 331 334 337 337 338
341 341 342 347 351 357 358 358
IV. LIQUID WASTE INJECTION Chapter 26. INJECTING BRINE AND INDUCING SEISMICITY AT THE WORLD’S DEEPEST INJECTION WELL, PARADOX VALLEY, SOUTHWEST COLORADO . . . . . . . . . . . . . . K. Mahrer, J. Ake, L. Block, D. O’Connell, and J. Bundy 26.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26.2. The Project . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26.3. Local Geology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26.4. Injection Well and Operations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26.5. Paradox Valley Seismic Network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26.6. PVSN Recording Sensitivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26.7. Seismicity and Injection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26.8. Seismicity and Local geology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26.9. Porosity and Reservoir Lifetime . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
361 361 361 362 363 364 365 366 369 373
Contents 26.10. 26.11.
Findings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Retrospective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Chapter 27. EVALUATION OF RESERVOIR INFORMATION IN RELATION TO EARTHQUAKES IN ASHTABULA, OHIO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . H. Gerrish and A. Nieto 27.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27.2. Historical Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27.3. Regional Geology and Tectonics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27.4. Chardon Earthquakes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27.5. Ashtabula Earthquakes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27.6. Possibility That Injection Triggered Earthquakes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27.7. An Alternative Source Mechanism for the Ashtabula Earthquakes . . . . . . . . . . . . . . . . . . . . . . . . . 27.8. Description of the Injection Zone . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27.9. Injection Activity in the Ashtabula Area . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27.10. Propagation of Injection-Induced Pressure Effects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27.11. Geological Framework for Seismicity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27.12. Sensitivity Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27.13. Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27.14. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Chapter 28. INJECTION OF BRINE FROM CAVERN LEACHING INTO DEEP SALINE AQUIFERS: LONG-TERM EXPERIENCES IN MODELING AND RESERVOIR SURVEY . . . . . . . . . . J. Zemke, M. Stöwer, and M. Borgmeier 28.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28.2. Historical Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28.3. Geological Requirements for Brine Disposal in Porous Aquifers . . . . . . . . . . . . . . . . . . . . . . . . . . 28.4. Preliminary Investigations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28.5. Technical Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28.6. Monitoring and Simulation Program for Brine Disposal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28.7. Results and Experiences Gathered . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Chapter 29. USE OF DEEP GEOLOGIC HORIZONS FOR LIQUID WASTE DISPOSAL AT POWER COMPLEXES IN CENTRAL RUSSIA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B.P. Gorbatenko, A.D. Turkovskiy, A.I. Rybalchenko, M.K. Pimenov, E.P. Kajmin, and E.V. Zacharova 29.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29.2. Geological Conditions for Waste Injection in the Central Part of Russia . . . . . . . . . . . . . . . . . . . . 29.3. Geology of the Kalinin Nuclear Power Plant . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29.4. Characteristics of Waste and Site Injection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29.5. Waste Treatment for Deep-Well Injection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29.6. Estimation of Deep-Well Injection Consequences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29.7. Equipment for Using a Natural Underground Solution in Moscow . . . . . . . . . . . . . . . . . . . . . . . . 29.8. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Chapter 30. CASE STUDY: EVALUATION OF OIL FIELD AND WATER-WELL DISPOSAL-WELL DESIGNS FOR OIL SANDS FACILITY IN NORTHERN ALBERTA, CANADA . . . . . . . . Y. Champollion, M.R. Gleixner, J. Wozniewicz, W.D. MacFarlane, and L. Skulski 30.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30.2. General Setting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30.3. Well Construction and Completion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
xi 373 374 374 374
377 377 377 378 379 380 381 385 389 390 391 393 395 397 398 398 398
403 403 403 404 405 406 407 412
413
413 413 414 415 416 416 417 417
419 419 419 423
xii 30.4. 30.5. 30.6.
Contents Test and Analysis Procedures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Test Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
424 427 428 429
Chapter 31. FLUID INJECTION NEAR THE WASTE ISOLATION PILOT PLANT . . . . . . . . . . . . . . . S. Ghose 31.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31.2. Mechanics and Practice of Fluid Injection in the Area . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31.3. Geology and Reservoir Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31.4. Mechanical Response and Environmental Effects of Fluid Injection . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
431
Chapter 32. INJECTION OF ORGANIC LIQUID WASTE IN A BASALTIC CONFINED COASTAL AQUIFER, REUNION ISLAND . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . J.-S. Martial, J.-L. Join, and J. Coudray 32.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32.2. The Case of the Savanna Distillery . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32.3. The Hydrogeological Setting of Bois Rouge—A Rare Asset . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32.4. Experiments and Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32.5. Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32.6. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Chapter 33. DEMONSTRATION OF PRESENCE AND SIZE OF A CO2-RICH FLUID PHASE AFTER HCL INJECTION IN CARBONATE ROCK . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . J.E. Clark, D.K. Bonura, C. Miller, and F.T. Fischer 33.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33.2. Wireline Logging Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33.3. Interference Testing and Pressure Calculations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33.4. Summary and Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Chapter 34. STABILITY ANALYSIS OF A SOLUTION CAVITY RESULTING FROM UNDERGROUND INJECTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . R.W. Nopper, Jr., C. Miller, and J.E. Clark, Jr. 34.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34.2. Methodology Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34.3. Stress Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34.4. Failure Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34.5. Model Results and Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
431 432 436 436 438
441 441 442 442 442 447 448 449 449
451 451 451 456 458 458
459 459 459 460 467 469 470 470
V. LIQUID RADIOACTIVE WASTE INJECTION Chapter 35. LIQUID RADIOACTIVE WASTE DISPOSAL INTO DEEP GEOLOGIC FORMATIONS BY THE RESEARCH INSTITUTE OF ATOMIC REACTORS (RUSSIA) . . . . . . . . . . . . . . V.V. Mironov, A.M. Ulyshkin, A.S. Ladzin, and V.I. Kuprienko 35.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35.2. Geologic-Hydrogeological Substantiation of the Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35.3. Physicochemical Investigations of Disposal Processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
473 473 474 475
Contents
xiii
35.4. 35.5.
476 478
LRW Disposal and Control Processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Chapter 36. SAFETY ASSESSMENT OF DEEP LIQUID-ORGANIC RADIOACTIVE WASTE DISPOSAL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B.G. Balakhonov, A.A. Zubkov, V.A. Matyukha, M.D. Noskov, A.D. Istomin, A.N. Zhiganov, and G.F. Egorov 36.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36.2. Mathematical Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36.3. Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36.4. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Chapter 37. RESULTS OF LONG-TERM DEEP LIQUID RADIOACTIVE WASTE INJECTION SITE OPERATION AT THE SIBERIAN CHEMICAL COMBINE . . . . . . . . . . . . . . . . . . . . A.A. Zubkov, A.S. Ryabov, V.A. Sukhorukov, V.V. Danilov, and A.I. Rybalchenko 37.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37.2. Characteristics of the Injection Sites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37.3. Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37.4. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Chapter 38. RADIONUCLIDE DISTRIBUTION IN A SANDSTONE INJECTION ZONE IN THE COURSE OF ACIDIC LIQUID RADIOACTIVE WASTE DISPOSAL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A.A. Zubkov, B.G. Balakhonov, V.A. Sukhorukov, M.D. Noskov, A.D. Istomin, A.G. Kessler, A.N. Zhiganov, E.V. Zakharova, E.N. Darskaya, and G.F. Egorov 38.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38.2. Phenomenological Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38.3. Mathematical Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38.4. Results and Discussion of the Simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38.5. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Chapter 39. DEEP-WELL INJECTION MODELING OF RADIOACTIVE AND NONRADIOACTIVE WASTES FROM RUSSIAN NUCLEAR INDUSTRY ACTIVITIES, WITH EXAMPLES FROM THE INJECTION DISPOSAL SITES OF SSC RF–NIIAR AND CHEPETSK MECHANICAL PLANTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . E.A. Baydariko, A.I. Rybalchenko, A.I. Zinin, G.A. Zinina, A.M. Ulyushkin, and A.L. Zagvozkin 39.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39.2. Model for Deep Injection Disposal of Industrial Waste at Glazov . . . . . . . . . . . . . . . . . . . . . . . . . . 39.3. Model of Deep Injection Disposal at Dimitrovgrad . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39.4. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Chapter 40. EFFECT OF ANTHROPOGENIC TRANSFORMATIONS OF DEEP LIQUID RADIOACTIVE WASTE REPOSITORY-CONTAINING ROCKS ON RADIONUCLIDE MIGRATION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . E.V. Zakharova, E.P. Kaimin, A.A. Zubkov, O.V. Makarova, and V.V. Danilov 40.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40.2. Experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40.3. Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40.4. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
481
481 482 483 485 485
487 487 487 488 490
491
491 491 492 497 499 499
501
501 502 507 508 509
511 511 511 512 520 520
xiv
Contents
Chapter 41. MOLECULAR BACTERIAL DIVERSITY IN WATER AT THE DEEP-WELL MONITORING SITE AT TOMSK-7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . M. Nedelkova, G. Radeva, and S. Selenska-Pobell 41.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41.2. Materials and Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41.3. Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41.4. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
521 521 521 523 530 531 531
VI. INJECTION OF SOLIDS Chapter 42. INTERNATIONAL DATABASE FOR SLURRY INJECTION OF DRILLING WASTES . . . J.A. Veil and M.B. Dusseault 42.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42.2. Slurry Injection Technology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42.3. Development of the Database . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42.4. Number of Injection Jobs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42.5. Location of Slurry Injection Jobs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42.6. Who is Doing the Injection? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42.7. Geological Information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42.8. Injection Depth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42.9. Duration of Injection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42.10. Injection Rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42.11. Injection Pressure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42.12. Type and Volume of Material Injected . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42.13. Slurry Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42.14. Pre-Injection Processing or Treatment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42.15. Problems Experienced . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42.16. Economics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Chapter 43. REGULATORY REQUIREMENTS AND PRACTICES GOVERNING SLURRY INJECTION OF DRILLING WASTES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . M.G. Puder, J.A. Veil, and W. Bryson 43.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43.2. Description of the Regulatory Compendium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43.3. Findings Presented in the Regulatory Compendium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43.4. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Chapter 44. ALASKAN UNDERGROUND INJECTION CONTROL OF SOLID WASTE DISPOSAL . T. Cutler 44.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44.2. Regulatory Framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44.3. Class I and Class V Wells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44.4. Underground Injection Reduces Surface Environmental Impacts . . . . . . . . . . . . . . . . . . . . . . . . . . 44.5. Aquifer Trends . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44.6. Geological Limits to Injection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44.7. Construction, Operation, and Management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44.8. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
539 539 539 540 541 541 542 542 543 543 543 544 544 544 545 545 546 547 547
549 549 550 550 555 556 556 557 557 558 558 560 561 563 564 566 567 567
Contents Chapter 45. DISPOSAL OF MEAT, BONEMEAL, AND RESIDUAL ASH BY INJECTION INTO DEEP GEOLOGICAL FORMATIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V. Brkic, I. Omrcen, S. Bukvic, H. Gotovac, B. Omrcen, and M. Zelic 45.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45.2. Historical Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45.3. Geological and Physical Properties of the Benicanci Oil Field . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45.4. Disposal of Meat and Bonemeal (MBM) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45.5. Disposal of Residual Ash . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45.6. Transport Modeling and Risk Evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45.7. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Chapter 46. THERMAL TREATMENT, CARBON SEQUESTRATION, AND METHANE GENERATION THROUGH DEEP-WELL INJECTION OF BIOSOLIDS . . . . . . . . . . . . . . M.S. Bruno, J.T. Young, O. Moghaddam, H. Wong, and J.A. Apps 46.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46.2. Proposed Technology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46.3. Hyperion Anaerobic Mesophilic and Thermophilic Digestion Pilot Test . . . . . . . . . . . . . . . . . . . . . . 46.4. Experimental Verification of Biodegradation and Methane Generation under Simulated Deep Subsurface Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46.5. Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46.6. Summary and Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
xv
569 569 570 570 572 574 576 584 585
587 587 588 589 591 593 602 604
VII. CO2 INJECTION Chapter 47. THE POTENTIAL FOR CO2 SEQUESTRATION IN LARGE AQUIFER STRUCTURES IN NORTHEASTERN GERMANY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . M. Stöwer, W. Gilch, and J. Zemke 47.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47.2. Sequestration in Germany . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47.3. Geological Conditions in Northeastern Germany . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47.4. Evaluating the Structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47.5. Parameter of Structures and Reservoirs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47.6. Case Study at Ketzin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47.7. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
607 607 608 609 612 613 616 620 621
Chapter 48. DEEP INJECTION OF ACID GAS IN WESTERN CANADA . . . . . . . . . . . . . . . . . . . . . . . S. Bachu, K. Haug, K. Michael, B.E. Buschkuehle, and J.J. Adams 48.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48.2. Surface Operations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48.3. Injection Well and Subsurface Requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48.4. Characteristics of Acid-Gas Injection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48.5. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
623
Chapter 49. UNDERGROUND INJECTION OF CARBON DIOXIDE IN SALT BEDS . . . . . . . . . . . . . S. Bachu and M.B. Dusseault 49.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49.2. Cavern Construction and Behavior . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49.3. CO2 Leakage and Monitoring . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49.4. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
637
623 624 627 629 633 634
637 638 645 646 647
xvi
Contents
Chapter 50. COUPLED HYDROMECHANICAL EFFECTS OF CO2 INJECTION . . . . . . . . . . . . . . . . . J. Rutqvist and C.-F. Tsang 50.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50.2. Fundamentals of Hydromechanical Interactions in Fractured Rock . . . . . . . . . . . . . . . . . . . . . . . . . 50.3. Natural and Industrial Analogs Related to Study of Caprock Integrity and Reservoir Leakage . . . . . 50.4. The TOUGH-FLAC THM Simulator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50.5. Application of the TOUGH-FLAC Code to CO2 Injection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50.6. Discussion and Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
649 649 650 657 658 660 675 676 677
Chapter 51. SUBSURFACE PROPERTY RIGHTS: IMPLICATIONS FOR GEOLOGIC CO2 SEQUESTRATION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . E.J. Wilson 51.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51.2. History of U.S. Property Rights . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51.3. The Negative Rule of Capture and Secondary Recovery . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51.4. Injecting Industries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51.5. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
681 682 683 685 692 693
Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Subject Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
695 697
681
CONTRIBUTING AUTHORS (LISTED ALPHABETICALLY) J.J. ADAMS Alberta Energy and Utilities Board, Edmonton, Alberta, Canada J. AKE U.S.A Bureau of Reclamation, Denver, Colorado, U.S.A J.A. APPS Lawrence Berkeley National Laboratory, Berkeley, CA, 94720, U.S.A (phone: (510) 486-5193, fax: (510) 486-5686,
[email protected]) A. BEHR Freiberg University of Mining and Technology, Freiberg, Germany J.D. ARTHUR Florida Department of Environmental Protection—Florida Geological Survey, Tallahassee, FL, 32304, U.S.A (phone: (850) 488-9380, fax: (850) 488-8086,
[email protected]) S. BACHU Alberta Energy and Utilities Board, Edmonton, AB, T6B 2X3, Canada, (phone: (780) 427-1517, fax: (780) 422-1459,
[email protected]) V.A. BAKSHEVSKAY Faculty of Geology, Moscow State University, Moscow, 119899, Russia, 7 (phone: (095) 939–2112) B.G. BALAKHONOV Siberian Chemical Combine, Seversk, Russia E.A. BAYDARIKO All-Russia Designing and Research Institute of Production Engineering (VNIPIPT), Moscow, Russia L. BLOCK U.S.A Bureau of Reclamation, Denver, Colorado, U.S.A F. BLOETSCHER Public Utility Management Planning Services, Inc., Hollywood, Florida, U.S.A D.K. BONURA Bonura Geological Consulting, Inc., Beaumont, TX, 77706, U.S.A (phone: 409-727-9430,
[email protected]) M. BORGMEIER E-On Hanse AG, Hamburg, Germany V. BRKIC INA Oil Industry Plc., Zagreb, 10000, Croatia (phone: +385-1-459-26-24, fax: +385-1-459-26-26,
[email protected]) C. BROWN South Florida Water Management District, West Palm Beach, Florida, U.S.A
xvii
xviii
Contributing Authors (Listed Alphabetically)
M.S. BRUNO Terralog Technologies USA, Inc., Arcadia, CA, U.S.A,
[email protected] W. BRYSON Argonne National Laboratory, Washington, D.C., U.S.A S. BUKVIC INA Oil Industry Plc., Zagreb, Croatia J. BUNDY Subsurface Technology, Inc., Houston, TX 77024, U.S.A (phone: 713-880-4640,
[email protected]) B.E. BUSCHKUEHLE Alberta Energy and Utilities Board, Edmonton, Alberta, Canada Y. CHAMPOLLION Golder Associates, Ltd., Calgary, Alberta, Canada J.E. CLARK E.I. du Pont de Nemours & Co., Beaumont, TX, 77704, U.S.A (phone: 409-727-9855, fax: 409-727-9389,
[email protected]) J. COUDRAY Laboratoire des Sciences de la Terre, Université de La Réunion, Saint-Denis, Ile de La Réunion, France J.B. COWART Department of Geological Sciences, Florida State University, Tallahassee, Florida, U.S.A T. CUTLER U.S.A Environmental Protection Agency, Seattle, WA, U.S.A (phone: 206-553-1673,
[email protected]) A.A. DABOUS Florida Department of Environmental Protection—Florida Geological Survey, Tallahassee, Florida, U.S.A E.N. DARSKAYA Institute of Physical Chemistry RAS, Moscow, Russia V.V. DANILOV Siberian Chemical Combine, Seversk, Russia K.E. DAVIS Subsurface Technology, Inc., Houston, Texas, U.S.A M. DEKKING Faculty of Civil Engineering and Geosciences, Delft University of Technology, Delft, The Netherlands M.B. DUSSEAULT University of Waterloo, Waterloo, ON, Canada J.A. DYER DuPont Engineering Research and Technology, Wilmington, DE, 19898, U.S.A (phone: (302) 774-2237, fax: (302) 774-1347,
[email protected])
Contributing Authors (Listed Alphabetically)
xix
G.F. EGOROV Institute of Electrochemistry RAS, Moscow, Russia A. ELFEKI Faculty of Civil Engineering and Geosciences, Delft University of Technology, Delft, The Netherlands C.R. FAUST GeoTrans, Inc., Sterling, Virginia, U.S.A F.T. FISCHER 217 Hidden Lake Rd., Hendersonville, Tennessee, U.S.A B.C. FRITZLER DuPont Engineering Research and Technology, Wilmington, Delaware, U.S.A H. GERRISH U.S.A Environmental Protection Agency, Chicago, IL, 60604-3590, U.S.A (phone: (312) 886-2939, fax: (312) 886-4235,
[email protected]) S. GHOSE U.S.A Environmental Protection Agency, Washington, D.C., U.S.A W. GILCH Untergrundspeicher- und Geotechnologie-Systeme GmbH Mittenwalde, Mittenwalde, Germany M.R. GLEIXNER Golder Associates, Ltd., Calgary, Alberta, Canada H. GOTOVAC Faculty of Civil Engineering, University of Split, CA, 21000, Croatia, (phone: 0038521303354, fax: 0038521465117,
[email protected]) B.P. GORBATENKO Kalinin Nuclear Power Plant, Udomlia, Russia K. HAUG Alberta Energy and Utilities Board, Edmonton, Alberta, Canada A.D. ISTOMIN Seversk State Technological Institute, Seversk, Russia J.-L. JOIN Laboratoire des Sciences de la Terre, Université de La Réunion, Saint-Denis, Ile de La Réunion, France E.P. KAIMIN Institute of Physical Chemistry of the RAS, Moscow, Russia E.N. KAMNEV All-Russia Designing and Research Institute of Production Engineering (VNIPIPT) Moscow, Russia A.G. KESSLER Seversk State Technological Institute, Seversk, Russia
xx
Contributing Authors (Listed Alphabetically)
R.R. KHAFIZOV Mining and Chemical Combine, Zeleznogorsk, Russia B. KNAPE Texas Commission on Environmental Quality, Austin, Texas, U.S.A B.J. KOBELSKI United States Environmental Protection Agency, Office of Ground Water and Drinking Water, Washington, DC, 20460, U.S.A (phone: (202) 564-3888,
[email protected]) P.K. KONOSAVSKY Institute of Environmental Geology of the RAS, St. Petersburg Division, St. Petersburg, Russia V.M. KOROTKEVICH Siberian Chemical Combine, Seversk, Russia V.I. KUPRIENKO State Scientific Center of Russian Federation, Research Institute of Atomic Reactors, Dimitrovgrad, Russia V.M. KUROCHKIN All-Russia Designing and Research Institute of Production Engineering (VNIPIPT) Moscow, Russia A.S. LADZIN State Scientific Center of Russian Federation, Research Institute of Atomic Reactors, Dimitrovgrad, Russia R.G. LARKIN R. G. Larkin Consulting, Austin, TX, 78749, U.S.A (phone: (512) 891-6742,
[email protected]) W.D. MACFARLANE Nexen Canada Ltd., Calgary, Alberta, Canada K. MAHRER U.S.A Bureau of Reclamation, Denver, CO, 80225, U.S.A (phone: (303) 445-3215, fax: (303) 445-6478,
[email protected]) O.V. MAKAROVA Siberian Chemical Combine, Seversk, Russia J.-S. MARTIAL Laboratoire des Sciences de la Terre, Université de La Réunion, Saint-Denis, Ile de La Réunion, France (
[email protected]) V.A. MATYUKHA Siberian Chemical Combine, Seversk, Russia L.K. MCDONALD Subsurface Technology, Inc., Houston, TX, 77024, U.S.A (phone: (713) 880-4640, fax: (713) 880-3248,
[email protected]) R.R. McGOWEN Terra Dynamics Inc., Austin, TX, 78759, U.S.A (phone: (512) 795-8183,
[email protected])
Contributing Authors (Listed Alphabetically)
xxi
J.W. MERCER GeoTrans, Inc., Sterling, VA, 20166, U.S.A (phone: (703) 444-7000, fax: (703) 444-1685,
[email protected]) K. MICHAEL Alberta Energy and Utilities Board, Edmonton, Alberta, Canada G. MICHEDLISHVILI Freiberg University of Mining and Technology, Freiberg, Germany C. MILLER Retired du Pont Experimental Station, E.I. du Pont de Nemours and Co., Wilmington, DE, U.S.A V.V. MIRONOV State Scientific Center of Russian Federation, Research Institute of Atomic Reactors, Dimitrovgrad, Russia A.V. MIRONOVA Institute of Environmental Geology of the RAS, St. Petersburg Division, St. Petersburg, Russia O. MOGHADDAM City of Los Angeles Department of Public Works, Los Angeles, CA, U.S.A A. MUNIZ Hazen and Sawyer, P.C., Boca Raton, FL, 33431-7343, U.S.A (phone: (561) 997-8070, fax: (561) 997-8159,
[email protected]) M. NEDELKOVA Institute of Radiochemistry, Dresden, 01314, Germany (phone: 00493512603138,
[email protected]) A. NIETO Departments of Geology and of Civil and Environmental Engineering, University of Illinois, UrbanaChampaign, Illinois, U.S.A R.W. NOPPER, JR. E.I. DuPont de Nemours & Co., Experimental Station, Wilmington, DE, 19880-0249, U.S.A (phone: (302) 695-3826, fax: (302) 695-8805,
[email protected]) M.D. NOSKOV Seversk State Technological Institute, Seversk, Russia B. OMRCEN Association of Petroleum Engineers and Geologists, Zagreb, Croatia I. OMRCEN INA Oil Industry Plc., Zagreb, 10000, Croatia, (phone: +38514592392, fax: +38514592626,
[email protected]) D. O’CONNELL U.S.A Bureau of Reclamation, Denver, Colorado, U.S.A E.B. PANKINA A.P. Alexandrov Technical Research Institute, Sosnovyi Bor, Russia P.W. PAPADEAS Sandia Technologies, Houston, TX, 77066, U.S.A (phone: (832) 286-0471, fax: (832) 286-0477,
[email protected])
xxii
Contributing Authors (Listed Alphabetically)
E. PARK Environmental Sciences Division, Oak Ridge National Laboratory, Oak Ridge, (Currently at Dept. of Geology, Kyungbuk National University, Daegu, Korea), TN, 37831-6036, U.S.A (phone: (865) 576-3978, fax: (865) 576-8543,
[email protected]) M.K. PIMENOV All-Russia Designing and Research Institute of Production Engineering (VNIPIPT) Moscow, Russia S. P. POZDNIAKOV Faculty of Geology, Moscow State University, Moscow, 119899, Russia, 7 (phone: (095) 939-2112,
[email protected]) M.G. PUDER Argonne National Laboratory, Washington, DC, 20024, U.S.A (phone: (202) 488-2484, fax: (202) 488-2471,
[email protected]) G. RADEVA Institute of Radiochemistry, Dresden, Germany D. RECTENWALD Safe Drinking Water Act Branch, U.S.A Environmental Protection Agency, Philadelphia, PA, 19103, U.S.A (phone: (814) 827-1952, fax: 814-827-3682,
[email protected]) E.A. REDKIN Siberian Chemical Combine, Seversk, Russia W.R. RISH Hull and Associates, Inc., 6397 Emerald Parkway, Dublin, Ohio 43016 (phone: (614) 793-8777,
[email protected]) A.I. RYBALCHENKO All-Russia Designing and Research Institute of Production Engineering (VNIPIPT) Moscow, Russia V.G. RUMYNIN Institute of Environmental Geology of the RAS, St. Petersburg Division, St. Petersburg, Russia J. RUTQVIST Earth Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, California, U.S.A A.S. RYABOV Siberian Chemical Combine, Seversk, Russia S.J. SANDERS Process Systems Enterprise, Denville, New Jersey, U.S.A N.C. SCRIVNER DuPont Engineering Research and Technology, Wilmington, DE, 19898, U.S.A (phone: (302) 774-2314, fax: (302) 774-2457,
[email protected]) S. SELENSKA-POBELL Institute of Radiochemistry, Dresden 01314, Germany (phone: 00493512602989,
[email protected]) V.M. SHESTSAKOV Faculty of Geology, Mascow State University, Mascow, Russia
Contributing Authors (Listed Alphabetically)
xxiii
D. SILIN Lawrence Berkeley National Laboratory, Berkeley, CA, 94720, U.S.A (phone: (510) 495-2215,
[email protected]) L.N. SINDALOVSKIY Institute of Environmental Geology of the RAS, St. Petersburg Division, St. Petersburg, Russia L. SKULSKI Nexen Canada Ltd., Calgary, Alberta, Canada R.E. SMITH United States Environmental Protection Agency, Office of Ground Water and Drinking Water, Washington, DC, 20004, U.S.A (phone: (202) 564-3895, fax: (202) 564-3756,
[email protected]) D.L. SPARKS Department of Plant and Soil Sciences, University of Delaware, Newark, Delaware, U.S.A N.F. SPYCHER Lawrence Berkeley National Laboratory, Berkeley, California, U.S.A S. STEPHEN PLATT Safe Drinking Water Act Branch, U.S.A Environmental Protection Agency, Philadelphia, Pennsylvania, U.S.A M. STÖWER Untergrundspeicher-und Geotechnologie-Systeme GmbH, UGS, Mittenwalde, 15749, Germany (phone: ++49 33764 82176, fax: ++49 33764 82290,
[email protected]) V.A. SUKHORUKOV Siberian Chemical Combine, Seversk, Russia D. THURSTON Minerals Management Service (MMS), Anchorage, Alaska, U.S.A (phone: 907-334-5338,
[email protected]) M. TOBON City of Fort Lauderdale, Fort Lauderdale, Florida, U.S.A P. TRIVEDI Department. of Civil and Environmental Engineering, University of Alaska, Fairbanks, AK, 99775, U.S.A (phone: (907) 978-0742, fax: (907) 978-6087,
[email protected]) C.-F. TSANG Lawrence Berkeley National Laboratory, Berkeley, CA, 94720, U.S.A (phone: 1 (510) 486-5782, fax: 1 (510) 486-5686,
[email protected]) A.D. TURKOVSKIY Kalinin Nuclear Power Plant, Udomlia, Russia A.M. ULYUSHKIN State Scientific Center of Russian Federation, “Research Institute of Atomic Reactors”, Dimitrovgrad, Russia. R.F. VAN VOORHEES Bryan Cave LLP, Washington, DC, 20005-3960, U.S.A (phone: (202) 508-6014, fax: (202) 508-6200,
[email protected])
xxiv
Contributing Authors (Listed Alphabetically)
J.A. VEIL Argonne National Laboratory, Washington, DC, 20024, U.S.A (phone: (202) 488-2450, fax: (202) 488-2413,
[email protected]) E.J. WILSON Humphrey Institute of Public Affairs, University of Minnesota, Minneapolis, MN 55455, U.S.A (phone (612)-(625-9505)
[email protected]) A.L. WHITEHURST United States Environmental Protection Agency, Office of Ground Water and Drinking Water, Washington, D.C., U.S.A H. WONG City of Los Angeles Department of Public Works, Los Angeles, CA, U.S.A J. WOZNIEWICZ Golder Associates, Ltd., Calgary, Alberta, Canada Y.-S. WU Earth Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, California, U.S.A J.T. YOUNG Terralog Technologies USA, Inc., Arcadia, CA, 91006, U.S.A (phone: (626) 305-8460,
[email protected]) A.L. ZAGVOZKIN Open Stock Company, Chepetsk Mechanical Plant Glazov, Udmurtia E.V. ZACHAROVA Institute of Physical Chemistry of the RAS, Moscow, Russia J. ZEMKE Untergrundspeicher-und Geotechnologie-Systeme GmbH, UGS, Mittenwalde, 15749, Germany (phone: ++49 33764 82-178, fax: ++49 33764 82290,
[email protected]) M. ZELIC Association of Petroleum Engineers and Geologists, Zagreb, Croatia G. ZIEGENBALG TU Bergakademie Freiberg—Freiberg University of Mining and Technology, Institute of Technical Chemistry, Freiberg, Germany A.I. ZININ State Scientific Center of Russian Federation, Institute of Physics and Power Engineering, Obninsk, Russia A.N. ZHIGANOV Seversk State Technological Institute, Seversk, Russia G.A. ZININA State Scientific Center of Russian Federation, Institute of Physics and Power Engineering, Obninsk, Russia A.A. ZUBKOV Siberian Chemical Combine, Seversk, Russia A.I. ZYKOV Siberian Chemical Combine, Seversk, Russia
PREFACE
Despite stringent regulations, the subsurface disposal or storage of liquids, gases, and slurries in deep sedimentary formations continues to expand, not only in the United States, but also in other countries throughout the world. Furthermore, new applications for deep underground disposal are under consideration, such as CO2 sequestration for alleviating climate change, and deep injection disposal of biosolids. Since the First International Symposium1 on Deep Injection Disposal of Hazardous and Industrial Wastes was held at Ernest Orlando Lawrence Berkeley National Laboratory (Berkeley Lab) in 1994, sub surface disposal technology and supporting science have continued to become more sophisticated and complex. Therefore, the community of engineers, scientists, and regulators involved in deep underground injection disposal believed that a second symposium was both timely and beneficial. Accordingly, Berkeley Lab hosted the Second International Symposium on Injection Science and Technology on October 22–25, 2003. This symposium provided an opportunity to take stock of developments over the nine-year interval since the first symposium. The second symposium, however, had a broader technical reach in that all aspects of underground injection were open for consideration, including those that underlie and crosscut the U.S.-defined classes: Class I (deep industrial/municipal and hazardous wastes), Class II (oil- and gas-related), Class III (solution mining), Class IV (not used), and Class V (other, generally shallow wells). The symposium also provided a forum for the exchange of ideas and clarification of scientific, technological, and regulatory issues of concern. Participants from over 10 countries attended, reflecting the broad international interest in the potential economic and environmental benefits of deep underground injection disposal. In this volume, key papers presented at the Second International Symposium have been revised, reviewed, and organized as book chapters in seven sections addressing specific topics of interest. The first section focuses on the history of deep underground injection as well as regulatory issues, future trends and risk analysis. The next section comprises 10 chapters dealing with well testing and hydrologic modeling. Well testing is conducted for a variety of reasons, among which is determining the response of the formation to the injection of the waste; testing corrosion and leaks in either the well itself or in the injection zone; and testing to identify and locate faults, impermeable barriers, or formation damage. Many well tests require suitable models for their interpretation, and therefore, well testing and model development must be seen together as an integrated process. Section 3, consisting of five chapters, addresses various aspects of the chemical processes affecting the fate of the waste in the subsurface environment. Consideration is given here to reactions between the waste and the geologic medium, and also to reactions that take place within the waste stream itself. All aspects of this subject are covered, including experimentation, field observation, theoretical modeling, and prediction. The remaining four sections deal with experiences relating to injection of, respectively, liquid wastes, liquid radioactive wastes in Russia, slurried solids, and compressed carbon dioxide. Chapters describing the injection of liquid wastes include two that deal with
1 Papers from this symposium were subsequently revised, edited, and published as a book entitled Deep Injection Disposal of Hazardous and Industrial Waste: Scientific and Engineering Aspects (John A. Apps and Chin-Fu Tsang, eds.), Academic Press, Inc., 1996, 775 pp.
xxv
xxvi
Preface
induced seismicity. The remainder cover a diverse range of issues. As with the First International Symposium, one of the highlights of the Second International Symposium was the presentation by Russian scientists of several papers revealing their knowledge and experience of the deep disposal of high-level radioactive liquid processing waste. They provide a unique perspective on the philosophy and implementation of radioactive waste disposal practices in Russia. The resulting chapters are compiled in Section 5. Section 6 consists of five chapters that cover the technology surrounding the injection disposal of waste slurries. Among the materials considered are drilling wastes, bone meal, and biosolids. This technology continues to develop and promises to provide a solution for the disposal of intractable wastes that are otherwise difficult to dispose of or treat economically. Finally, the four chapters in Section 7 deal with questions relating to carbon dioxide sequestration in deep sedimentary aquifers. This subject is particularly topical, as nations grapple with the issue of controlling the buildup of carbon dioxide in the atmosphere with attendant concerns regarding climate change. We are deeply indebted to the sponsors of the Second International Symposium: the U.S. Environmental Protection Agency (Office of Ground Water and Drinking Water), and the U.S. Department of Energy (Office of Fossil Energy, National Energy Technology Laboratory; Office of Science, Office of Basic Energy Sciences, Chemical Sciences and Geosciences Division; and Office of Science and Technology in the Office of Environmental Management). In particular, we would like to thank EPA’s Bruce Kobelski and Robert E. Smith, who provided encouragement and advice throughout the organization of the symposium. We would also like to express our appreciation to our co-sponsors: the Ground Water Protection Council, the Solution Mining Research Institute, the International Association of Hydrologists, the International Association of Hydraulic Research and the American Institute of Hydrology. We are particularly appreciative of the assistance provided by members of the Symposium Advisory Committee in suggesting symposium topics and assisting with the arduous task of reviewing and critiquing individual chapters of the book to ensure quality and consistency. We wish to especially thank Julie McCullough, who not only provided invaluable help in organizing and facilitating the smooth running of the Symposium, but also, in coordination with our contributors, shepherded all of the individual chapters through the laborious editing and formatting process. Last, but not least, we would also like to acknowledge the help of our conference coordinators, Kathleen Brower and Pat Butler, for organizing the myriad behind-the-scenes details of the symposium; Bruce Balfour, Theresa Duque, and Dan Hawkes for their editorial assistance; Kryshna Avina, Donald Nodora, and Alice Ramirez for their assistance in organizing and formatting the document; and Maria Atkinson and Flavio Robles for providing their graphical expertise to the cover art and illustrations that are contained in this comprehensive volume. Chin-Fu Tsang John A. Apps Editors
ELSE_TSANG_CH001.qxd
9/28/2005
9:25 PM
Page 1
I. HISTORY, REGULATION, AND RISK ASSESSMENT
1
This page intentionally left blank
2
ELSE_TSANG_CH001.qxd
8/22/2005
7:07 AM
Page 3
Chapter 1
AN OVERVIEW OF INJECTION WELL HISTORY IN THE UNITED STATES OF AMERICA J.E. Clarka, D.K. Bonurab, and R.F. Van Voorheesc a
E.I. du Pont de Nemours & Co., Inc., Beaumont, TX, USA Bonura Geological Consulting, Inc., Beaumont, TX, USA c Bryan Cave LLP, Washington, DC, USA b
1.1 INTRODUCTION Disposal of liquids into underground formations through injection wells was started in the 1930s by the U.S. petroleum industry, which, as a common practice, disposed of produced brine in this manner. The first report of shallow industrial waste injection was in the mid-1930s. Since the early 1950s, injection wells have been used for fluids associated with industrial facilities. Injection wells were regulated by the various states with no national oversight program. The Safe Drinking Water Act (SDWA) was passed in 1974 to address underground injection issues from a national approach and includes all types of injection wells. Class I wells are used to inject hazardous and nonhazardous fluids below any underground sources of drinking water (USDW). Class II wells inject brine fluids associated with oil and gas production. Class III wells pertain to in situ mining wells. Class IV wells (banned except for remediation) handled disposal of hazardous liquids into or above USDWs. Class V wells relate to geothermal and other wells that do not fall into the previous categories. The United States Environmental Protection Agency (EPA) has implemented Underground Injection Control (UIC) rules and regulations since the early 1980s as an outcome of the SDWA, to protect citizens from exposure and reduce risk to human health and the environment. In 1984, Congress passed an expansion of the Resource Conservation Recovery Act (RCRA). This Act, in essence, banned hazardous disposal unless the demonstration was made that the injected fluid would be protective of human health and the environment. In 1988, EPA promulgated rules and regulations dealing with the land disposal ban for Class I injection wells (40 CFR §124, 144, 146, and 148). These regulations established a mechanism for making the demonstration of 10,000-year flow and containment of injected fluid or chemical fate transformation within the injection zone. The primary objective of deep-well disposal is to permanently isolate injected fluids from the biosphere. In 1989, the EPA did a qualitative and comparative risk study and found that Class I injection is a safe and effective technology because of its very low risk to human health and the environment. In this study, the EPA also found that underground injection of hazardous fluids was rated the lowest risk in comparison with other operations such as municipal waste combustion. Based on EPA regulations, Class I injection wells are constructed and monitored to assure protection against any toxic releases into the environment. A quantitative risk analysis by Rish et al. (1998) agrees with EPA studies that deep-well injection is a low-risk management practice. The risk associated with a Class I hazardous
ELSE_TSANG_CH001.qxd
4
8/22/2005
7:07 AM
Page 4
An Overview of Injection Well History in the United States of America
injection well for the loss of waste containment to the lowermost USDW is less than one in 1 million. The loss of injectate isolation probability is low because of redundancies in well construction barriers and geological requirements that provide multiple safety factors.
1.2 PRIOR TO EPA UIC REGULATIONS Underground injection is the disposal of liquid waste material into isolated geologic strata, placing the wastes in portions of the earth’s crust that are free from the usual effects of the hydrologic cycle regulated under 40 CFR §267, Subpart G and §146 and 148 (EPA, 1989, p. 5). The primary objective of deep-well injection is to permanently isolate disposed fluids from the biosphere. Injection of fluids into underground formations in the U.S. through wells began in the 1930s by the petroleum industry for disposal of produced brines associated with oil and gas production (Brasier and Kobelski, 1996, p. 1). The first report of shallow industrial waste injection was in the mid-1930s. However, that practice lasted only a few days because injected fluid found its way back to the surface where other wells penetrated the 800foot-deep sand (Harlow, 1939). DuPont drilled the first deep industrial waste injection well in Texas in 1949 and began operations in the early 1950s. In 1950, there were four injection wells, and by the early 1960s there were 30 injection wells (Smith, 1996, p. 10). Texas was the first state to adopt regulations (1961) regarding industrial injection wells (Warner and Orcutt, 1973, p. 692). Early regulation of underground injection was traditionally a state responsibility under specific disposal-well statutes, water-well statutes, oil and gas regulations, or surface waste pollution-control statutes (Walker and Cox, 1973, pp. 5–6). State regulations were not uniform in water-quality-levels’ protection for potential usable groundwater (Fig. 1.1). Federal control over underground disposal of radioactive wastes was under the direction of the Atomic Energy Commission under the Atomic Energy Act of 1954, and pre-empted state control of underground injection (Walker and Cox, 1973, p. 9). By the early 1970s, the number of injection wells was approximately 250 (Warner and Orcutt, 1973, p. 688), nearly a 10-fold increase over the 1960 well total (Fig. 1.2). Concerned about the increasing number of injection facilities that might be avoiding surface waste treatment, EPA published an Administrative Decision Statement No. 5 guidance in 1973 regarding
Fig. 1.1. Historical levels of water quality protection (after Walker and Cox, 1973, p. 7).
ELSE_TSANG_CH001.qxd
8/22/2005
7:07 AM
Page 5
1.3 EPA UIC Regulations
5
Fig. 1.2. Approximate number of Class I injection wells by decade.
EPA policy for placement of fluid in the subsurface to prevent contamination of groundwater (Hall and Ballentine, 1973, pp. 786, 790). Passage of the Federal Water Pollution Control Act Amendments (Public Law 92-500) in 1972 gave EPA control of surface waters. Some regulation and permitting of underground injection occurred under this statute, but the authority for control of injection was uncertain. This law did not have clear legal standards for regulating injection. It did, however, require states to regulate injection wells as a prerequisite for federal funding of area-wide waste-treatment management of surface waters. Brine produced during petroleum extraction and injection of fluids for facilitating oil and gas production were exempt from federal control (since these materials were not classified as pollutants under the Amendments), provided that the oil and gas operations were subject to state regulations (Walker and Cox, 1973, p. 9).
1.3 EPA UIC REGULATIONS Enactment of the SDWA in 1974 ratified the EPA’s underground injection policy position and required the Agency to promulgate minimum injection well requirements of state programs to prevent endangerment of USDWs (Brasier and Kobelski, 1996, p. 2). The EPA and state agencies conducted detailed reviews of injection practices during the late 1970s, which were incorporated into the final UIC regulations promulgated by the EPA in 1980 (Brasier and Kobelski, 1996, p. 3). With the 1980 regulations, a national standard was established protecting current and potential drinking water sources with ⬍ 10,000 mg/L total dissolved solids (TDS) that could serve as a public water system (EPA, 1980). Minimum technical requirements for siting, construction, operation, testing, monitoring, and plugging and abandonment
ELSE_TSANG_CH001.qxd
6
8/22/2005
7:07 AM
Page 6
An Overview of Injection Well History in the United States of America
Fig. 1.3. EPA injection well classification system (modified from EPA, 1994a).
were established. Additionally, five classes of injection wells were established (Fig. 1.3). Class I wells are used to inject hazardous and nonhazardous fluids below any USDW. These wells may be industrial or municipal. Class II wells inject brine fluids associated with oil and gas production. Class III wells pertain to in situ mining. Class IV wells (banned except for remediation) handled disposal of hazardous or toxic liquids into or above USDWs. Class V wells relate to geothermal and other wells that do not fall into the previous categories. This paper primarily addresses Class I wells, excluding municipal wells. The 1980 UIC regulations strengthened well standards by requiring multiple layers of protection between injected fluid and USDWs. Before UIC regulations came into effect, one of the few problem wells had well construction materials that were incompatible with unpermitted low pH injectate. Pre-1980 EPA regulations did not require packers, injection tubing, an annulus system, an alarm system, or monitoring of well parameters such as pH. Figure 1.4 is an event-tree for this 1975 incident, which shows that the problem would not have occurred after implementation of the 1980 UIC regulations. In this case, injected fluids entered an unpermitted saline aquifer. The problem was remediated by using the injection well and additional wells to pump fluids out (EPA, 1985, p. 11). A majority of states approved and codified the 1980 regulations from 1982 to 1984. As of 2002, 33 states and 3 territories have UIC primacy. The EPA retained primacy for 10 states,
ELSE_TSANG_CH001.qxd
8/22/2005
7:07 AM
Page 7
1.4 Class I Hazardous Well Regulations
7
Fig. 1.4. Event-tree for a 1975 injection well leak pre- and post-1980 EPA well regulations.
2 territories, Washington, DC, and all Indian tribes. The EPA and the states share primacy for 7 states (EPA, 2002).
1.4 CLASS I HAZARDOUS WELL REGULATIONS In 1984, the Hazardous and Solid Waste Amendments (HSWA) prohibited land disposal of hazardous waste, including underground injection (the “land-ban” restriction), unless the EPA could determine that the disposal would not adversely affect human health and the environment (Smith, 1996, p. 9). In a 1985 Report to Congress on injection of hazardous waste, the EPA Office of Drinking Water stated that underground injection “was considered a method to isolate wastes (that could not be easily treated) from the accessible environment by placing them into deep formations where they would remain for geologic time” (EPA, 1985, p. 3). The report included an inventory of hazardous wells and also looked at hydrogeology, engineering, mechanical integrity tests, monitoring waste characteristics, and noncompliance incidents. From 1986 to 1988, state and federal agencies, environmental groups, and industry participated in negotiated rulemaking (“Reg-Neg”) to implement the land-ban provision of HSWA (EPA, 1991, p. 10). Although the Reg-Neg group did not achieve complete consensus, the EPA (1988) strengthened the regulatory requirements for hazardous injection wells by establishing the no-migration demonstration for hazardous constituents. “The 1988 UIC
ELSE_TSANG_CH001.qxd
8/22/2005
8
7:07 AM
Page 8
An Overview of Injection Well History in the United States of America
regulations . . . offer additional protection by requiring operators of Class I hazardous wells to complete no-migration petitions to demonstrate that the hazardous constituents of their wastewater will not migrate from the injection zone for 10,000 years, or that characteristic hazardous wastewater will no longer be hazardous by the time it leaves the injection zone.” (EPA, 2001, p. xiii). The EPA also stated, “After 10,000 years of containment constituents would either be immobilized or otherwise be at nonhazardous levels throughout the injection zone.” (EPA, 1988, p. 28122). An environmental group that had withdrawn from the Reg-Neg process in the final stages challenged the 1988 EPA UIC Hazardous Waste Disposal Injection Restrictions and Requirements. The U.S. Court of Appeals for the D.C. Circuit ruled in the EPA’s favor and upheld the 1988 regulations, leaving the No-Migration Exemption program for Class I hazardous waste injection wells in place (EPA, 1990).
1.5 RISK ANALYSIS Risk assessment is based on actual exposure as related to concentration and time. Human health or environmental risk from underground injection is extremely low because the potential exposure is removed—that is, injected waste is confined for at least 10,000 years or rendered nonhazardous (EPA, 1997, p. E-6). Figure 1.5 shows the results of a 1989 EPA qualitative and comparative risk study by the Office of Solid Waste and Emergency Response (OSWER). This study determined that injection of hazardous waste in Class I wells is safe and effective because of its very low risk to
Fig. 1.5. Office of Solid Waste and Emergency Response (OSWER) risk assessment (EPA, 1989).
ELSE_TSANG_CH001.qxd
8/22/2005
1.5 Risk Analysis
7:07 AM
Page 9
9
human health and the environment. The EPA study of Class I wells found that injection of waste is safer than burying them in landfills, storing them in tanks, or burning the waste in incinerators (EPA, 1994b). EPA conducted an “Analysis of the Effects of EPA Restrictions on the Deep Injection of Hazardous Waste” (1991). This report concluded that hazardous deep-well injection under EPA’s current regulations is a safe technology, and that the UIC regulations would have prevented the few reported incidents regarding underground injection (1991, pp. 8–9). This report describes in detail how EPA regulations prevent Class I hazardous wells from endangering USDWs. The Land Disposal Program Flexibility Act of 1996 (Public Law 104-119) required EPA to conduct a study regarding the risks associated with Class I nonhazardous injection. The 2001 Report to Congress “Class I Underground Injection Control Program: Study of the Risks Associated with Class I Injection Wells” was their response. The study found that multiple safeguards exist against failure of Class I nonhazardous and hazardous industrial waste wells or the migration of injected fluids (EPA, 2001, p. xii). Siting criteria minimize the potential for waste migration, and inspections, well testing, and passive monitoring systems can detect malfunctions before fluids escape the injection system (EPA, 2001, p. xiii). After several decades of Class I well operations, only four significant cases of injectate migration have been documented, and none of these affected a drinking water source (EPA, 2001, p. xiii). Historical problems were the result of practices not allowed under current UIC regulations. Redundant monitoring systems and multiple protective construction layers reduce failure possibilities. Furthermore, in the unlikely event a well should fail, the geologic and siting criteria are additional safety factors in preventing the movement of injectate toward USDWs (EPA, 2001, p. xiii). Rish et al. (1998) quantitatively estimated the risk of loss of waste containment and movement of injectate into a USDW from a Class I hazardous injection well to be less than one in 1 million. This risk category agrees with EPA studies that deep-well injection is a low-risk management practice. The two failure scenarios dominating risk that waste isolation is lost are: (1) the possibility that a transmissive microannulus develops in the cemented borehole outside of the long string casing, and extends from the injection zone up past the confining zones; and (2) the possibility of inadvertent future extraction of injected waste. The loss of injectate isolation would be low, owing to EPA regulations requiring proper geological siting, buffer aquifer(s), multiple layers of well construction barriers, continuous monitoring systems, and annual mechanical testing. Rish et al. (1998) determined that the annulus pressure system is a critical barrier in preventing contamination to USDWs, but displays high reliability because of the presence of automatic alarms, shut-offs, and fulltime operators. Figure 1.6 is a fault tree that begins with the assumption that the annulus pressure is less than the injection pressure (probability 1.0E⫹00; the actual probability of this occurrence is 5.8E⫺04). Then, the chances of an automatic alarm failing to function (probability 3.0E⫺04) in combination with a full-time operator failing to respond to the alarm (probability 5.0E⫺05) results in a loss of injectate containment probability of 1.5E⫺08. Therefore, an automatic alarm system and a full-time operator are the keys to preventing loss of injectate containment. An automatic alarm system and a full-time operator are required by UIC regulations for hazardous wells, and many states have adopted this requirement for nonhazardous wells by regulatory requirement (e.g., Texas) or by permit requirement (e.g., Louisiana).
ELSE_TSANG_CH001.qxd
8/22/2005
10
7:07 AM
Page 10
An Overview of Injection Well History in the United States of America
Fig. 1.6. Annulus pressure fault tree for Class I hazardous wells. The risk of loss of containment (injected fluid moves into a USDW) is less than one in a million (1.5E-08) (Rish et al., 1998). Table 1.1 UIC timeline 1930 1935 1949 1961 1970 1972 1974 1980 1982–1984 1984 1985 1988 1989 1991 1996 2001
Petroleum industry injection disposal of saltwater from oil and gas production Dow injects spent brine into shallow industrial well DuPont drills first industrial deep well Texas first state to enact injection well laws EPA Subsurface Emplacement Policy Federal Water Pollution Control Act Amendments Safe Drinking Water Act with Federal UIC Program First EPA UIC regulations promulgated State primacy programs; EPA direct implementation Hazardous and Solid Waste Amendments with Land Disposal Ban Report to Congress on Injection of Hazardous Waste EPA No-Migration Exemption Regulations EPA OSWER Comparative Risk Project Report to Congress on Restrictions of Deep Injection of Hazardous Waste Land Disposal Program Flexibility Act Report to Congress on Land Disposal Program—Study of the Risks Associated with Underground Injection Wells
1.6 SUMMARY Prior to UIC regulations in 1980, only four significant cases of injectate migration occurred as a result of Class I hazardous well operations, and none of these affected a drinking water source. Since 1980, with the implementation of the UIC program of the SDWA,
ELSE_TSANG_CH001.qxd
8/22/2005
References
7:07 AM
Page 11
11
no cases of USDW contamination have occurred, due to stringent siting, construction, operation, and testing requirements for Class I hazardous and nonhazardous wells. Those few instances of contamination prior to 1980 would not have occurred had the 1980 regulations been in place. Injection of hazardous and nonhazardous waste into Class I injection wells since 1980 has been, and continues to be, a low-risk method management of liquid wastes that has proven to be safe and effective. Table 1.1 summarizes important events in the history of underground injection, primarily Class I injection. Additional information about the UIC program in the U.S. may be found at: http://www.epa.gov/safewater/uic.html.
REFERENCES Brasier, F.M. and Kobelski, B.J., 1996. Injection of industrial wastes in the United States. In: J.A. Apps and Chin-Fu Tsang, (Eds), Deep Injection Disposal of Hazardous and Industrial Waste. Academic Press, San Diego, pp. 1–8. EPA, Federal Register v. EPA, 1980. 45, No. 98, pp. 33290–33418, May 19, 1980. EPA, Office of Drinking Water 1985. Report to Congress on Injection of Hazardous Waste (EPA 570/9-85-003), May 1985. EPA, 40 CFR §124,144, 146, and 148. 1998. Federal Register Vol. 53, No. 143, pp. 28117– 28157, July 26, 1988. EPA, Office of Solid Waste and Emergency Response, 1989. OSWER Comparative Risk Project: Executive Summary and Overview (OSWER) (EPA 540/1-89/003). EPA, Natural Resources Defense Council v. U.S. EPA, 1990. 907, F. 2d 1146 (D.C. Cir. 1990). EPA, Office of Ground Water and Drinking Water, 1991. Analysis of the Effects of EPA Restrictions on the Deep Injection of Hazardous Waste (EPA 570/9-91-031). EPA, Office of Water, 1994a. Underground Injection Wells and Your Drinking Water (EPA 813-F-94-001). EPA, Office of Water, 1994b. Class I Injection Wells and Your Drinking Water (EPA 813-F94-002). EPA, Office of Pollution Prevention and Toxics, 1997. Toxic Release Inventory Relative Risk-Based Environmental Indicators Methodology. EPA, Office of Water, 2001. Class I Underground Injection Control Program: Study of the Risks Associated with Class I Underground Injection Wells (EPA 816-R-01-007) 76p (includes the Land Disposal Program Flexibility Act of 1996, Public Law 104-119, as Appendix A). EPA, Office of Ground Water and Drinking Water, 2002. Protecting Drinking Water through Underground Injection Control (EPA 816-K-02-001), January 2002. Hall, C.W. and Ballentine, R.K., 1973. U.S. Environmental Protection Agency policy on subsurface emplacement of fluids by well injection. In: Underground Waste Management and Artificial Recharge Vol. 2. Preprints of papers presented at the Second International Symposium on Underground Waste Management and Artificial Recharge, New Orleans, LA, September 26–30, 1973, American Association of Petroleum Geologists, U.S. Geological Survey, and International Association of Hydrological Sciences, pp. 783–789. Harlow, I.F., 1939. Waste problems of a chemical company. Ind. eng. chem., 31: 1346–1349. Rish, W.A., Ijaz, T. and Long T.F., 1988. A Probabilistic Risk Assessment of Class I Hazardous Waste Injection Wells (draft). Smith, R.E., 1996. EPA mission research in support of hazardous waste injection, 1986–1994. In: J.A. Apps and Chin-Fu Tsang (eds), Deep Injection Disposal of Hazardous and Industrial Waste. Academic Press, San Diego, CA, pp. 9–24.
ELSE_TSANG_CH001.qxd
12
8/22/2005
7:07 AM
Page 12
An Overview of Injection Well History in the United States of America
Walker, W.R. and Cox, W.E., 1973. Legal and institutional considerations of deep-well waste disposal. In: Underground Waste Management and Artificial Recharge Vol.1. Preprints of papers presented at the Second International Symposium on Underground Waste Management and Artificial Recharge, New Orleans, Louisiana, September 26–30, 1973, American Association of Petroleum Geologists, U.S. Geological Survey, and International Association of Hydrological Sciences, pp. 3–19. Warner, D.L. and Orcutt, D.H., 1973. Industrial wastewater-injection wells in United States status of use and regulation, 1973. In: Underground Waste Management and Artificial Recharge Vol.2. Preprints of papers presented at the Second International Symposium on Underground Waste Management and Artificial Recharge, New Orleans, LA, September 26–30, 1973, American Association of Petroleum Geologists, U.S. Geological Survey, and International Association of Hydrological Sciences, pp. 687–697.
ELSE_TSANG_CH002.qxd
8/19/2005
6:24 PM
Page 13
Chapter 2
DEEP INJECTION DISPOSAL OF LIQUID RADIOACTIVE WASTE IN RUSSIA, 1963–2002: RESULTS AND CONSEQUENCES A.I. Rybalchenkoa, M.K. Pimenova, V.M. Kurochkina, E.N. Kamneva, V.M. Korotkevichb, A.A. Zubkovb, and R.R. Khafizovc a
All-Russia Designing and Research Institute of Production Engineering (VNIPIPT), Moscow, Russia b Siberian Chemical Combine, Seversk, Russia c Mining and Chemical Combine, Zeleznogorsk, Russia
2.1 INTRODUCTION Deep-well injection of liquid radioactive waste obviates the need for surface construction of additional liquid radioactive waste and industrial waste storage sites. Such surface storage facilities could contaminate the environment by discharging industrial waste discharge into lakes and rivers. Injection also leads to significant cost savings, preserving funds that would otherwise be spent on construction of surface storage equipment with antifiltration barriers and other protective constructions. June 2003 marked the 40th anniversary of experimental deep-well injection of mediumlevel radioactive waste at the Siberian Chemical Combine (Tomsk-7, or Seversk). Two sand reservoir horizons were used at a depth of 270–320 m and 314–386 m (see Table 2.1). This experimental deep injection project was developed out of preliminary investigations of the area’s geological structure, near the Siberian Chemical Combine. Deep-well injection was accompanied by observations confirming the predictions and data of preliminary investigations. Positive results from these experimental activities and additional investigations led to the creation of a deep-well injection facility (at an industrial scale) for three categories of waste: low level, medium level, and high level. In 1966, a few years after the beginning of the experimental deep-well injection project, deep-well injection was further developed at the State Scientific Center of Russian Federation “Research Institute of Atomic Reactors” (at Dimitrovgrad, in the Ulianovskaia Region) and at the Mining and Chemical Combine (Krasnojarsk-26, or Zeleznogorsk). Then, in 1967, the experience gained from deep-well injection of radioactive waste was used at an injection site for nonradioactive waste (albeit created by the atomic energy industry) at the Kalinin atomic power plant (see Table 2.1). While investigating sites for proposed deep-well injection (feasibility studies) in 1955–1960, and site operations in 1963–1990, new data and practical experience were gained. This information was summarized at a previous International Symposium in 1994 on deep-well injection of toxic liquid waste (Rybalchenko et al., 1998). Operating designs for injection of radioactive waste were completed during the period 1995–1998. Recent analyses of international deepwell injection projects (European Commission, 1999, 2003), as well as discussion of these projects with foreign experts at international conferences, has indicated the feasibility of continuing deep-well injection at functioning sites into the 21st century and up to 2010. This report considers those new analyses for different aspects of deep-well injection of liquid radioactive waste, which substantiate the continuing use of deep-well injection in Russia.
ELSE_TSANG_CH002.qxd
8/19/2005
14
6:24 PM
Page 14
Deep Injection Disposal of Liquid Radioactive Waste in Russia, 1963–2002
Table 2.1. Deep injection sites for liquid radioactive waste and nonradioactive waste from enterprises of Minatom, Russia Enterprise
Liquid radioactive waste Siberian Chemical Combine (Tomsk-7) Mining-Chemical Combine (Krasnojarsk-26) Institute of Nuclear Reactors (Dimitrovgrad)
Depth of Type of reservoir injection (m) horizon and underground water
Commencement of injection (year)
Volume of injected waste (mln ⫻ m3)
270–320 314–386
Sand, sandstones, freshwater
1963
43.5
180–280 355–500
Sand, freshwater
1967
6.1
Limestones, brines
1966
2.5
Limestones, brines
1992
4.3
1260–1440
Limestones, brines
1987
4.1
1200–1400
Sand, brines
2005
––
1130–1410 1440–1550
Liquid nonradioactive waste Chepetsk Mechanical 1435–1600 Plant (Glazov) Kirovo-Chepetsk Chemical Combine (Kirovo-Chepetsk) Kalinin Atomic Power Plant* (Udomlia)
*Site injection is being constructed; waste containing tritium can be injected in it.
2.2 CHARACTERISTICS OF DEEP-WELL INJECTION SITES AND PRELIMINARY INVESTIGATION 2.2.1 Tomsk-7 and Krasnojarsk-26 in 1963–2003 Deep-well injection of liquid radioactive waste in a freshwater horizon at Tomsk-7 and Krasnojarsk-26 was begun in the early 1960s, when the Soviet government granted a permit for use of this horizon. The context surrounding this permit was exceptional because of the context of contemporary events connected with storing waste on the surface. One such event occurred at “Mayak” Combine (South Ural), where a surface storage tank for radioactive waste exploded and contaminated vast areas, including the Techa River. However, as later evaluations showed, contamination of freshwater horizons would only slightly reduce the usability of underground water for the water supply in the Tomsk-7 and Krasnojarsk-26 region, if limitations on the exploitation of the geologic medium were observed. The geological structure of the deep injection site at Krasnojarsk-26 has a distinctive feature: the reservoir horizons are located in a sinkline depression, the right edge of which is limited by a tectonic fault (a hydrodynamic barrier). Waste has more density than underground water, and that works to retard waste migration in the central part of the depression.
ELSE_TSANG_CH002.qxd
8/19/2005
6:24 PM
Page 15
2.2 Characteristics of Deep-Well Injection Sites and Preliminary Investigation
2.2.2
15
Dimitrovgrad in 1966–2003
Geological conditions for deep-well injection of liquid radioactive waste at Dimitrovgrad were quite favorable. The reservoir horizon is located at a depth of more than 1000 m, contains water with salt concentration of 220–240 g/kg, and is completely isolated from shallow-bedded fresh water. The pore space of the reservoir horizon presents generally open and oriented fractures. However, the flow structure in such a horizon is rather complex, and that creates difficulties in interpreting the data acquired by investigations and observations of injection processes. The establishment of injection sites at Dimitrovgrad in 1957–1965 was preceded by investigations of geological structure and hydrogeology, to determine site suitability for waste injection. These investigations included well drilling, complex geophysical investigations, filtration tests, and sampling of rock samples, underground water, and waste. Considerable material was gathered and processed, and these materials were used to design the deep injection sites. The project was approved by experts and control bodies, and quickly implemented. Surface accumulation of liquid radioactive waste was ceased. Results from preliminary investigations were confirmed by observations during the injection process. 2.2.3 Conclusions of Preliminary Investigation From the perspective of our recent findings and experience, we can make the following conclusions: ● The number of drilled exploratory wells on the preliminary stage was evidently excessive. For example, while conducting exploratory activities in the Tomsk-7 region, 128 wells were drilled. The work there could be performed with many fewer wells. ● Test-filtration investigations—pumping out and injection allowed us to define some parameters of the reservoir horizon and to evaluate the isolating properties of above-lying low, penetrable horizons rather reliably. Because of this, some objective data were accumulated. However, the sensitivity of the employed methods for defining underground water levels should be higher, given the capabilities of the latest technology. ● At Dimitrovgrad, while investigating the conditions for injection of waste with low salt content into a rock formation with high clay content, we did not take into account the possibility of sharp decreases in filtration properties when exchanging natural salt water for waste that did not contain salt. This led to complications in the initial stage of investigation, but this problem was in time overcome—by elevating the interval for injection into the limestone above the original injection location. ● The application of radioactive indicators as tracers for determining characteristics of a reservoir horizon was not effective, and in several cases were a reason for erroneous decisions. This last conclusion needs further explanation. The 1960s saw the beginning of intensive application of radioactive isotopes for different spheres (for example, medical technology). This development was considered as part of a trend toward peaceful use of atomic energy. In geology, radioactive isotopes were used to determine underground water movement. However, a variety of factors affecting the predictability of radioactive isotopes within a geological medium impeded foolproof interpretation of data. The classic example of this was a geological exploration in the region of the “Mayak” combine (South Ural region). In the area where the Tech-Brodskay structures are located, there is a syncline depression analogous to the synclinal structure within the Krasnoyarsk-26 region, but greater in area (~two times larger) and containing sedimentary rocks that are much thicker (~four times thicker).
ELSE_TSANG_CH002.qxd
16
8/19/2005
6:24 PM
Page 16
Deep Injection Disposal of Liquid Radioactive Waste in Russia, 1963–2002
Permeable rock in the lower part of the depression at a depth of 1000–1400 m contains salt water. Even so, radioactive isotopes injected at a depth of 1400 m were found 4 hours later at a depth of 400 m. From this, we concluded that there was a tight interconnection between the upper and lower horizon. Modern analysis has shown that the horizons under investigation were connected through wells whose annuli (between rock and casing) were not cemented. Unfortunately, because of this, plans to create a deep-well injection site within “Mayak” were cancelled, which resulted in the development of a dangerous ecological situation.
2.3 INVESTIGATIONS OF WASTE BEHAVIOR AND NEW DATA RECEIVED WHILE OPERATING DEEP-WELL INJECTION SITES Scientists involved in deep-well injection projects envisaged observation wells in which measurements and sampling of underground waters would be conducted. These activities would be aimed at checking data from preliminary investigations and making decisions on that basis. The observation wells were located both in the vicinity of the waste injection location at 10–50 m and at a distance of hundreds to thousands of meters. They were equipped for both reservoir and upper horizon control. In the wells, samples of underground water were taken to determine their composition, including waste components, geophysical measurements of radioactive radiation and temperature, determination of underground water levels, and their depth. Observations confirmed that these hypotheses concerning the sites for injection, about their geological structure, hydrogeology, and rock properties, were correct. At the same time, some variances were discovered, and these necessitated some adjustments in the injection sites. These variances, listed below, are interesting from the perspective of applying previous deep-well-injection experience to other regions: ● While injecting the waste, layers (in a sand-reservoir horizon) and fractured zones (for a carbonate reservoir horizon) became filled. The total thickness of those layers or zones is significantly less than it was supposed to be, according to data from geological exploration. At the same time, effective porosity was higher, owing to the real specific capacity of reservoir horizons being close to the designed specific capacity. ● Completeness of reservoir horizon injection (or number of layers being filled) depends on injection pressure. Under pressure close to or above hydraulic fracturing, separate thin zones are filled, resulting in accelerated waste spreading. In this connection, injection pressures must be lower than hydraulic fracturing pressures. ● Clay layers and horizons effectively prevent vertical redistribution of waste. Clay layers with a thickness of several meters included within a sand reservoir horizon play the role of local confining layers. Regularities of penetration and movement of waste in clay differ from regularities for sand (Darcy’s law). ● Radioactive nuclides (waste components) are intensively retarded in sand–clay rock containing fresh water. These conditions cause deceleration of radioactive migration. An increase in the salt content of underground water reduces the retarding properties of rocks. Operating injection sites for liquid radioactive waste requires solving a number of problems, the most important of which is the technical condition of wells and geodynamic phenomena, including seismicity. While creating the first injection wells for the site injection at Tomsk-7, casings of largediameter (11 inches and more), double columns were established within the interval of the reservoir horizon and the above-lying horizon to create an additional barrier between the well
ELSE_TSANG_CH002.qxd
8/19/2005
6:24 PM
Page 17
2.4 Safety Requirements and Criteria while Performing Site Injection
17
casing and geological medium. Later on, a problem was discovered with these wells: specifically, the movement of underground water into the space between the external wall of a casing and the wall of rock, caused by nonuniform distribution of cement in the space between the external wall and the large-diameter casing. The quality of the cement improved when well constructions were used with smaller-diameter casings of 5 or 6 inches. Investigation of special cements, used for grouting of the well, showed that the stone formed from such cement is rather stable within the intervals of sandy rocks, containing waters with low salt concentration. Isolating properties of cement stone improved over time. In the Krasnojarsk-26 region, at the Western complex of sedimentary rocks containing reservoir horizons, investigations of geodynamic conditions for deep-well injection of liquid radioactive waste were conducted to evaluate the possible influence of limitations imposed by tectonic faults. (Precedents have occurred in the United States where injection of waste in a zone of tectonic faults resulted in small earthquakes.) Investigations included seismic observations employing a stationary system located in the area of tectonic faults and injection wells, highly precise geodesy, and observations over the position of the earth’s surface. The following results were obtained: ● While conducting seismic observations, no events were registered as a consequence of waste injection, though distant earthquakes and nuclear weapons tests were noted. This was to be expected, since the area for deep-well injection, characterized by small natural seismicity and tectonic faults, is an almost-vertical clay screen and practically impermeable. ● According to data for determining surface position, positive increments were marked near injection wells during waste injection at the level of first few millimeters, it being known that near wells on the upper reservoir horizon (second horizon, 180–280 m intervals) increment are registered more confidently than the lower reservoir horizon (first horizon, 355–500 m intervals). Geodynamic events connected with surface uplift could be a reason for the deteriorating integrity and isolating properties of the grout in the space outside of the well casing. For the lower horizon, such phenomena were not predominant. The operational resources of wells on the upper horizon were reduced. The results of these control observations were used to choose optimum conditions for waste injection, conducting other geologic-technical measures, repair, and well shutdown. It was stated that waste components were within the limits of forecast boundaries, with processes occurring in reservoir horizons also corresponding to forecasts. The results of observations were used to specify geomigration models for time periods following injection. In accordance with the calculations, waste will be localized within the stated period of a minimum 1000 years. Consequently, the safety of liquid radioactive waste injection has been confirmed. 2.4 SAFETY REQUIREMENTS AND CRITERIA WHILE PERFORMING SITE INJECTION FOR LIQUID RADIOACTIVE WASTE AT THE PRESENT TIME The main problem of deep-well injection of liquid radioactive waste consists of limiting its safety hazards during injection and after shutdown—in other words, limiting the consequences of deep-well injection for future generations. The possibility of direct contact between liquid forms of radioactive waste and the underground waters of deep-reservoir horizons is a clear cause for concern, and requires detailed investigations of injection consequences and confirmation of its safety, taking into account modern requirements and criteria. These requirements and criteria frequently differ from those used when these injection sites were created, back in the 1950s and 1960s.
ELSE_TSANG_CH002.qxd
8/19/2005
18
6:24 PM
Page 18
Deep Injection Disposal of Liquid Radioactive Waste in Russia, 1963–2002
When these injection sites were created, requirements were geared toward injection safety, the main requirement of which was isolation of waste within reservoir horizons during the period sufficient for decay of radionuclides (i.e., for the products of fission to diminish to concentrations not exceeding that allowed in potable waters, according to norms established by law). Such isolation was provided for if appropriate conditions existed at the site of waste injection. Moreover, the reservoir horizon had to possess sufficient specific capacity. Horizon filtration properties had to be sufficiently high; at the same time, the rate of natural movement of underground waters could not exceed several meters per year. Above a reservoir horizon, there should be a clay horizon, with its low-permeability sediments limiting the vertical movement of water. The numerical characteristics of geological section properties and their ratios were the criteria for such sites. However, in due course, concern about the potential danger of radioactive waste led to the toughening of requirements protecting the environment and population. This increased concern was reflected in the adoption of new laws and norm documents, and made it necessary to improve safety criteria for deep-well injection of liquid radioactive waste. At present, deep-well injection of liquid radioactive waste is regulated by Federal Act “On Mineral Resources” (2000). According to this law, geological formations are granted into use for disposal of different kinds of wastes, including radioactive waste, based on observing a number of requirements for waste localization in reservoir horizons. The Federal Act “On Radiation Safety for the Population” (1997) lays down limitations on the radioactivity exposure to the population, stated in terms of radiation dose. The main criteria for waste localization is that waste must be contained within the limits of established boundaries for a geological medium, and dose criterion, i.e., radiation dose for a human, must not exceed a normalized value. The localization criterion is based on regional allotment of waste injection, within the boundaries of which waste can be located. A probability criterion was also used—the probability of exceeding a normalized value. Based on the indicated criteria in every individual case, secondary criteria are formed, including those concerning migration characteristics and depths of reservoir horizon bedding. A number of requirements for waste management are included in Federal Law “On the Environment” (2002). “Discharge” of waste into geological formations and wells is forbidden. At the same time, disposal or burying (injection of liquid waste) is not equated to “discharge,” and corresponding requirements are claimed to it. Unfortunately, Russian legislation has a distinct definition for “discharge” and that makes it difficult to regulate. Similar difficulties were overcome by the creation of a special norm regulating documents on the basis of federal legislation and practical experience. In the sphere of deep-well injection of liquid radioactive waste, there are some guidances and rules in effect, which were developed in cooperation with the Russian Ministry of Public Health. 2.5 CONCLUSION From the 1960s into the 21st century, deep-well injection of liquid radioactive wastes has played a significant role in preventing environmental contamination. Injected wastes were localized within the stated boundaries. According to forecasts, any direct effects on humans or on any other forms of life are not expected. Still, deep-well injection is considered a temporary measure, to be changed through the technology of waste solidification. At the same time, continued developments show that it is reasonable to combine solidification of specific
ELSE_TSANG_CH002.qxd
8/19/2005
References
6:24 PM
Page 19
19
waste categories with liquid-form disposal, if favorable geological conditions are present for those purposes. Forecast accounts and modeling were carried out to support the continued deep-well disposal of liquid radioactive wastes up to 2010–2115. Injection projects have been prepared, evaluated ecologically, and judged positively. The continuation of deep-well injection of liquid radioactive wastes, at operational sites, will help to avoid the significant economic expenses associated with the application of alternative technologies—and will simultaneously ensure safe waste disposal. REFERENCES Federal Act, 1997. On Radiation Safety for the Population, Collection of the Acts of the President and Government of the Russian Federation, Edition of the Administration of the President of the Russian Federation. Federal Act, 2000. On Mineral Resources, Collection of the Acts of the President and Government of the Russian Federation. Edition of the Administration of the President of the Russian Federation. Federal Act, 2002. On the Environment, Collection of the Acts of the President and Government of the Russian Federation. Edition of the Administration of the President of the Russian Federation. Rybalchenko, A., Pimenov, M. and Kostin, P., 1998. Injection Disposal of Hazardous and Industrial Wastes, Scientific and Engineering Aspects. In: Deep Injection Disposal of Liquid Radioactive Waste in Russia. Academic Press, New York. Vieveg, M., Denecke, C., Neerdal, B., Schneider, L., Lopatin, V., Kamnev, E., Rybalchenko, A., Sigaev, B., Zacharova, E. and Tichkov, V., 1999. Evaluation of the Radiological Impact Resulting from Injection Operations in Tomsk-7 and Krasnoyarsk-26, Final Report, European Commission, EUR 18189 EN. Wickham, S., Galson, D., Sillen, X., Wang, L., Marivoet, J., Beaucaire, C., Artinger R., Klenze, R., Selenska Pobbel, S., Rybalchenko, F., Zubkov, A., Aranovich, L., Zakharova, E., Torras, J., White, M., Beker, A., and Knight, L., 2003. Building Confidence in Deep Disposal: The Borehole Injection Sites at Krasnoyarsk-26 and Tomsk-7 (BORIS). Final Report, European Commission, Euratom, EUR 20615 EN.
This page intentionally left blank
20
ELSE_TSANG_CH003.qxd
8/19/2005
6:25 PM
Page 21
Chapter 3
APPLICATIONS OF DEEP-WELL INJECTION OF INDUSTRIAL AND MUNICIPAL WASTEWATER IN TEXAS B. Knape Texas Commission on Environmental Quality, Austin, TX, USA
3.1 INTRODUCTION A number of uses of deep-well injection for disposal of wastewater have been successfully made by industries and municipal governments in Texas over the past 50 years. All of these uses of injection wells for wastewater disposal have in common a very high rate of success in isolation of wastewater below underground sources of drinking water (USDW), and in the consequent protection of human health and the environment. An overview of the uses and regulations of such wells for industrial and municipal wastewater disposal in Texas is provided here. Deep-well injection of industrial and municipal waste, as an outgrowth of Frasch-process sulfur mining, originated in the early twentieth century with the use of injection of superheated water and the development of deep-well injection disposal of produced oilfield brines in the East Texas Basin during the 1930s. The first recorded use of deep-well injection for disposal of industrial waste in Texas was initiated by DuPont Chemical Company in 1951. As a point of interest, although the average lifetime of such waste disposal wells is considered to be approximately 30 years, this DuPont well is still in use for safe disposal of hazardous waste. In the 1960s, Monsanto Chemical Company, followed by Celanese Chemical Company and others, began similar wastewater injection operations in Texas. The list of companies that have used deep-well injection for industrial waste includes many internationally known companies involved in chemical manufacturing, refining of petroleum and metals, electric power generation, food processing, commercial waste management, and environmental remediation. 3.2 REGULATORY JURISDICTION AND FEDERAL PROGRAM AUTHORIZATION As provided by the Texas Injection Well Act codified as Chapter 27, Texas Water Code, protection of freshwater from pollution through regulation of underground injection of industrial and municipal waste is within the regulatory jurisdiction of the Texas Commission on Environmental Quality (“the Commission”). In 1974, the federal Safe Drinking Water Act authorized creation of the Underground Injection Control (UIC) Program. This act mandated the United States Environmental Protection Agency’s (EPA’s) development of regulations for protecting USDWs from contamination by injection wells, and regulations for authorizing states to administer the UIC Program (Title 40, Code of Federal Regulations [CFR], Parts 144, 145, 146, 147, and 148). Upon meeting requirements for federal authorization of the parts of the UIC Program for the wells under its jurisdiction, the Texas Department of Water Resources, the Commission’s predecessor, received federal authorization of its UIC Program in 1982.
ELSE_TSANG_CH003.qxd
22
8/19/2005
6:25 PM
Page 22
Applications of Deep-Well Injection of Industrial and Municipal Wastewater in Texas
3.3 DEFINITION OF TERMS Under the federal classification system, injection of industrial and municipal waste is designated as either Class I or Class V injection, depending on whether the injection zone is below all USDWs (Class I), or within or above USDW (Class V), respectively (40 CFR §144.6). For purposes of this chapter, “deep-well injection of industrial and municipal wastewater” is generally synonymous with Class I injection under the federal UIC Program. Class V injection wells are generally much shallower than Class I wells, include miscellaneous types and purposes of injection beyond waste disposal, and consequently, their use is not generally described as “deep-well disposal.” The term “USDW,” as provided in federal regulation, is defined as the area of the subsurface strata containing groundwater with a total dissolved solids (TDS) concentration less than 10,000 mg/L (40 CFR §144.3). The term “freshwater,” as defined in Texas rules, generally has the same TDS standard as defined for USDWs, but may include water with TDS concentration greater than 10,000 mg/L if it can be demonstrated that such water has a beneficial use (30 TAC §331.2).
3.4 WELL NUMBERS AND LOCATIONS Figure 3.1 provides a map of Texas indicating the general distribution of Class I injection wells. Each dot on the map generally indicates the presence of a cluster of wells. At present, there are 102 Class I injection wells active in the state. Just over half (approximately 56%) are authorized to inject hazardous waste; the remainder is limited to nonhazardous waste. The Class I well inventory presently includes no wells for municipal waste. There are recent proposals, however, for using Class I injection wells for disposal of municipal waste from desalination operations for public-water supply. Figure 3.1 shows the concentration of Class I wells within the Gulf Coast and western regions of the state, and a belt through the center of the state devoid of such wells. The areas with Class I wells correspond to deep sedimentary basins providing porous and permeable injection zones to hold wastewaters, and impermeable confining strata to ensure permanent separation of injected wastes from USDWs. That these same areas are noted for the production of oil and gas is not surprising, because the same geologic conditions favorable to trapping oil and gas for geologically long periods of time also provide conditions for safe disposal of industrial and municipal waste for similarly long time periods. The area through the central part of the state without Class I wells is indicated, in geotechnical reports, to be less conducive generally to underground injection of waste, as a result of geologic features, including the Balcones Fault Zone, the Llano Uplift (Precambrian basement), important sole-source aquifers for drinking-water supplies, and in the southwestern end of the trend, tertiary volcanics (Knape et al., 1984).
3.5 TECHNICAL REQUIREMENTS OF RULES The Commission’s Underground Injection Control rules are provided in Title 30, Texas Administrative Code (TAC), Chapter 331. Chapter 331 provides technical standards for Class I injection wells. These standards provide basic definitions; considerations for permit issuance; and requirements for construction, operation, testing, monitoring, record keeping, reporting,
ELSE_TSANG_CH003.qxd
8/19/2005
6:25 PM
Page 23
3.6 General Characteristics of Injected Waste Streams
23
Fig. 3.1. Locations of Class I injection wells in Texas. Well locations represent the conjunction of industrial operations with suitable geology and other siting factors. Presently, there are 102 active permitted wells in the state.
corrective action, closure, and postclosure care of Class I wells. Figure 3.2 shows the standard design elements of a Class I well as required by the rule. The rules emphasize the importance of redundant protective barriers in the well and in the host geologic formations. As a result of the rigorous requirements for redundant protection barriers in Class I injection well operations, and for monitoring well operations to ensure the wells are maintaining the required levels of safety for drinking-water resources, there have been no incidents of pollution of USDW from Class I injection wells in Texas since 1982, the year the state’s UIC Program was federally authorized. The Texas UIC Program experience in regulating Class I injection wells is consistent, therefore, with the 1989 EPA study finding that “injecting wastes in Class I wells is safer than burying them in landfills, storing them in tanks, or burning the waste in incinerators” (EPA Office of Solid Waste and Emergency Response, 1989). Furthermore, the Commission’s UIC Program experience has been in line with the conclusion by the EPA Office of Water that “the probability of Class I well failures, both nonhazardous and hazardous, has been demonstrated to be low” (EPA Office of Water, 2001). 3.6 GENERAL CHARACTERISTICS OF INJECTED WASTE STREAMS The waste streams injected in Class I injection wells in Texas are aqueous streams with inorganic chemicals and usually minor amounts of organic chemicals. Wastes injected in Class I wells include hazardous and nonhazardous waste. Hazardous waste streams possess the characteristics of ignitability, corrosivity, reactivity, or toxicity, as provided under 40 CFR Part 261, Subpart C, or are listed as hazardous waste under 40 CFR Part 261, Subpart D.
ELSE_TSANG_CH003.qxd
24
8/19/2005
6:25 PM
Page 24
Applications of Deep-Well Injection of Industrial and Municipal Wastewater in Texas
Fig. 3.2. Profile of a Class I disposal well (from Rish, et al., 1998).
3.7 INDUSTRIES WITH CURRENT OR HISTORICAL USE OF CLASS I INJECTION WELLS 3.7.1 Petroleum Refining and Chemical Manufacturing Injection of wastewater from petroleum refining and chemical manufacturing comprises the largest used category of Class I injection wells in Texas. Currently there are 74 such wells. Approximately two-thirds of these wells are authorized to inject hazardous waste; the remainder are limited to nonhazardous waste. As a result of federal hazardous waste land disposal restrictions (LDR), all Class I wells injecting restricted hazardous waste require EPA approval of petitions for exemption from the LDR, in addition to requiring permits issued by the Commission (40 CFR Part 148 and 30 TAC §331.7). 3.7.2 Metals Refining Class I injection wells are used for wastewater disposal at the Asarco Copper Refinery in Amarillo. The refinery uses an electrolytic process to purify copper, with the production of
ELSE_TSANG_CH003.qxd
8/19/2005
6:25 PM
Page 25
3.7 Industries with Current or Historical Use of Class I Injection Wells
25
silver and gold as by-products. Asarco operates four Class I wells for disposal of low-pH hazardous wastewater from these metal-refining processes. 3.7.3 In Situ Uranium Mining Since 1982, the Commission has regulated 18 in situ uranium mines in south Texas. Each of these Class III injection well mine operations has had one or more permitted Class I wells for disposal of wastewater. The wastewater is regulated as a by-product waste rather than a radioactive or hazardous waste, with radium being one of the most significant constituents. After mining ceases, rules require restoration of the groundwater quality in the mine zone to near background condition. Class I wells play a key role in this restoration process. The process involves sweeping mining solutions from the ore body by producing water from the affected portion of the aquifer. Groundwater produced during aquifer restoration is injected in the Class I wells on site. 3.7.4 Electric Power Generation Class I injection wells have been used at three electric power plants for disposal of water for steam generation and cooling. Water is typically recycled 6–10 times through the steam generation and cooling systems, progressively becoming more concentrated in TDS, and necessitating disposal to maintain optimum system efficiency. 3.7.5 Animal and Food Processing Two facilities in the Texas Panhandle use Class I wells for injection of wastewater from beef processing: Tyson and ConAgra. Also, Pilgrim’s Pride has recently received Class I permits for wastewater injection from a proposed poultry processing plant in northeast Texas. 3.7.6 Commercial Waste Management and Disposal Since the Commission’s UIC Program federal authorization in 1982, 12 Class I wells have been permitted for commercial disposal of industrial wastewater including hazardous waste, and five wells have been permitted for commercial disposal limited to nonhazardous waste. At present, only three wells are used for commercial hazardous waste. The other wells authorized for commercial hazardous waste have been voluntarily removed from commercial operation as a result of internal business decisions or negotiated agreements with local residents, have stopped operating as a result of bankruptcy, or have lost permit authorization because of noncompliance with Commission rules. 3.7.7 Environmental Remediation (Site Cleanup) In a number of instances, such as those described for Class I wells at in situ uranium mines, Class I wells have been used for environmental remediation and site closure. Class I wells have also been used for environmental remediation and facility closure in three other projects: injection of pond water containing tin and other metals at the Textin smelter in Texas City; site cleanup and closure of the American Ecology commercial hazardous waste management facility in the northeast Texas city of Winona; and site cleanup and closure of the Malone Services commercial hazardous waste management facility in Texas City.
ELSE_TSANG_CH003.qxd
26
8/19/2005
6:25 PM
Page 26
Applications of Deep-Well Injection of Industrial and Municipal Wastewater in Texas
3.8 PROPOSED USES OF CLASS I INJECTION WELLS 3.8.1 Carbon Dioxide Sequestration The Commission is currently reviewing an application for authorization of injection wells as part of a pilot study on geologic sequestration of carbon dioxide (CO2). The pilot study is being developed and coordinated by the Texas Bureau of Economic Geology with funding by the U.S. Department of Energy (DOE). The study’s objective is to determine environmental safety of the process, its effectiveness in keeping CO2 out of the atmosphere, and whether the technology can be effective in large-scale reduction of waste CO2 release in the Gulf Coast region. The study proposes construction of an injection well to Class I standards. The well will be located near Houston in the South Liberty field, with injection into the Frio Formation at a subsurface depth of 5500 ft. Surface and subsurface monitoring in the pre-injection phase to establish baseline conditions will be continued through the CO2 injection and post-injection phases. Monitoring data will be used to develop a computer model of the actual reservoir performance for use in predicting outcomes of large-volume CO2 injection in the subsurface near refineries or power plants. 3.8.2 Desalination of Produced Water There is currently a very high level of interest in Texas in developing projects for desalination of groundwater or seawater. Interest is strongest in areas with concerns about being able to provide adequate public water supplies for growing populations. The Office of the Governor has therefore advocated development and implementation of desalination projects to provide new drought-resistant water supplies for the more arid regions of the state. The Commission has already approved Class V injection of desalination waste for a residential development on south Padre Island. Two larger desalination projects proposing to use either Class I or Class V injection for the waste stream are being developed: one involves the City of El Paso in the far western part of the state in partnership with the Army Corps of Engineers; the other is being proposed by the City of Corpus Christi for operation on north Padre Island. 3.8.3 Salt Cavern Disposal of Industrial Solid Waste Based on the success of hydrocarbon storage in salt, and on studies evaluating the potential for radioactive waste disposal in salt, proposals began to be made in the early 1980s to the Commission for permitting engineered salt caverns developed by salt dissolution for disposal of industrial hazardous wastes. A number of unique properties of salt have led to its consideration as a containment medium for hazardous wastes. These properties include its highly impermeable nature, its lack of chemical reactivity with other substances, and at depths greater than 300 m, its viscoplastic behavior. Ever since the adoption of federal regulations prohibiting free liquid hazardous wastes in salt caverns, permit applications for such disposal of hazardous waste in Texas have included plans to (1) stabilize (solidify) liquid wastes and (2) inject the solidified wastes into dry (dewatered) caverns. All such proposals have involved salt domes (salt diapirs) in the Gulf Coast and east Texas regions; no permit applications have been received for industrial waste disposal in caverns in bedded salt characteristic of west Texas. In the Commission’s
ELSE_TSANG_CH003.qxd
8/19/2005
6:25 PM
Page 27
3.9 Conclusion
27
1992 adoption of the nation’s first rules providing technical standards for salt cavern injection of industrial solid waste in domal salt, such cavern operations in bedded salt were prohibited (30 TAC 331, Subchapter J). The Commission’s rules for salt cavern waste disposal feature a “no escape” performance standard, which requires a demonstration that no hazardous constituents will escape from the salt cavern injection zone (30 TAC §331.162). The standard provides that such a demonstration include modeling simulations based on a 15,000-year time period. Figure 3.3 shows a cross section of a salt dome and a cavern for solid waste disposal. The diagram illustrates the critical issue recognized in the Commission’s rules: assurance of proper spacing between caverns and the edge of the salt stock. The left edge of the salt stock in the diagram indicates the presence of an anomaly in the salt–sediment interface, which, if undetected, could result in an insufficient thickness of salt wall for any cavern constructed too close to this anomaly. The commission’s rules, therefore, require the permit application to contain a thorough characterization of the geology of the salt dome, including a threedimensional seismic survey sufficient to demonstrate at least 500 ft of salt between the boundary of any salt cavern injection zone and the edge of the salt stock, 30 TAC §§331.121(d) and 331.164(b). Five proposed salt cavern disposal projects with permit applications for industrial hazardous waste disposal have been developed since the early 1980s. As a result of strong public and political opposition, with concerns about completeness and technical adequacy of the applications, geologic suitability of the host salt dome, adequacy of the waste stabilization/solidification technology, and the degree of public need for such a facility, no permits have been issued in response to these applications. In 2001, the Texas legislature prohibited hazardous waste disposal in salt dome caverns.
3.9 CONCLUSION Over the past 50 years, deep-well injection of industrial waste has developed a very significant record of protecting human health and the environment by protecting freshwater from pollution. The protectiveness of such injection operations is the result of rules requiring
Fig. 3.3. Salt dome with salt cavern disposal wells.
ELSE_TSANG_CH003.qxd
28
8/19/2005
6:25 PM
Page 28
Applications of Deep-Well Injection of Industrial and Municipal Wastewater in Texas
redundant protective barriers, both in the design of wells and in the host geologic formations. The absence of instances of pollution of drinking water by Class I injection wells during the history of the Commission’s federally authorized UIC Program is indicative of well operators’ and regulators’ diligence in complying with these rules. In protecting the environment, the wells have demonstrated success in meeting the wastewater management and disposal needs of a wide range of industrial operations. While the various industrial uses that have been made of deep-well injection in Texas might be expected to continue in the future, the most significant areas for new application of the technology of underground injection appear to be in providing geologic sequestration of CO2, and safe disposal of wastewater from desalination projects for public water supply. Also possible as a future application of deep-well injection for solidified wastes, salt cavern disposal wells may offer one of the most secure forms of geologic waste isolation available. While the Texas Commission on Environmental Quality strongly advocates the reduction in volume of generated waste, and the recycling and reuse of waste materials to the fullest practical extent, it is recognized that there will remain a need in the future for environmentally protective wastewater disposal for residual waste volumes. It is concluded, therefore, that deep-well injection of industrial and municipal waste will continue to play a very important supporting role in the state’s comprehensive program for pollution prevention.
REFERENCES Code of Federal Regulations (CFR), 1996. Title 40, Chapter 1, Environmental Protection Agency, Parts 144, 145, 146, 147, 148, 261, and 268. Knape, B.K., et al., 1984. Underground Injection Operations in Texas—A Classification and Assessment of Underground Injection Activities, Texas Department of Water Resources Report 291. Rish, W.A., Ijaz, T., and Long, T.F., 1998. A Probabilistic Risk Assessment of Class I Hazardous Waste Injection Wells. Draft, Chemical Manufacturers Association. Texas Administrative Code (TAC), 1996. Title 30, Chapter 331, Underground Injection Control Rules. Texas Water Code, 1981. Title 2, Chapter 27, Injection Wells. U.S. Code, 2003. Title 42, Chapter 6A, Public Health Service, §300h. U.S. Environmental Protection Agency, 1989. Office of Solid Waste and Emergency Response, Executive Summary and Overview, OSWER Comparative Risk Project, EPA/ 540/1-89/003. U.S. Environmental Protection Agency, 2001. Office of Water, Class I Underground Injection Control Program—Study of the Risks Associated with Class I Underground Injection Wells, EPA 816-R-01-007.
ELSE_TSANG_CH004.qxd
9/8/2005
4:46 PM
Page 29
Chapter 4
WHY CURRENT REGULATIONS PROTECT FLORIDA’S SUBSURFACE ENVIRONMENT A. Muniza, M. Tobonb, and F. Bloetscherc a
Hazen and Sawyer, P.C., Boca Raton, FL, USA City of Fort Lauderdale, Fort Lauderdale, FL, USA c Public Utility Management Planning Services, Inc., Hollywood, FL, USA b
4.1 INTRODUCTION Municipalities in Southeast Florida produce over 500 million gallons per day (mgd) of effluent and concentrate that must be disposed of in a cost-effective and environmentally sound manner. With limited options, utilities have selected ocean outfalls and deep-well injection as the most economical, environmentally sound, and reliable choice for managing effluent and concentrate. Hence, Class I Injection Wells have been used in Florida for over 30 years and have proven to be an excellent means of effluent and concentrate management. At the outset of the 20th century, waterborne disease was a persistent public health threat in the United States. There were many outbreaks of typhoid fever, cholera, dysentery, and other diseases that are rarely encountered today. The near elimination of these diseases has been attributed to improvements in wastewater and water treatment, and to an increased awareness of the need to separate potable water and wastewater to the extent practicable. In spite of a century of advances in sanitary engineering, safeguarding the public water supply remains as difficult today as ever, and promises to become more so. The future assuredly will include increasing stress on the nation’s water supplies. These stresses will stem from both quantity and quality concerns. According to a projection by the U.S. Bureau of the Census, the United States must find water for an additional 2.5 million people per year over the next 50 years (Fig. 4.1). That is equivalent to adding a city the size of the San Diego metropolitan area to the nation each year. Those additional people, and the agriculture and industry necessary to sustain them, will generate wastewaters, and the vast majority of the wastewaters generated will ultimately impact a water resource. The State of Florida must address these issues. The U.S. Bureau of the Census projects a Florida population increase of more than 5 million people over the next 25 years. Providing water for this population can be accomplished. By any measure, Florida is considered a water-rich area, receiving an average rainfall approaching 60 in/year. The challenge is to capture that water for the most effective use (Davis et al., 2001). The coincident challenge is to safely handle the wastewaters attendant to the existing and future populations.
ELSE_TSANG_CH004.qxd
30
9/8/2005
4:46 PM
Page 30
Why Current Regulations Protect Florida’s Subsurface Environment
Fig. 4.1. Population projection.
4.2 CURRENT RAW WATER SUPPLY Southeast Florida has historically relied on the highly prolific, sole source Biscayne Aquifer for its water supply. This aquifer is surficial, with a piezometric surface within a few feet of grade. Publicly owned water utilities, as well as state and local regulatory agencies, have recognized this resource’s vulnerability to contamination and have taken measures to ensure protection. Such measures include wellhead protection ordinances; elimination of inland surface-water wastewater discharges; restrictions on waste injection into groundwaters; septic tank replacement; and stringent underground storage tank regulations. The surficial aquifer system provides over 90% of all raw water for South Florida. Below the surficial aquifer system, which is a water table aquifer, lies a thick sequence of clays known as the Hawthorn Group that acts as a confining interval between the upper Floridan Aquifer System and the overlying water table aquifers. The Hawthorn Group is typically several hundred feet thick and serves as a barrier between the surficial aquifer freshwater and the brackish waters of the Floridan Aquifer. Figure 4.2 presents a generalized cross section of South Florida’s unique underground formations. The Floridan Aquifer System consists of both productive and confining strata (Meyer, 1989). The upper portion is productive and contains brackish waters. Permeability is moderate, and yields are typically increased by utilizing more than one horizon. This portion of the Floridan Aquifer System is used for water supply (i.e., Floridan blending wells) and aquifer storage and recovery, and as a monitoring zone for injection wells. Use of the upper Floridan Aquifer as a raw water supply source is slowly being developed as utilities exhaust the surficial aquifer. Costs for supply and treatment are significantly more for development of this deeper production zone. Monitor wells usually monitor the first productive interval
ELSE_TSANG_CH004.qxd
9/8/2005
4:46 PM
Page 31
4.3 Effluent Disposal Alternatives
31
Fig. 4.2. A generalized cross section of south Florida’s unique underground formations.
(upper monitor zone) above the potential underground source for drinking water (USDW), as defined by the 10,000 mg/L total dissolved solids isochlor. A lower monitor zone is also mandated to monitor the first permeable zone above the confining units. The most transmissive part of the Floridan Aquifer System is found near its base. A formation that contains huge boulders, known as the “Boulder Zone,” exists in South Florida. This formation contains water similar to seawater, and has been theorized to be connected to the ocean at similar depths. The Boulder Zone is very transmissive and accepts injected waters at low pressures (i.e., usually less than 60 psi). 4.3 EFFLUENT DISPOSAL ALTERNATIVES Do real options exist for disposal of treated effluent, and what safeguards are there to protect the state’s potential underground sources of drinking water? The State of Florida has three primary options for disposal of treated effluent in South Florida (Bloetscher et al., 2001; Englehardt et al., 2001). These options are: ● Ocean outfall ● Injection wells ● Reuse The elimination of inland surface-water wastewater discharges (and attendant impacts on the community water supply) was essentially accomplished in a decade of reengineering, from the mid-1970s to the mid-1980s. This advance in source separation was feasible due to two unique features of the Southeast Florida environment: (1) the existence of the Gulf Stream in the Atlantic Ocean, and (2) the “Boulder Zone” at a depth of approximately 3000 ft below land surface. Currently, disposal to the ocean and deep wells is about evenly distributed in southeast Florida (Fig 4.3).
ELSE_TSANG_CH004.qxd
32
9/8/2005
4:46 PM
Page 32
Why Current Regulations Protect Florida’s Subsurface Environment
Fig. 4.3. Treated effluent disposal methods in Southeast Florida (tricounty area).
Fig. 4.4. Gulf Stream.
The Gulf Stream (Fig. 4.4) is a fast-moving northerly current located within a few thousand feet offshore. The current runs parallel to the coast until reaching Cape Hatteras where it diverts to deeper water. The transport of the Gulf Stream off Florida’s coast has been estimated at over 1,000,000,000 cu. ft of water per second, with peak velocities near 6.5 ft/s. This transport volume is approximately 1700 times as large as the average discharge of the Mississippi River. Six open ocean outfalls discharge secondary effluent into the Gulf Stream at depths ranging from 89 to 107 ft. Approximately 290 mgd (annual average) is discharged through the outfalls. Extensive oceanographic studies directed by the National Oceanic and Atmospheric Administration (NOAA) have been conducted over a 10-year period beginning in 1989. These comprehensive investigations, entitled Southeast Florida Outfall Experiment (SEFLOE) Phase I and II, demonstrate that the open ocean outfall conditions are environmentally acceptable. The treated effluent is rapidly diluted in the zone of initial dilution; bioassay results clearly indicate that the effluent is not acutely toxic to the test marine organisms; and surveys indicate a balanced and richly diverse marine community present in the outfall dispersal area (Fergen et al., 1994). Perhaps most importantly, extensive study and measurement of plume dynamics indicate that a pathway to human interface is extremely remote.
ELSE_TSANG_CH004.qxd
9/8/2005
4:46 PM
Page 33
4.3 Effluent Disposal Alternatives
33
Figure 4.5 shows the existing ocean outfalls in south Florida. Presently, there are six outfalls that discharge into nearshore waters of the Atlantic Ocean. Treatment and dilution of fluids discharged have made this option the most cost-effective. However, ongoing monitoring continues to increase, and strict compliance to permit conditions is mandatory to avoid penalties by regulators. Cost of such monitoring is expensive and is anticipated to increase as regulations get stricter. Use of Class I injection wells provide a practical means of this disposal and could be viewed as a vertical ocean outfall. Injection wells are designed to dispose of treated effluent into formations that are separated from potential sources of drinking water. In addition, the receiving formations contain waters with qualities similar to seawater, have very high transmissivities that are unique to south Florida, and allow injection of large volumes at low pressures. Injection wells typically inject approximately 18 mgd into 24 in. diameter wells at pressures of approximately 50–60 psi. Class I injection wells are similarly protective of human health. The popularity of Class I wells for use in effluent disposal is shown by the fact that they are used in 19 states. It should be noted, however, that only Florida uses such wells for municipal (domestic wastewater) disposal purposes, due to favorable hydrogeologic conditions (EPA, 2001). In addition to being cost-effective, a major advantage to injection wells is that they can operate for as long as 24 hours, 7 days a week, which is critical to support other disposal options such as reuse. The major disadvantage of injection wells is the potential threat associated with vertical migration of injected fluids and subsequent contamination of potential USDW. The State of Florida has approximately 90 injection well sites. While most are for disposal of treated effluent, a growing number are used for disposal of concentrate from membrane water treatment plants. Injection wells that dispose of concentrate require an industrial design that consists of a tubing and packer in addition to the final cemented casing string. A final, commonly used disposal practice in south Florida is reuse. Reuse continues to be attractive, since it assists in completing the water cycle and replenishes the surficial aquifer system. This benefit is also a disadvantage, as some systems could become direct potable reuse, since the spray irrigation results in infiltrating directly into the drinking water
SCRWTDB (24 mgd) Outfall: 30-inch, 24 mgd West Palm Beach
Fort Lauderdale
Miami
Boca Raton WWTP (17.5 mgd) Outfall: 36-inch, 22 mgd BCOES NRWWTP (80 mgd) Outfall: 54-inch, 66 mgd Hollywood SRWWTP (42 mgd) Outfall: 60-inch, 42 mgd
Miami-Dade WASD CDWWTP (143 mgd) Outfall: 120-inch, 143 mgd
Fig. 4.5. South Florida outfalls.
Miami-Dade WASD NDWWTP (120 MGD) Outfall: 90-inch, 112 mgd
ELSE_TSANG_CH004.qxd
9/8/2005
34
4:46 PM
Page 34
Why Current Regulations Protect Florida’s Subsurface Environment
supplies. Irrigation also presents an issue during rainfall events, making backup alternatives or storage reservoirs mandatory to manage disposal during wet conditions. Injection wells are typically backup disposal methods for reuse water systems.
4.4 RISK ISSUES Deep-well injection waste migration risks are hard to quantify. There are two known wells that demonstrate some migration, but in neither case do the plumes rise to the top of the Floridan Aquifer, let alone penetrate the Hawthorn or Biscayne formations. These wells have existed for nearly 30 years, so the risk of surficial contamination appears minimal. Experts on the Miami team concurred that contamination of the surficial system was highly unlikely given the known drilling logs and ongoing experience with the Hawthorn zone. Since the plumes have remained in the Floridan Aquifer System, the Florida Water Environmental Association (FWEA) contracted to have a relative assessment, comparing deep wells with ocean outfalls and developed surface discharges. The relative risk analysis developed by the University of Miami research team used several tracer constituents, and compared the relative likelihood of those indicators showing up in quantities that exceeded the receiving or drinking water standards (depending on which was more applicable). To select appropriate constituents that could be used to measure health risks, the following were evaluated: ● Presence in wastewater ● Concentration higher in wastewater than ambient receiving waters ● Potential for health impacts Human health risk indicators that were selected for evaluation included arsenic, Nnitrosodimethylamine, and Crytposporidium or rotavirus. Total Kjeldahl nitrogen was used as an indicator of ecological risk. In general, the collected data did not indicate significant health concerns associated with the injection of treated effluent. Of the measured constituents with specific toxicity or infectivity, only antimony and total coliforms were higher in the effluent with respect to both ambient water and regulatory drinking water standards.
4.5 ENVIRONMENTAL REGULATIONS Underground injection programs are regulated under the Underground Injection Control (UIC) regulations (40 CFR 146) as Class I municipal wells. These regulations were established under the authority of the Safe Drinking Water Act approved in 1974, and amended in 1986 and 1996, setting forth standards for underground injection control programs. Florida received national primary enforcement responsibility for the UIC program for Class I, III, IV, and V wells on March 9, 1983; however, significant issues have required the continued involvement of the United States Environmental Protection Agency (EPA) in the underground injection program in Florida. Florida Chapter 62–528 F.A.C. governs underground injection. Florida regulations are similar to federal rules, with minor variances that are stricter than the federal criteria. After groundwater monitoring revealed migration of injected or native formation fluids into USDW, violating current Federal UIC regulations, EPA proposed changes to federal rules. Proposed changes would “allow for continued injection by existing Class I municipal wells that have caused or may cause such fluid movement
ELSE_TSANG_CH004.qxd
9/8/2005
4:46 PM
Page 35
4.6 Injection Well Construction and Testing
35
into USDWs in specific areas of Florida if certain requirements are met that provide adequate protection for underground sources of drinking water” (Federal Register, 2000). Federal rules define five classes of injection wells in Sec. 144.6. Class I wells are defined as wells that inject fluids beneath the lowermost formation containing, within one-quarter mile of the well bore, a USDW. Class I wells can be hazardous, industrial, or municipal waste disposal wells. Thus, injection wells used for disposal of treated municipal wastewater are regulated as Class I wells. Class II wells are those used to inject fluids that are brought to the surface in connection with oil and natural gas production, or to enhance the recovery of oil and natural gas, and the storage of hydrocarbons that are at liquid temperature. Class III wells are used for extracting minerals. Class IV wells are used for generators of hazardous radioactive waste that inject water below the lowermost drinking water zone. Class V wells are wells not included in Class I, II, III, or IV; they include air conditioning and cooling water return wells, drainage wells, dry wells for injecting wastes, recharge wells for replenishing water wells, saltwater intrusion barrier wells, wells to inject water into freshwater aquifer wells, wells to inject mixtures of water and sand, aquifer storage and recovery wells, sand backfill wells, and septic system wells. Federal regulations contain formulas and descriptions of test methods for determining well operations; corrective actions in cases of well failure; and requirements for mechanical integrity tests to ensure detection of leaks in the casing, tubing, or packer (when used), and to ensure that there is no significant fluid movement into an underground source of drinking water through vertical channels adjacent to the well. Subparts B, C, D, and F set up construction, operating, monitoring, and reporting requirements; and information to be considered in permitting wells. This may include information on the proposed operation of the well (such as maximum daily rate of flow and volume of fluids to be injected in the average injection pressure), the source of the water, analysis of the characteristics of the injected fluids, appropriate geological data, and the construction details of the well. The regulations require that an applicant include a certificate that the applicant has assured, through performance bond or other appropriate means, that the permittee has the resources necessary to close, plug, and abandon the well as required by the federal regulations.
4.6 INJECTION WELL CONSTRUCTION AND TESTING The depth for the injection horizon varies, but requires one or more confining units that separate the receiving formation from potential potable water supplies. Because of the geological characteristics of the formation, injection zones are typically limited to southern Florida for the most part (i.e., south of the Tampa/Daytona Beach latitude). Secondary treatment is required in Florida prior to deep-well injection, and injectate may be chlorinated as well. Without chlorination, there is the potential for enhanced microbial growth and fouling in the wells and surrounding formation, resulting in long-term damage to well casings. Design and testing of injection wells has advanced in recent years with the creation of more sophisticated drilling and testing equipment (Geraghty and Miller Inc., 1984). Procedures such as cementing have reached new heights and provide added assurances that wells are fully cemented. In addition, geophysical logging techniques have been developed to monitor and confirm underground conditions and construction methods. For example, cement bond logs are refined to the point whereby they can easily detect bonding
ELSE_TSANG_CH004.qxd
36
9/8/2005
4:46 PM
Page 36
Why Current Regulations Protect Florida’s Subsurface Environment
around casings to ensure adequate cement seals. Radioactive tracer tests conducted as part of mechanical integrity demonstrations are also used to measure possible vertical migration around the final casing string. Two typical construction designs are presently employed for construction of Class 1 injection wells in Florida. Figures 4.6 and 4.7 depict the features associated with municipal and industrial designs. Figure 4.6 shows the typical design for disposal of municipal wastewater; this design uses one final casing string that is fully cemented from the bottom up to the land surface.
Fig. 4.6. Typical municipal design.
ELSE_TSANG_CH004.qxd
9/8/2005
4:46 PM
Page 37
4.6 Injection Well Construction and Testing
37
Industrial-design wells are similar to municipally designed wells, except they contain one additional casing string known as the tubing inside the final casing string. The annular space between the final casing string and the tubing can be fluid-filled with a corrosion inhibitor, or cemented. A fluid-filled annulus is pressurized to approximately 10 psi higher
Fig. 4.7. Typical industrial design.
ELSE_TSANG_CH004.qxd
9/8/2005
38
4:46 PM
Page 38
Why Current Regulations Protect Florida’s Subsurface Environment
than the anticipated pressure needed to maintain a positive head for added environmental protection.
4.7 SUMMARY Limited options exist for disposal of treated effluent and concentrate from membrane water treatment plants. Of the three viable options—ocean outfalls, injection wells, and reuse—injection wells appear to be the most favored by utilities. They provide a reliable, environmentally sound, and cost-effective disposal method, when compared to other options and risks. Design, construction, and testing of Class I injection well systems continue to get stricter to ensure environmentally sound effluent disposal practices. Protection of water quality remains paramount to regulators and engineers alike. Monitoring systems, including multizone monitor wells, are used daily to ensure injection systems are operating safely. The EPA delegated primacy of the UIC program to Florida, as they have done in many states. The Florida Department of Environmental Protection carefully manages this program via its strict permitting program. Routine inspection of systems is accomplished through demonstration of mechanical integrity to confirm the performance of the entire injection well system. With the anticipated growth and associated wastewater production, use of safe, reliable, and cost-effective effluent management practices that include Class I injection wells will be key in sustaining South Florida’s unique ecosystem.
REFERENCES Bloetscher, F., Englehardt, J.D., Amy, V.P., Chin, D.A., Solo-Gabriele, H., Fleming, L.E., Rose, J.B., Gokgoz, S., Tchobanoglous, G., 2001. Comparative assessment of human and ecological impacts from municipal wastewater disposal methods in Southeast Florida: Deep wells, ocean outfalls and canal discharges. In: Groundwater Protection Council Annual Conference Proceedings—Reno NV, GWPC, Oklahoma City, OK. Davis, P.A., Hui, A.M., Brant, W.M., Dernlan, G.D., 2001. The utility dilemma or recognizing interdependencies with water resource management strategies. Florida Section American Water Works Association 2001 Conference Proceedings, Kissimmee, FL. Englehardt, J.D., Amy, V.P., Bloetscher, F., Chin, D.A., Fleming, L.E., Gokgoz, S., SoloGabriele, H., Rose, J.B., and Tchobanoglous, G., 2001. Comparative Assessment of Human and Ecological Impacts for Municipal Wastewater Disposal Methods In Southeast Florida: Deep Wells, Ocean Outfalls, and Canal Discharges, University of Miami, Coral Gables, FL. EPA, Class I Underground Injection Control Program: Study of the Risks Associated with Class I Underground Injection Wells. EPA 815-R-01-007, March 2001. Fergen, R., Cooke, J.P. and Huang, H., 1994. An overview of the Southeast Florida Ocean Outfall Experiment (SEFLOE). Marine Technology Society 94 Advanced Program: Challenges and Opportunities in the Marine Environment. September 7–9, 1994. Geraghty and Miller, Inc., 1984. Construction and Testing of Disposal Wells 1, 2 and 3 at the George T. Lohmeyer Plant, Fort Lauderdale, FL, Consultant’s Report. Meyer, F.W., 1989. Hydrogeology, Ground-Water Movement and Subsurface Storage in the Floridan Aquifer System in Southern Florida, Regional Aquifer-System Analysis— Floridan Aquifer System. United States Geological Survey Professional Paper 1403-G.
ELSE_TSANG_CH005.qxd
9/10/2005
11:16 AM
Page 39
Chapter 5
AN INTERPRETATION OF THE SAFE DRINKING WATER ACT’S “NON-ENDANGERMENT” STANDARD FOR THE UNDERGROUND INJECTION CONTROL (UIC) PROGRAM B.J. Kobelski, R.E. Smith, and A.L. Whitehurst United States Environmental Protection Agency, Office of Ground Water and Drinking Water, Washington, DC, USA
5.1 INTRODUCTION The Safe Drinking Water Act (SDWA) established the Underground Injection Control (UIC) program to protect underground sources of drinking water (USDWs1) from contamination from the injection of fluids. Two central aspects of that protection are: (1) the definition of the resource to be protected, and (2) the level of protection to be provided. The United States Environmental Protection Agency (EPA) and the states have addressed those issues in a variety of ways to reflect the variable risks, local or regional underground hydrogeology, economic factors, and other public health considerations. This chapter is an attempt to describe the methodology for varying approaches and considerations.
5.2 STATUTORY AND REGULATORY AUTHORITY Section 1421(d)(2) of the SDWA defines the term “endangerment” as follows: Underground injection endangers drinking water sources if such injection may result in the presence in underground water which supplies or can reasonably be expected to supply any public water system of any contaminant, and if the presence of such contaminant may result in such system’s not complying with any national primary drinking water regulation or may otherwise adversely affect the health of persons. The UIC regulations, as developed to implement the requirements to protect USDWs (Part C of the SDWA) clarify the statutory requirements further, and read at 40 Code of Federal Regulations Section 144.12(a) as follows: No owner or operator shall construct, operate, maintain, convert, plug, abandon, or conduct any other injection activity in a manner that allows the movement of fluid 1 An USDW is an aquifer or a portion of an aquifer that supplies any public water system or contains a sufficient quantity of groundwater to supply a public water system; currently supplies drinking water for human consumption or contains fewer than 10,000 milligrams per liter (mg/l) total dissolved solids; and is not an exempted aquifer (i.e., exempted from UIC regulations).
ELSE_TSANG_CH005.qxd
40
9/10/2005
11:16 AM
Page 40
An Interpretation of the Safe Drinking Water Act’s “Non-Endangerment” Standard
containing any contaminant into underground sources of drinking water, if the presence of that contaminant may cause a violation of any primary drinking water regulation under 40 CFR Part 142 or may otherwise adversely affect the health of persons. The regulations at 40 CFR §144.12(b) go on to explicitly direct that UIC programs prevent movement of contaminants into USDWs from Class I, II, and III injection wells: . . . if any water quality monitoring of an underground source of drinking water indicates the movement of any contaminant into the underground source of drinking water, except as authorized under part 146, the Director shall prescribe such additional requirements for construction, corrective action, operation, monitoring, or reporting (including closure of the injection well) as are necessary to prevent such movement. Class V wells are presumed to cause fluid movement into USDWs. Therefore, the regulations are quite different and read as follows: . . . if at any time the Director learns that a Class V well may cause a violation of primary drinking water regulations under 40 CFR part 142, he or she shall: (1) require the injector to obtain an individual permit; (2) order the injector to take such actions (including, where required, closure of the injection well) as may be necessary to prevent the violation . . .; or (3) take enforcement action. [40 CFR §144.12(c)] and Whenever the Director learns that a Class V well may be otherwise adversely affecting the health of persons, he or she may prescribe such actions as may be necessary to prevent the adverse effect. [40 CFR §144.12(d)].
5.3 APPROACHES FOR PREVENTING ENDANGERMENT To determine the most appropriate course of action for preventing endangerment of USDWs from the threats posed by all injection practices, the EPA evaluated key considerations within the framework of the SDWA. These considerations included: ● Public health. The result or likelihood of human exposure to contaminants from certain practices. ● Programmatic and practical. The costs and benefits of alternatives to the injection practice and the difficulty in monitoring different practices’ impact on public health and USDWs. ● Legal. The legal basis for EPA actions and remedies for protecting human health and USDWs. After evaluating these factors in light of the varying risk posed by the full range of injection activities, the EPA believes that all permissible injection practices should fall within five categories labeled A through E, as illustrated in Figure 5.1. Three additional practices labeled F through H are not believed to be protective of public health, because there is no assurance that such practices would prevent endangerment of USDWs.
ELSE_TSANG_CH005.qxd
9/10/2005
11:16 AM
Page 41
5.3 Approaches for Preventing Endangerment
41
Fig. 5.1. Range of injection practices.
A general description of each possible injection approach is as follows: A. Prohibit injection. Establish a specific regulatory (or statutory) prohibition on a class of well or within-class injection practice, thereby achieving the highest level of certainty for protection of public health and USDWs. B. Isolate injection. Achieve protection by careful sitting that provides confinement between the injection zone and USDWs, and by establishing specific construction, operation, testing and monitoring, maintenance, and closure standards. C. Meet DW standards at point of injection. Through pre-injection treatment and a carefully tailored monitoring program, maintain a specific waste stream quality prior to injection at the wellhead. D. Meet DW standards at edge of USDWs. By a combination of setting specific limits on injectate quality and through an accurate monitoring or modeling program, meet limits (e.g., maximum contaminant levels (MCLs) or other health-based levels) at the top (or base) of the USDW to prevent endangerment. E. Meet DW standards in USDW. Achieve protection of public health and USDWs based on knowledge of injectate quality—determined by accurate monitoring or modeling of the “point of compliance” some distance from the point of injection—and locational standards to ensure drinking water is not withdrawn from within those “zones of impact.” F. Meet standards at usable resource. Establish standards or a monitoring program that will only protect waters of less than 3000 ppm total dissolved solids. G. Meet DW standards at water well. Allow contamination of USDWs to occur, but monitor source water quality and link any contaminants to injection well sources. H. Meet DW standards at the tap. Allow contamination of USDWs to occur and monitor drinking water after treatment, linking contamination to injection well sources and requiring control of injectate quality only where it cannot be treated at public water systems.
ELSE_TSANG_CH005.qxd
42
9/10/2005
11:16 AM
Page 42
An Interpretation of the Safe Drinking Water Act’s “Non-Endangerment” Standard
These UIC approaches for implementing the “non-endangerment” standard are indicated in Figure 5.2, and each set of approaches are grouped to indicate the level of certainty that the approach will be effective in implementing the UIC program’s goal of protecting human health and preventing endangerment of USDWs. From Figure 5.2, one can determine that Approaches A, B, and C are the most protective of public health and USDWs; are consistent with the mandate of the SDWA; and are appropriate for higher-risk injection practices. These approaches are easiest to ensure compliance, but they are, in general, more costly approaches. Approaches D and E can be equally protective of public health and USDWs, provided that these practices are closely monitored. These approaches are more appropriate for lower-risk injection practices, but are more challenging to ensure compliance. Typically, these approaches are less costly, but still may require significant resources at the program level to be effective. All other approaches for preventing endangerment have been suggested by various proponents, but as in Approaches F, G, and H, they appear not to fall within the statutory framework of the SDWA and are believed to be unacceptable injection well practices. Besides being inconsistent with the SDWA, these practices are exceedingly difficult to assure compliance, and there is a high degree of uncertainty over whether public health protection is being achieved.
5.4 CONCLUSIONS The authors believe that injection well approaches to preventing endangerment of USDWs generally fall within five different categories. All of these approaches take into UIC - Approaches for Preventing Endangerment of USDWs Note: "DW Stds." = Drinking Water Strandards D. Meet DW Stds. at Usable Resource
A. Prohibit Injection
B. Isolate Injectate
D. Meet DW Stds. at Edge of USDW
No Fluid Movement
C. Meet DW Stds. at Injection
D. Meet DW Stds. at Well Intake
E. Meet DW Stds. in USDW E. Meet DW Stds. at Tap
Less Protection Certainty
More Protection Certainty
Meets SDWA Mandate
Meets SDWA Mandate
Appropriate For Higher Risk
Appropriate For Lower Risk
Inconsistent with SDWA Mandate
Generally Higher Cost
Generally Lower Cost
Easier to Assure Compliance
Difficult to Assure Compliance
Fig. 5.2. UIC Approaches for preventing endangerment of USDWs.
ELSE_TSANG_CH005.qxd
9/10/2005
11:16 AM
Page 43
References
43
account specific factors and considerations in how the appropriate course of action is taken by the UIC program. Some of these approaches have a low assurance of preventing endangerment of USDWs and the public health. The approaches in question do not appear to fall within the statutory framework of the Safe Drinking Water Act; therefore, they are not acceptable injection well practices.
ACKNOWLEDGMENTS AND DISCLAIMER The authors wish to thank their colleagues in the EPA Regional Offices for their comments on the graphical representations and recommendations on portions of the language specific to non-endangerment. The opinions of the authors are based on their technical expertise and experience in the federal regulation of underground injection wells and do not necessarily reflect the official policies of the EPA, nor has this chapter received formal EPA peer review.
REFERENCES United States Congress, 1996. Safe Drinking Water Act Amendments of 1996. Pub. L. Nos. 104–182, 110 Stat. 1613, 1996. U.S. Environmental Protection Agency, 1980. Statement of Basis and Purpose: Underground Injection Control Regulations. Office of Drinking Water (unpublished document), May 1980. U.S. Environmental Protection Agency, 2002. Title 40, Code of Federal Regulations, Parts 144 to 148, 2002. Protection of Environment, July 1, 2002.
This page intentionally left blank
44
ELSE_TSANG_CH006.qxd
8/19/2005
6:29 PM
Page 45
Chapter 6
THE APPROPRIATE METHODOLOGY FOR DETERMINING THE USE OF A FIXED-RADIUS AREA OF REVIEW OR ZONE OF ENDANGERING INFLUENCE, WHEN CONDUCTING AN AREA-OF-REVIEW ANALYSIS FOR UNDERGROUND INJECTION CONTROL OPERATIONS S. Stephen Platt and D. Rectenwald Safe Drinking Water Act Branch, U.S. Environmental Protection Agency, Philadelphia, PA, USA
6.1 INTRODUCTION Underground Injection Control (UIC) regulations promulgated under the Safe Drinking Water Act (SDWA) in 1980, and amended in 1983, provide a number of protective standards to prevent endangerment of underground sources of drinking water (USDWs) from the subsurface emplacement of fluids. USDWs are defined as aquifers or portions of aquifers that supply any public water system, or contain a sufficient quantity of groundwater to supply a public water system, and contain fewer than 10,000 mg/L total dissolved solids. UIC Regulations, 40 CFR, Section 144.12, as well as Sections 1421(b)(1)(A)–(D) and 1425 of the SDWA, mandate the protection of USDWs from underground injection operations. Section 144.12 provides the protective standards that EPA must follow when directly implementing a UIC Program, whereby Sections 1421(b)(1)(A)–(D) and 1425 provide the protective standards that States must follow when implementing the Class II portion of the UIC Program (e.g., the injection of fluids related to the production of oil and gas). Although the language between these regulatory provisions varies slightly, their common theme is the prevention of underground injection that could endanger drinking water sources. One of the protective standards designed to protect USDWs is the requirement associated with determining the area of review (AOR) around an injection well or injection well field. During injection operations, significant pressure buildup often occurs in the injection zone. This pressure buildup can extend a considerable distance from the injection well, resulting in the lateral movement of the injection and formation fluids, and the potential for the vertical movement of fluids. The AOR is conducted to prevent injection or formation fluids from migrating out of the injection zone and upwards into a USDW. The regulations permit an operator to choose between a specified fixed-radius AOR (generally one-quarter mile for Class II oil- and gas-related injection wells) or a calculated zone of endangering influence (ZEI). Both the fixed-radius AOR and the ZEI calculation require the operator, as part of the permitting process, to take into account any improperly abandoned or unplugged wells, inactive wells, fractures, faults, etc., within the AOR that penetrate the injection zone. Fluid migration can occur into USDWs, through these open conduits during the operation of injection wells, if proper corrective action is not taken. Although using a fixed-radius AOR or calculated ZEI are both acceptable under the UIC regulations, under certain circumstances, one methodology should take precedence over the other.
ELSE_TSANG_CH006.qxd
46
8/19/2005
6:29 PM
Page 46
The Appropriate Methodology for Determining the Use of a Fixed-Radius Area of Review
6.2 METHODOLOGIES In theory, both a fixed-radius AOR and a calculated ZEI provide useful methodologies under the present regulatory framework to protect USDWs. However, history has shown that problems can arise when the fixed-radius AOR is used instead of a calculated ZEI under certain circumstances. Figure 6.1 provides an example of when a ZEI calculation should take precedence over the use of a fixed-radius AOR. The figure shows how the ZEI could extend past a one-quarter mile fixed-radius area of review—this is the region where injection pressure could force fluid out of the injection zone and into a USDW. In Figure 6.1, the ZEI, represented by the injection pressure curve, equates to the pressure necessary to cause a column of fluid to rise a certain distance above the injection zone should an open conduit (i.e., an unplugged well) exist. In Figure 6.1, the ZEI extends some distance beyond the one-quarter mile fixed radius (between well number 2 and well number 3), and could allow fluid migration into the lowermost USDW through an open conduit. The selection of a fixed-radius AOR or a calculated ZEI depends on a number of factors. However, as stated above, the intent of the UIC regulations is to prevent any fluid migration into USDWs. Therefore, when deciding whether to use one or the other methodology, the determining factor should be based on which one of the two will provide the necessary protection to USDWs. Every operator is responsible for determining whether corrective action will be necessary in their injection well field, regardless of whether a fixed-radius AOR or a ZEI calculation is used. If an operator intends to develop or expand their operation in an area that has seen extensive oil or gas development, it would make sense for that operator to consider using a calculated ZEI. In this situation, a number of wells could potentially require corrective
Fig. 6.1. Example of fixed-radius AOR versus calculated ZEI.
ELSE_TSANG_CH006.qxd
8/19/2005
6.2 Methodologies
6:29 PM
Page 47
47
action. By calculating a ZEI, the operator will be able to determine the number of wells that require corrective action under the proposed operating conditions. The ZEI calculation could achieve one of two things: (1) it could identify key wells outside a fixed one-quarter mile AOR that could act as conduits for fluid migration and thereby be subject to corrective action requirements, or (2) it could, just as easily, indicate that the injection operation does not present a problem to wells some distance inside a fixed one-quarter mile radius, thereby reducing the area to be reviewed for corrective action. Conversely, if an operator intends to operate in an area with little past drilling history, a fixed one-quarter mile AOR may be satisfactory. In this situation, because few if any wells exist within the AOR, corrective action requirements could be minimal. A ZEI calculation may not always prove feasible even in areas where a significant amount of oil and gas development has occurred. In some instances, lack of geologic information may hinder the calculation of a ZEI. Such a calculation requires knowledge of factors such as the porosity, permeability and thickness of the injection formation, reservoir pressure, injection formation compressibility, injection rate, specific gravity of the injection fluid, time of injection, and length of injection. ZEI calculations are also based on the assumptions that the injection zone is homogeneous, isotropic, of sufficient extent, and exhibits radial flow in all directions. This may not always be the case in actual subsurface conditions. Under circumstances like this, the use of a fixed-radius AOR may prove more advantageous, as long as appropriate corrective action is performed. As stated, both of these methodologies can be useful strategies in preventing the migration of fluid out of an injection zone and into USDWs during injection well operations. However, uncertainty can still exist with the use of either methodology. So what can an operator do, in situations where the use of a fixed-radius AOR or a calculated ZEI cannot entirely eliminate the possibility of fluid migration out of the injection zone and into USDWs? It is recommended that the operator incorporate the use of monitoring wells into the AOR or ZEI methodology. By measuring fluid levels in these monitoring wells, the operator can obtain a continuous record of how injection affects pressure in the injection zone and how that pressure extends laterally outward away from the injection well. Typically, wells that exist within the field of operation and penetrate the injection zone can be utilized for monitoring purposes. Older unplugged wells or production wells within the field are common choices for monitoring fluid levels. If necessary, monitoring wells can be drilled during project development. It is of practical interest in this chapter to examine a selected example of injection and associated pressure response, and compare that response with a theoretical response that would be predicted on the basis of a ZEI model calculation. The example is based on a UIC Class II permit that EPA Region 3 issued for an enhanced recovery development in Taylorstown, Pennsylvania (Fig. 6.2). Both reservoir data and monitoring data from injection and monitoring wells were available, allowing for easy computer simulation and comparison with the actual data. The UIC permit required fluid-level monitoring when the location of old production wells, shown on old lease maps, could not be substantiated through plugging records, newer lease maps, or field survey. The permit requires monitoring wells to be located between the injection wells and the suspected locations of potential unplugged wells, and that the fluid level be monitored to ensure protection of USDWs. Project injection began prior to production, so the fluid-level monitoring provided a continuous record of the injection formation’s response prior to production. Additionally, a ZEI calculation was conducted to project what the fluid level response would be over time.
ELSE_TSANG_CH006.qxd
48
8/19/2005
6:29 PM
Page 48
The Appropriate Methodology for Determining the Use of a Fixed-Radius Area of Review
Fig. 6.2. Location map for Taylorstown, PA, enhanced recovery project injection wells, Noble 2 monitoring well, and hypothetical well X, West Middletown, PA, USGS.
The Gordon Sandstone is found at an average depth of 2500 ft. Depending on topography and subsurface relief, reservoir depth varies from 2313 to 2719 ft. The Gordon Sand is Upper Devonian in age and of shallow marine origin. Average gross thickness of the sand as computed from logs, cores, and drillers logs is 11 ft, with average porosity of 19% and average permeability of 100 millidarcies. After 9 months of injection, several of the monitoring wells exhibited static fluid levels above the USDW, which is located at a depth of approximately 500 ft. Table 6.1 provides the data used for the calculation of the ZEI for one of the monitoring wells (Noble 2) after only 286 days of injection. The results of this calculation show the effects of injection from three injection wells located less than one-half mile from Noble 2. Totaling the pressure influence at the monitoring well location from the three injection wells, the reservoir pressure after 286 days equaled 937 psi. This pressure equated to a fluid column of 2158 ft. The top of the Gordon Sand in the monitoring well is at a depth of 2330 ft, resulting in a calculated fluid level of 178 ft below land surface. The observed fluid level measured in the monitoring well was 125 ft below land surface. Reasonable agreement was achieved between the predicted and observed results during the early phase of the injection operation, lending confidence that long-term pressure-buildup projections would apply at other project locations. Using the same model and parameters listed in Table 6.1, the reservoir pressure was calculated at a hypothetical point (X) located equidistant, one-quarter mile, from each of the same three injection wells. Table 6.2 lists the results of that calculation. The resulting reservoir pressure at hypothetical well X would equate to a fluid column of 1963 ft. Once again, using an estimated injection formation top of 2363 ft, the depth to fluid
ELSE_TSANG_CH006.qxd
8/19/2005
6:29 PM
Page 49
6.3 Conclusion
49
Table 6.1. ZEI calculation for Nobel 2 monitoring well, Taylorstown, PA, project Parameter
011
033
034
Initial pressure Injection rate Viscosity Specific gravity Formation volume factor Permeability Reservoir thickness Compressibility Porosity Distance to monitoring well Calculated reservoir pressure at monitoring well
100 psi 590 STB/D 1 1 1 100 md 12 ft 0.0000032 psi⫺1 0.19 745 ft 346 psi
100 psi 255.6 STB/D 1 1 1 100 md 12 ft 0.0000032 psi⫺1 0.19 1834 ft 180 psi
100 psi 731.5 STB/D 1 1 1 100 md 12 ft 0.0000032 psi⫺1 0.19 701 ft 411 psi
Table 6.2. Hypothetical calculation of pressure influence at one-quarter mile Well number
Distance to hypothetical monitoring well (ft)
Calculated reservoir pressure at hypothetical monitoring well (X) (psi)
011 033 034
1320 1320 1320
306 189 356
Total
851
from ground surface in well X would be 400 ft. This would imply that the fluid level, after only 286 days of injection, would be into or above the lowermost USDW, outside the fixed radius of one-quarter mile, if an unplugged or improperly abandoned well existed in this location.
6.3 CONCLUSION This injection operation in the Gordon Sandstone provides a good example of why a fixed radius of one-quarter mile cannot always be assumed to be protective of USDWs. Less than one year after the start of the injection operation, the potential for fluid migration into USDWs from the injection developed. Additionally, if fluid-level monitoring at this project was unavailable, then injection would have continued at unacceptable rates, and the zone of endangering influence would have continued to grow. Monitoring enabled injection rates to be reduced and effectively controlled the injection operation, preventing endangerment to USDWs. The use of a fixed-radius AOR or a ZEI calculation depends on a number of physical and operating factors. Operators as well as regulators need to ensure that the methodology selected provides the greatest protection of USDWs. Therefore, some type of monitoring should be incorporated into the permit conditions. Continuous or, minimally, periodic monitoring of the injection reservoir fluid level is the only sure way of determining the potential for endangerment of USDWs.
This page intentionally left blank
50
ELSE_TSANG_CH007.qxd
8/22/2005
7:07 AM
Page 51
Chapter 7
ANALYSIS OF INJECTATE LOCATION AT DUPONT BEAUMONT WORKS J.W. Mercera, C.R. Fausta, C. Brownb, and J.E. Clarkc a
GeoTrans, Inc., Sterling, VA, USA South Florida Water Management District, West Palm Beach, FL, USA c E.I. du Pont de Nemours & Co., Inc., Beaumont, TX, USA b
7.1 INTRODUCTION DuPont Beaumont Works (Fig. 7.1) is located adjacent to the Neches River, about 5 miles south of Beaumont, Texas. The plant site encompasses approximately 728 acres. Three wells, located in the north-central portion of the plant property, have been used for the injection into deep strata (Fig. 7.2) below the Beaumont Works facility. Injection Well No. 1 has been used for injection, first, into the Catahoula sand, later into the upper Oakville sand, and then (and currently) into the lower Oakville sand since it was completed in 1972. Injection Well 2, also completed in 1972, was first used for injection into the Catahoula sand and later (and currently) for injection into the lower Oakville sand. Injection Well 3 was completed in 1987 and has only been used for injection into the Frio sand. Beaumont Works manufactures acrylonitrile, aniline, and other specialty chemicals. The wastewater streams that are injected are composed of approximately 95% water. The remainder is primarily dissolved salts with traces of organic chemicals. The wastewater stream from the acrylonitrile plant is currently injected into Injection Wells 1 and 2. The wastewater stream from the aniline system is currently injected into Injection Well 3. As of the end of 1999, a total of approximately 6648 million gallons (mgal) of injectate had been injected into the three wells at the site (approximately 2252 mgal into Injection Well 1, approximately 3657 mgal into Injection Well 2, and approximately 740 mgal into Injection Well 3). The lower Oakville sand has received the most fluid (approximately 3666 mgal, starting in 1980). Injection into the Catahoula sand from 1972 to 1982 totaled approximately 1370 mgal, and injection into the upper Oakville sand from 1982 to 1991 totaled approximately 872 mgal. Total injection into the Frio sand is approximately 740 mgal, starting in 1987. The analysis presented in this paper focuses on the Oakville sands. The SWIFT code (Reeves et al., 1986) is used to determine the present-day (beginning of year 2000) extent of injectate in these sands. DuPont (2000) has performed extensive modeling of the Oakville sands, providing calibration and predictive simulations. The modeling performed here takes advantage of the existing knowledge base developed as part of the regulatory modeling. However, in the regulatory no-migration demonstration modeling performed by DuPont, conservative data and assumptions were used to show that the deep wells are safe and protective of human health and the environment, even under worst-case assumptions. The modeling performed in this analysis is based primarily on the most likely data and assumptions to determine where the injectate is actually located. Some conservative assumptions were made in this analysis.
52
8/22/2005
7:07 AM
Page 52
Analysis of Injectate Location at Dupont Beaumont Works
Fig. 7.1. Beaumont Works.
ELSE_TSANG_CH007.qxd
ELSE_TSANG_CH007.qxd
8/22/2005
7:07 AM
Page 53
7.1 Introduction
53
Fig. 7.2. General site stratigraphy (Dupont, 2000).
An important issue in defining the injectate plume is the outermost contour. The worstcase regulatory approach has determined that the injectate plume be defined using the 10−6 contour, where health-based limits (HBLs) for drinking water are used. Yet the brine into which the wastewater is injected cannot be used for drinking water, and this worst-case approach is not appropriate for this application. Alternatively, three different methods are used to define a realistic injectate plume: (1) mass-in-place calculations, (2) an HBL
ELSE_TSANG_CH007.qxd
8/22/2005
54
7:07 AM
Page 54
Analysis of Injectate Location at Dupont Beaumont Works
approach adjusted for brine, and (3) use of detection limits. Injectate modeling was performed for the Oakville sand, and plume contours were determined. This modeling was based on recent seismic data, as well as other information, that aided in characterization of nearby faults. The three independent approaches lead to the same conclusion that a realistic injectate plume definition in a brine environment is achieved using the 10−4 contour. If chemical fate (degradation and sorption) is considered, the contour will be greater (10−3 or higher).
7.2 GEOLOGY The following discussion is summarized from DuPont (2000). DuPont Beaumont Works is located in the Gulf Coast geosyncline, a thick wedge of interlayered sands and shales/clays1 (confining beds). Approximately 2 miles west of the centroid of the injection wells is the Spindletop Dome. The confining beds are composed of clay. The general stratigraphy including the injection sands is shown in Figure 7.2. In descending order of depth, the injection sands historically used at the site are upper Oakville, lower Oakville, Catahoula, and Frio sands. The top of the upper Oakville sand is at a depth of about 4200 ft, approximately 3200 ft below the base of the underground sources of drinking water (USDWs), which is less than 1000 ft deep. Between the base of the USDW and the injection sands are over 2000 ft of clay confining beds, including the Lagarto shale and buffer aquifers. The vertical distance between the Oakville and the deeper Frio sands is more than 3000 ft, which includes additional confining clays (e.g., Anahuac shale). The Oakville sand is continuous laterally across the area and has sufficient permeability, porosity, and thickness to accept the injectate and to prevent migration of fluids into USDWs. The confining zone for injection into the Beaumont Works injection wells is the interval located approximately 3200–3800 ft subsurface. This interval includes part of the Lagarto Formation shale, which is also continuous laterally. The confining zone is separated from the base of the lowermost USDW by multiple saline aquifers occurring in the interval from 1000 to 3200 ft subsurface. These saline aquifers are buffers and offer additional protection against potential vertical migration of injectate by providing a bleed-off area for any pressure or upward flowing fluids that would ever reach them. That is, the buffer aquifers are dominated by lateral flow that would help divert vertical migration. The clay strata between injection sands provide an effective barrier to vertical migration between injection intervals. The upper and lower Oakville sands are separated by a 10 ft clay based on electric logs. In 1987, during the drilling of Injection Well 3, fluid samples were collected using Schlumberger’s Repeat Formation Tester (RFT) from both the upper Oakville sand (4085 ft depth) and the lower Oakville sands (4133 ft depth). Injectate reaction products were recovered from the lower Oakville sands, whereas no injectate reaction products were recovered from the upper Oakville sand, indicating that the 10 ft clay at this location is an effective barrier to upward migration. The confining zone for Beaumont Works Injection Wells 1 and 2 is 600 ft thick, providing a thick barrier to upward migration. As indicated, additional clays are present above the confining zone, providing further protection. The injection site is located between two large salt domes: Spindletop Dome to the west and Port Neches Dome to the east. The salt at the top of Spindletop Dome is approximately 1000 ft deep, whereas the Port Neches Dome is even deeper (approximately 7000 ft deep).
1
The fine-grained layers are clays, but, locally, are referred to as shales. Both terms are used in this chapter.
ELSE_TSANG_CH007.qxd
8/22/2005
7.3 Analysis
7:07 AM
Page 55
55
The Spindletop Dome is a barrier to flow. A radial fault (Fault A) originating from Spindletop Dome passes through the subsurface in an east–west trending direction, and passes near the north side of the injection site at the Oakville injection sand level, where approximately 850 ft of offset occurs at the injection level. The same fault passes near the south side of the injection site at the Frio and Catahoula levels. The up-thrown side of Fault A is north, i.e., it is a down-to-the-south fault. Fault H, also originating at Spindletop Dome, is a minor down-to-the-north fault, with approximately 70 ft of vertical displacement in the Oakville sand (both upper and lower) that formed as an antithetic fault to the larger regional Fault A. The location and extent of Fault H have been determined by recent seismic data. In terms of the potential for vertical fluid migration, Fault A has been determined to be nontransmissive, i.e., the fault does not transmit pressure or fluid vertically. The same lack of injectate reaction products in the upper Oakville sand when Injection Well 3 was drilled is consistent with the lack of communication between the two Oakville sands and the nontransmissive nature of Fault A, Fault H, or other local faults. The impact of Fault A on horizontal fluid flow depends upon the juxtaposition of sediments across the fault. Fault A is a sealing fault (a no-flow horizontal boundary with a noncommunicating nature) in the Catahoula and Frio sands, based on the contact of sand abutting against clay and/or the available clay for clay smear, DuPont model calibration, and reservoir test results. This sealing nature in the Catahoula sand was further supported during the 1987 drilling of Injection Well 3, when a fluid sample was collected from the Catahoula sand (at a 4712 ft depth). No injectate reaction products were recovered from the Catahoula sand. Injectate migration to the south, in Frio and Catahoula sands, is limited by Fault A, which, as indicated, is a sealing fault at the depths of these sands. Therefore, injectate placed into the Frio sand through Injection Well 3 is limited to migration north of Fault A. In addition, migration of injectate in the Catahoula sand is limited also to the north of Fault A. The throw on Fault A at the Oakville level is approximately 850 ft. This amount of throw juxtaposes the lower Oakville sands against Catahoula sands and clays, on the opposite (north) side of the fault, in the vicinity of the injection wells. A review of the geology and reservoir test analysis indicates that Fault A is a communicating fault in the portion of the lower Oakville sand nearest the site where the lower Oakville sand abuts sand across the fault; Fault A is noncommunicating, away from the site where the lower Oakville sand abuts clay across the fault. This interpretation is supported by the small pressure buildup in Wells 1 and 2 (if Faults A and H were both sealed, the observed pressure buildup would be much greater). Fault H is another fault with a west-to-east orientation from Spindletop Dome, sub-parallel to Fault A. Like Fault A, Fault H is nontransmissive (i.e., does not allow vertical fluid flow). This interpretation is supported by the lack of injectate reaction products in the upper Oakville sand when Injection Well 3 was drilled, which is consistent with a lack of communication between the two Oakville sands and the nontransmissive nature of Fault H. The sealing nature of Fault H is supported by annual injection/falloff tests in the Oakville sands that identify a noflow boundary at the approximate location of Fault H. These test results and analysis of the stratigraphy across the fault indicate that Fault H is noncommunicating (i.e., it is a sealed fault and a no-flow boundary) in the lower Oakville in the vicinity of the deep wells.
7.3 ANALYSIS The approach used to determine the extent of injectate migration consisted of groundwater modeling using the Sandia Waste-Isolation Flow and Transport Model (SWIFT).
ELSE_TSANG_CH007.qxd
8/22/2005
56
7:07 AM
Page 56
Analysis of Injectate Location at Dupont Beaumont Works
SWIFT is a fully transient, three-dimensional model that simulates the flow and transport of fluid, heat, and solutes in geological media (Reeves et al., 1986). The primary equations are coupled by fluid density, fluid viscosity, and porosity. SWIFT was preceded by the code, Survey Waste Injection Program (SWIP), which was designed specifically to evaluate deepwell injection (Intercomp, 1976). The injection sands at the Beaumont Works are all greater than 4000 ft deep. Vertical containment, including the impact of faults, and oil and gas wells referred to as artificial penetrations (APs), was evaluated by DuPont (2000). The conclusion was reached that the injectate would be contained. The lack of transmissive faults is consistent with Gulf Coast clays, known to exhibit viscoelastic deformation behavior, which causes any natural fractures to close and heal very rapidly under the action of in situ compressive stresses. The lack of vertical migration is supported by (1) sample results, taken from the time Injection Well 3 was drilled, showing injection reaction products in the lower Oakville sand and no injection reaction products in the upper Oakville sand; (2) over 2000 ft of clay confining beds overlying the injection sands; (3) results from Core Laboratories brine flow-through permeability tests on cores from Injection Well 3 of the Anahuac shale (depth range of 6687–7356 ft) that established an average permeability to liquid of 5.2 × 10–6 md; (4) vertically oriented core plugs from the Lagarto shales (depth range of 3602–4055 ft) that established an average permeability of 1.1 × 10–5 md; and (4) modeling by DuPont that showed only a few feet of vertical permeation distance into the overlying aquitards. Therefore, vertical migration is not considered in this analysis. As indicated, the acrylonitrile wastewater is injected into the Oakville sands via Injection Wells 1 and 2. The hazardous components for this injectate are cyanide, acetonitrile, acrylonitrile, and acrylamide. These chemicals hydrolyze, and their decomposition rate is affected by the pH (the higher the pH, the faster the decomposition) and temperature (the higher the temperature of the injection formation, the faster the decomposition). Independent field evidence, based on an acrylonitrile waste sample collected at a depth of 4133 ft when Injection Well 3 was drilled, supports the theory of attenuation. Injection Well 3 is approximately 860 and 1400 ft away from Injection Wells 2 and 1, respectively. It was estimated that the encountered injectate was injected in late 1984 and was approximately 2.5 years old. Based on the injectate sample that indicated decomposition, it was observed that (1) the acrylonitrile concentration had decreased to below the detection limit (20 ppm) compared with an average of 222 ppm in the injected wastewater; and (2) the cyanide concentration had dropped from 542 to 22 ppm. Although this degradation is occurring, it is not included in this modeling. 7.3.1 Input Data An important geological consideration for the lower Oakville sand model is the faulting. In plan view, the lower Oakville sand is south of Fault A, and the juxtaposed sand (Catahoula sand) is north of Fault A. All of Fault H is sealed, whereas a portion of Fault A is open where the lower Oakville sand is in contact with the juxtaposed sand, on the north side of the fault in the area of the three deep wells. The basis for Fault H being sealed is its (1) geology, where shale is on the opposite side of the fault, and (2) reservoir test results indicating a barrier near its location. The basis for Fault A being open is its (1) geology, where sand is on the opposite side of the fault; and (2) low-measured pressure buildup at injection wells (if the fault were sealed, higher pressure buildup would occur). At the Oakville level, Fault A is within approximately 400–500 ft of Injection Wells 1 and 2, whereas Fault H is approximately 4000 ft away from these wells. The lower Oakville sand has received injectate only from the acrylonitrile plant through Injection Wells 1 and 2. Injection Well 2 began injection into the lower Oakville in April
ELSE_TSANG_CH007.qxd
8/22/2005
7:07 AM
Page 57
7.3 Analysis
57
1980, and by December 1999, it had injected approximately 2915 mgal. Injection Well 1 began injection into the lower Oakville sand in February 1992, and by December 1999, it had injected approximately 751 mgal. The total injectate in the lower Oakville sand is approximately 3666 mgal. Injection histories for these wells were used in the model. The model treated the lower Oakville sand as a single layer, confined above and below. This is a conservative assumption with respect to lateral migration and is consistent with analysis showing little or no (1–2 ft) vertical migration of the injectate into the overlying confining bed. The top of the lower Oakville sand is located at a depth of approximately 4250 ft near the injection wells. Near the injection wells, the net sand thickness is between 75 and 100 ft. Unlike DuPont’s regulatory modeling that assumed a uniform thickness, the modeling here uses actual thickness that varies spatially. According to temperature logs and other data from DuPont (2000), the entire thickness of the lower Oakville sand receives the injectate. The parameters used in the model are similar to those used by DuPont and are listed in Table 7.1. Formation fluid density and viscosity were computed by DuPont as functions of depth, determined from observed variations of temperature and total dissolved solids (TDS). Injectate density is measured prior to injection. The natural background velocity is on the order of only in./year (DuPont, 2000). This parameter does not have a significant effect during the injection time period. Sorption effects would retard migration and were not taken into account. This is a conservative assumption; i.e., given the clay content of the sand, some sorption that would reduce injectate constituent migration likely occurs. Degradation through hydrolysis is not incorporated into the model. This is a conservative assumption, given that hydrolysis occurs and reduces injectate constituent concentrations. The porosity was determined from core samples (the average of 17 measurements that ranged from 0.274 to 0.339). The compressibility values are based on literature estimates and reservoir test analyses. Hydrodynamic dispersion is a process that occurs at the leading edge of dissolved plumes. The values of dispersivity are based on other field studies. The values for dispersivity used in this modeling are larger than those determined from smallerscale field experiments, because the scale of the injectate plume at Beaumont Works is larger than the field experiments. This is a conservative assumption and results in a larger predicted plume. The pressure of 1920.29 psia is based on a measured value from Injection Well 2, made on May 24, 1994, for a depth of 4261 ft KB. This is the lowest measured pressure for the lower Oakville sand. Adjusting the reference depth to ground surface yields 4234.8 ft below ground surface (bgs). In addition, injectate in the upper Oakville sand was
Table 7.1. Model parameters Parameter Compressibility (fluid) Compressibility (sand) Formation fluid density Injectate fluid density Fluid viscosity Porosity Permeability Reference pressure Reference depth Longitudinal dispersivity Transverse dispersivity
Value 3.035E-06/psi 2.81E-06/psi 65.83 lb/ft3 (1.055) 65.83 lb/ft3 (1.055) 0.62 cP 0.31 11 darcies 1920.29 psia 4248.6 ft bgs 50 ft 5 ft
ELSE_TSANG_CH007.qxd
8/22/2005
58
7:07 AM
Page 58
Analysis of Injectate Location at Dupont Beaumont Works
modeled also. The same parameters were used, and a reference pressure of 1891 psia, at a reference depth of 4184 ft, was used. The permeability estimate is based on reservoir tests. Reservoir data are summarized by DuPont (2000). The lower Oakville has a transmissibility of 1,209,677 md-ft/cp, and a storativity of 0.000136 ft/psi. In addition, the reservoir test data indicate a single no-flow boundary at 2700 ft. Using the porosity and viscosity values in Table 7.1 and thicknesses determined from seismic surveys and wells, yield permeability and compressibility values are similar to those used in the model and listed in Table 7.1. The single no-flow boundary indicated from the reservoir tests corresponds approximately to Fault H. The actual distance of Fault H from Injection Wells 1 and 2 is greater than 2,700 ft. The smaller distance was estimated from the reservoir tests using a uniform thickness for the lower Oakville sand. The thickness of the lower Oakville sand thins toward Fault H. Not accounting for this thinning has resulted in an underestimate of the distance to the no-flow (Fault H) boundary. The finite-difference grid used for the lower Oakville sand model is 320 × 320 nodes. The grid is uniform, consisting of cells 100 ft × 100 ft horizontally, with variable thickness and variable elevation. Fault H is treated in the model as sealed, whereas Fault A is treated as open near the injection wells, and sealed away from the injection wells (based on the geology). The initial conditions for the model consist of the initial relative concentration (0.0) and the initial pressure (1920.29 psia at datum). The boundary conditions for the lower Oakville sand model, shown in Figure 7.3, are no-flow at Spindletop Dome; the boundaries at the edge of the grid blocks are treated as an infinite aquifer using the Carter–Tracy method (using a thickness of 30 ft, radius of 15,000 ft, and 360°). The Carter–Tracy method is a boundary condition that allows the effects of the injection sand to be extended beyond the grid. Faults A and H are shown, where the blackand-white portions of the faults indicate sand against shale across the fault, and therefore, that portion of the fault is sealed. The injection rates are used in the model, and the injectate’s relative concentration is assumed to be 1.0. For the numerical solution, centered-in-time and centered-in-space were selected, and the matrix solution technique of L2SOR was used. 7.3.2 Simulation Results The maximum pressure buildup in the lower Oakville sand occurs in the first quarter of 1998 at Injection Well 1, and is an increase of only 37.75 psi or a 1.97% increase over the initial pressure (datum of 4248.6 ft). This pressure declines rapidly once injection ceases, based on falloff reservoir tests. This small computed pressure buildup is consistent with the observed measured pressure buildup. This is consistent with the conclusion that both Faults A and H cannot be sealed. The predicted extent of injectate migration is shown in Figure 7.4. The present-day (beginning of year 2000) extent of injectate migration is displayed using five contours. The mass of injectate contained within each plume contour was calculated. The results are as follows: ■ The 0.5 contour (50% injectate/50% brine, by vol.) contained 91.81% of the injectate mass. ■ The 0.1 contour (10% injectate/90% brine, by vol.) contained 99.08% of the injectate mass. ■ The 10–2 contour (1% injectate/99% brine, by vol.) contained 99.94% of the injectate mass. ■ The 10–3 contour (0.1% injectate/99.9% brine, by vol.) contained 99.99% of the injectate mass. ■ The 10–4 contour (0.01% injectate/99.99% brine, by vol.) contained 99.999% of the injectate mass. This corresponds to the HBL contour in Figure 7.4. Therefore, the 10–4 contour contains almost all of the injectate mass and volume. If chemical fate (sorption and degradation) were considered, the contour that contains the
ELSE_TSANG_CH007.qxd
8/22/2005
7:08 AM
Page 59
7.3 Analysis
59
Fig. 7.3. Boundary conditions for the lower Oakville sand model.
plume would be even larger (10–3 or larger). If a health-based approach is used instead, then the following steps should be taken into account for the brine. First, chemical fate should be considered when selecting the appropriate injectate chemical; i.e., a persistent chemical that has the lowest HBL should be selected. This leads to the selection of acrylonitrile, which has an HBL of 0.00006 mg/L. A concentration reduction factor (CRF) can be calculated by dividing the HBL by the injection concentration. The actual yearly average concentration of acrylonitrile in the injectate for 1980 (the first year the lower Oakville was used for injection) was 83.6 mg/L. Performing this division yields a CRF of 7.18 × 10–7. A final adjustment to the CRF, based on the recognition that HBLs apply to drinkable water, whereas the Oakville brine is not drinkable, even before any injectate is added, is needed. Water in Oakville sands, as measured in DuPont Injection Well 3 at a depth of 4085 ft, contains 73,000–85,000 mg/L TDS. In contrast, the EPA-recommended secondary drinking water standard for TDS is 500 mg/L. Therefore, to make Oakville brine drinkable, it would have to be diluted by a factor of 146–170. For computational purposes, a factor of 150 is used. Diluting the brine in its native state, making it drinkable water, would also dilute any injectate contained in the brine. Applying the factor of 150 to convert the Oakville brine into a
ELSE_TSANG_CH007.qxd
60
8/22/2005
7:08 AM
Page 60
Analysis of Injectate Location at Dupont Beaumont Works
Fig. 7.4. Plume for the lower Oakville sand.
drinking-water equivalent, the CRF (ignoring chemical fate) becomes 1.08 × 10–4, which is consistent with the contour determined using mass-in-place calculations. Finally, adjusting detection limits based on brine concentrations leads to the conclusion that the organic chemicals contained in the injectate only can be detected inside the 10–4 contour. A similar plot is provided for the upper Oakville sand in Figure 7.5. As may be seen, the plume for the upper Oakville sand is smaller than that for the lower Oakville sand, because less injectate was placed in the upper Oakville sand. Grid cell sizes and time steps were selected using standard Courant and Peclet number criteria to control numerical dispersion (Anderson and Woessner, 1992). Because a uniform grid was used, cell-aspect ratio was not an issue. The Peclet number is 2; the Courant number was as high as 0.7. These are within acceptable limits (Spitz and Moreno, 1996; Woessner and Anderson, 1992). The extent of injectate migration depends primarily on the volume of injectate, amount of dispersion, amount of hydrolysis and sorption, sand thickness, and sand porosity. Of these parameters, volume of injectate is measured during injection, porosity is measured from core samples, and thickness is determined from well logs and geophysical surveys. Thus, the estimates for these parameters have a high degree of confidence. Reasonable and conservative values for dispersion have been used, and hydrolysis and sorption, which would limit migration, have not been simulated. In addition, the injectate migration
ELSE_TSANG_CH007.qxd
8/22/2005
7:08 AM
Page 61
7.3 Analysis
61
Fig. 7.5. Plume for the upper Oakville sand.
is influenced by the presence and behavior of local faults, which have been studied through a variety of logs, seismic surveys, and hydraulic testing. Although changes in permeability affect computed pressure, the extent of injectate migration is not very sensitive to changes in permeability. 7.3.3 Verification A comparison of simulated pressure falloff compared with measured pressure falloff from the 1992 interference test with the gauge in Well 2 is shown in Figure 7.6. Pressure was monitored in Well 2 while an injection and falloff test was conducted in Well 1. The data used in this simulation is contained in DuPont (2000). As shown in Figure 7.6, the model’s predicted pressures closely track the measured pressures. A comparison is also made with the 1996 falloff test, which was conducted in Well 1 and is shown in Figure 7.7. Again, the comparison is good, providing confidence that the conceptual model, which is the basis for the numerical model, is correct. Data used to verify the simulated results include injectate measurements at Injection Well 3 in 1987, when that well was drilled. As shown in Figure 7.8, the model predicted the presence of injectate in the lower Oakville sand at that location in 1987, and injectate was detected in samples collected when Injection Well 3 was drilled in 1987. At the time that
ELSE_TSANG_CH007.qxd
62
8/22/2005
7:08 AM
Page 62
Analysis of Injectate Location at Dupont Beaumont Works
Fig. 7.6. Falloff pressure versus elapsed time at Injection Well 2 for the 1992 interference test. Solid line: simulated result; and circles: measured pressures.
Fig. 7.7. Falloff pressure versus elapsed time at Injection Well 1 for the 1996 falloff test.
Injection Well 3 was drilled, a sample also was collected from the upper Oakville sand. This sample indicates no injectate at this location at that time. A time-series plot for injectate concentration in the upper Oakville for the Well 3 location is shown in Figure 7.9. Consistent
ELSE_TSANG_CH007.qxd
8/22/2005
7:08 AM
Page 63
7.3 Analysis
63
Fig. 7.8. Concentration increase in the lower Oakville sand at Injection Well 3.
Fig. 7.9. Concentration increase in the upper Oakville sand and at Injection Well 3.
with the observed data, the model predicted that no injectate was present in the upper Oakville sand at the time that Well 3 was drilled. Simulations where Fault A was sealed (not shown) forced injectate to migrate south of Fault A; for this simulation, injectate was present at the Well 3 location in April 1987. These observations support the conceptualization that Fault A is open in the vicinity of the injection wells.
ELSE_TSANG_CH007.qxd
8/22/2005
64
7:08 AM
Page 64
Analysis of Injectate Location at Dupont Beaumont Works
7.4 CONCLUSIONS Based on borehole information and seismic data, there are two nearly parallel faults (Faults A and H) in the Oakville formation in the vicinity of Injection Wells 1 and 2, with Fault A being the closer fault. Based on reservoir tests, a no-flow boundary is present near the vicinity of Fault H. Therefore, based on the reservoir test results, Fault H is considered a sealed fault. The best evidence that Fault A is not sealed in the vicinity of the injection wells is the small observed pressure buildup. This conceptualization was used in the SWIFT modeling performed in this chapter, resulting in a small computed pressure increase that is consistent with observed data. The conceptual model of Fault A being open is further supported by injectate data collected at Injection Well 3, when it was drilled in 1987. The SWIFT model computed an injectate location that is consistent with observed injectate data, whereas treatment of Fault A as being sealed would have caused injectate to migrate further south in the upper Oakville sand than had been observed. The present-day location of the injectate plume in the Oakville sands has been determined using SWIFT. A realistic injectate plume definition in a brine environment was developed using several different methods. Three different methods were used to define a realistic injectate plume: (1) mass-in-place calculations, (2) an HBL approach adjusted for brine, and (3) use of detection limits. The three independent approaches lead to the same conclusion that a realistic injectate plume definition in a brine environment is achieved using a 10–4 or larger contour if chemical fate is considered.
REFERENCES Anderson, M.P. and Woessner, W.W., 1992. Applied Groundwater Modeling, Simulation of Flow and Advective Transport. Academic Press, New York. DuPont Beaumont Works, 2000. EPA No-Migration Petition, Lower Oakville Sand Reissuance Request. Prepared by DuPont, Beaumont, TX, and submitted to EPA. Intercomp, 1976. A Model for Calculating Effects of Liquid Waste Disposal in Deep Saline Aquifer. U.S. Geological Society, WRI 76-61, Reston, VA. Reeves, M., Ward, D.S., Johns, N.D., and Cranwell, R.M., 1986. Theory and Implementation of SWIFT II: the Sandia Waste-Isolation Flow and Transport Model for Fractured Media, Release 4.84. NUREG/CR-3328, SAND83-1159, Sandia National Laboratories, Albuquerque, NM. Spitz, K. and Moreno, J., 1996. A Practical Guide to Groundwater and Solute Transport Modeling. Wiley, New York.
ELSE_TSANG_CH008.qxd
8/19/2005
6:33 PM
Page 65
Chapter 8
AQUIFER STORAGE AND RECOVERY WELLS IN FLORIDA: HOW AND WHEN WILL IMPACT BE REGULATED? A. Muniza, M. Tobonb, and F. Bloetscherc a
Hazen and Sawyer, P.C., Boca Raton, FL, USA City of Fort Lauderdale, Fort Lauderdale, FL, USA c Public Utility Management Planning Services, Inc., Hollywood, FL, USA b
8.1 INTRODUCTION Aquifer storage and recovery (ASR) wells have become one of the most controversial issues in Florida. One reason that it has become the center of controversy is that ASR has been touted as the savior of South Florida’s water supply problems. Known for its tropical climate and environment, South Florida has been experiencing tremendous population growth, which has been of great concern to the state, local entities, utility companies, and environmentalists. One of the biggest concerns is that of water supply, which seems ironic in a state that is second only to Louisiana in annual rainfall. The shortfall occurs because Florida receives rain when it is not needed and the state’s flat topography does not lend itself to efficient storage methods. In addition, competing demands from the environment have created a situation that demands better and more efficient water management. The proposed solution is implementation of the ASR concept at an unprecedented and unproven level. Is ASR the panacea to solve South Florida’s water supply problems? According to current plans by the South Florida Water Management District (District) and the United States Army Corps of Engineers (COE), to restore the Everglades and provide urban water supplies, ASR is the solution. Because it has been proposed as a panacea, the proposed plan has raised great concern among many scientists, engineers, and environmentalists. Most are concerned about the impacts to Florida’s hydrogeologic environment that may result from the proposed storage of over 1.6 billion gallons of water in the Floridan Aquifer System (FAS). Degradation of the native groundwater is a critical issue for many who oppose using Florida’s hydrogeologic setting for water resource management. This permanent solution will supposedly improve South Florida’s water supplies for the next 50 years. Figure 8.1 shows the area being considered for the wide-scale implementation of ASR. In addition to concerns over the water quality impacts associated with large-scale implementation of ASR, additional questions exist regarding the implementation of this relatively new concept on such a massive scale. No large-scale pilot projects have been completed to date, and none will be completed during the next several years. In fact, the application of ASR has a very brief record in South Florida. Recovery efficiencies of stored water must be carefully addressed, because this one issue could make ASR inadequate to meet the goals of the restoration plans.
ELSE_TSANG_CH008.qxd
66
8/19/2005
6:33 PM
Page 66
Aquifer Storage and Recovery Wells in Florida: How and When will Impact be Regulated?
Fig. 8.1. Proposed ASR program for the Everglades Restoration Plan.
8.2 THE CONCEPT ASR is a relatively new concept in the management of water supplies, in both potable and nonpotable water systems. The concept is to inject water into a suitable aquifer system (i.e., storage horizon) during times when water is available, and to recover that water later to meet future needs. Needs might vary from emergency demands, to peak shaving at water treatment plants. At public water utilities, the injection period occurs when plant capacity is underutilized, so that excess plant capacity can be used to create water supplies that can be successfully stored below grade and recovered for use in the potable distribution system without a significant amount of additional treatment. Figure 8.2 illustrates the ASR concept. ASR wells are constructed and operated differently from either production wells or injection wells. Beneath the surface, the injected freshwater displaces native brackish water that naturally exists in the aquifer. This scenario creates an underground storage reservoir or “bubble.” The stored water can be then withdrawn to meet peak demands for short periods of time. It should be clearly understood that withdrawals could only occur for a maximum of 50% of the time under ideal conditions. In reality, the recovery period is significantly less, since time for recharge must be allotted to create a reserve. In addition, demands might not dictate recovery at will.
ELSE_TSANG_CH008.qxd
8/19/2005
6:33 PM
Page 67
8.3 Floridan Aquifer System
67
Fig. 8.2. Aquifer storage and recovery concept.
In the utility business, employing the ASR technology can improve water supply management, increase the efficiency of system operations, and increase water supply availability during drought periods. It has been demonstrated that effective use of ASR can easily defer or possibly eliminate expansion of small water-treatment facilities. Considerable expense can also be recognized by the more efficient overall operation of the treatment facility. This is especially true for membrane facilities that are designed to operate 24 hours a day. An important element in understanding the function of the ASR concept is to clearly comprehend what ASR can and cannot do. For instance, ASR can serve as an underground storage tank, but it cannot completely replace aboveground storage that can pump large volumes of water to meet emergency demands, since underground formations have limited yield. Another issue that is often misunderstood is the fact that ASR does not “produce” water. An ASR system stores water that was produced or withdrawn from a production well or surface-water intake. The water recovered simply augments for a short period of time (i.e., the time that it takes to empty the stored water) the yield of a system. Many incorrectly assume that ASR can increase average plant capacity. The peak shaving benefit is for short-term high-demand periods, and should not be considered to be permanent, since excess water is needed to create a reserve. 8.3 FLORIDAN AQUIFER SYSTEM Southeastern Florida is underlain by a thick sequence of carbonate rocks, limestone and dolomite, and lesser amounts of unconsolidated clastics consisting of sand silt and clay and minor amounts of evaporites (gypsum and anhydrite). Carbonate rocks are the principal rock types. The evaporites are present in the lower (deeper) part, and the clastics are present in the upper (shallow) part. The movement of groundwater occurs principally through the carbonate rocks (Englehardt et al., 2001).
ELSE_TSANG_CH008.qxd
68
8/19/2005
6:33 PM
Page 68
Aquifer Storage and Recovery Wells in Florida: How and When will Impact be Regulated?
Ranging from the oldest to youngest, the various geologic formations comprising the sequence are the Cedar Keys, Oldsmar, and Avon Park Formations; the Ocala and Suwannee Limestones; the Tampa Limestone; and the Hawthorn Formation (Group). These formations constitute the various elements of the FAS. Evaporite deposits present in the Cedar Keys Formation constitute a lower confining unit, marking the base of the active groundwater flow system (Meyer, 1989). The permeable limestones and dolomites of the various formations are hydraulically interconnected, to a degree. The degree of interconnection varies, as does the permeability. In general, the rocks comprising the Floridan aquifer resemble a layered cake composed of numerous zones of alternating high and low permeability (Meyer, 1989). In southeastern coastal Florida, the base of the Floridan aquifer system occurs at an approximate depth of 3500 ft; its top is present at a depth of ±900 ft. Clay, marl, and claystone present in the Hawthorn Formation (Group) constitute the confining sequence for the FAS, which isolates the Floridan from the beds forming the Biscayne and shallow aquifers in Southeast Florida (Miller, 1986). These relationships are shown in the cross section given in Figure 8.3, which has been reproduced from U.S. Geological Survey Professional Paper 1403-G (Meyer, 1989). The line of the cross section is east-west through the City of Fort Lauderdale. In South Florida, the upper FAS, which contains brackish water, exists at depths ranging from approximately 900 to 1800 ft. This portion of the upper FAS has interbedded layers of horizons that have high to low hydraulic conductivities. As a result, it is the area that has been selected for the aquifer storage and recovery zone(s) (Englehardt et al., 2001). Hydraulic properties of the FAS, from samples at depths of 1360 to 2993 ft can be summarized in Table 8.1.
Fig. 8.3. Hydrogeologic cross section through South Florida (Meyer, 1989).
ELSE_TSANG_CH008.qxd
8/19/2005
6:33 PM
Page 69
8.4 ASR Development in Florida
69
The transmissivity of the Upper Floridan Aquifer is estimated to range from 10,000 to 60,000 ft2/day (Meyer, 1989). Transmissivities are higher in the more productive Lower Floridan Aquifer, with values ranging from 3,200,000 to 24,600,000 (Meyer, 1989) and 13,000,000 (Geraghty and Miller, 1984).
8.4 ASR DEVELOPMENT IN FLORIDA Development of ASR in Florida has been overstated in some instances due to the attractiveness of the concept as a cure-all. Figure 8.4 presents a listing and map of ASR wells in the State of Florida. First glance at this map is misleading, as one would assume that all of these facilities are in operation or under construction. For example, the Florida Keys Aqueduct Authority ASR well is shown, yet this well has been abandoned for some time. A recent survey by the American Water Works Association (AWWA) painted a totally different picture. The AWWA survey indicated only four operating systems in Florida. The real answer lies somewhere in the middle. In reality, there appear to be 11 operating systems with another 10 under development. In some instances, systems are touted as ASR systems, Table 8.1. A summary of hydraulic properties of the FAS, from samples at depths of 1360 to 2993 ft Mean
maximum
minimum
std. deviation
median
no. of samples
Vertical hydraulic conductivity (cm/s) 2.83E-4 5E-3 9.6E-10
7.15E-4
4.6E-5
131
Horizontal hydraulic conductivity (cm/s) 2.56E-4 4E-3 2.5E-9
5.82E-4
7.4E-5
83
Porosity (fraction) 0.317 0.45
0.091
0.33
0.034
Fig. 8.4. ASR facilities in Florida (FDEP, 2001).
127
ELSE_TSANG_CH008.qxd
70
8/19/2005
6:33 PM
Page 70
Aquifer Storage and Recovery Wells in Florida: How and When will Impact be Regulated?
when in fact, the water being injected is very similar to the water of the native formation. Systems that store water in horizons of water quality similar to the recharge water might not meet the true definition of ASR. Such wells may be viewed as production wells and not as ASR wells. The Miami-Dade ASR facility actually operates in a dual mode whereby the well is used for recovery of stored water, and when the reserve is exhausted, the operation changes to production where the well operates as a Floridan Aquifer production well. The City of Boynton Beach was one of the few sites that has achieved high-recovery efficiencies using potable water. Other systems, such as the City of Fort Lauderdale, selected a different storage horizon and have yet to achieve the desired recovery efficiencies. Plots of recent recovery curves for the City of Fort Lauderdale and the City of Boynton Beach are presented in Figures 8.5 and 8.6, respectively. The Lake Okeechobee ASR demonstration used a broader horizon that included several production zones as the targeted storage interval. A comparison of the storage horizons employed at these three sites is presented in Figure 8.7. Recovery rates from this well could easily yield 5–10 million gallons per day (mgd) per well. Unfortunately, testing was cut short at this site, and full-scale testing as recommended was never performed. Data from the Lake Okeechobee ASR well would be invaluable for future assessment of a large-scale ASR program.
8.5 EVERGLADES RESTORATION—ASR COMPONENT To meet the goals of the Comprehensive Everglades Restoration Project, cowritten by the District, ASR is proposed to store water during rainfall events and recover that water during drought periods. As a result, the water managers are implementing plans to use hundreds of ASR wells to store nutrient-rich waters in pristine underground formations, which tend to occur in 2-year cycles.
Fig. 8.5. City of Fort Lauderdale Fiveash WTP ASR cycle testing results.
ELSE_TSANG_CH008.qxd
8/19/2005
6:33 PM
Page 71
8.5 Everglades Restoration—ASR Component
71
Fig. 8.6. City of Boynton Beach ASR cycle testing results.
Fig. 8.7. Comparison of ASR storage zone horizons in South Florida.
Water deliveries for environmental, urban, and agricultural purposes are to come in the form of ASR wells. The plan specifically targets the upper FAS as the zone of injection, which is also the formation from where some reverse osmosis supplies are derived, thereby setting a potential future conflict. The plan further states that using ASR technology might provide greater storage efficiency when compared to the land requirements, and to high seepage and evapotranspiration rates associated with aboveground reservoir storage. The COE notes that water quality concerns, particularly regarding untreated surface water, currently limit the ability to use ASR wells in the area. The quality of untreated runoff
ELSE_TSANG_CH008.qxd
72
8/19/2005
6:33 PM
Page 72
Aquifer Storage and Recovery Wells in Florida: How and When will Impact be Regulated?
may preclude its injection for ASR purposes under current regulations. Retention facilities to capture and hold excess water for injection into the aquifer might be required at some sites, both of which might make ASR unavailable for meeting water supply demands in South Florida, as conceived in the COE Central and South Florida Restudy project. It is not surprising then that public comment has indicated that ASR wells should be tested to evaluate technical uncertainties with high-capacity applications (GCSSF Technical Advisory Committee Aquifer Storage and Recovery Report, May 23, 1996). The COE identifies several issues that need to be addressed in planning for the regional ASR programs: 1. Environmental and health concerns regarding water quality. 2. Current regulatory constraints. 3. Costs of the project. 4. Potential benefits of having additional clean water at the chosen site. The COE does acknowledge that ASR should be investigated to determine its feasibility at a regional scale as well as its environmental impacts, as might be expected in a conceptual plan. If large-scale ASR is shown to be feasible, more extensive regional-scale facilities using untreated surface water runoff and Lake Okeechobee discharges could be beneficial in meeting additional demands within the region. Potential locations chosen for the regional, high-capacity ASR pilot projects include sites on the fringe of Lake Okeechobee, the Lower East Coast, and the Water Conservation Areas. A final conclusion on ASR from the Restudy indicates that the COE “recognizes that water injected into the aquifer may not meet appropriate water quality standards.” Hence, ASR facilities are most useful at the site of water treatment plants, which is not where it is proposed that the facilities be located, but is where clean treated water can be injected, plant operation economies can be realized, and conveyance losses can be eliminated.
8.6 CONCERNS ABOUT AQUIFER STORAGE AND RECOVERY ON A LARGE SCALE Given the pronouncement that ASR is South Florida’s water-supply solution, concerns about the proposed ASR systems in South Florida regarding the following issues remain: 1. Many of the ASR projects store freshwater in relatively freshwater zones for relatively short periods of time (i.e., less than 60 days). The Boynton Beach project is the only active ASR project in Southeast Florida that stores water from 30 to 45 days before they begin withdrawal. For ASR to work on a regional basis, or to be a long-term water supply supplement, utilities need to be sure that ASR can be stored for months at a time. This has been done in the Peace River Project on the west coast of Florida (although this is basically freshwater in a freshwater zone), and is done to a certain extent on Cocoa Beach; however, those are the only major projects where the long-term-storage concept has been applied successfully in Florida, and because both are located remotely from the Restudy area, aquifer conditions will not be the same. The definition of success or efficiency of an ASR system has to include a minimum storage time, with a certain percent of recovery as a sliding scale. Figure 8.8, which presents a proposed definition of success based on District and other data (Bloetscher, 2001), shows that a near-90% recovery efficiency is possible initially, but recovery efficiency decreases over time, with only approximately 40% anticipated after 2 years. 2. The transition zone between the freshwater and the saltwater must be defined. It has been theorized (Missimer, 1969) that pumping the injectate into a zone that is relatively thin and
ELSE_TSANG_CH008.qxd
8/19/2005
6:33 PM
Page 73
8.6 Concerns about Aquifer Storage and Recovery on a Large Scale
73
Recovery
90%
40%
Year 1
Year 2
Year 3
Fig. 8.8. Proposed definition of success.
has well-confined strata could clear all of the water out of the zone. Unfortunately, many of the ASR projects pump into “thick” zones, so the native water cannot be fully displaced. Other ASR projects have indicated that there is a significant mixing zone between the injected and native water. In the Collier County well, the mixing zone was estimated to be between 200 and 250 million gallons. The curve to define this amount, which is presented in Figure 8.9, shows an increase as the amount of water injected increased (Bloetscher, 2001), and that storage volumes above 200–250 million gallons show little benefit in terms of recovery efficiency. Cycle 4, for example, exhibited a lower efficiency, with water decreasing at a rate of 0.07 mg/L chloride per 1000 gallons of water recovered. Cycle 6 results, on the other hand, show a much better recovery efficiency with water deteriorating at a slower rate (i.e., approximately 0.037 mg/L chloride per 1000 gallons of recovered water). As a result, 100% recovery of the water is not a reasonable goal, as the rate of recovery curve shows a “flattening” when storage volumes reach around 250 million gallons. 3. The injection of 1.0–1.5 mgd has been proven to work for ASR wells in Florida. The suggestion that 5 or 10 mgd could be used for wells has not been demonstrated. The impact of injection of this quantity of water into the aquifer is unknown. If significant pressures build up in the aquifer, what is the long-term impact to the formation? And if the water pressure is reduced by withdrawal, is there a potential for collapse of the formation as a result of fracturing during injection of water? The answers to these questions are simply unknown, and probably can only be evaluated through an actual test project. 4. There are no rules to define, or to limit, competition between water supply and ASR in the Upper Floridan Aquifer. Rules need to be defined, or the bubble that is created with ASR wells in the Florida Aquifer could move toward a Floridan water-supply well. The Floridan Aquifer is known to have significant drawdowns in order to act on reasonable withdrawals (i.e., 1–1.5 million gallons per day per well can translate to over 100 ft of drawdown). As a result, the cone of influence spans a significant distance. Little is known about how the FAS operates, nor have any significant models been developed for Southeast Florida. The concern about the water supply/ASR competition needs to be resolved. 5. The questions about raw water or water of less than pristine quality has long been debated by EPA. EPA advocates the prohibition of contaminants that do not exist naturally in the aquifer system from the aquifer. Yet with any ASR project, this is not possible. Logic and data collected to date do not support the impression that the Floridan is a pristine aquifer. Both the surficial and Biscayne aquifers are known to have microbiological activity as a result of the total organic carbon. The introduction of raw water with the associated
ELSE_TSANG_CH008.qxd
6:33 PM
Page 74
Aquifer Storage and Recovery Wells in Florida: How and When will Impact be Regulated? Rate of Chloride Increase/1000 Gallons Recovered
74
8/19/2005
0.10 Cycle 2 0.09 Cycle 3 0.08 Cycle 4
0.07 0.06 0.05
Cycle 5 Cycle 6
0.04 0.03
Cycle 7 0.02 0.01 0.00
0
50
100
150
200
250
Net Injected Water (Millions of Gallons)
Fig. 8.9. Collier County ASR recovery efficiency with successive cycles.
6.
1. 2. 3. 4.
microbiology, without some degree of control of the potential for growth as a result of introduced total organic carbon of the raw water, should be questioned. Even deep injection wells that do not inject chlorinated effluent showed deterioration with time and the potential for fouling. (Looking at pictures or videos of the wells, one can see all of the microbiological growth on the side of the wells.) However, the injection of chlorine to keep the wells clean and to provide some control of the microbiological activity close to the well creates a concern related to the formation of trihalomethanes with the Florida Department of Environmental Protection (FDEP). There is an AWWA Research Foundation report that indicates the microbiological activity will reduce haloacetic acids (HAAs) almost immediately, and will remove the trihalomethanes (Pyne et al., 1995) over a 30–90 day period. Likewise, the bacteria will reduce the injected total organic carbon. The protocol for large-scale injection has not been fully presented in a manner that would make many utilities fully comfortable. If one assumes a formation of a reasonable thickness could displace all native water over time, the injectate would create an ever-increasing bubble that would expand outward from the wells. If the concept of large bubbles with clusters of ASR wells (Fig. 8.10) were pursued (Bloetscher, 2001), then it would not be helpful to inject into all of the ASR wells at the same time; it would, however, make sense to start at the center and work outward so that aquifer pressure would be minimized, while at the same time displacing as much of the native water as possible. Likewise, withdrawals would occur only from the center wells and not from the exterior wells. Such a protocol could be composed of the following steps: Turn on well 1 to pump/inject water. When bubble reaches wells 2, 3, and 4, begin injecting into wells 2, 3, and 4, and discontinue pumping into well 1. When bubble reaches wells 5–10, begin injecting into wells 5–10, and discontinue pumping into wells 2, 3, and 4. This forces the bubble to always move outward. Withdraw only from the center wells.
ELSE_TSANG_CH008.qxd
8/19/2005
6:33 PM
Page 75
8.7 Governing Regulations
75
Fig. 8.10. Illustration of large bubbles with clusters of ASR wells.
8.7 GOVERNING REGULATIONS 40 CFR Sections 144 and 146 are the federal regulations for underground injection control. The rules were established under the authority of the Safe Drinking Water Act, which was approved in 1974 and amended in 1986 and 1996. The purpose of the Underground Injection Control (UIC) rules is to protect the quality of underground sources of drinking water (USDW) and to prevent degradation of the quality of other aquifers adjacent to the injection zone, both vertically and horizontally, that may be used for other purposes. This regulatory intent is achieved through rules that govern the construction and operation of injection wells in such a way that the injected fluid does not migrate into the USDW, defined as an aquifer having less than 10,000 mg/L of total dissolved solids (TDS). During the Reagan administration (1980–1988), the federal government delegated a series of programs for administration by states. The UIC program was delegated to 34 states, and partially delegated to six others. The rules set forth standards for the Federal UIC program mirrored in many states. Injection programs primarily focus on ASR at the state level, and often all injection programs are deemed to be ASR. Some states permit injection projects to have an extended zone of discharge included in the permit to meet secondary criteria. The extended zone of discharge applies to parameters listed as secondary drinking water standards and for sodium. Zones of discharge are not generally provided for parameters listed as primary drinking water standards (except for sodium). Each state, where allowed, will define this differently. As an example, in Florida, the extended zone of discharge can extend radially to the permittee’s property line, which may be greater than the 100 feet normally allowed for a zone of discharge. The Groundwater Rule, approved in 1992, may require that the water be disinfected upon withdrawal unless the water meets the requirements for “natural disinfection,” or if the system qualifies for a variance. It is often pointed out that this issue is where the disinfection by-products concerns come into play, as outlined above. Other rules involve the methods to regulate the stored water. ASR wells are different from production and/or injection wells because they are dual-purpose wells. During recharge, an ASR well is an injection well and is permitted under the EPA and state criteria governing Underground Injection Control. Other agencies may also be involved during recharge, since the recharge water has to be withdrawn from a source. During storage, ASR wells are treated
ELSE_TSANG_CH008.qxd
76
8/19/2005
6:33 PM
Page 76
Aquifer Storage and Recovery Wells in Florida: How and When will Impact be Regulated?
as facilities that store and impact native water quality. The EPA and state also govern this mode. During recovery, the permitting issue gets complex depending on the water quality of the recovered water, the receiving zone water quality, and the intended use. Local regulatory agencies in addition to EPA and the state regulate this process. The district is also involved, since withdrawals from an underground source are being made.
8.8 CONCLUSIONS AND RECOMMENDATIONS It is certain that ASR is a concept that will play an important role in the solution to the watersupply concerns in South Florida, not only for urban users, but also for agricultural and environmental users. ASR is a viable method (subject to proper geologic conditions) for improved water management. Regulators will need to be flexible in allowing demonstration projects to proceed in a cost-effective manner to enable full understanding and possible assessment of the role of ASR in the Everglades Restoration Plan. Much can be concluded from historical lessons learned from existing and ongoing ASR projects. More knowledge would be available with additional research to better describe scenarios where ASR designs experienced unique challenges. Below is a list of conclusions that can be drawn from results of ASR studies: 1. In South Florida, the best results are achieved when design and testing are coordinated with project objectives. For example, the Florida Keys Aqueduct Authority demonstration project tested the ASR concept to investigate its feasibility. The result was that storage in a thin, well-confined zone was capable of achieving approximately 70% recovery, but each well could store volumes of 12–15 million gallons only. 2. For small ASR systems, use of the upper FAS (i.e., the Suwannee Limestone) has proved to be the most favorable for successful ASR. The Boynton Beach ASR is a typical example. 3. Inadequate data exist to evaluate ASR performance on a large scale. The Lake Okeechobee ASR demonstration project was the only large-scale test performed, but the testing was limited. The question now is: What can we do better to improve our understanding of Florida’s unique hydrogeologic setting to allow safe development of the ASR concept on a broader scale? While ASR appears to be moving forward as a major component of the Restudy, much more information is needed to definitively assess its value and possible environmental impacts. It should be noted that the reliance on large-scale regional ASR projects are of concern to many utility systems, because there are no successful large-scale demonstration ASR projects in South Florida, as suggested in the Restudy. Since geology and water quality play such a significant role in the success of ASR projects, before moving toward the assumption that ASR should be a significant component for South Florida’s water supply needs, it is hoped that these ideas will provide some thoughts on the implementation of ASR on a contained basis in Southeast Florida. The following are some recommendations: ● Each ASR well should be designed, constructed, and tested with the unique goals of the project in mind (e.g., injection rate, storage volume, storage period, recovery rate, water quality of recharge, water quality of native, and water quality of intended recovered use). ● Proper peer review and multi-jurisdictional input, in an effort to maximize the information gained, will serve to help clarify the magnitude of ASR’s future in the South Florida regional water-supply picture. ● Cluster pilot testing should be fast-tracked to gain knowledge of system performance and for refinement of final designs.
ELSE_TSANG_CH008.qxd
8/19/2005
References ●
●
●
6:33 PM
Page 77
77
More emphasis should be placed on regulatory flexibility and monitoring to ensure pilot projects collect the data needed to demonstrate environmental soundness of large-scale implementation. Groundwater modeling should be developed and calibrated with test data from pilot projects to model the impacts of the proposed ASR program. A time-certain schedule should be mandated to all of the above.
REFERENCES American Water Works Association, 2002. Survey and Analysis of Aquifer Storage and Recovery (ASR) Systems and Associated Regulatory Programs in the United States, AWWA, Denver, CO. Bloetscher, F., 2001. Does the current knowledge of ASR answer all the necessary questions? Groundwater Protection Council Annual Conference Proceedings—Reno, NV, GWPC, Oklahoma City, OK. Englehardt, J.D., Amy, V.P., Bloetscher, F., Chin, D.A., Fleming, L.E., Gokgoz, S., SoloGabriele, H., Rose, J.B. and Tchobanoglous, G., 2001. Comparative Assessment of Human and Ecological Impacts for Municipal Wastewater Disposal Methods in Southeast Florida: Deep Wells, Ocean Outfalls, and Canal Discharges. University of Miami, Coral Gables, FL. Geraghty & Miller, Inc., 1984. Ground Water Consultants, Construction and Testing of Disposal Wells 1, 2 and 3 at the George T. Lohmeyer Plant, Fort Lauderdale, Florida. Consultant’s Report. Meyer, Frederick W., Hydrogeology, 1989. Ground-Water Movement and Subsurface Storage in the Floridan Aquifer System in Southern Florida, Regional Aquifer-System Analysis—Floridan Aquifer System. United States Geological Survey Professional Paper 1403-G. Miller, J.A., 1986. Hydrogeologic Framework of the Floridan Aquifer System in Florida, and in Parts of Georgia, Alabama, and South Carolina. United States Geological Survey Professional Paper 1403-B. Parker, G.G., Ferguson, G.E., Love, S.K., et al., 1955. Water Resources of Southeastern Florida, with Special Reference to Geology and Ground Water of the Miami Area. United States Geological Survey Water Supply Paper 1255. Pyne, R.D.G., Singer, P.C. and Miller, C.T., 1995. Aquifer Storage Recovery of Treated Drinking Water. AWWA Research Foundation, Denver, CO. U.S. Army Corps of Engineers (CERP), 1999. C&SF Comprehensive Review Study Final Integrated Feasibility Report and Programmatic Impact Statement (PEIS). ACOE, Washington, DC.
This page intentionally left blank
78
ELSE_TSANG_CH009.qxd
8/19/2005
6:34 PM
Page 79
Chapter 9
ALASKA-SAKHALIN 2002 SYMPOSIUM DISCUSSION OF UNDERGROUND INJECTION CONTROL IN ARCTIC OILFIELDS* T. Cutlera and D. Thurstonb a
U.S. Environmental Protection Agency, Seattle, Washington, DC, USA Minerals Management Service (MMS), Anchorage, Alaska, USA
b
9.1 INTRODUCTION The exchange of information at the 2002 Sakhalin Seminar on the management of oilfield wastes through injection in Alaska, and the proposed actions in Sakhalin to manage oilfield wastes by injection, is summarized. A description of the history and development plans of Sakhalin is outlined. Oilfield waste management using underground injection is an alternative that Sakhalin may choose to apply in their sensitive arctic and subarctic settings around Sakhalin Island. Questions posed at the seminar are listed to highlight current needs and concerns as they apply to underground injection as a future waste management tool in Sakhalin. The sharing of Alaska’s and Sakhalin’s Underground Injection Control (UIC) program experiences with the people of Sakhalin has been beneficial to all participants from the United States and Russia. Two multidisciplinary teams of Alaskan experts visited Sakhalin Island in April 2001 and March 2002. This was part of an Offshore Oil and Gas Environmental Management seminar series sponsored by USAID and organized under the Alaska–Sakhalin working group. These teams were invited by the Sakhalin Regional Administration’s Oil and Gas Office to share Alaska’s oil- and gas-related environmental and safety management experience—both good and bad—with a multi-stakeholder Russian audience. The Alaska teams consisted of specialists from federal and state regulatory agencies, industry, and nongovernmental organizations. The Russian audience consisted of federal, regional, and local government representatives, industry experts, environmental organizations, academia, and the press. The program was a series of seminars and workshops followed by interactive sessions. One of the many topics requested and discussed during both visits was disposal of drilling wastes. As expressed by the Assistant to the Director of the Sakhalin Oil and Gas Department (Arseniev, 2000) on the main ecological problems facing Sakhalin as a result of offshore drilling, “... no less serious is the problem posed by the discharge and recycling of liquid and hard residues used to drill at sea, particularly the solutions and the slime.” The solutions of ecological problems are a priority of the administration of Sakhalin Island as they rise to meet future challenges. Those challenges include balancing the development of hydrocarbon resources with rich fisheries, preservation of the environment, preservation of * Views expressed in this chapter do not necessarily represent the views of the United States Government or any of its agencies.
ELSE_TSANG_CH009.qxd
8/19/2005
80
6:34 PM
Page 80
Alaska-Sakhalin 2002 Symposium Discussion of Underground Injection
native habitats, and the conservation of natural and archeological sites. The siting, operation, and decommissioning of offshore platforms, the management of oilfield wastes, the choice of pipeline routes, and the integration of current technologies are all issues that Sakhalin must soon manage. In the near future, Sakhalin will be requiring reinjection of drilling fluids, produced water, sand, and drill cuttings. We were asked to address these issues at the seminar and did this by making presentations on Alaska’s Class I and II wells for all stakeholders, and by holding an UIC workshop, with government and industry experts, to discuss our UIC program and future steps in the development of a Sakhalin management regime.
9.2 BACKGROUND Sakhalin is the largest island in Russia and is located just north of the Japanese island of Hokkaido. Sakhalin lies between the Sea of Okhotsk on the east and the Tartar Straits on the west. There are approximately 700,000 people living on the island. Sakhalin has a rich history and is of great historic and nostalgic importance to the Russians. The great playwright Anton Chekhov toured the island extensively and wrote about his adventures in 1890. Many Russians came to know the hardship of frontier life, the visions of great wilderness, and incredible beauty of Sakhalin from his writings. Nine years later, in 1889, oil was discovered on Sakhalin. Sakhalin oil was first produced onshore in 1928, and there was evidence of oil offshore, but the Soviet Union did not have the money or the technology for marine exploration. In the 1970s, a Japanese–Russian joint venture discovered the first offshore fields, but this project was abandoned due to lower oil prices and prevailing international politics in the early 1980s. Early marine surveys indicated potential resources of at least 1000 million metric tons (mmt) of oil and 3600 billion cubic meters (bcm) of gas, far greater than the oil and gas onshore. But harsh environmental conditions, including subarctic climate, 6 months of sea ice cover, typhoons, and active seismicity, made it difficult for the USSR to produce these hydrocarbons. That task required new and expensive technologies, which Russian companies in the 1990s did not have. In 1992, after the collapse of the Soviet Union, offshore exploration began again, this time with the participation of western oil companies and Russian Joint Venture companies. Several prospective areas identified in earlier surveys were opened for tender, but two areas off the northeastern shore drew particular attention from industry—the Sakhalin I and II areas (see Fig. 9.1). As large reserves have been discovered offshore, and production has come online, many Russians worry that drilling will harm the marine environment. Russians are particularly concerned with the impacts of releases from oil spills and overboard disposal of drilling wastes. This concern led to new legislation and regulations that will require underground injection of drilling muds and cuttings, and possibly other waste by 2004, resulting in Russian interest in Alaska’s UIC program. 9.2.1 Sakhalin I Consortium Sakhalin I signed their Production Sharing Agreement (PSA) in 1996. Total recoverable reserves are estimated at 2.3 billion barrels of oil (307 million tons) and 17.1 trillion cubic feet (tcf) of natural gas (485 bcm) in three fields—Arkutun-Dagi, Chaivo, and Odoptu—all located about 20 miles off the northeast coast of Sakhalin in 150 ft of water. Oil production
ELSE_TSANG_CH009.qxd
8/19/2005
9.2 Background
6:34 PM
Page 81
81
Fig. 9.1. Location of major oil and gas deposits on the Sea of Okhotsk, offshore Sakhalin Island (Exxon Neftegaz, 2003).
will reach a plateau of about 250,000 barrels daily (33,000 tons/day). First, oil will be produced from Chayvo at the end of 2005 and from Odoptu in early 2007. Future phases of development include construction of a natural gas pipeline to Japan and development of the Arkutun-Dagi field (Exxon NefteGaz, 2003). The operator of Sakhalin I is Exxon NefteGaz, which is a consortium of India’s Oil and Natural Gas Corporation (ONGC) 20%; Rosneft 20%; ExxonMobil 30%; and Japan’s SODECO 30%. In 2000, the Chaivo 6 test well flowed at a rate of 6000 barrels (800 tons) of crude oil/day (ExxonMobil, 2000). The Sakhalin I Consortium intends to develop the project in four phases. The first phase will focus on the major oil zones in the Chayvo and Odoptu fields, with limited gas production to help meet Russian domestic demand.
ELSE_TSANG_CH009.qxd
8/19/2005
82
6:34 PM
Page 82
Alaska-Sakhalin 2002 Symposium Discussion of Underground Injection
9.2.2 Sakhalin II Consortium The Sakhalin II project agreement was signed in 1994 and consists of two fields: the PiltunAstokhskoye field and the Lunskoye field. Estimated resources at the Piltun-Astokhskoye field, located 16 km offshore in 30 m water depth, is 775 million barrels (100 million tons), with production at 120,000 barrels/day (16,000 tons/day) (SEIC, 2002). Drilling is from the mobile drilling platform, Molikpaq, with first oil production in 1999 using a floating tankerloading system until a pipeline to shore is built. Piltun-Astokhskoye will require two platforms in full production. The Lunskoye field is estimated to contain 325 million barrels (42 million tons) of oil, 11 tcf (0.3 trillion cubic meters) of gas, and is located in an area where the water is 15 m deep. Lunskoye will only need a monopod rig. The Operator is Sakhalin Energy Investment Company. Shell controls 55%, Mitsui Sakhalin Holding BV 25%, and Mitsubishi 20%. 9.3 WELL DRILLING AND OPERATION 9.3.1 Piltun-Astokhskoye Field (PA-B Platform) Drilling operations in the Piltun field will involve drilling both oil producing and waterinjection wells (SEIC, 2002). The PA-B platform has 45 well slots designed for 20–30 producing wells including gas-lift wells, and for 15–22 water-injection wells (including one spare slot). 9.3.2 Lunskoye Field (LUN-A Platform) The LUN-A platform is equipped with 32 well slots, which are all exploitable. Following drilling, the Lunskoye field will be comprised of gas producing wells, reinjection wells, and at least one producing oil rim well. Depending on the successful production of the first oil rim well, an additional 10 oil rim wells may be drilled (SEIC, 2002). 9.3.3 Drilling Program The drilling program for the PA-B platform will encompass the drilling of 16 wells during the period of October 2006 to December 2008: one well in 2006, seven wells in 2007, and eight wells in 2008. Drilling at LUN-A will commence in December 2005. It is expected that there will be 192 days of drilling during 2005 and 2006, and an additional 60 days of drilling activity in 2007 (SEIC, 2002). 9.3.4 Drilling Fluids Drilling fluids, commonly termed “muds,” serve two primary purposes, including pressure control and transport of drill cuttings to the surface. Platform PA-B has two tank systems, each with a capacity of 500 cubic meters (m3): one tank for storage of drill water for water-based drilling mud, and a second tank for storage of base oil for oil-based drilling muds. The tank systems will be equipped with solution mixing and transportation circuits that include pumps, pipelines, and balance tanks. The LUN-A platform will use the same engineering technical with an additional 400 m3 capacity storage-tank system. The platforms will also have drilling fluid additive bulk storage capacity for 300 m3 barite, 75 m3 bentonite, and 225 m3 cement.
ELSE_TSANG_CH009.qxd
8/19/2005
6:34 PM
Page 83
9.4 Water Discharge Disposal Routes
83
Other drilling fluids to be stored on the platform in drums include chemical additives (e.g., biocides, corrosion inhibiters, and anti-foaming agents), degreasers (e.g., rig wash), and lubricants (e.g., pipe dope). Drill cuttings and muds are screened onboard the platform to separate the drilling muds from cuttings. Drilling muds will be recycled. The drill cuttings, and a certain amount of waste drilling muds, will be collected in a tank and subsequently disposed (SEIC, 2002). Well design on the platforms will allow for the reinjection of drill cuttings and waste muds. Overboard discharge of drill cuttings and wastewater-based muds will occur only during the drilling of the first well for the PA-B platform, the first four wells for the LUNA platform, and for the drilling of the conductor string of each subsequent well on both platforms. Cuttings and waste muds for all other well sections will be reinjected. Oilbased muds, drilling cuttings, produced sand, and run-off/wash water contaminated with drilling mud returns will be collected in a tank, diluted, and subsequently reinjected. When water-based muds are used to drill through shale, they may form a gumbo of hydrated shale with the consistency of heavily saturated clay. The material with this gumbo consistency is difficult to slurry for reinjection. Overboard discharge may be permitted, depending on the section of the well being drilled. If the overboard discharge option of heavy saturated clays is exercised, then an equal amount of the 30 in. conductors’ cuttings will not be discharged. 9.3.5 Produced Water Treatment Platftorms will be equipped with facilities capable of treating and injecting 19,078 m3 of produced water per day at the PA-B platform, and 3180 m3 of produced water per day at the LUN-A platform. In early field life, the volumes of produced water will be small and seawater will be injected to supplement volumes. The water-injection system will incorporate equipment for seawater filtration, de-aeration, and injection (SEIC, 2002).
9.4 WATER DISCHARGE DISPOSAL ROUTES 9.4.1 Produced Water Disposal SEIC will reinject produced water for disposal and reservoir-pressure maintenance. Offshore facilities will be designed to prevent overboard discharge of produced water.
9.4.2 Drain Water Disposal Offshore Closed drain systems will be provided to allow safe disposal of hydrocarbon liquids during normal operation and maintenance of the facility. Drainage will collect here, and the separated hydrocarbon stream will be routed to a slop drum prior to returning to the process stream. Hazardous liquids captured at drains will be injected into the reservoir, and water from nonhazardous drains and storm drains (i.e., rainwater and firewater) will be treated for hydrocarbon contamination to meet effluent standards before discharge to the sea. The injection of seawater cooling water, mixed with pressure-maintenance system water, will be considered to minimize overboard discharges.
ELSE_TSANG_CH009.qxd
84
8/19/2005
6:34 PM
Page 84
Alaska-Sakhalin 2002 Symposium Discussion of Underground Injection
9.4.3 Muds and Cuttings Disposal Waste muds, cuttings, and other drilling wastes will not be discharged overboard from the platforms, with the exception of the first conductor string of each well, the first PA-B well, and the first four LUN-A wells. Oil-based muds will not be used during these specific activities to reduce environmental impacts. Drains from the drilling module will be collected into either hazardous drains or nonhazardous drain tanks as required, and then routed to the cuttings-injection system for injection into the well annulus. During Phase I of Sakhalin II operations, SEIC discharged a small amount of drilling muds and cuttings in accordance with its water-use license. Based on a 3-year monitoring program, using parameters agreed upon by the regulatory authorities prior to its initiation, no environmental impacts were detected. For Phase II, SEIC’s goal is reinjection of 100% of the drilling waste. The equipment necessary to reinject muds and cuttings is on the PA-A Molikpaq platform, and has been successfully tested. One exception to this target will be the water-based drilling muds and cuttings generated in the first well. Another exception will be the conductor string (or top-hole section) for each subsequent well at the PiltunAstokhskoye field. In addition, the water-based drilling muds and cuttings generated during the first and subsequent wells (up to a maximum of four, depending on operational and technical feasibility), and the conductor string for each subsequent well at the Lunskoye field, will not be injected; however, there will be no overboard disposal of oil-based muds. This is consistent with international practice, including those adopted by companies operating in Alaska. 9.4.4 Gray Water Disposal Living quarters wastewater such as sanitary, kitchen, and shower wastewaters, will be discharged overboard after treatment in a bioreactor.
9.5 THE UNDERGROUND INJECTION CONTROL PRESENTATION TO SAKHALIN In the United States, the U.S. Environmental Protection Agency’s (EPA, 2001) UIC Program is intended to support clean water resources by ensuring that waste fluids are safely reinjected back into the ground with zero discharge of harmful substances to the surface environment. The basic program components include a framework of regulations, authority to enforce the regulations, and a stringent permit program requiring mandatory injection well testing. Due to the harsh environmental conditions, carrying out this program in Alaska presents unique challenges. Alaska’s total production for the year 2000 was 388 million barrels (54 million tons) of oil. More than 3 trillion cubic feet (0.08 trillion cubic meters) of natural gas, and 949 million barrels of fluids, were reinjected. Class 1 injection capability is critical for North Slope oil and gas development. Although no underground sources of drinking water (USDW) are presently identified in oil extraction areas, stringent permit requirements, including annual mechanical integrity tests (MITs), ensure protection of sensitive environments. Class I injection is an important and proven component in the oilfield to maintain an integrated waste program, with zero discharge, in harsh environments with the presence of permafrost and the absence of year-round roads.
ELSE_TSANG_CH009.qxd
8/19/2005
6:34 PM
Page 85
9.5 The Underground Injection Control Presentation to Sakhalin
85
9.5.1 Regulations The statutory authority for UIC is based on the Safe Drinking Water Act (SDWA) of 1974 (as amended in 1996), codified as Title 42, Chapter 6A, Subchapter XII, Part C, of the U.S. Code, preventing underground injection that endangers, or potentially pollutes, potential underground sources of drinking water (42 USC, 1996). Federal UIC roles include setting national standards and program requirements, providing state assistance, support, and sound science, plus approving and overseeing state-delegated programs. The EPA also directly implements the program in 17 states and most federal Indian tribal lands. The roles of states and tribes in UIC include submitting applications and assuming program primacy authority for all or part of the UIC program. Currently, 33 states have been granted primacy. State and tribal programs must meet or exceed federal standards, and must demonstrate adequate enforcement capability. Well operators have the role of managing wells so they are in compliance with all requirements. In the UIC arena, public interest groups have the role of public involvement. In Alaska, EPA implements the Class I and Class V program, oversees the state, and provides grant money to Alaska Oil and Gas Conservation Commission, a state agency that directly manages the Class II (enhanced oil recovery, storage, and disposal) UIC well program. The UIC program is codified in the U.S. Code of Federal Regulations (CFR) as follows: 40 CFR Part 124 for Public Participation, 40 CFR Part 144 for Permitting and Program, 40 CFR Part 145 for Requirements and Procedures for State Program Approval, 40 CFR Part 146 for Technical Criteria and Standards, 40 CFR Part 147 for state-administered program requirements, and 40 CFR Part 148 regarding Hazardous Waste Injection Restrictions (40 CFR, 1996). Legally, the laws protect USDWs, which are defined as follows: an aquifer or portion of an aquifer which supplies any public water system, or contains a quantity of groundwater sufficient to supply a public water system, and either contains fewer than 10,000 mg/L total dissolved solids (TDS) or is not an exempted aquifer. An aquifer may be exempted from UIC regulations when that aquifer contains TDS greater than 3000 mg/L, but less than 10,000 mg/L, and it is not expected to supply a public water system; or is not a current source of drinking water and is not expected to ever serve as a source of drinking water because recovery is impracticable (costs too much); or is so contaminated that it is economically and technically impracticable to be fit for drinking water. The aquifer-exemption process begins with the operator who proposes aquifer exemptions as part of the permit application for a Class I or Class II injection well. The proposal is reviewed and recommended by the state UIC program (Alaska Oil and Gas Conservation Commission, AOGCC, 2002), and the EPA must approve aquifer exemptions. Public health risks are of concern, as the nature of injected fluids poses a risk to groundwater quality and public health if managed improperly. In addition, the deep Class I and Class II wells must be properly sited and constructed to avoid contaminating underground sources of groundwater. Deep injection is often the best waste disposal alternative for remote sites; especially sites not accessible by road or located offshore. Six Class I (“class one”) injection wells are permitted to legally accept most wastes generated at remote production sites in the North Slope. All Class I wells require permits approved by an EPA regional office. The Class I permit application may take 6 months for EPA to review and approve, because a 30-day public comment period is required, and applications may be in excess of 100 pages in length. Deep-well control measures identified in the permit include site construction requirements, operation and maintenance requirements, monitoring and testing, reporting and record keeping, and permitting and closure requirements. Class I permit application reports include project description; waste sources, characteristics, and alternative handling and disposal plans; geologic setting and operating characteristics; subsurface aquifers identification; an
ELSE_TSANG_CH009.qxd
8/19/2005
86
6:34 PM
Page 86
Alaska-Sakhalin 2002 Symposium Discussion of Underground Injection
area-of-review study; performance standards and compliance; well construction, testing, monitoring, and closure; waste minimization, segregation, and analysis; existing permits and fiscal responsibility; and summary and conclusions. This assures that a complete Class I permit application defines quality and pressure of groundwater with respect to depth; injection properties of naturally saline aquifers; thickness and area extent of potential confining zones; presence of faults, fractures, boreholes, or other possible fluid conduits; and proof of financial responsibility. The responsibilities of the operator are that they do not pollute; attain a permit, observe, measure, and record injection parameters, and any other permit parameters; maintain mechanical integrity and periodically demonstrate MIT; report to UIC Director as required including monitoring parameters, loss of MIT within 24 hours, and any noncompliance. Injection parameters and injectate must be monitored. All injected fluids must be monitored, and requirements vary by well type. Monitoring parameters include the injection rate—commonly 5 barrels per minute (BPM) in Class I wells and up to about 25 BPM in Class II wells. The injection pressure is commonly up to 3000 pounds per square inch (psi) at depths of 6000–8000 ft. Monthly and cumulative injected volumes vary, based on the activities in the particular field. The offshore Northstar injection well, for example, injects up to 16,000 barrels/day. Class I well permits require that data on annulus pressure and volume; waste characteristics such as density, and pH, are recorded and submitted to EPA. Currently, the six “Class I-NH” disposal injection wells in Alaska may legally inject industrial and municipal waste beneath the lowermost formation containing a USDW. EPA’s “Exploration and Production Exemption” policy allows for injection of waste that shows hazardous characteristics into Class I wells, provided that the fluids came from down hole. Although Sakhalin has no permafrost, on the Alaska North Slope, permafrost is 800–2000 ft thick and fresh waters are commonly limited to the surface. In the North Slope areas where no USDWs exist, with an EPA variance, injection pressures above formation-fracture pressures are allowed. This is significant because this applies to all six Class I wells which, with grind-and-inject systems, allows for both solids and liquids to be disposed in formations commonly located near the oil-bearing formation, keeping the surface impacts to the environment at a minimum. Class I injection wells require a rigorous permitting process, continuous annulus monitoring, frequent reporting (quarterly reports), and regular MITs every 1–5 years. Active Class I facilities include BPXA’s Pad 3 Facility (Prudhoe Bay Unit), which includes three wells that receive predominantly fluids, BPXA’s Badami Facility that receives solids or liquids, ConocoPhillips’ Colville River Field Facility that receives mostly liquids and some solids, and BPXA’s Northstar Facility that receives solids and liquids. Class II wells may dispose of fluids (mostly salt water) and solids that were brought to the surface in connection with oil or natural gas production. Class II wells may also be used for enhanced oil or gas recovery or to store liquid hydrocarbons. Class II wells usually exhibit three layers of protection; must pass a MIT every 5 years, and may be permitted in an areawide permit. Regarding inspections, EPA is authorized to inspect any facility subject to the UIC program under Section 1445(b) of SDWA (42 USC, 1996). The types of inspections commonly include preoperational and prepermitting inspections; MIT; operational and compliance inspections, plugging and abandonment; and inspection to attain general information. 9.5.2 Enforcement Enforcement serves as an important deterrent where statutory maximums are set at $27,500 per day per violation. The EPA penalty policy distinguishes between severities of violations, and the emergency powers authority under Section 1431 of the SDWA allows EPA to take
ELSE_TSANG_CH009.qxd
8/19/2005
6:34 PM
Page 87
9.5 The Underground Injection Control Presentation to Sakhalin
87
rapid action if necessary to protect aquifers. An example of an enforcement action in Alaska is a $20 million Civil/Criminal Settlement Enforcement action in 1999 (42 USC, 1996). The violation settlement was reached with EPA for injection of hazardous waste (from 1992 to 1995) into a BP Endicott oilfield Class II injection well. EPA levied a $6 million fine against BPXA and required BPXA to spend $14 million for a Supplemental Environmental Project that was an “environmental enhancement system.” In addition, BPXA is required to self-disclose noncompliance during a 5-year probation period (1999–2004). 9.5.3 Well Integrity Testing The definition of mechanical integrity includes no significant leak in casing, tubing, or packer, and no significant fluid movement into the USDW through vertical channels adjacent to an injection-well bore. Common pathways of contamination are caused by external and internal failure. Conversely, prevention is based on mechanical integrity that includes an absence of leaks in the well and an absence of upward flow outside of casing; and an areaof-review, which is a permit study that locates potential pathways of upward migration. An example of a disposal injection well failure occurred in 1997 at a Class II injection well that spilled over 18,000 barrels of brine to the surface in 4 days. From this example, much was learned and applied to future well designs. The well had been converted into a Class II injection well from a test well and it had a history of construction problems. A poor choice became an expensive problem. The casing had been patched. The cement squeeze job had failed, and other wells without continuous cement in similar intervals were located near the disposal well and corrected and so they served as conduits to the surface for the injected brine. Class I MIT procedures include internal MIT standard annulus pressure tests (SAPT), a 30-minute pressure test that shows less than 10% loss and stabilizing tendency, to confirm the absence of leaks in casing, tubing, or packer. Class I MIT procedures also include an external well test to verify a lack of fluid movement behind the casing. Also, the cement is evaluated for cement continuity, commonly with an ultrasonic imaging tool. The MIT methods commonly used for external mechanical integrity on the North Slope include the borax pulse neutron tracer, the oxygen activation tracer, and the radioactive tracer (RAT). However, due to the challenges posed by the management of both radioactive source material and radioactive waste, the RAT is less popular. The borax pulse neutron log (borax PNL) uses 7 lb of sodium tetraborate pentahydrate mixed with each barrel of hot seawater to make the tracer injectate. The well is logged and then the solution is injected. A second logging is run and compared with the first run prior to solution injection. When evaluating Pad 3 wells, temperature logs are run after the borax is injected. The oxygen-activation tracer, or water-flow log, is based on the principle that a pulsed neutron log activates the oxygen in formation water, forming an unstable nitrogen isotope with a 7.35 s half-life. The burst is followed by a measurement of decay of the nitrogen back to oxygen. The movement of fluid is detected by excessive counts observed on the detectors. At the offshore Northstar injection well, and at ConocoPhillips’ disposal well, temperature logs are run and caliper logs are used to observe for tubular corrosion. Radioactive tracer logs utilize a radioactive Iodine-131 with an 8.5 day half-life. The radioactive material calls for additional logistics and requires special handling. Extra efforts and challenges include the use of specialized service companies to handle the radioactive source and radioactive waste disposal. Historically, the RAT was used at the Badami injection well with temperature logs. Recently, the borax PNL has been used to attain the same results without the challenges of managing radioactive material (Syed and Cutler, 2001; Talib Syed Associates, 2002).
ELSE_TSANG_CH009.qxd
88
8/19/2005
6:34 PM
Page 88
Alaska-Sakhalin 2002 Symposium Discussion of Underground Injection
9.5.4 Discussion of Specific Wells All Class I wells on the North Slope are below permafrost. “Pad 3” injection wells were originally drilled as permafrost test wells in the 1970s. The base of the permafrost was discovered to be at 1835 ft measured depth (md) below surface. The Pad-3 injection zone is into the Sagavanirktok Formation, and perforations are at 1978–2093 ft md. Common injection rates are 3500–4500 barrels/month/well at 900–1100 psi, where the maximum permitted wellhead-injection pressure (WHIP) is 1400 psi. The BPXA’s Badami Class I well was permitted in August 1997, and the permit is valid for 10 years. As Badami is an isolated oilfield with no all-weather roads, a small grind-andinject system is utilized at the facility to allow for solids injection. The injection zone is into the Ugnu Formation, and perforations are at 8390–8420 ft md below the surface. The maximum permitted WHIP is 3000 psi, and as drilling is no longer ongoing, the average injection rates are about 600 barrels/day or less at injection pressures of 1350–1450 psi. The ConocoPhillips’ Alpine Colville River Field Class I injection-disposal well was permitted in February 1999, and the permit is valid for 10 years. The well injects into the Sag River Formation, and the lower injection zone is into the Ivishak Formation. Perforations are at 9459–10,047 ft md below the surface. The well is permitted for WHIP to 3200 psi. Monthly injection volumes have seen 15,000 barrels at 1450–1800 psi. A portable grindand-inject system is on site. However, due to the formation-limiting characteristics, the solids volumes have been kept to a minimum and utilized only as a last resort to assure the well remains open for fluids disposal. The BPXA’s Northstar Class I well was drilled offshore in January 2001 as the first joint state/federal offshore arctic field. The base of permafrost is at 1512 ft md below the surface, and the upper injection zone is at a depth of 5007 ft md. The lower injection zone is an openhole design located at 8029–8246 ft md, and the maximum permitted WHIP is 3000 psi. Monthly average injection volumes range from 28,000 to 49,000 gallons/day as drilling continues under way. The permit is written for a second injection well that is scheduled for construction before 2004. Over the total 20-year life expectancy of the Northstar well, the total estimated volume of fluid disposal is 120 million barrels, of which solids will constitute a major percentage of the waste stream during the first 3–5 years, and eventually the waste stream will be predominantly fluid requiring little grinding. Estimates are that grinding will be needed for 140,000 barrels, predominantly consisting of rock cuttings, fracture sand, vessel sludge, and sand. Little or no grinding will be needed for the following fluids and solids including 600,000 barrels of camp sewage and gray water, 400,000 barrels of well work-over fluids, 360,000 barrels of drilling muds and fluids, and 40,000 barrels of industrial nonhazardous waste fluids. It is anticipated that no grinding is needed for the following fluids that represent 98% of all injectate for the life of the well: produced water (oilreservoir brine estimated to be 118,500,000 barrels) and storm water (onsite rain and snowmelt waters estimated at 182,500 barrels). Most fracture-slurry injection at the Northstar well is expected in the first 2 years. The average fracture-slurry injection rate is about 28,000 barrels/month, and the maximum fracture-slurry injection rate is 65,000 barrels/month from backlogs including plant upsets, scheduled shutdowns, well treatments, and workovers (redrilling). Furthermore, the Northstar well produced-water injection rates are projected to increase after the first 2 years. Produced water is expected to reach 16,000 barrels/day after the first few years and remain at that level for the final 18 years. In closing, Class I injection capability is critical for North Slope oil and gas development. Integrated waste management is necessary to attain zero surface-discharge targets in the
ELSE_TSANG_CH009.qxd
8/19/2005
6:34 PM
Page 89
9.5 The Underground Injection Control Presentation to Sakhalin
89
harsh environment with permafrost and the absence of year-round roads. Although no reported USDWs are present at Class I locations, permit requirements, including annual MIT, are stringent to ensure protection of the arctic tundra environment. 9.5.5 Russian Questions During the seminar, we addressed the following questions that are listed below. Questions posed by the general symposium participants: ● How many tons of solids are injected underground? ● Why is there injection into open holes? ● How many production wells and injection wells are there? ● What size are the solids (cuttings) ground into? ● What percentage of waste is injected and what percentage is managed in other ways? ● You have 1000 wells and they are tested annually. That’s a lot of work. Does EPA pay for this? ● Besides environmental issues, how are legal and social impacts addressed? ● You mentioned several methods for testing. What about environmental controls? ● Please elaborate on materials that you can’t inject. ● What about the disposal of municipal wastes? ● Who does the geological study before the permit is issued? ● Are jail sentences a possibility as a result of the sanctions and enforcement regime? ● According to what you have said, everything that is being done is to protect the environment—is there anyone who actually monitors this? ● Can offshore drilling cause earthquakes? ● Can offshore drilling affect any kind of underground processes? ● Can you provide a list of substances that can be disposed of in Class II wells? ● Suppose you have waste—how do you choose which way to dispose of it? Does the grinding and injection facility service the entire North Slope? Industry questions for the symposium workshop (Exxon-Neftegaz, Sakhalin Energy): ● Please explain to us, briefly and more clearly, the difference between Class I, Class II, and Class III wells. ● What information is obtained on the field geology of an injection well’s location? ● What are limitations of the injection rate? ● What is the purpose of the injection well? What are other reasons? ● Why are the injection wells of Class I supervised/controlled by EPA, while the injection wells of Class II are under supervision of AOGCC? ● What are the additional EPA requirements for issuing a special permit for carrying out injection under pressure exceeding the formation-fracturing pressure value (besides the condition of a USDW source’s absence)? What evaluation techniques are used for determination of the maximum height of the fracturing fault? ● What is the minimal validity term (period) of the issued permit for drilling/construction and operation of the injection well? ● What are the reporting requirements regarding content and volume of information and how is it presented? What is the reporting schedule or period? ● Are there any royalties to be paid by Subsurface Resources Users in case they are used for field operational and waste fluids/cuttings injection into the formations?
ELSE_TSANG_CH009.qxd
8/19/2005
90 ●
●
●
●
6:34 PM
Page 90
Alaska-Sakhalin 2002 Symposium Discussion of Underground Injection
MIT Monitoring: Is there any difference in the inner and outer MIT techniques between normal injection wells (w/o) pressures exceeding the formation-fracturing pressure) and injection wells with pressures above the formation-fracturing pressure values? As we have understood, waste fluids and cuttings injection into the annular space can be used only for exploration/appraisal-well drilling, and the injection procedure anticipates limitations to the injection volumes: not more than 35,000 barrels per one well of clean waste fluids without cuttings. What is the method of solid waste disposal in this case (including cuttings)? In case waste-fluid injection is carried out with pressure values in excess of the formation-fracturing pressure, what methods of fracture-fault monitoring are used for the whole well-drilling license area? What monitoring techniques are used for fracture-fault propagation control? What methods and techniques were used for obtaining the Young modulus data (formation rocks strength/elasticity data) from the injection target layer, localizing and underlying (basement) clay rocks? (Whether core examination or logging data interpretation techniques were used for this purpose?)
9.6 CONCLUSIONS In conclusion, large reserves in Sakhalin have been discovered offshore, and production has come online. Russians are concerned about the protection of the marine environment from drilling wastes, particularly the impacts of releases from oil spills and overboard disposal of drilling wastes. In response to that concern, new legislation requiring underground injection of drilling muds and cuttings, and possibly other wastes, will be required by 2004. The sharing of Alaska’s UIC program experiences with the people of Sakhalin has been beneficial to all participants from the United States and Russia. Future exchanges would continue to be fruitful.
ACKNOWLEDGMENTS We thank Sakhalin II Symposium participants for contributions, support, and review.
REFERENCES Alaska Oil and Gas Conservation Commission (AOGCC), 2002. Alaska UIC Prog. Rev. Arseniev, N.V., 2000. Speech by N.V. Arseniev (Feb. 2000). Sakhalin Oil and Gas Department, to the Pacific Rim Construction, Oil and Mining (PACCOM) Exposition, Anchorage, AL. ExxonMobil, 2000. Exxon Neftegaz appraises oil accumulation on Sakhalin I, News Release (9/28/2000). Exxon NefteGaz, 2003. Sakhalin I Project on Exxon NefteGaz (http://www.sakhalin1.com/). Sakhalin Energy Investment Company (SEIC), 2002. Environmental Impact Assessment, SEIC (http://www.sakhalinenergy.com/). Syed and Cutler, 2001. Alaska Underground Injection Control Class I Program 2001, presented at Ground Water Protection Council.
ELSE_TSANG_CH009.qxd
References
8/19/2005
6:34 PM
Page 91
91
Talib Syed Associates, 2002. Alaska UIC Program Class I Report. U.S. Environmental Protection Agency, 2001. Northstar Permit, BPXA Waste Analysis Plan. 40 CFR, 1996. Underground Injection Control Program, Chapter 1. Environmental Protection Agency, Parts 124, 144, 145, 146, 147, and 148. 42 USC, 1996. Safe Drinking Water Act, §§, 1431, 1445[b], Chapter 6A, Subchapter XII, Part C.
This page intentionally left blank
92
ELSE_TSANG_CH010.qxd
9/15/2005
3:45 PM
Page 93
Chapter 10
A PROBABILISTIC RISK ASSESSMENT OF CLASS I HAZARDOUS WASTE INJECTION WELLS W.R. Rish Hull and Associates, Inc., Dublin, OH, USA
10.1 INTRODUCTION The disposal of large volumes of industrial and municipal wastes has been a source of ongoing concern throughout the latter half of the twentieth century. Over the past 20 years, increasingly stringent waste-disposal regulations have improved environmental quality while limiting disposal options and raising costs. Because waste reduction techniques are equally subject to the law of diminishing returns, some waste will always result from human activities, and disposal issues will remain to be addressed. From a societal viewpoint, the ideal disposal method should be (virtually) infinite, cheap, permanent, and result in no human or ecological exposures in the foreseeable future. Most current regulated methods of disposal, for example, landfills or incineration, fail in one or more of these areas. Only deepwell injection appears to satisfy all four requirements; however, the environmental risks associated with Class IH disposal technology remains a source of controversy. Approximately 150 underground injection wells exist in the United States that are categorized by the United States Environmental Protection Agency (EPA) as Class IH (U.S. Environmental Protection Agency, 1996) wells that inject hazardous liquid waste. The majority of Class IH wells are located in the Great Lakes Region and the Gulf States, due to the favorable geology in these regions. Over half of these wells are located in Texas and Louisiana, and almost 90% are in EPA Regions V and VI (U.S. Environmental Protection Agency, 1996). Based on figures from the EPA’s TRI (U.S. Environmental Protection Agency, 1996), the volume of hazardous waste disposed of through Class IH deep-well injection is about 220 million pounds. This quantity is somewhat deceptive, since the practice of deep-well injection involves dilution of the waste with large amounts of water before it is pumped into the subsurface. Industries that practice deep-well injection are sometimes singled out as major sources of pollutant releases to the environment. Since the intent of deep-well injection is the permanent isolation of waste from the biosphere, it is unclear if the use of deep-well injection can be properly considered a release to the environment. While problems resulting from deep-well injection have occurred, these incidents took place in the past, and the conditions that caused them do not occur under current regulations and practices. In 1980, the EPA promulgated regulations governing all injection wells, including those injecting hazardous waste (53 FR 28131). In 1988, EPA passed additional regulations requiring operators of Class IH wells to demonstrate that no migration of the waste constituents will occur from the injection zone while the waste remains hazardous (or
ELSE_TSANG_CH010.qxd
94
9/15/2005
3:45 PM
Page 94
A Probabilistic Risk Assessment of Class I Hazardous Waste Injection Wells
for 10,000 years) (40 CFR Parts 146 and 148). Waste isolation is accomplished by a combination of: ● The application of strict siting criteria. ● The presence of multiple redundant engineered and geological barriers. ● Practices to ensure chemical compatibility of waste with geology. ● Operating restrictions and preventive maintenance during active injection operations. ● Continual monitoring and testing of performance and confinement integrity. ● The presence of alarms and a full-time operator. These factors combine to assure that waste will be prevented from entering the accessible environment, i.e., that portion of the environment where human or ecological exposure can occur. In the absence of such exposure, no risk to human health or welfare exists. Studies published by both industry and the EPA in the past 10 years have concluded that the current practice of deep-well injection is both safe and effective, and poses an acceptably low risk to the environment (CH2M Hill, 1986a; Clark, 1994; Department of Energy and Natural Resources et al., 1989; Underground Injection Practices Council, 1987; U.S. Environmental Protection Agency, 1985, 1989, 1991; Ward et al., 1987). Nonetheless, various advocacy groups have challenged the effectiveness of deep-well-injection regulations, and have opposed the practice on principle (Gordon and Bloom, 1985; MacLean and Puchalsky, 1994; Sierra Club Legal Defense Fund, 1989). Studies purporting to examine the risks from deep-well injection take as their starting point the assumption that release of waste from confinement to a drinking water aquifer has already occurred and then model the transport time to a receptor well and the dose received by that receptor (The Cadmus Group, Inc., 1995). None of these studies to date has assessed the probability of the release even occurring in the first place. Since the primary risk associated with deep-well injection is that isolation from the accessible environment might fail, this probability must be examined before drawing any conclusions regarding health or environmental risks from such a release. The purpose of this paper is to examine the risk from such a failure of isolation, and to provide an objective and quantitative analysis of the risk of waste isolation loss from Class IH underground injection wells that will allow meaningful identification and comparison of waste isolation subsystems as contributors to that risk. Areas of uncertainty will be identified and quantified as to their possible contribution and importance to the risk estimates for the purposes of collecting additional data, identifying new sources of data, or stimulating new research to reduce these uncertainties. In doing so, we hope to provide all stakeholders with the type of rigorous scientific support needed to make appropriate decisions regarding deep-well injection.
10.2 BACKGROUND A review of available studies on Class I injection well failures over the past 20 years was conducted. These studies originated from a variety of sources, including industry studies, peer-reviewed studies, trade association reports, as well as reports from advocacy groups. Case studies and accident reports involving injection wells were reviewed as well. The relevant regulations were also carefully reviewed to determine the ways that regulatory requirements and restrictions affect siting, design, construction, and operations. Numerous discussions and interviews were held with injection well operators and regulators. Based on this information, the critical factors to maintaining waste isolation were identified.
ELSE_TSANG_CH010.qxd
9/15/2005
10.2 Background
3:45 PM
Page 95
95
An important concept that appears throughout injection well risk studies and regulations is that of USDW. Releases from injection wells into the accessible environment (i.e., that portion of the environment where human or ecological exposures can occur) may occur either at the ground surface, or at subsurface groundwater zones that have potential human use. These subsurface groundwater zones are typically called USDWs in studies and regulations. While surface releases are readily observed and remedied, and as such do not result in chronic exposures and have not been included in risk assessments, potential releases to USDWs are the primary focus of risk assessments and regulations. Accordingly, this assessment assumes the relevant release point to be the lowermost USDW (i.e., closest to the injection zone). In general, previous studies fall into four categories. The first category is case studies of injection-well failures that have resulted in releases (CH2M Hill, 1986b; Clark, 1987; Engineering Enterprises, Inc. et al., 1986; Ken E. Davis Associates, 1986; Paque, 1986; Underground Resource Management, Inc., 1984). There are relatively few cases of this sort and none involving a release from a Class I well to a USDW since the EPA regulations took effect in 1980 (U.S. Environmental Protection Agency, 1985, 1991). These historical incidents are confined without exception to issues of well siting, design, and operation practices that are no longer allowed under today’s regulations, nor do they exist in today’s population of Class I wells (Clark, 1994; Engineering Enterprises Inc. et al., 1986; Ken E. Davis Associates, 1986; Paque, 1986; Underground Resource Management, Inc., 1984; U.S. Environmental Protection Agency, 1991). The second category is geologic fate and transport modeling studies (Buss et al., 1984; Davis, 1987; Don L. Warner, Inc. and Engineering Enterprises, Inc., 1984; Goolsby, 1972; Meritt, 1984; Miller et al., 1986; Morganwalp and Smith, 1988; Scrivner et al., 1986; U.S. Environmental Protection Agency, 1990a, 1990b; Ward et al., 1987). These studies assume a release from an injection well, and model the fate and transport of contaminants as they migrate through the typical geologic formations associated with injection wells. These include modeling efforts performed for the “no-migration petition” required for an operating permit. In general, such studies demonstrate that the proper selection of the geologic formation creates an effective means to achieve waste isolation. While such studies can provide useful information on geologic factors important for maintaining waste isolation, and on the potential for failure of geologic barriers, they assume that a release has already occurred and do not account for waste isolation provided by engineered barriers of the well system. These studies can help with understanding mechanisms and the likelihood of failure of geologic formation as one component of the loss of waste isolation, and can help in developing estimates of release volumes and concentrations to USDWs. The third category is properly characterized as exposure studies (The Cadmus Group, Inc., 1995). One study of this type was found. In this study, it was assumed that a release occurred from the injection well to the USDW. The transport of this release into the USDW aquifer was modeled to a point of withdrawal for potable use. As with other modeling studies, a release was assumed without providing any information on how the release occurred and the probability of that release mechanism. Additionally, such studies do not take into account the effect of the containment or attenuation factors posed by geologic features (e.g., layers of low-permeability rock) between the point of release and the USDW. The final category is regulatory reviews and comparative risk studies. A 1989 EPA comparative risk evaluation of waste management alternatives by experts in the field concluded that deep-well injection posed among the lowest environmental risks on a relative scale
ELSE_TSANG_CH010.qxd
96
9/15/2005
3:45 PM
Page 96
A Probabilistic Risk Assessment of Class I Hazardous Waste Injection Wells
(U.S. Environmental Protection Agency, 1989). A 1991 EPA analysis of their restrictions on Class IH wells concluded that since 1980, Class IH wells are safer than virtually all other waste disposal practices (U.S. Environmental Protection Agency, 1991). EPA studied over 500 Class I wells in operation from 1988 to 1991 and found no failures known to have affected a USDW. In response to a 1992 House of Representatives subcommittee inquiry, EPA (U.S. Environmental Protection Agency, 1993) provided state-by-state summaries of reported Class I well failure incidents between 1988 and 1992. This was defined as a breakdown or operational failure of components of the well system, whether waste isolation loss occurred or not. Although component failures were reported during the survey period, no waste isolation failure occurred, and no waste from a Class I injection well reached a USDW. While these studies indicate the waste isolation effectiveness of current injection practices, they do not quantitatively address future risk. In summary, no studies were identified that provide full quantitative characterization of the risk of Class I hazardous waste injection wells. Some describe release incidents for well systems that cannot and do not exist under today’s regulations. Others characterize only a portion of the risk, for example, estimating exposures that might occur after presuming a release (often by mechanisms that have never occurred). Others demonstrate that releases have not occurred under current practices, but do not characterize the likelihood that releases might occur in the future. To properly assess the environmental risks posed by Class I injection wells, it is critical that the probability of loss of waste isolation be quantitatively assessed. Waste volumes and concentrations corresponding to realistic release scenarios should be included in the assessment.
10.3 METHODOLOGY To quantitatively evaluate environmental risks posed by Class IH well injection, it was necessary to develop a detailed characterization of how the siting, construction, design, operation, testing, and maintenance of a Class IH well system function as a whole to ensure waste isolation (Buttram, 1986; CH2M Hill, 1986a; Underground Injection Practices Council, 1986; SCS Engineers, 1985; Warner and Lehr, 1977). The critical elements of this system that are important in maintaining waste isolation are singled out for special attention. Inherent in this approach is a systematic identification and depiction of events and conditions that could result in loss of waste isolation. This information was gathered from historical records on well failure events, and obtained from interviews with injection well construction, maintenance, and testing practitioner; operators of injection wells; and the agencies that regulate them. From this information, a comprehensive set of scenarios depicting the ways in which a typical Class IH injection well system could fail to isolate waste was developed. The probability of waste isolation loss in each of these scenarios was then quantified. Uncertainties in the analysis were given explicit quantitative treatment using Monte Carlo analysis. More specifically, the techniques of probabilistic risk assessment (PRA) were employed. PRA is a generally accepted approach for analyzing risks that arise through failure of engineered systems. In this case, PRA was used to identify sequences of events by which waste isolation could fail and result in waste reaching the lowermost USDW, and to characterize the probabilities of these event sequences. The results quantitatively and probabilistically demonstrate the degree of certainty that waste injected in this manner will effectively remain isolated and pose no future risk. The outcome of interest to this study was that the loss of
ELSE_TSANG_CH010.qxd
9/15/2005
3:45 PM
Page 97
10.4 Class IH Injection Well System Definition
97
waste isolation by release to the lowermost USDW could be due to any cause. Factors considered included: ● Errors in site selection or characterization, such as inappropriate or incompatible geology, unidentified abandoned wells, undetected geologic faults, or incorrect characterization of waste migration potential. ● Geologic or engineered system failures, such as seismic fracturing of confining zones, tubing, or casing breaches, annulus fluid pressure loss, or alarm failures. ● Operator errors, such as failure to respond to alarms, failure to detect leaks during testing, overpressurizing, or injecting incompatible waste. ● Other possible human errors, such as inadvertent extraction of waste in the future. The following steps were taken, and detailed discussion of each follows: 1. The Class IH well system, individual components, and conditions on which the PRA is based were defined. 2. FMEA was performed with the assistance of injection well experts. 3. Based on FMEA results, event and fault trees were developed, depicting the sequence of events that must occur for waste isolation to be lost. 4. Based on historical or expert information, probability distributions characterizing the uncertainty in the frequency of occurrence of the various failures and other events were developed. 5. Boolean logic and Monte Carlo analysis were used to combine the frequencies of independent and dependent events as depicted in the event and fault trees to estimate the overall probability of waste isolation loss for a Class IH well.
10.4 CLASS IH INJECTION WELL SYSTEM DEFINITION In order to quantitatively assess the risk of loss of waste isolation from Class IH injection wells, the injection well system must be defined at a high enough level of detail so that specific event sequences can be identified and their frequencies quantified. At the same time, the system definition must not be so unique that its methodologies and conclusions cannot be generalized to the population of Class IH wells as a whole. The Class IH well system definition used for this study was based on the minimal design and operation features allowed under current regulations; this ensures the broadest applicability of this study’s results and conclusions. The regulatory system is sufficiently effective to eliminate the possibility of any Class IH injection wells that do not at least meet the system definition. This conclusion was verified by discussions with state and EPA officials, a review of the current EPA injection well database (U.S. Environmental Protection Agency, 1996), and a random survey of Class I injection well operators of about 20% of the currently operating Class IH wells (Woodward Clyde Consultants, 1995). It was nonetheless appropriate to evaluate the possible failure of certain elements of the regulatory process that influence the effectiveness of waste isolation; this was done (e.g., the possibility that an unplugged well in the area is unaccounted for in the site review was included in the study). The design and operation features of the system analyzed are listed in Table 10.1, and a diagram of the system is shown in Figure 10.1. As a standard Class IH injection well, the system is assumed to comply with the requirements of the Code of Federal Regulations, Chapter 40, Parts 146 and 148, and Part 267 (Subpart G). The salient features of these requirements with respect to waste isolation are listed in Table 10.1. It is assumed that the
ELSE_TSANG_CH010.qxd
9/15/2005
98
3:45 PM
Page 98
A Probabilistic Risk Assessment of Class I Hazardous Waste Injection Wells Table 10.1. Class IH well system definitions—design and operating features
Waste isolation element
Design or operating feature
Applicable regulation Site selection and characterization
Complies with 40 CFR 146 Subpart G Area of Review: 2-mile radius. “No-migration petition” for injection of restricted wastes. Two confining layers between the injection zone and the lowermost USDW Surface casing set below lowermost USDW. Casing completed with continuous cement. Liquid-based annulus pressure barrier Equipped with auto-alarm and a full-time operator. Annual Radioactive Tracer survey or OA log for fluid movement temperature, and noise logs once every 5 years
Geologic barriers
Engineered barriers
Testing, monitoring, and inspection
well operator has prepared a no-migration petition, which is required to receive a permit to inject restricted wastes. The no-migration petition results in a marked increase in site and system scrutiny by both the industry and the regulators. The operator must demonstrate through modeling that no migration of the waste will occur from the injection zone while the waste remains hazardous (or for 10,000 years). Petitions such as the no-migration petition extensively document the local geology and faults, the well design, the operation and maintenance procedures, comprehensive local well surveys, and fate and transport through mathematical modeling. In the process of characterizing the proposed injection site, an area of review (AOR), extending for a two-mile radius around the site, must be investigated. The impact of these extensive analyses and investigations need to be considered in assessing the probability of release. The geologic features of the system analyzed are depicted in Figure 10.1. The injection zone is the permeable subsurface rock that receives the waste. Class I injection well depths range from 1700 to 9500 ft nationwide (U.S. Environmental Protection Agency, 1996). Typically, the USDW and injection zone are separated by several thousand feet (U.S. Environmental Protection Agency, 1996). The injection zone is required to be separated from the USDW by at least two confining zones consisting of dense rock or other geologic formations impermeable to fluid migration. For this assessment, it was assumed that only two confining zones exist. In actual practice, Class I injection wells have more than two confining layers (U.S. Environmental Protection Agency, 1996), which are separated by nonpotable water-bearing zones called “buffer zones.” Studies have shown that if waste fluid were to migrate through a confining zone, there would be significant dilution in each successive buffer (Don L. Warner, Inc. and Engineering Enterprises, Inc., 1984; U.S. Environmental Protection Agency, 1990). This phenomenon has not been accounted for in exposure assessments to date (The Cadmus Group, Inc., 1995), which generally assume that the waste inventory is released directly to a USDW. Injection wells are constructed by extending concentric pipes or casings down the drilled well boring. Corrosion-resistant materials such as steel alloy or fiberglass are used in the
ELSE_TSANG_CH010.qxd
9/15/2005
3:45 PM
Page 99
10.4 Class IH Injection Well System Definition
99
Fig. 10.1. Simplified Class I injection well system assumed for PRA.
casings. The upper and outermost casing (Fig. 10.1) is called the surface casing, and is required by regulation (Table 10.1) to extend below the base of the lowermost USDW. As shown in Figure 10.1, the surface casing might not extend into the uppermost confining zone. This could result in a section of the well without surface casing to pass through an area of non-confining rock, below the lowermost USDW, but above the confining zones (see Location A in Fig. 10.1). This area is important in the PRA, because it is the location with the least number of barriers to loss of waste isolation. Within the surface casing is the long string casing, which extends to the injection zone. Chemically resistant cement or epoxy resin is used to fill the borehole space outside the surface casing, between the surface and long string casings; and the borehole space outside
ELSE_TSANG_CH010.qxd
100
9/15/2005
3:45 PM
Page 100
A Probabilistic Risk Assessment of Class I Hazardous Waste Injection Wells
the long string casing, from top to bottom. These casings were assumed to be completed with continuous cement (Table 10.1); this effectively binds the casings together and seals the well boring along its entire length, creating a single unit. Nonetheless, in this conservative assessment, the cement was considered a barrier for vertical but not horizontal fluid migration. A smaller steel or fiberglass pipe, the injection tube, extends the length of the casings through a lower seal (the packer) into the injection zone. Waste pumped from above flows into and is forced out of the portion of the borehole that extends into the injection zone. This is known as the injection interval, and may be uncased or fitted with a perforated section to prevent loose material from entering and potentially clogging the borehole or injection tube. The space between the long string casing and the injection tube (the annulus) is sealed at the surface by the wellhead and at the base by the packer, and filled with a noncorrosive fluid under positive pressure in excess of the injection tube pressure. In Class IH wells, the annulus fluid is required to function as an additional pressure barrier to prevent waste fluid from leaking through the injection tube or the packer. Measurement of the fluid pressure and volume within the annulus is used to monitor the mechanical integrity of the injection tube, long string casing, and packer. An operating Class IH injection well system incorporates the redundancy of safety systems that typically characterize safe engineering design. The long string casing is continuously cemented from top to bottom. Along with the annulus fluid pressure, the casing is a barrier to an injection tube or packer leak, and the cement provides a barrier to vertical migration of any fluid that would escape along the outside of the casing or the borehole. The surface casing presents another barrier to waste migration in the portion of the well passing through USDWs. Finally, the annulus is sealed at both ends and is pressurized. Because the pressure in the annulus is higher than the pressure used to inject the waste (positive pressure), any leaks in the injection tube would result in annulus fluid being forced into the tube rather than waste fluid escaping into the annulus. The fluid pressure is required to be continuously monitored both by automated alarm systems and manually by a full-time operator for loss of pressure or volume. Such loss could indicate that the system integrity is compromised (e.g., pump failure, packer failure, casing failure). Most Class IH systems include automatic shutdown of the injection pumps upon alarm, although it was conservatively assumed that the system assessed did not have an auto-shutdown feature. Of course, the injection pumps shut down upon loss-of-power events. Class IH wells are monitored annually for a number of factors related to waste isolation, including injection zone pressure buildup, water quality monitoring in lower USDW in some cases, and required mechanical integrity testing to detect fluid movement outside the long string casing. Such testing includes annual radioactive tracer or oxygen activation logging, as well as temperature and noise logging at least once every 5 years. Casing inspection logs are required whenever the injection tube is removed. When migration or flaws are detected, they are repaired. In summary, the system assessed was a Class I hazardous waste injection well that minimally complies with 40 CFR 146 Subpart G requirements. The system components included in the PRA included geologic, engineered, and human elements. Finally, the system was assumed to be operating, with an operating lifetime of 30 years. Post-operating risks analyzed included the possibility of inadvertent human extraction of waste and migration through breached geologic confining zones.
ELSE_TSANG_CH010.qxd
9/15/2005
3:45 PM
Page 101
10.6 Event and Fault Tree Development
101
10.5 FAILURE MODES AND EFFECTS ANALYSIS FMEA was performed on the Class IH injection well system defined above. This is a systematic technique for identifying all means by which the injection well components could fail, and what the effect could be with respect to waste isolation. Each component and activity identified as important was evaluated by: ● Identifying all possible failure modes of the component (e.g., injection tube leaks, injection tube crushes, injection tube plugs, etc.). ● Identifying possible reasons for these failure modes (e.g., corrosion, improper installation, etc.). ● Assessing possible consequences of the failure mode (e.g., loss of annulus pressure, fracturing of injection zone, etc.). ● Identifying system features that serve to prevent the failure or mitigate its consequences (e.g., the annulus fluid is under positive pressure). The FMEA process is a brainstorming activity that does not exclude events based on the probability of their occurrence. All plausible events are considered even if they are considered to be of very low probability. The results of the FMEA are qualitative in nature and are not in themselves suitable for quantifying risk. Because the FMEA identifies all potential failure modes for the system, failure mechanisms of the components, and the safety systems designed to prevent or mitigate failures, it creates a level of understanding that can be used to develop the probabilistic framework to quantify risk (i.e., the event and fault trees). The FMEA process in this assessment was one through a series of workshops with deepwell injection operators and expert consultants. In addition, FMEA results were presented at a number of Ground Water Protection Council national meetings, and refined through input obtained from injection well operators, maintenance and testing professionals, and state and EPA regulatory staff who attended the meetings.
10.6 EVENT AND FAULT TREE DEVELOPMENT Based on the understanding gained from the FMEA, event trees were developed that identify potential sequences of events that could result in a release to the lowermost USDW. Seven possible initiating events were identified that characterize the overall risk of waste isolation loss for the Class IH injection well system defined. The seven initiating events identified were: 1. Packer leak 2. Major packer failure 3. Injection tube leak 4. Major injection tube failure 5. Cement microannulus leak 6. Confining zone(s) breach 7. Inadvertent injection zone extraction Once initiated, the likelihood of waste isolation loss depends on the subsequent failure of additional components, barriers, and backup systems within a relevant time domain. The event tree is a diagram that depicts the sequence of events and component failures that must follow for a release to the lowermost USDW to occur. A pathway can be traced through the event tree along its branches, depicting different combinations of failures and successes of system components and operational events that function together to prevent or result in waste isolation loss.
ELSE_TSANG_CH010.qxd
102
9/15/2005
3:45 PM
Page 102
A Probabilistic Risk Assessment of Class I Hazardous Waste Injection Wells
Fault trees were developed for three events of sufficient complexity, involving multiple events themselves. These three events were: loss of the annulus pressure barrier, lower geologic confining zone breach, and upper geologic confining zone breach. The event and fault trees for each initiating event sequence are discussed in more detail below, with estimated frequencies of occurrence for events in the trees described first. 10.7 EVENT-FREQUENCY-DISTRIBUTION DEVELOPMENT Perhaps the most problematic part of this PRA was estimating frequencies of occurrence for events in the trees. For many of these events, occurrence was so rare and data were so sparse that a confident point estimate for the frequency of occurrence could not be established. Consequently, uncertainty about occurrence frequencies was given explicit quantitative treatment in the assessment. Probability distributions of event occurrence frequencies were developed, either based on available occurrence data or expert judgment. These distributions are shown in Table 10.2, where the event names correspond to event names appearing on the event and fault trees in Figures 10.2–10.11. Simultaneous occurrence of the events in a sequence is required for a release to occur. The period of time during which simultaneous occurrence could feasibly happen before detection and remedy would occur was assumed to be one day. Thus, the frequencies shown in Table 10.2 are based on a daily time frame, unless they are on-demand probabilities of a failed state or response once a sequence is in progress (e.g., the probability that an alarm fails or the probability that a discontinuity is present in the confining zone). 10.8 QUANTITATIVE ANALYSIS OF EVENT TREES In PRA, event frequencies are combined according to the logic of the event and fault trees using Boolean algebra. The result is the estimated frequency (or probability) of a release to the lowermost USDW over the lifetime of the Class I hazardous waste injection well. Since uncertain event frequencies in this assessment were characterized by probability distributions, these distributions were propagated through the Boolean algebra calculations using Monte Carlo analysis. The result is expressed as a distribution of the probability that waste isolation will be lost during the lifetime of the injection well. This approach enables one to draw conclusions as to the certainty of the waste isolation loss risk estimates, and to conduct sensitivity analyses to identify which individual events contribute the most uncertainty to the risk estimates. To facilitate such analyses, both fault and event tree probabilities were placed into Microsoft Excel™ spreadsheets while the random sampling and generation of stochastic results were performed using Crystal Ball™. Latin Hypercube Sampling (LHS) was used to generate input values for all distributions. The analysis was performed with 5000 iterations to provide the best possible estimate of the percentiles. For operator errors likely to involve the same operator or similarly trained operators, the frequency distributions were correlated. A parametric sensitivity analysis was also performed based on percent contribution of uncertain event frequencies to the overall variance in the loss of waste isolation probability distribution. 10.9 PROBABILISTIC RISK ASSESSMENT (PRA) RESULTS Using the event and fault trees, the risk of waste isolation loss and release to the USDW over the 30-year life of a Class IH waste injection well was characterized quantitatively. Most of the
Automatic alarm fails Annulus pressure drops below injection pressure Loss of injection zone capacity results in overpressurization Annulus check valve fails to open Transmissive breach occurs through lower confining zone Transmissive breach occurs through upper confining zone Annulus pressure control system fails, resulting in underpressurization Injection pressure control system fails, resulting in overpressurization Failure to identify abandoned well in AOR Presence of unidentified transmissive discontinuity Extraction of injection zone groundwater Testing fails to detect injection fluid migration along outside of long string casing Waste injected chemically incompatible with geology or previously injected waste Sudden/major failure and breach of injection tube Injection tube leak Injected fluid is sufficiently buoyant to penetrate lower confining zone breach Long string casing leak located between surface
ALARM ANNPRESSLO CAPLOSS
LOCATION A
ITUBFAIL ITUBLEAK LBUOYANCY
INCOMPWASTE
EXTRACT FLUIDTEST
DETECTWELL DISCONT
CONTROLPI
Uniform
Poisson Poisson Single value
Uniform
Uniform Uniform
Uniform Uniform
Uniform
Uniform
From fault tree
1E−02
3E−07 3E−05 1E+00
1E−05
1E−05 5E−04
1E−03 1E−04
1E−06
1E−06
6E−04
1E−04 6E−04
5E−05 9E−14 1E−05
Lower bound
3E−02
6E−07 6E−05 1E+00
5E−05
1E−04 3E−03
5E−03 1E−03
1E−05
1E−05
3E−03
3E−04 3E−03
3E−04 7E−12 1E−04
Median
5E−02
8E−07 8E−05 1E+00
1E−04
1E−03 5E−03
1E−02 1E−02
1E−04
1E−04
1E−02
1E−03 1E−02
5E−04 8E−11 1E−03
Upper bound
3:45 PM
CONTROLPA
CONFINEBRCHU
Triangular From fault tree
Uniform From fault tree Uniform
Probability distribution type
9/15/2005
CHECKPA CONFINEBRCHL
Description
Event name
Table 10.2. Event probability distributions for a Class I hazardous waste injection well
ELSE_TSANG_CH010.qxd Page 103
10.9 Probabilistic Risk Assessment (PRA) Results 103
Long string casing leak located above base of surface casing Long string casing leak is located below confining zone(s) Sudden/major failure and breach of long string casing Long string casing cement microannulus allows fluid movement along casing Long string casing leak Waste migrates up microannulus to Location A between surface casing and upper confining zone Failure to recognize groundwater extraction located within injection waste zone Operator fails to recognize changes in confining zone capacity Operator fails to detect/respond to unacceptable pressure differential Operator error results in induced transmissive fracture through the lower confining zone Operator error causes annulus pressure below injection pressure
LOCATION B
OPERRPA
OPERRFRAC
OPERRDET
OPERINJ
NORECOGNIZE
LSTRINGLEAK MIGRATION A
LSCEMLEAK
LSCASEFAIL
1E−03 5E−05 5E−05 5E−05 5E−05
Uniform* Uniform* Uniform* Uniform*
2E−05 1E−04
2E−06
2E−07
9E−01
1E−02
Lower bound
Uniform
Poisson Uniform
Poisson
Poisson
Uniform
Uniform
Probability distribution type
3E−04
3E−04
3E−05
3E−05
5E−03
3E−05 1E−03
6E−06
3E−07
9E−01
5E−02
Median
5E−04
5E−04
5E−04
5E−04
1E−02
5E−05 1E−02
1E−05
5E−07
1E+00
1E−01
Upper bound
3:45 PM
LOCATION C
Description
9/15/2005
Event name
104
Table 10.2. (continued)
ELSE_TSANG_CH010.qxd Page 104
A Probabilistic Risk Assessment of Class I Hazardous Waste Injection Wells
Uniform
Same as OPERRDET
Single value
Note: Frequencies are per day or per demand. * Operator error event probability distributions are correlated (r = 0.5) to account for same operator or similar training.
WASTEPRESENT
UBUOYANCY
TRANSUSDW
Poisson Single value
Uniform
Poisson Triangular Single value
Poisson Poisson Uniform
Uniform
Uniform*
1E−02
1E−05
1E−01
2E−06 1E−01
1E−05
2E−04 5E−05 5E−01
2E−07 2E−05 1E−05
1E−05
5E−05
1E−01
5E−05
1E−01
3E−06 1E−01
5E−05
8E−04 5E−04 5E−01
4E−07 4E−05 1E−04
5E−05
3E−04
1E+00
1E−04
1E−01
5E−06 1E−01
1E−04
2E−03 5E−03 5E−01
6E−07 6E−05 1E−03
1E−04
5E−04
3:45 PM
SURFCASELEAK TRANSLCZ
Operator error causes injection pressure above annulus pressure Injection waste has migrated outside of Area of Review to unconfined zone Sudden/major failure and breach of packer Packer leak Confining zone has unexpected transmissive permeability Identified abandoned well plug fails Annulus pump fails Groundwater monitoring fails to detect waste release outside injection zone Seismic event induces a transmissive fault or fracture Surface casing leak Unidentified abandoned well transmissive from injection zone through lower confining zone Unidentified abandoned well transmissive through upper confining zone to USDW Injected fluid is sufficiently buoyant to penetrate upper confining zone breach Injected waste has not transformed into nonwaste
9/15/2005
SEISMFAULT
PLUGFAIL PUMPPA RELDETECT
PACKFAIL PACKLEAK PERMEA
OUTAOR
OPERRPI
ELSE_TSANG_CH010.qxd Page 105
10.9 Probabilistic Risk Assessment (PRA) Results 105
ELSE_TSANG_CH010.qxd
9/15/2005
106
3:45 PM
Page 106
A Probabilistic Risk Assessment of Class I Hazardous Waste Injection Wells
trees represent the daily probability of the event sequence, and their results were converted into 30-year probabilities for presentation (see Table 10.3). Events that are independent of time (i.e., inadvertent injection zone extraction) are presented as event probabilities. The cumulative percentile results of the analysis for each event sequence are presented in Table 10.3. Values shown in Table 10.3 are probabilities of the loss of waste isolation (i.e., release to the lowermost USDW) over the lifetime of the well. The cumulative percentile is the likelihood of being less than or equal to (i.e., likelihood of not exceeding) the corresponding loss of isolation risk. 10.9.1 Packer Leak The initiating event in this sequence is the development of a leak in the packer at the base of the injection tube and pressurized annulus (see Fig. 10.2). If the packer leaks during injection, containment is maintained as long as the annulus pressure is greater than the injection pressure. If the annulus pressure drops, containment will still be maintained by the long string casing. A leak in the long string casing might occur, but its location will be critical since this determines what additional failures must occur to lose containment. A long string casing leak in the area between the bottom of the surface casing and the upper confining zone (Location A) was assumed to result in a release to the lowermost USDW, even though current regulations require the surface casing to be set below the base of the lowermost USDW, into a confining bed. In addition, there actually may be significant geologic interaction between this point and the USDW. If the long string casing leak is located above the base of the surface casing, a release to the USDW requires either a leak in the surface casing or a crack (microannulus) in the long string cement casing that opens to Location A. A leak below the confining layer(s) requires a breach of the geologic barrier(s) or a microannulus that opens to Location A. Two component failures in the event tree are described by fault trees: the first quantifies the probability that the annulus pressure is less than the injection pressure, while the second addresses the probability that the confining zone is breached. These fault trees are presented in Figures 10.3 and 10.4, respectively, while the event probabilities associated with these fault trees are shown in Table 10.2. The PRA results of the packer leak scenario indicate that the probability of waste isolation loss over the life of the well from this initiating event is on the order of 10−17–10−18 (see Table 10.3). The annulus pressure is the primary barrier to loss of containment, and the probability of pressure loss is extremely low, since it would require simultaneous alarm and full-time operator failures. In fact, a difference in pressure between the annulus and injection fluids does occur, but the high reliability of the redundant auto-alarm and full-time operator keeps the probability of this extremely low, resulting in a pressure barrier loss during injection. Additionally, the location of a long string casing leak is a critical factor in waste isolation loss, as it determines the presence or absence of additional barriers. 10.9.2 Major Packer Failure This event is distinguished from the packer leak event in that it involves a complete and sudden loss of the packer and the subsequent rapid loss of annulus pressure (see Fig. 10.5). Without the annulus pressure barrier, the containment now depends on the integrity of the long string casing and associated components. The sequence of component failure leading to waste isolation loss thereafter is similar to the packer leak tree, except there is no annulus pressure barrier.
2.05E−20 5.35E−19 1.18E−18 2.67E−18 5.76E−18 1.11E−17 9.12E−17
0 10 25 50 75 90 100
3.31E−20 8.46E−19 1.85E−18 4.19E−18 8.98E−18 1.77E−17 1.09E−16
Injection tube leak 1.15E−09 3.22E−09 4.45E−09 6.35E−09 8.54E−09 1.06E−08 2.08E−08
Sudden injection tube failure 0.00E+00 1.78E−08 4.33E−08 1.35E−07 4.50E−07 1.04E−06 4.57E−06
Cement microannulus
Cumulative percentile is the likelihood of being less than or equal to (i.e., not exceeding) the corresponding loss of isolation risk.
7.73E−10 2.05E−09 2.82E−09 4.08E−09 5.53E−09 7.00E−09 1.32E−08
Sudden packer failure
5.05E−12 6.37E−11 1.20E−10 2.38E−10 4.80E−10 8.98E−10 6.39E−09
Confining zones fail
2.35E−10 3.55E−09 1.22E−08 4.79E−08 1.94E−07 6.41E−07 8.64E−06
Inadvertent extraction
3:45 PM
*
Packer leak
Cumulative percentile*
9/15/2005
Table 10.3. Cumulative percent results for each loss of waste isolation event in a Class I hazardous waste injection well
ELSE_TSANG_CH010.qxd Page 107
10.9 Probabilistic Risk Assessment (PRA) Results 107
108
9/15/2005
3:45 PM
Page 108
A Probabilistic Risk Assessment of Class I Hazardous Waste Injection Wells
Fig. 10.2. Event tree for packer leak in a Class I hazardous waste injection well.
ELSE_TSANG_CH010.qxd
9/15/2005 3:45 PM
Fig. 10.3. Fault tree for an annulus pressure barrier failure in a Class I hazardous waste injection well.
ELSE_TSANG_CH010.qxd Page 109
10.9 Probabilistic Risk Assessment (PRA) Results 109
110
9/15/2005
3:45 PM
Page 110
A Probabilistic Risk Assessment of Class I Hazardous Waste Injection Wells
Fig. 10.4. Fault tree for a lower confining zone breach in a Class I hazardous waste injection well.
ELSE_TSANG_CH010.qxd
ELSE_TSANG_CH010.qxd
9/15/2005
3:45 PM
Page 111
111
Fig. 10.5. Event tree for a major packer failure in a Class I hazardous waste injection well.
10.9 Probabilistic Risk Assessment (PRA) Results
ELSE_TSANG_CH010.qxd
112
9/15/2005
3:45 PM
Page 112
A Probabilistic Risk Assessment of Class I Hazardous Waste Injection Wells
A major packer failure is a lower probability event than a packer leak. Despite this, the assumed absence of annulus pressure eliminates an important barrier to waste isolation loss and results in a higher risk than for a simple packer leak on the order of 10−8–10−9 (see Table 10.3). With the loss of pressure, the waste is assumed to mix with the annulus fluid in the column. As above, the location of the long string casing is a critical factor in waste isolation loss, as it determines the presence or absence of additional barriers. 10.9.3 Injection Tube Leak This initiating event involves a leak in the injection tube above the packer (see Fig. 10.6). Since it is not a catastrophic failure, annulus pressure is maintained. Aside from the location of the leak, the events and the sequence leading to containment loss is identical to that of the packer leak scenario. Similar to the packer leak, the results indicate that the probability of waste isolation loss over the life of the well is extremely low, on the order of 10−17–10−19 (see Table 10.3). As with the packer leak, the annulus pressure is the primary barrier to loss of containment. Additionally, the location of the long string casing remains a critical factor in waste isolation loss to the accessible environment, as it determines the presence or absence of additional barriers. 10.9.4 Major Injection Tube Failure This initiating event is similar to the major packer failure, and is characterized by a catastrophic failure of the injection tube above the packer, with the resulting loss of annulus pressure (see Fig. 10.7). Aside from the location of the failure, the sequence of events leading to possible containment loss is identical to that of the major packer failure scenario discussed above. A major injection tube failure has a lower probability of occurring than an injection tube leak. As with the major packer failure, the assumed immediate loss of annulus pressure eliminates an important barrier to waste isolation loss and results in a higher risk than a simple leak of the injection tube, on the order of 10−8–10−9 (see Table 10.3). With the loss of positive pressure, it is assumed that the waste mixes with the annulus fluid and escapes through the leak in the long string casing. As in all these scenarios, the location of the long string casing is a critical factor to waste isolation loss. 10.9.5 Cement Microannulus Failure Radiotracer studies are performed annually on Class IH wells to detect migration. This event sequence involves the possibility that an extended vertical opening (i.e., microannulus) in the cement surrounding the long string casing remains undetected and results in waste isolation loss (see Fig. 10.8). The cement extends from the surface through all confining layers to the injection zone. Should a microannulus crack open in the cement, extend from the injection zone through the upper confining zone, and remain undetected, waste injected under pressure could possibly migrate up to Location A and then to the USDW. Alternatively, waste could migrate only up to a location below the upper confining zone, and then the upper confining zone could breach. An additional fault tree is needed to estimate the probability that the upper confining zone will be breached. This fault tree is presented in Figure 10.9, with the corresponding probabilities presented in Table 10.2. The probability that loss of waste isolation will result under this scenario was calculated to be on the order of 10−6–10−8 (see Table 10.3). The event sequence is controlled by the location
9/15/2005 3:45 PM
Fig. 10.6. Event tree for an injection tube failure in a Class I hazardous waste injection well.
ELSE_TSANG_CH010.qxd Page 113
10.9 Probabilistic Risk Assessment (PRA) Results 113
114
9/15/2005
3:45 PM
Page 114
A Probabilistic Risk Assessment of Class I Hazardous Waste Injection Wells
Fig. 10.7. Event tree for major injection tube failure in a Class I hazardous waste injection well.
ELSE_TSANG_CH010.qxd
ELSE_TSANG_CH010.qxd
9/15/2005
3:45 PM
Page 115
115
Fig. 10.8. Event tree for a cement microannulus in a Class I hazardous waste injection well.
10.9 Probabilistic Risk Assessment (PRA) Results
ELSE_TSANG_CH010.qxd
116
9/15/2005
3:45 PM
Page 116
A Probabilistic Risk Assessment of Class I Hazardous Waste Injection Wells
Fig. 10.9. Fault tree for upper confining zone breach in a Class I hazardous waste injection well.
to which the microannulus extends. In this case, it was assumed to extend from the injection zone to the USDW. The greatest uncertainty lies in whether such an extended and transmissive microannulus will occur, and if the waste fluid can travel that far given that the injection zone represents the path of least resistance to the pressurized waste stream. Additionally, the annual testing for fluid migration also limits the risk to loss through this mechanism. 10.9.6 Confining Zone Breach The initiating event in this scenario is a transmissive breach of the lower confining zone (directly above the injection zone) (see Fig. 10.10). The probability of this event is based on the fault tree analysis first developed for the packer leak (see Fig. 10.4). Once the lower confining zone is breached, the remaining barriers to waste isolation loss are: 1. The waste is sufficiently buoyant to penetrate the lower confining zone breach. 2. Groundwater monitoring fails to detect waste outside of the injection zone. 3. The upper confining zone is breached. 4. The waste is sufficiently buoyant to penetrate the upper confining zone breach. A breach in the confining zone requires that all confining zones must be completely breached with transmissive openings. This must remain undetected in spite of ongoing monitoring of pumping pressure and volumes, injection zone pressure, and groundwater quality. Additionally, the waste must have a driving force in all zones to be sufficiently buoyant to penetrate the USDW above, and there must be no bleed-off into the buffer aquifers between the confining zones. This scenario has a probability of waste isolation loss on the order of 10−10 (see Table 10.3).
9/15/2005 3:45 PM
Fig. 10.10. Event tree for lower and upper confining zone breaches in a Class I hazardous waste injection well.
ELSE_TSANG_CH010.qxd Page 117
10.9 Probabilistic Risk Assessment (PRA) Results 117
ELSE_TSANG_CH010.qxd
118
9/15/2005
3:45 PM
Page 118
A Probabilistic Risk Assessment of Class I Hazardous Waste Injection Wells
10.9.7 Inadvertent Injection Zone Extraction Given the depth of most injection wells, future human intrusion into the injection zone is unlikely (see Fig. 10.11). An extraction scenario also does not rely on any additional components of the operating system. This initiating event assumes extraction of injected waste with the additional sequence probabilities included to assess the possibility that the extraction of the injection zone material goes unnoticed by the well user. The time domain is not relevant, as all such activities are assumed to have occurred after a system closure. This scenario is the most difficult to estimate the probability of occurrence. Nonetheless, the possibility that extraction of isolated waste will occur after closure was calculated to be less than 10−6 (see Table 10.3). Since injection zones are more than 1000 ft deep and presumably underlie most accessible and higher quality aquifers, it is unclear why water from the injection zone would be extracted by anyone. Depending on timing and location, the waste may no longer present a potential hazard, or the plume may not be intersected by the extraction wells. 10.9.8 Incompatible Waste Injection The issue of incompatibility of wastes and well components or geologic formations was covered under the outcomes of the other event trees. Carbon dioxide or other gas formation may result in packer blowout, rupture of the injection tube, transmissive geologic fracturing, or wellhead blowout. Each of these events are covered by the event trees for packer or injection tube failure or by the fault tree for confining zone breaches, or are considered spills and not relevant to this evaluation. Corrosion of rock or other system components are covered under the fault tree for the lower confining zone breach or the event tree for the relevant system component (i.e., injection tube leak or failure). A chemical interaction may also result in a plug forming in the system, resulting again in packer blowout, failure of the injection tube, or fractures of the different confining zones in response to a pressure buildup. These are addressed by event trees for the confining zone breach and the packer or injection tube failure, or by the fault tree for the breach of the lower confining zone.
10.10 OVERALL LOSS OF WASTE ISOLATION RESULTS Based on the PRA conducted for Class IH wells, the 90th percentile risks for the individual scenarios detailing the potential loss of waste isolation range from a low of 10−17 (packer leak) to a high of 10−6 (cement microannulus) (see Fig. 10.12). The probability for all events combined (assuming that these risks are additive) resulting in a loss of waste isolation is between 10−6 and 10−8 (Fig. 10.12). The event sequences that are predominant contributors to overall risk are the microannulus failure and the possibility of inadvertent future injection zone extractions. The sensitivity analysis (Fig. 10.13) identified the following contributions to overall uncertainties about probability of loss of waste isolation: ● Distance that waste migrates along a vertical cement microannulus (52% of the variance). ● Likelihood of future extraction from the injection zone (17% of the variance). ● Probability that at the time of future extraction the waste is no longer hazardous or the plume is not present (15% of the variance). ● Likelihood that the fluid testing fails to detect migration (8% of the variance). ● Likelihood that the extracted material is unrecognized as waste by the well user (3% of the variance).
9/15/2005 3:45 PM
Fig. 10.11. Event tree for inadvertent extraction from an injection zone in a Class I hazardous waste injection well.
ELSE_TSANG_CH010.qxd Page 119
10.10 Overall Loss of Waste Isolation Results 119
ELSE_TSANG_CH010.qxd
120
9/15/2005
3:45 PM
Page 120
A Probabilistic Risk Assessment of Class I Hazardous Waste Injection Wells
Fig. 10.12. Probability distribution for total loss of waste isolation risks in a Class I hazardous waste injection well.
Fig. 10.13. Sensitivity chart of relative contributions to overall uncertainties for loss of waste isolation risks.
10.11 CONCLUSIONS AND RECOMMENDATIONS Because of the conservative assumptions used for failure event probabilities and the explicit treatment given to uncertainties in this analysis, we believe that the risk of loss of waste isolation from Class IH wells is less than 10−6. The low risk is due in large measure to the use of redundant engineered systems and geology to provide multiple and diverse barriers to prevent release of waste to the accessible environment. This is aided in part by the fact that deep-well injection is a simple design relying on passive systems to minimize failure modes and frequencies. The annulus pressure is a critical barrier and performance monitor, and it displays a high reliability due to the presence of automatic alarms and shutoffs, and full-time operators.
ELSE_TSANG_CH010.qxd
9/15/2005
3:45 PM
Page 121
10.11 Conclusions and Recommendations
121
The risk of waste isolation loss is dominated by two failure scenarios: The possibility that a transmissive microannulus develops in the cemented borehole outside the long string casing, and extends from the injection zone up past the geologic confining zones, and 2. The possibility of inadvertent future extraction of injected waste. Uncertainty about the overall risk to waste isolation is also dominated by events associated with these two scenarios. For example, in developing the frequency distribution for the microannulus initiating event (LSCEMLEAK in Fig. 10.8), it was conservatively assumed that “vertical migration detected” events in the well failure database (U.S. Environmental Protection Agency, 1993) were equivalent to the occurrence of a transmissive microannulus extending from the injection zone through one or both of the confining layers; however, Class IH well operators contend that evidence of a microannulus extending from the injection zone through the confining layers has not been found. Thus, a highly uncertain event initiates the highest-risk sequence, and is therefore treated with significant conservatism in the PRA; this points to the need for more complete data on the location, duration, and length of detected microannuli, rather than just noting the number of times that vertical migration is detected. Numerous conservative assumptions were used in this PRA that, combined with the explicit treatment of uncertainties (i.e., the Monte Carlo analysis), lend confidence to the conclusions of low risk. Credit was not taken for cement as a horizontal barrier to waste migration. Likewise, in using the well failure database (U.S. Environmental Protection Agency, 1993), all events termed “failure” for packers, tubing, and casing were assumed to be breaches of sufficient size and duration to transmit waste. As explained above, “vertical migration detected” events were similarly assumed to represent a complete transmissive pathway from the injection zone, and up past the geologic confining layer(s). In the event of a breach of the confining layers, the buoyancy of the waste and the injection pressure were assumed to be high enough to drive migration through breaches of multiple confining layers. Significant bleed-off and attenuation that would occur in the intervening buffer aquifers were not taken into account. Only two geologic confining layers were assumed throughout this PRA, although survey information indicates that three or more confining zones are usually present. Published human-error data were used as the lower bound on probability distributions for events that assumed an equal probability for error rates to be an order of magnitude higher than published rates. While automatic shutdown of the injection well pumps is a typical operating feature of most Class IH wells, no automatic shutdown was assumed for this PRA. It was further assumed that a release between the surface casing and the upper confining zone was equivalent to a release to the USDW, and that releases below the confining zones involved only one confining zone barrier to the USDW. Finally, the timing between independent occurrences in the various event and fault trees was assumed to be coincident for sufficient duration prior to detection and corrective action for a release to the USDW to occur. Since the failure location and timing of the individual events are critical to the development of these release scenarios, uncertainty would be reduced and knowledge improved if this information were collected and included in the databases maintained on Class I well failures. The presence, degree of training, and diligence of the operator is important in preventing system failure and loss of waste isolation. This is especially critical in maintaining the annulus pressure, which is a major barrier to loss of waste from the system. Uncertainty over the existence and transmissivity of extended vertical cement breaches is important. Experimental or field data on the microannulus assumed to exist in these scenarios would assist in reducing this uncertainty and improving the risk estimates. Finally, we recommend that future assessments of the potential environmental risks associated with deep-well injection explicitly take into 1.
ELSE_TSANG_CH010.qxd
9/15/2005
122
3:45 PM
Page 122
A Probabilistic Risk Assessment of Class I Hazardous Waste Injection Wells
account the probability of release and the amount of waste that could be released by the mechanisms of feasible system failure scenarios.
REFERENCES Buss, D.R. et al., 1984. A Numeric Stimulation Study of Deep-Well Injection in Two Hydrogeologic Settings—Texas Gulf Coast, Great Lakes Basin. GeoTrans, Inc., June 20. Buttram, J.R., 1986. Operation and maintenance of underground injection wells. Ground Water Monitoring Review, 6(3): 64–67. CH2M Hill, 1986a. A Class I Injection Well Survey. Phase II Report—Performance Evaluation. Prepared for Underground Injection Practices Council, Oklahoma City, OK, D19976.S2.00, May. CH2M Hill, 1986b. A Class I Injection Well Survey Phase I Report. Survey of Selected Sites. Prepared for Underground Injection Practices Council, Oklahoma City, OK, D19976.S1, April. Clark, J.E., 1987. Factors that can cause abandoned wells to leak as verified by case histories from class II injection, Texas Railroad Commission files. Proceedings of the International Symposium on Subsurface Injection of Oilfield Brines, May 4–6, New Orleans, LA, pp. 166–223. Sponsored by the U.S. Environmental Protection Agency and the Underground Injection Practices Council. Published by the Underground Injection Practices Council, Oklahoma City, OK. Clark, J.E., 1994. Environmental scoring without risk assessment. Presented at Clean Texas 2000—Environmental Trade Fair, E.I. du Pont de Nemours and Co., Inc., April 14. Clemen, R.T., 1991. Making Hard Decisions: An Introduction to Decision Analysis. PWSKent Publishers, Boston, MA, pp. 253–254. Davis, S., 1987. Responses to Questions from the Chemical Manufacturers Association. Submitted to Chemical Manufacturers Association by Department of Hydrology and Water Resources, University of Arizona, Tucson, AZ, April 10. Davis, M.L. and Satterwhite, D.G., 1988. Summary Procedures Guide for Application of Risk and Hazards Analysis Techniques. EG&G, IA. Department of Energy and Natural Resources et al., 1989. Illinois Scientific Surveys Joint Report 2. Evaluation of Underground Injection of Industrial Waste in Illinois. Don L. Warner, Inc. and Engineering Enterprises, Inc., 1984. Confining Layer Study. Prepared for EPA Region V, December. Engineering Enterprises, Inc., Geraghty & Miller, Inc. and Ken E. Davis Associates, 1986. Class I Hazardous Waste Injection Wells Evaluation of Non-Compliance Incidents. Prepared for U.S. Environmental Protection Agency, Washington, DC, September. Envirosphere Company, 1988. Draft Task Report. Below Regulatory Concern Waste Accident Scenario Dose Assessment Task 1: Unexpected Event/Accident Identification. Prepared for Electric Power Research Institute, Palo Alto, CA, February. Goolsby, D.A., 1972. Geochemical Effects and Movement of Injected Industrial Waste in a Limestone Aquifer. U.S. Geological Survey, Tallahassee, FL. Gordon, W. and Bloom, J., 1985. Deeper Problems: Limits to Underground Injection as a Hazardous Waste Disposal Method. Natural Resources Defense Council, Inc. Ken E. Davis Associates, 1996. Conclusions and Recommendations on the Assessment of Fourteen Alleged Class I Well Failures. Keda Project No. I 001633, Houston, TX, May 1986. Lannoy and Procaccia, 1996. European Industry Reliability Databank.
ELSE_TSANG_CH010.qxd
References
9/15/2005
3:45 PM
Page 123
123
MacLean, A. and Puchalsky, R., 1994. Where the Wastes Are: Highlights from the Records of the More than 5,000 Facilities that Receive Transfers of TRI Chemicals. OMB Watch and Unison Institute, April. McCormick, N.J., 1981. Reliability and Risk Analysis, Chapter 3. Academic Press, Inc., San Diego, CA. Meritt, M.L., 1984. Digital Simulation of the Regional Effects of Subsurface Injection of Liquid Waste Near Pensacola, Florida. U.S. Geological Survey, prepared in cooperation with the Florida Dept. of Environmental Regulation, Tallahassee, FL. Miller, C., Fischer, T.A., II, Clark, J.E., Porter, W.M., Hales, C.H. and Tilton, J.R. 1986. Flow and containment of injected wastes. Ground Water Monitoring Review, 6(3): 37–47. Morgan, P.G., 1985. A Closer Look at “Deeper Problems”—A Response to Those Who Would Ban Hazardous Waste Disposal by Underground Injection: The New Mexico Experience. New Mexico Environmental Improvement Division, Underground Injection Control Program. Morganwalp, D.W. and Smith, R.E., 1988. Modeling of Representative Injection Sites. Paque, M.J., 1986. Class I injection well performance survey. Ground Water Monitoring Review, 6(3): 68–69. Scrivner, N.C., Bennett, K.E., Pease, R.A., Kopatsis, A., Sanders, S.J., Clark, D.M. and Rafal, M. 1986. Chemical fate of injected wastes. Ground Water Monitoring Review, 6(3): 53–57. SCS Engineers, 1985. Final Report, Summary of Chemical Manufacturers Association Underground Injection Well Survey. Prepared for CMA UIC Work Group. Washington, DC, February. Sierra Club Legal Defense Fund, 1989. In: E.P. Jorgensen (Ed.), The Poisoned Well: New Strategies for Groundwater Protection. Island Press, Washington, DC. Swain, A.D., 1987. Accident Sequence Evaluation Program: Human Reliability Analysis Procedure. NUREG/CR-4772, SAND86-1996, February. Swain, A.D. and Guttman, A.L., 1980. Handbook of Human Reliability Analysis with Emphasis on Nuclear Power Plant Applications. NUREG/CR-1278, Sandia National Laboratories. The Cadmus Group, Inc., 1995. Regulatory Impact Analysis of Proposed Hazardous Waste Disposal Restrictions for Class I Injection of Phase III Wastes. Prepared for EPA, Office of Ground Water and Drinking Water, January 12. Underground Injection Practices Council, 1986. Journal of the Underground Injection Practices Council, No. 1. Available from the Ground Water Protection Council Library. Underground Injection Practices Council, 1987. A Class I Injection Well Survey. Phase II Report—Survey of Operations. Oklahoma City, OK, December. Underground Injection Practices Council, 1989. Injection Well Bibliography. Oklahoma City, OK, January. Underground Resource Management, Inc., 1984. Evaluation of a Subsurface Waste Injection System near Vickery, Ohio. Prepared for the Ohio EPA, March. U.S. Environmental Protection Agency, 1985. Report to Congress on Injection of Hazardous Waste. EPA 570/9-85-003, EPA Office of Drinking Water, May. U.S. Environmental Protection Agency, 1989. OSWER Comparative Risk Project Executive Summary and Overview Report. Washington, DC, September. U.S. Environmental Protection Agency, 1990a. Assessing the Geochemical Fate of DeepWell-Injected Hazardous Waste, A Reference Guide. EPA/625/6-89/025a, EPA Office of Research and Development, June.
ELSE_TSANG_CH010.qxd
124
9/15/2005
3:45 PM
Page 124
A Probabilistic Risk Assessment of Class I Hazardous Waste Injection Wells
U.S. Environmental Protection Agency, 1990b. Assessing the Geochemical Fate of DeepWell-Injected Hazardous Waste, Summaries of Recent Research. EPA/625/6-89/025b, EPA Office of Research and Development, July. U.S. Environmental Protection Agency, 1991. Analysis of the Effects of EPA Restrictions on the Deep Injection of Hazardous Waste. EPA/570/9-91-031, EPA Office of Ground Water and Drinking Water, October. U.S. Environmental Protection Agency, 1993. Letter from Martha G. Prothro, Acting Assistant Administrator, EPA Office of Water, to the Honorable John D. Dingell, Chairman, Subcommittee on Oversight and Investigations, Committee on Energy and Commerce, U.S. House of Representatives, Attachment W, April 19. U.S. Environmental Protection Agency, 1996. Draft UICWELLS Database, EPA Office of Water, Underground Injection Control Branch, April. U.S. Environmental Protection Agency, 1999. 1997 Toxics Release Inventory Public Data Release Report. EPA Office of Prevention, Pesticides, and Toxic Substances, April. Visocky, A.P., Nealon, J.S., Brower, R.D., Krapac, I.G., Hensel, B.R. and Guthrie, M.A. 1986. Study of current underground injection control regulations and practices in Illinois. Ground Water Monitoring Review, 6(3): 59–63. Ward, D.S. et al., 1987. A Numerical Evaluation of Class I Injection Wells for Waste Containment Performance. GeoTrans, Inc. Prepared for EPA Office of Drinking Water, Underground Injection Control Program, September 30. Warner, D.L. and Lehr, J.H., 1977. An Introduction to the Technology of Subsurface Wastewater Injection. University of Missouri—Rolla and National Water Well Association, prepared for Robert S. Kerr Environmental Research Lab, Ada, OK, December. Wesson, R.L. and Nicholson, C., 1987. Earthquake Hazard Associated with Deep-Well Injection. Prepared for EPA, U.S. Geological Survey, June. Woodward Clyde Consultants, 1995. Underground Injection Well Questionnaire. Survey prepared for Chemical Manufacturers Association, August.
APPENDIX: BASIS FOR EVENT FREQUENCY PROBABILITY DISTRIBUTIONS There are 42 events identified in the PRA (Table 10.2) for which failure rates are needed to calculate event- and fault-tree probabilities. For many of these events, occurrence is so rare and data are so sparse that a confident point estimate for the frequency of occurrence cannot be established. Directly applicable data on the frequency of most events were not found. In common practice, most component failure modes are identified and corrected during required testing and maintenance, and thus may not be recorded as a failure event per se. More than one-third of the events involved some type of human error. Human error frequency data are available (Swain and Guttman, 1980; Swain, 1987); however, their direct applicability to the human tasks involved in Class IH wells is uncertain. Consequently, uncertainty about occurrence frequencies was given explicit quantitative treatment in the PRA. Probability distributions of event occurrence frequencies were developed, either based on available occurrence data or expert judgment. In general, probability distributions for event frequencies were derived as follows: 1. A 1993 EPA reply to a House of Representatives subcommittee inquiry (U.S. Environmental Protection Agency, 1993) provided state-by-state summaries of certain
ELSE_TSANG_CH010.qxd
9/15/2005
3:45 PM
Page 125
Appendix: Basis for Event Frequency Probability Distributions
125
reported types of Class I injection well failure events between 1988 and 1992. Numbers of events were reported for 469 Class I wells (hazardous and nonhazardous) located in 12 states. Events reported included tubing, casing, and packer leaks; and waste migration on the outside of the long string casing (i.e., cement microannulus). The number of reported events was divided by 855,925 well days (469 wells × 5 yr × 365 days/yr) to derive an estimate of the average daily occurrence rate for each type of event. Because nonhazardous wells have less regulatory restrictions than hazardous, it was a conservatism to include these data. 2. Modeling these failure rates with a binomial distribution, it is possible to determine the confidence intervals for a given average failure rate. Estimations of the 90th percentile upper confidence limit of the average failure rates were calculated using methods outlined by McCormick (1981). These are shown in Table A.1. Table A.1. Component
Number of reported failures
90th percentile confidence limit of average failure rate (day−1)
Tube Casing* Packer* Waste migration†
48 28 31 5
6.80E−05 4.20E−05 4.60E−05 1.10E−05
*Three recorded “annulus leak” events were included because it could not be determined if these were casing- or packer-related. † This category is assumed to be a surrogate for casing cement leak events.
Probability distributions representing uncertainties about frequency rates of these events (ITUBLEAK, LSTRINGLEAK, PACKLEAK, LSCEMLEAK) were developed by using these upper confidence limits for the average rate as the rate parameter in a Poisson distribution. The Poisson distribution is commonly used in reliability analyses to describe random failures in a system that cause irreversible transitions in the system (Clemen, 1991) such as a loss of waste isolation. The Poisson distribution requirements (Clemen, 1991), which are met for this application, include: ● Events can happen at any time within the day. ● The probability of an event is small. ● Events can happen independently of other events. ● The average number of events per day does not change with time. 3. For events involving typical components of any industrial system, such as valve, pump, control system, or alarm failures, occurrence frequencies were obtained from available industrial reliability databases (Davis and Satterwhite, 1988; Envirosphere Company, 1988; Lannoy and Procaccia, 1996). 4. Most human-error rates were derived from available human reliability data for similar activities. Usually, these human error data have been compiled for highly trained and scrutinized occupations such as nuclear power plant operators (Swain and Guttman, 1980; Swain, 1987) and firemen (Davis and Satterwhite, 1988; Envirosphere Company, 1988). While Class I hazardous waste injection well operators arguably fall into this same category, this assessment conservatively assigned human-error rates as the lower bound of the distribution, with an upper bound set at a higher order of magnitude.
ELSE_TSANG_CH010.qxd
126
9/15/2005
3:45 PM
Page 126
A Probabilistic Risk Assessment of Class I Hazardous Waste Injection Wells
5. For events in which data are entirely lacking, the authors relied on professional judgment, shaped in part by the experience of deep-well operators and regulators elicited during workshops held in conjunction with Ground Water Protection Council national meetings. To account for uncertainty in professional judgment, relatively large bounds of uncertainty were applied to frequencies derived in this manner. When the uncertainty was high, the range of the distribution would span several orders of magnitude. In some cases, the frequency was set at a maximum value; for example, the probability that injected fluid is sufficiently buoyant to penetrate a lower confining zone breach was assumed to be 1. The probability distributions representing uncertainty about event frequencies are summarized in Table 10.2 of the chapter and discussed individually below. Event: Description: Probability: Basis:
ITUBLEAK Injection tube leak. Poisson distribution with 6.8E−05/day rate. This event quantifies the probability that the injection tube carrying waste to the injection zone will develop a leak. Based on compilation of stateby-state data analyzed as discussed above.
Event: Description: Probability: Basis:
ITUBFAIL Sudden major failure and breach of the injection tube. 1/100th of ITUBLEAK probability. ITUBFAIL assumes a sudden and major failure of the injection tube such that the annulus pressure is lost simultaneously. Based on professional judgment, the likelihood of the injection tube failing catastrophically was estimated to be 1/100th the probability of a leak. Thus the ITUBFAIL probability was assigned a value 0.01 times the ITUBLEAK probability.
Event: Description: Probability: Basis:
ANNPRESSLO Annulus pressure drops below injection pressure. Determined by fault tree analysis. Due to the multiple components associated with this failure event, an ANNULUS PRESSURE BARRIER FAILURE FAULT TREE (Fig. 10.3) was developed and used to evaluate the event probability. The resulting cumulative distribution for this event frequency is: 10th percentile 20th percentile 30th percentile 40th percentile 50th percentile 60th percentile 70th percentile 80th percentile 90th percentile
Event: Description: Probability: Basis:
1.5E−12 2.6E−12 3.8E−12 5.2E−12 7.0E−12 9.3E−12 1.2E−11 1.7E−11 2.4E−11
LSTRINGLEAK Long string casing leak. Poisson distribution with 4.2E−05/day rate. Based on compilation of state-by-state data analyzed as discussed above.
ELSE_TSANG_CH010.qxd
9/15/2005
3:45 PM
Page 127
Appendix: Basis for Event Frequency Probability Distributions
127
Event: Description: Probability: Basis:
LSCASEFAIL Sudden and major failure and breach of the long string casing. 1/100th of LSTRINGLEAK probability. LSCASEFAIL assumes a sudden and major failure of the long string casing such that the annulus pressure is lost simultaneously. Based on professional judgment, the likelihood of the long string casing failing catastrophically was estimated to be 1/100th the probability of a leak. Thus the LSCASEFAIL probability was assigned a value 0.01 times LSTRINGLEAK.
Event: Description: Probability: Basis:
SURFCASELEAK Surface casing leak. Poisson distribution with 4.2E−06/day rate. The surface casing surrounds the long string casing and provides one of the final engineered barriers to the USDW. Failure probabilities are derived from LSTRINGLEAK with a correction of 0.1 to account for the fact that the surface casing is subject to less stress than the long string casing, and it is shorter and closer to the surface, making it less likely to be subject to construction failure modes.
Event: Description:
LSCEMLEAK Long string casing cement microannulus allows fluid movement along casing. Poisson distribution with 1.1E−05/day rate. Surrounding the entire length of the long string casing is cement, which fills the void between the casing and the surrounding geology. Given that there may be discontinuities in the cement pack, there is the probability that waste may migrate up the outer length of the casing through a microannulus discontinuity in the cement. Based on the state-by-state data responses for “waste migration,” a failure rate parameter for the distribution was determined using the methodology described above.
Probability: Basis:
Event: Description: Probability: Basis:
Event: Description: Probability: Basis:
LOCATION A Long string casing leak is located between surface casing and uppermost confining zone. Uniform distribution from 1.0E−02 to 5.0E−02. Given that a long string casing leak has occurred, the exact location along its entire length determines the likely migration route. If the leak occurs within the bounds defined by LOCATION A, migration to the USDW is assumed to be immediate and complete. Estimation of probability is based on professional judgment and takes into account the length of casing in this location relative to the typical overall long string casing length. In addition, consideration was given to the fact that stresses on the casing increase with depth. LOCATION B Long string casing leak is located above the bottom of the surface casing. Uniform distribution from 1.0E−02 to 1.0E−01. The same logic applied to the determination of LOCATION A probability is used here.
ELSE_TSANG_CH010.qxd
128
9/15/2005
3:45 PM
Page 128
A Probabilistic Risk Assessment of Class I Hazardous Waste Injection Wells
Event: Description: Probability: Basis:
LOCATION C Long string casing leak is located below the confining zone(s). 1-Prob(LOCATION A)–Prob(LOCATION B). The final section of the casing string extends from the top of the uppermost confining zone to the injection zone. This represents the largest fraction of the casing length, and stresses increase with depth, so the likelihood for a casing leak is higher in this location. Given that a long string casing leak has occurred, the probabilities for LOCATION A, LOCATION B, and LOCATION C must sum to unity. Thus, an algorithm is included in the event tree for the Monte Carlo simulation that calculates the probability of LOCATION C based on the probabilities selected at each iteration for LOCATION A and LOCATION B.
Event: Description:
MIGRATION A Waste migrates up the microannulus to Location A between the surface casing and the upper confining zones. Uniform distribution from 1.0E⫺04 to 1.0E⫺02 Radiotracer studies are performed annually on Class IH wells to detect migration. It is assumed that these studies do not always detect the formation of an extended vertical opening, i.e., a microannulus, in the cement surrounding the long string casing. If a microannulus extends from the injection zone through the upper confining zone, waste under pressure could migrate to Location A, and ultimately to a USDW. The probability of loss of waste isolation by this scenario is calculated to be on the order of 10⫺6 to 10⫺8.
Probability: Basis:
Event: Description: Probability: Basis:
PACKLEAK Packer leak. Poisson distribution with 4.6E−05/day rate. This event quantifies the probability that the packer will develop a leak. The packer seals the bottom of the annulus between the long string casing and the injection tube. The probability is based on compilation of stateby-state data analyzed as discussed above.
Event: Description: Probability: Basis:
PACKFAIL Sudden and major failure and breach of packer. 1/100th of PACKLEAK probability. Using the same basis applied to other catastrophic failure events, a professional judgment of 1/100th of the probability of a leak was used for complete packer failure.
Event: Description:
FLUIDTEST Testing fails to detect injection fluid migration along outside of long string casing. Uniform distribution from 5.0E−04 to 5.0E−03. Regular testing is required to detect migration fluid along the outside of the casing material. Generally, the probability of failing to detect a leak is most likely due to operator error, either in the procedure or in the interpretation of results. Thus, the probability of failing to detect fluid migration is based
Probability: Basis:
ELSE_TSANG_CH010.qxd
9/15/2005
3:45 PM
Page 129
Appendix: Basis for Event Frequency Probability Distributions
129
on the probability of operator—hence, human error. Studies prepared for nuclear power plant reliability analyses (Swain and Guttman, 1980; Swain, 1987) are primary sources for human-error rates. These studies show that errors of omission for nonpassive tasks (maintenance, test, or calibration) occur at a rate of approximately 1.0E−03 per demand, with a range from 5.0E−04 to 5.0E−03. It is assumed that a single failure to detect on demand (i.e., at the time of the test) results in significant fluid migration. Event: Description: Probability: Basis:
CONFINEBRCHL Transmissive breach occurs through lower confining zone. Determined by fault tree analysis. Due to the multiple components associated with this failure event, a LOWER CONFINING ZONE BREACH FAULT TREE (Fig. 10.4 in chapter) was developed and used to evaluate the event probability. The resulting cumulative distribution for this event frequency is: 10th percentile 20th percentile 30th percentile 40th percentile 50th percentile 60th percentile 70th percentile 80th percentile 90th percentile
Event: Description: Probability: Basis:
CONFINEBRCHU Transmissive breach occurs through upper confining zone. Determined by fault tree analysis. Due to the multiple components associated with this failure event, an UPPER CONFINING ZONE BREACH FAULT TREE (Fig. 10.9) was developed and used to evaluate the event probability. The resulting cumulative distribution for this event frequency is: 10th percentile 20th percentile 30th percentile 40th percentile 50th percentile 60th percentile 70th percentile 80th percentile 90th percentile
Event: Description: Probability:
1.7E−03 1.9E−03 2.2E−03 2.5E−03 2.9E−03 3.4E−03 4.3E−03 5.8E−03 8.2E−03
1.6E−03 1.8E−03 2.1E−03 2.4E−03 2.7E−03 3.3E−03 4.2E−03 5.6E−03 7.9E−03
LBUOYANCY Injection fluid is sufficiently buoyant to penetrate the lower confining zone breach. 1.0
ELSE_TSANG_CH010.qxd
130
9/15/2005
3:45 PM
Page 130
A Probabilistic Risk Assessment of Class I Hazardous Waste Injection Wells
Basis:
Because fluid is being injected under pressure below the lower confining zone, it is conservatively assumed that this provides sufficient buoyancy to penetrate a breach. In general, in the absence of active injection pressure, it is unlikely that buoyancy would be sufficient to transmit injected fluid completely through a breach.
Event: Description:
UBUOYANCY Injection fluid is sufficiently buoyant to penetrate upper confining zone breach. Uniform distribution from 1.0E−05 to 1.0E−04. It is assumed that fluid injection would need to be maintained (while losing pressure to the breach in the confining zones) or even overpressurized to provide a sufficient force to drive fluid through breaches in both the lower and upper confining zones. For this to occur, there would need to be an operator error in failing to detect an injection pressure loss or overpressurization. As explained above, human reliability data show that errors of omission for nonpassive tasks occur within a range of 5.0E−04 to 5.0E−03 per demand. While pressure is checked continuously during injection, it is conservatively assumed that a single failure to detect a pressure change results in significant fluid movement up through the breaches.
Probability: Basis:
Event: Description: Probability: Basis:
RELDETECT Groundwater monitoring fails to detect waste release outside injection zone. 0.5 This probability is based on professional judgment. Given a release of waste fluid through postulated confining zone breaches, required groundwater monitoring should detect a release. When the release is detected, the injection would be ceased, and the driving force for upward fluid movement would be eliminated. This sequence could fail if the monitoring locations are not at or down gradient of the location of the breach in the confining zone, or if the time between release and detection is long enough that a significant release occurs before corrective action is taken.
Event: Description: Probability: Basis:
EXTRACT Extraction of groundwater from same saturated zone as injection zone. Uniform distribution from 1.0E−05 to 1.0E−03. This probability is based on professional judgment. Deep-well injection zones contain nonpotable water, usually of high salinity, with no attractive resource value. A number of more useful water-bearing zones occur at shallower depths that can be accessed much more cost-effectively. The probability of this event occurring near an existing or former deep-injection well at any time in the foreseeable future is considered to be very low.
Event: Description:
NORECOGNIZE Failure to recognize that groundwater extraction is located within injected waste plume.
ELSE_TSANG_CH010.qxd
9/15/2005
3:45 PM
Page 131
Appendix: Basis for Event Frequency Probability Distributions
131
Probability: Basis:
Uniform distribution from 1.0E−03 to 1.0E−02. Assuming that someone in the future screens an extraction well at injection zone depth, there is the probability that they do not recognize the well has intercepted an injected waste plume. This event would require both the failure to recognize that the well is located within a documented Class I hazardous waste injection well AOR and that the extracted water contains waste. The distribution is based on professional judgment, taking into consideration significant uncertainties associated with time frames in the thousands of years as well as the small area of the plume relative to the entire saturated zone.
Event: Description: Probability: Basis:
OUTAOR Injection waste has migrated outside the AOR to an unconfined zone. Uniform distribution from 1.0E−05 to 1.0E−04. Migration of the injected waste plume outside the AOR is assigned a low probability of occurrence given the extensive characterization efforts required for the no-migration petition. It is conservatively assumed in the PRA that if this event occurs and the injected material is still characteristically hazardous, then a release to a USDW occurs. Horizontal and upward migration of injected fluid far from predicted ranges would be necessary for this to occur.
Event: Description: Probability: Basis:
WASTEPRESENT Injected waste has not transformed into nonwaste. Uniform distribution from 1.0E−02 to 1.0. This event addresses the probability that injected waste has not transformed into a nonhazardous form at a future time when either (a) groundwater is inadvertently extracted from the injected waste plume or (b) the plume has migrated outside the AOR to an unconfined zone. The assigned probability distribution takes into consideration (a) it is not uncommon to render the waste nonhazardous by pretreatment and dilution prior to or during injection, (b) injected waste attenuates in the plume, and (c) biodegradation and other transformation/loss processes may decrease hazardous constituents over time. Inadvertent extraction and migration outside the AOR are events with long time frames, and there is reasonable likelihood that these factors could have transformed the waste by the time of these event sequences.
Event: Description: Probability:
PUMPPA Annulus pump fails. Triangular distribution with min = 5.0E−05; mode = 3.0E−04; max = 5.0E−03. The European Industry Reliability Data Bank (Lannoy and Procaccia, 1996) provides a resource of compiled data for equipment failure rates. Based on the failure rates per hour (5.0E−07 to 5.0E−04) for pumps with long operating times, the daily (assuming a 10 hour daily operating period) probability of pump failure is between 5.0E−06 and 5.0E−03 day–1. These data are generally supported by similar mechanical failure rates from
Basis:
ELSE_TSANG_CH010.qxd
132
9/15/2005
3:45 PM
Page 132
A Probabilistic Risk Assessment of Class I Hazardous Waste Injection Wells
PRAs performed for the nuclear power industry. Range estimates for pump failures from a number of nuclear industry resources (McCormick, 1981) provide a median value of 3.0E−05 failures/hour (3.0E−04 failures/day). For the nuclear industry, redundancies and routine replacement ensure that the failure rates and consequences of pump failure are minimal. A triangular distribution was used for annulus pump failure rates, using the nuclear power industry value of 3.0E−04 failures/day as the mode and assigning the European database values as the extreme range values. Event: Description: Probability: Basis:
CHECKPA Annulus check value fails to open. Triangular distribution with min = 1.0E−04; mode = 3.0E−04; max = 1.0E−03. Given that the annulus pump fails, CHECKPA is the probability that the check valve, designed to keep the annulus fluid contained and pressurized in the annulus, stays open. This is an on-demand failure rate in that failure only occurs when the component is called to function. Data from McCormick (1981) give an on-demand failure rate for check values (fail to open) of 1.0E−04 to 1.0E−03 per demand (median of 3.0E−04). Because CHECKPA is conditional upon PUMPPA, and both are represented by the same AND gate within the fault tree, the on-demand probability is used directly.
Event: Description: Probability: Basis:
CONTROLPA Annulus pressure control system fails, resulting in underpressurization. Uniform distribution from 1.0E−06 to 1.0E−04. Control system failures are usually the result of electronic or electrical failures resulting from loss of signal function. Lannoy and Procaccia (1996) list the range of electrical/electronic failures from the compiled databases to be between 5.00E−08 and 1.00E−05 hour−1. For a one-day operating period, this range converts in to a failure probability of 1.2E− 06–2.4E−04 day−1. Since this range has no point of central tendency, a uniform distribution is selected for the PRA.
Event: Description: Probability: Basis:
CONTROLPI Injection pressure control system resulting in overpressurization. Uniform distribution from 1.0E−06 to 1.0E−04. This is a similar control system failure, as was described for CONTROLPA. Similar logic is used to specify a probability distribution.
Event: Description: Probability: Basis:
OPERRPA Operator error causes annulus pressure to drop below injection pressure. Uniform distribution from 5.0E−05 to 5.0E−04. Swain (1987) provides data on human error, showing a frequency of 1.0E− 05 error per action. Assuming the operator is performing five critical actions per day that could lead to a potential pressure drop, the daily failure rate is 5.0E−05. A uniform distribution that assumes this estimate is that lower bound was used; it is equally likely to be up to an order of magnitude of higher frequency of human error. Since all operator errors in this
ELSE_TSANG_CH010.qxd
9/15/2005
3:45 PM
Page 133
Appendix: Basis for Event Frequency Probability Distributions
133
PRA may be performed by either the same or a similarly trained operator, this and the other operator error event probability distributions were correlated in the Monte Carlo simulation using a correlation coefficient of 0.5. Event: Description: Probability: Basis:
OPERRPI Operator error causes injection pressure to rise above annulus pressure. Uniform distribution from 5.0E−05 to 5.0E−04. The same basis applies as for event OPERRPA, above.
Event: Description: Probability: Basis:
OPERRDET Operator fails to detect/respond to unacceptable pressure differential. Uniform distribution from 5.0E−05 to 5.0E−04. The same basis applies as for event OPERRPA, above.
Event: Description: Probability: Basis:
OPERRFRAC Operator error results in induced transmissive fracture through lower confining zone. Uniform distribution from 5.0E−05 to 5.0E−04. The same basis applies as for event OPERRPA, above.
Event: Description: Probability: Basis:
OPERINJ Operator fails to recognize changes in confining zone capacity. Uniform distribution from 5.0E−05 to 5.0E−04. The same basis applies as for event OPERRPA, above.
Event: Description: Probability: Basis:
CAPLOSS Loss of injection zone capacity results in overpressurization. Uniform distribution from 1.0E−05 to 1.0E−03. The capacity of injection zone rock is carefully studied for a Class I well as part of the site-selection process and no-migration petition. Given the extent of the characterization efforts involved, it is unlikely that a lack of capacity will be overlooked. This would be the result of a human error of omission, which occurs at a rate of approximately 1.0E−03 per demand. Because at least one additional independent review of this factor would be performed (e.g., by the regulatory agency), this frequency is assumed to be the upper bound of the distribution.
Event: Description: Probability: Basis:
PERMEA Confining zone has unexpected transmissive permeability. Uniform distribution from 1.0E−05 to 1.0E−03. The permeability of confining zone rock is carefully studied for a Class I well as part of the site-selection process and no-migration petition. Given the extent of the study efforts involved, it is unlikely that permeability will be incorrectly characterized. This would be the result of a human error of omission, which occurs at a rate of approximately 1.0E−03 per demand. Since at least one additional independent review of this factor would be performed (e.g., by the regulatory agency), this frequency is assumed to be the upper bound of the distribution.
ELSE_TSANG_CH010.qxd
134
9/15/2005
3:45 PM
Page 134
A Probabilistic Risk Assessment of Class I Hazardous Waste Injection Wells
Event: Probability: Description: Basis:
DISCONT Uniform distribution from 1.0E−04 to 1.0E−02. Presence of unidentified transmissive discontinuity. As per the discussion on the characterization efforts outlined above for PERMEA, it is unlikely that the geologic properties of the confining zone were not completely described. However, irregularities in the geological characteristics of the confining zone are possible, given the lateral extent of the injection zone. Thus a factor of 10 higher probability than was assigned to PERMEA was used.
Event: Description: Probability: Basis:
DETECTWELL Failure to identify abandoned well in AOR. Uniform distribution from 1.0E−03 to 1.0E−02. Based on similar arguments used for PERMEA and DISCONT, it is unlikely that the presence of abandoned wells within the AOR would remain undetected. However, records for abandoned wells could be missing or incorrect. The distribution range used is higher in error frequency to reflect this added consideration.
Event: Description: Probability: Basis:
ALARM Automatic alarm fails. Uniform distribution: 1.00E−05 to 1.00E−03. The frequency of alarm failures were analyzed by Davis and Satterwaite (1988) for fire hazards associated with the management and storage of radioactive waste. A failure probability of 5.00E−05 was determined. However, this assessment was based on alarms with high-reliability requirements specified for nuclear facilities. To account for the possibility that less-reliable equipment might exist at an injection well facility, this value was used as the lower bound of a uniform distribution that includes an equal probability that the alarm failure rate could be as much as a factor of 100 higher.
Event: Description: Probability: Basis:
SEISMFAULT Seismic event induces a transmissive fault or fracture. Uniform distribution: 1.00E−05 to 1.00E−04. Avoidance of areas prone to seismic activity is carefully studied for a Class I well as part of the site-selection process and no-migration petition. In addition, seismic factors are part of the design criteria for the well. Given the extent of the study efforts involved, it is unlikely that the well will be located where seismic activity has been incorrectly characterized. The event would more likely be a rare event that heretofore had not occurred at such a magnitude in the region of the well site, and therefore is not reflected in historical seismic event data. In addition, the seismic event would need to result in a transmissive fault or a fracture penetrating entirely the confining zone. This event was assigned, by judgment, a probability of occurrence in the range of 1 in 100,000 to 1 in 10,000.
ELSE_TSANG_CH010.qxd
9/15/2005
3:45 PM
Page 135
Appendix: Basis for Event Frequency Probability Distributions
135
Event: Description: Probability: Basis:
PLUGFAIL Identified abandoned well plug fails. Poisson distribution with 8E−04/well rate. Assignment of failure probability is based on TRC proper plug hearing files in Clark (1987). In this study, 2531 oil and gas fields were examined for plug leakage incidents in abandoned wells. Two leakage incidents were found. The number of abandoned wells could exceed the number of fields by a factor of 10. A conservative failure rate was estimated as 2 plug failures per 2531 fields, or 8E−04 plug failures per abandoned well (assuming only one well per field). Since this event meets the Poisson distribution requirements (see above in introductory remarks), a Poisson distribution was assumed using the failure rate determined here.
Event: Description:
TRANSUSDW Unidentified abandoned well is transmissive through upper confining zone to USDW. 0.1 There are no data on which to base this event frequency. The assumed probability of 0.1 assumed is believed to be very conservative considering that the event requires the abandoned well to provide a pathway, other than plug failure, to transmit injected waste through the entire confining zone.
Probability: Basis:
Event: Description: Probability: Basis:
Event: Description: Probability: Basis:
TRANSLCZ Unidentified abandoned well is transmissive from injection zone through lower confining zone. 0.1 There are no data on which to base this event frequency. The assumed probability of 0.1 is believed to be very conservative considering that the event requires the abandoned well to provide a pathway, other than plug failure, to transmit injected waste through the entire confining zone. INCOMPWASTE Injected waste is incompatible with previously injected material. Uniform distribution from 1.00E−05 to 1.00E−04. Material that is injected is well characterized to ensure that no chemical or physical reactions that could sufficiently alter the properties of the material in the injected zone take place. In addition, the no-migration petition process requires study of waste–host rock compatibility. This event also assumes sufficient waste volume and reaction with the confiningzone rock to result in a complete breach of the confining zone. This event was assigned, conservatively by judgment, a probability of occurrence in the range of 1 in 100,000 to 1 in 10,000.
This page intentionally left blank
136
ELSE_TSANG_CH011.qxd
8/20/2005
9:20 PM
Page 137
II. WELL TESTING AND HYDROLOGIC STUDIES
137
This page intentionally left blank
138
ELSE_TSANG_CH011.qxd
8/20/2005
9:20 PM
Page 139
Chapter 11
REPLACING ANNUAL SHUT-IN WELL TESTS BY ANALYSIS OF REGULAR INJECTION DATA: FIELD-CASE FEASIBILITY STUDY D. Silina, C.-F. Tsanga, and H. Gerrishb a b
Lawrence Berkeley National Laboratory, Berkeley, CA, USA U.S. Environmental Protection Agency, Chicago, IL, USA
11.1 INTRODUCTION Regulations governing deep injection of industrial wastes for disposal require regular tests for monitoring changes in formation hydraulic properties in the vicinity of the wellbore. Transient-pressure well testing, a procedure routinely used in the environmental and oil industries, is used to determine those changes. During such tests, the injection pressures and rates are recorded and analyzed to estimate the transmissivity and storativity of the rock in the vicinity of the wellbore. Numerous methods for analyzing such data have been developed since the pioneering paper by Theis (1935). The well test analysis methods are summarized in several monographs—e.g., Matthews and Russel (1967) and, Earlougher (1977). Traditional well test analysis is often based on estimating the slope of the pressure falloff curve in a special time scale, e.g., using the Horner plot method (Horner, 1951). This method is derived from the Theis’ radial flow solution, which is valid only if the initial pressure distribution is uniform. However, because of the operations preceding the test, such an initial condition may not hold true. For example, in Silin and Tsang (2002, 2003), it has been demonstrated that in the Horner plot method, this circumstance partially explains the deviation of the data points from the theoretically predicted straight line. In addition, in regular operations, the flow rates may not be constant, as is required for conventional analysis. In the new analysis method proposed in Silin and Tsang (2002, 2003), the well-test data are analyzed accounting for the pre-test operations and arbitrary flow rates. This method has been validated using synthetic and field well-test data. In this chapter, we demonstrate how this method can be applied to analyze regular pressure and flow-rate data from an injection well to estimate the formation’s hydraulic properties without interrupting operations. In this estimation, we use the code ODA (Operations Data Analysis) developed at Lawrence Berkeley National Laboratory. This code implements the methods and algorithms developed by Silin and Tsang (2002, 2003). The chapter is organized as follows: In the next section, we present a brief overview of the method and describe the procedure used in the analysis. Then, in the following section, we present the analysis of data from several injection wells. The results of this analysis are summarized in our conclusions.
ELSE_TSANG_CH011.qxd
8/20/2005
140
9:20 PM
Page 140
Replacing Annual Shut-in Well Tests By Analysis of Regular Injection Data
11.2 DESCRIPTION OF THE METHOD The procedure we use in this study was designed to estimate formation hydraulic properties from an injection pressures/injection rates data set obtained from an active injection well. Specifically, we assume that the operation conditions prior to the data time intervals are not known. The recovered parameters include formation transmissivity and storativity, skin factor, and an average reservoir pressure. Additionally, the method estimates an effective pre-test injection rate. This parameter approximately characterizes the nonuniform pressure distribution near the wellbore with a single number. The procedure consists of several steps. First, a data set has to be selected. Each data point has three components: the time of measurement, the injection pressure, and the injection rate. Any selected set must include variations in the pressures and flow rates. The selected data interval is further split into two parts: the beginning phase and the test phase. This splitting is imposed by the method of analysis only and is not related to the physical operations at the well. The test-phase data points of the pressure curve are used in a best-fitting procedure to estimate formation parameters, whereas the beginning-phase data interval is used for intermediate calculations. Note that we do not require a constant flow rate or well shut-in for either of the selected time intervals. In many practically important cases, the bottomhole pressure can be calculated from the instantaneous wellhead pressure. In fact, if an average reservoir pressure estimate is not required, the wellhead injection pressures can be used without any adjustments. A distinctive feature of the method is that an effective injection rate, denoted by Q−1, is introduced to account for the pre-test operations. Although this parameter plays an intermediate role in the fitting procedure, it can be used for an a posteriori assessment of the quality of the analysis results. If some information about the flow rates before the data interval is available, the discrepancy between the actual rates and Q−1 can be evaluated. A small discrepancy confirms a good quality of fitting, whereas a large discrepancy may indicate that either some minimization parameters need to be changed or a different data interval has to be selected for analysis. Let us denote by t0 and t2, respectively, the beginning and end times of the selected data set, and denote by t1 the splitting time between the beginning and test phases. The modified radial flow solution has the following form (Silin and Tsang, 2002, 2003):
冕
exp(B/(tτ)) B p(t) p(t0) AQ1Ei( ) A Q(τ )dτ sAQ(t), t1 tt2. (11.1) tτ tt0 t0 t
Here p and Q are the injection pressures and rates, respectively. We adopt the convention that the flow rate is positive if the fluid is injected. The coefficients A and B are related to the transmissivity T and storativity S as follows: 1 T , 4π A
2B S , π r 2A
(11.2)
where r is the effective wellbore radius. If the pressures and rates were measured between t1 and t2 at points θ1 , θ 2 ,ᠮ, θΝ , then the quality of fitting can be estimated using a quadratic criterion
冤
冥
2 1 N J 冱 p(θi)pdata(θi) , N i1
(11.3)
ELSE_TSANG_CH011.qxd
8/20/2005
9:20 PM
Page 141
11.3 Analysis of Field Data
141
where pdata is the measured injection pressure. The skin factor s, pre-test flow rate Q−1, and the coefficients A and B are the fitting parameters. An efficient best-fitting algorithm proposed in Silin and Tsang (2002, 2003) has been implemented in the code ODA, which was the main tool in this study.
11.3 ANALYSIS OF FIELD DATA The method described in the previous section is tested by the following procedure. First, the regular injection operation data are analyzed using the ODA code to estimate the formation parameters. Then, these results are compared with results obtained independently from conventional analyses of fall-off tests. The discrepancies are studied to understand the sources of the differences and to determine which method is more accurate. This procedure has been performed for the data from four deep waste disposal wells in Ohio, which have been in operation for over 10 years. The injected fluid is water with a dilute solution of chemicals. The injectant is treated by removal of particles before injection, so that the density and viscosity of the fluid practically equal those of water. The injection zone includes part of the Middle Run Formation and all of the Mt. Simon and Eau Claire Formations at depths 3223 to 2430 ft below the surface (Fig. 11.1). Each of these formations extends laterally far beyond the vicinity of the injection site. The Mt. Simon Formation is composed of sandstone and is between 2813 and 3153 ft deep, with porosities as high as 20% and averaging 12% for the entire formation. Permeabilities at Mt. Simon range from 0.0005 to 695 md for core samples and average as much as 64 md for the lowermost 183 ft thick division. The underlying Middle Run Formation is composed of argillaceous sandstone and siltstone, with porosity of 2% and permeability about 10 md in the uppermost 70 ft layer. The Eau Claire portion of the active injection interval is between 2775 and 2813 ft deep. This layer is composed mainly of sandstones with porosities ranging from 3.5 to 17% and an average permeability of 300 md. We illustrate the location of the wellbore by presenting Well B in Figure. 11.1.
1438’ Secondary vertical aquitard
Black River group
Eau Claire formation Mount Simon sandstone Middle Run
Confining zone Injection zone
Knox dolomite
Secondary containment sequence
1837’ 1855’ 2100’
Primary containment sequence Arrestment strata
2430’ 2640’ 2813’
Effective injection interval 3153’ Lower flow barrier
3223’ 3409’
Well B
Fig. 11.1. Schematic of the formation structure near the injection interval.
ELSE_TSANG_CH011.qxd
8/20/2005
142
9:20 PM
Page 142
Replacing Annual Shut-in Well Tests By Analysis of Regular Injection Data
The injection zone is subdivided into an effective injection interval and an arrestment interval. Fluid is injected directly into the active injection interval from the open-hole portions of the wells. The passive injection interval is between 2640 and 2775 ft deep and consists of two layers of the Eau Claire Formation, which are composed of sandstone and silty sandstones with generally moderate porosity (8%) and low average permeability (1−5 md horizontal and 0.003−0.004 md vertical), caused by occlusion of pore spaces by shale and dolomite. The portion of the injection zone serving as the arrestment interval consists of three layers of the Eau Claire Formation between the depths of 2430 and 2630 ft. These layers consist mostly of dolomite with some interbedded shale and sand, and contain confining units (dense carbonates and shales). Analyses of core samples indicate that the effective vertical permeability of the dolomite is 0.00005 md. Above the injection zone is a confining zone made up of two layers in the lower part of the Knox Dolomite from 2100 to 2430 ft deep. These layers are continuous for hundreds of square miles. Information about the depths and thicknesses of the injection intervals of the wells used in this study is summarized in Table 11.1. The average injection interval depth was used to calculate the downhole pressure. For ODA data analysis, we used hourly records of injection pressures and rates collected over time intervals of 3–4 days. No information about the operations immediately before or after data intervals is available. In some cases, a running averaging over a 3-hour window was applied. Fitting of such smoothed data is slightly more stable with respect to the selection of the intervals (t0, t1) and (t1, t2). How this averaging may affect the results is discussed below. The principal output parameters of the best-fitting procedure are the transmissivity and storativity of the near-wellbore formation, the skin factor, and the ambient reservoir pressure. In the runs where t0 was selected inside the data set, the quality of fitting was additionally confirmed by good agreement between this actual rate and the estimated effective pre-test injection rate Q−1. Here, a run means an instance of analysis of the selected data interval.
Table 11.1. The depths of injection zones for Wells A–D
Well A Well B Well C Well D
Minimal injection interval depth (ft)
Maximal injection interval depth (ft)
Injection interval thickness (ft)
Average injection interval depth (ft)
2783 2813 2810 2885
3077 3125 3140 3159
294 312 330 274
2930 2969 2975 3022
Table 11.2. Results of regular data analysis using code ODA for Well A Run no. ⇓
Data set
t0 (hours)
t1 (hours)
t2 (hours)
Transmissivity (d ft/cP)
Skin factor
Ambient pressure (psi)
1 2 3 4 5 6
Jun01 Jun01 Jun01 Jun01 Aug02 Aug02
0 15 30 45 0 15
23 19 40 49 17 18
95 55 80 85 71 51
4.44 4.28 3.84 4.0 3.8 5.07
0.26 0.06 −0.28 −0.8 1.4 1.82
1302 1322 1266 1267 1217 1353
ELSE_TSANG_CH011.qxd
8/20/2005
9:20 PM
Page 143
11.3 Analysis of Field Data
143
For Well A, several data intervals extracted from two data sets denoted by “Jun01” and “Aug02,” respectively, have been analyzed. The data sets are 72 and 96 hours long and approximately 13 months apart from each other. The results are presented in Table 11.2. The estimated skin factor is small in all runs. Results from Runs 1 through 4 are presented in Figure. 11.2. Note that in Runs 2–4, the time t2 is inside the entire data set Jun01. Therefore, the part of the respective curve between t2 and the end of the data set at 95 hour is a prediction based on fitting the data only between t1 and t2. All four matching curves practically coincide with the data. The results obtained from June 2001 data on different test intervals are in a good agreement, but they are different from the results obtained from August 2002 data (Runs 5–6). This data set is characterized by irregular fluctuations of the injection pressures and flow rates, and hourly sampling of the data may be insufficient for deriving reliable conclusions. Table 11.2 results can be compared with the results of the fall-off tests from previous years analyzed independently using conventional methods (Table 11.3). Note that the variation of the results of conventional analysis performed in different years also is significant: the transmissivity estimate of October 1996 is almost two times higher than that of December 1992. The transmissivity estimates obtained by the conventional method are consistently higher than those obtained by our method. This difference is partially compensated for by the difference in the skin factor. 2100 2050
Pressure [psi]
2000 1950 1900 Data Run 1 Run 2 Run 3 Run 4
1850 1800 1750 1700 1650 0
10
20
30
40 50 60 Time [hours]
70
80
90
100
Fig. 11.2. Examples of matching the Jun01 data set from Well A. Pressure fitting curves for different data intervals are in a good agreement and produce similar results (Runs 1–4, Table 11.2).
Table 11.3. Results of fall-off well test analysis for Well A Date of the test ⇒
Apr 1991
May 1992
Dec 1992
Jan 1994
Nov 1994
Nov 1995
Oct 1996
Oct 1997
Oct 1998
Transmissivity (d ft/cP) Skin factor Extrapolated pressure (psi)
12.26
9.13
7.89
12.06
12.05
15.57
17.59
13.88
12.28
−1.7 1324
−1.8 1350
−2.4 1336
−2.7 1394
−2.8 1406
−3.4 1423
−3.5 1422
−2.3 1445
−1.9 1451
ELSE_TSANG_CH011.qxd
8/20/2005
144
9:20 PM
Page 144
Replacing Annual Shut-in Well Tests By Analysis of Regular Injection Data Table 11.4. Results of regular data analysis using code ODA for Well B
Run # no. ⇓
Data set
t0 (hours)
t1 (hours)
t2 (hours)
Transmissivity (d ft/cP)
1 2 3* 4* 5 6 7* 8*
Jun01 Jun01 Jun01 Jun01 Aug02 Aug02 Aug02 Aug02
0 30 0 30 0 33 0 33
23 44 23 44 17 36 17 36
95 80 95 80 71 66 71 66
6.88 5.0 7.44 5.17 9.02 7.93 15.99 12.0
Skin factor
Ambient pressure (psi)
1.51 1.19 -0.47 0.11 1.76 −0.6 0.83 3.64
1383 1311 1389 1319 1450 1427 1493 1457
* Results for smoothed data.
Table 11.5. Results of fall-off well-test analysis for Well B Date of the test ⇒
Aug 1991
Dec 1991
Mar 1993
Mar 1994
May 1995
Jul 1996
Apr 1997
Apr 1998
Apr 1999
Apr 2000
Transmissivity (d ft/cP) Skin factor Extrapolated pressure (psi)
6.04
6.73
10.00
9.43
16.71
9.74
17.44
12.50
13.53
13.41
−4.2 1462
−4.2 1462
−2.0 1376
−1.9 1391
−1.9 1412
0.89 1401
−1.9 1348
−2.6 1376
−2.6 1368
−4.23 1381
2100 2000
Pressure [psi]
1900 1800 1700 1600 1500 1400 1300 1200
t0 0
10
20
t1 30
40 50 Time [hours]
t2 60
70
80
90
Fig. 11.3. Matching the Jun01 data from Well B (Run 4, Table 11.4). Note two different injection pressure regimes: between 0 and 50 hours and between 62 and 95 hours.
Results of analyses performed on Well B data are gathered in Tables 11.4 and 11.5. Analysis of smoothed data does not significantly affect the results: compare Runs 1–2 with 3–4, and Runs 5–6 with 7–8. However, there is a noticeable difference between the transmissivity estimates obtained at different time intervals in July 2001. This difference could be caused by the changed injection regime: the pressures on the interval analyzed in Runs 3
ELSE_TSANG_CH011.qxd
8/20/2005
9:20 PM
Page 145
11.3 Analysis of Field Data
145
and 4 are lower than the average injection pressure (Fig. 11.3). As for Well A, the results of conventional analysis for Well B also vary significantly between different years: the transmissivity estimate obtained in April 1997 is almost three times higher than the estimates obtained in 1991 (Table 11.5). The data from Well C also were collected over two time intervals: 96 hours in July 2001 and 168 hours in August 2002. In the 2001 data, the injection pressures at later times are higher than those in the beginning of the interval. This can explain the consistent difference between the transmissivity estimates in Runs 1 and 3, and 2 and 4. The 2002 data interval includes a fall-off test followed by resumption of regular operations. The ODA results are presented in Table 11.6. As with Well B, the difference between estimated parameters obtained from averaged and “raw” data is not large. Figure. 11.4 presents the 2002 data. The injection pressures on the intervals preceding the test data in Runs 5 and 6 (Fig. 11.4a) are higher than the pressures during the shut-in before the testing interval in Runs 7 and 8 (Fig. 11.4b). The transmissivity corresponding to the later times is smaller than the one obtained from the analysis of the fall-off curve. Results of conventional fall-off test analyses from previous years are presented in Table 11.7. Note that the transmissivity estimates obtained by analyzing the fall-off curve using ODA (Runs 5 and 7, Table 11.6) are lower than the estimates of the same data by conventional methods (Aug 2002 column in Table 11.7). This difference is partially compensated by the skin factor. We believe that the reason for this discrepancy is that in conventional analysis, the persistent residual influence of injection before the test, which is rigorously accounted by ODA, is attributed exclusively to the skin effect. The same circumstance could be the reason why the ambient pressures estimates obtained by ODA are, on average, higher than the extrapolated pressures in Table 11.7. The credibility of ODA results is confirmed by the fact that in both Runs 5 and 6, the obtained value of Q−1 was 164 gpm, whereas the actual pre-test flow rates fluctuated around 175 gpm, as shown in Figure. 11.4c. Operations data from Well D were collected over a 72-hour time interval in August 2002. Analysis results are presented in Tables 11.8 and 11.9. The flow rates fluctuate between 86 and 140 gpm (Fig. 11.5); the corresponding fluctuations of the injection pressures are within 160 psi, as shown in Figure. 11.6. The difference between the estimated values of the skin factor in Runs 2 and 8 is quite significant. In this case, the data averaging does not significantly affect the quality of curve fitting. However, the transmissivity and skin factor estimates are not stable (Fig. 11.5). We explain this instability by the fact that the data set does Table 11.6. Results of regular data analysis using code ODA for Well C Run # no. ⇓
Data set
t0 (hours)
t1 (hours)
t2 (hours)
Transmissivity (d ft/cP)
Skin factor
Ambient pressure (psi)
1 2 3* 4* † 5 6† 7* 8*
Jun01 Jun01 Jun01 Jun01 Aug02 Aug02 Aug02 Aug02
25 55 25 55 37 120 37 120
29 59 29 59 40 125 40 125
62 92 62 92 120 155 120 155
6.05 6.3 5.42 5.78 10.08 5.92 8.67 6.03
4.56 3.53 4.43 3.24 3.21 0.49 4.02 0.69
1527 1618 1507 1572 1391 1451 1398 1450
* Results for smoothed data. † Analysis of a fall-off test.
ELSE_TSANG_CH011.qxd
8/20/2005
146
9:20 PM
Page 146
Replacing Annual Shut-in Well Tests By Analysis of Regular Injection Data 2200 2100
Pressure [psi]
2000 1900 1800 1700 1600 1500 1400 1300 1200
t1
t0 0
20
40
t2 60
80
100
120
140
160
Time [hours]
(a) 2200 2100
Pressure [psi]
2000 1900 1800 1700 1600 1500 1400 1300 1200
t0 0
20
40
60
80
100
t1
120
t2
140
160
Time [hours]
(b) 200 180 Injection rate [gpm]
160 140 120 100 80 60 40 20 0
t0 0
20
t1 40
t2 60
80
100
120
140
160
Time [hours]
(c)
Fig. 11.4. 2002 injection rate data for Well C: (a) matching the pressure fall-off curve (Runs 5 and 6); (b) matching pressure curve at the resumption of regular operations (Runs 6 and 7); (c) flow-rate plot of—the rates before t0 fluctuates between 173 and 175 gpm. Table 11.7. Results of fall-off well-test analysis for Well C Date of the test ⇒
Sep 1992
Aug 1993
Aug 1994
Aug 1995
May 1996
Jul 1997
Jul 1998
Aug 1999
Aug 2002
Transmissivity (d ft/cP) Skin factor Extrapolated pressure (psi)
6.20
9.32
15.57
8.77
17.28
9.52
11.88
11.3
14.32
−2.8 1441
−1.3 1414
−2.3 1359
−4.4 1352
−2.5 1370
1.3 1406
−3.7 1300
−3.3 1372
−3.9 1377
ELSE_TSANG_CH011.qxd
8/20/2005
9:20 PM
Page 147
11.3 Analysis of Field Data
147
Table 11.8. Results of regular data analysis using code ODA for Well D Run # no. Data ⇓ set
t0 (hours)
t1 (hours)
t2 (hours)
Transmissivity (d ft/cP)
Skin factor
Ambient pressure (psi)
1 2 3* 4*
10 20 10 20
18 28 18 28
60 70 60 70
8.83 22.1 7.7 17.76
2.56 9.71 3.21 0
1457 1542 1420 1489
Aug02 Aug02 Aug02 Aug02
* Results for smoothed data.
Table 11.9. Results of fall-off well-test analysis for Well D Test Date ⇒
June 1995
June 1996
June 1997
June 1998
Transmissivity (d ft/cP) Skin factor Extrapolated pressure (psi)
13.92 −4.2 1299
12.35 −3.6 1220
15.43 −4.1 1260
7.29 −3.7 1180
(a)
(b)
Fig. 11.5. Well D: The quality of fitting is stable with respect to the selection of the data set and regardless of whether the data are smoothed (b) or not (a). However, due to the almost-steady-state character of flow, the results of analysis are not stable (Table 11.8).
ELSE_TSANG_CH011.qxd
148
8/20/2005
9:20 PM
Page 148
Replacing Annual Shut-in Well Tests By Analysis of Regular Injection Data
Fig. 11.6. Well D: Flow-rate fluctuations are well correlated with the pressure fluctuations (Fig. 5), which suggests that the character of flow is close to steady state.
not include transient flow interval. More frequent data sampling may help to resolve this instability.
11.4 CONCLUSIONS A new well test analysis method accounting for pre-test operations has been applied to analyze regular injection data during normal operations in real field conditions. We used the code ODA as the main tool, which implements the new method developed by Silin and Tsang (2002, 2003). Key parameters are the transmissivity and storativity of the formation in the vicinity of the wellbore. As by-products, the method also produces estimates of the skin factor and the effective pre-test flow rate parameter. The latter can be used for additional verification of the quality of analysis. Results from data analysis confirm the possibility of estimating the formation hydraulic properties and monitoring their changes over time, using regular operations data instead of or in conjunction with conventional well tests. The ODA method is based on analysis of large data intervals, and the flexibility in the selection of such intervals makes possible detection of variations of formation properties caused by changing the regime of operations. The recovered transmissivity factor is stable with respect to the selection of the data interval, so that a value of reasonable confidence can be obtained. The effective pre-test flow rate is close to the average actual rate prior to the test interval, which provides additional confirmation of the results. Results from conventional well test analyses vary significantly from year to year. However, they consistently show a higher transmissivity with a negative skin factor. The reason for this could be that both conventional methods and the method discussed here evaluate only the effective formation properties. Local heterogeneities can be interpreted in different ways. A lower transmissivity with a larger positive skin factor yields similar effects to higher transmissivity with a smaller or negative skin. The results obtained by the method
ELSE_TSANG_CH011.qxd
8/20/2005
References
9:20 PM
Page 149
149
discussed here are confirmed by comparison between the data and the theoretical curve extended beyond the fitting interval. To summarize, the applicability of this new method for estimating formation hydraulic properties based on regular operations data has been confirmed. The code ODA makes such analysis simple and inexpensive. Implementation of this method in the field can lead to automation of the process of formation-properties monitoring without interrupting regular operations.
ACKNOWLEDGMENTS This research has been supported by the U.S. Environmental Protection Agency (EPA), Office of Ground Water and Drinking Water, Underground Injection Control Program, under an Interagency Agreement with the U.S. Department of Energy under Contract No. DEAC03-76SF00098. The authors are thankful to BP Chemicals of Lima, Ohio, for providing the field data. The authors also thank the anonymous reviewer for useful suggestions.
REFERENCES Earlougher, R.C., 1977. Advances in Well Test Analysis, Monograph Series, 5. Society of Petroleum Engineers, New York. Horner, D.R., 1951. Pressure buildup in wells. In: Proceedings of Third World Petroleum Conference, The Hague, The Netherlands, pp. 503–523. Matthews, C.S. and Russell, D.G., 1967. Pressure Buildup and Flow Tests in Wells, Monograph Series. Society of Petroleum Engineers, New York. Silin, D.B. and Tsang, C.-F., 2002. Estimation of formation hydraulic properties accounting for pre-test injection or production operations. J. Hydrol. 265(1): 1−14. Silin, D.B. and Tsang, C.-F., 2003. A well-test analysis method accounting for pre-test operations. SPE Journal 8(1): 22−31. Theis, C.V., 1935. The relationship between the lowering of the piezometric surface and the rate and duration of discharge of a well using ground-water storage. Trans. AGU, 2: 519−524.
This page intentionally left blank
150
ELSE_TSANG_CH012.qxd
8/19/2005
6:47 PM
Page 151
Chapter 12
EXPERIMENTAL STUDY OF INJECTION-INTERVAL HYDRAULIC ISOLATION FROM OVERLYING FORMATION AT THE DISPOSAL SITE OF THE SIBERIAN CHEMICAL COMPLEX, USING HIGH-ACCURACY HYDRAULIC HEAD MEASUREMENTS A.A. Zubkova, V.A. Sukhorukova, A.I. Zykova, E.A. Redkina, V.M. Shestsakovb, S.P. Pozdniakovb, V.A. Bakshevskayb, and V.M. Kurockinc a
Siberian Chemical Combine, Seversk, Russia Faculty of Geology, Moscow State University, Moscow, Russia c All-Russia Designing and Research Institute of Production Engineering (VNIPIPT), Moscow, Russia b
12.1 INTRODUCTION Since 1963, radioactive waste has been injected into disposal Areas 18 and 18a in deep aquifers of the Cretaceous terrigenous formation at the Siberian Chemical Combine, which is located in the southwest corner of the Western Siberian Artesian Basin (Rybalchenko et al., 1996, 1998). In recent decades, extensive field investigation has been performed for hydrologic and geologic characterization of this formation. A grouping based on all geological attributes and hydrogeologic measurement defines a system of seven aquifers stacked one on top of the other with six intervening leaky-confining aquitards. Individual aquifers are identified as Aquifers I, II, III, IV, IVa, V, and VI, with Aquifer I being the deepest and Aquifer VI being the shallowest. The aquitards are identified as Aquitards A through F, with Aquitard A being positioned between Aquifers I and II. Radioactive waste is injected into Aquifers II and III at depths of about 300–400 m below land surface. Aquitard D, between Aquifers III and IV, is the protective confining layer that must prevent upward waste migration into the shallow groundwater system, where Aquifer V is the main aquifer used for water supply. Recent studies (Shestakov et al., 2002; Pozdniakov et al., 2005, this volume) indicate that the formation is heterogeneous and the mentioned aquitards are not perfectly confining clay layers. Sand and clay anisotropic bodies, with a typical width of 100 m and a typical depth of 1 m, form the internal heterogeneity of the aquitards as well as the aquifers. The volumetric portion of sand in the overall formation thickness is about 0.5. Potential migration of wastes through the aquitards into shallow groundwater zones presents a risk of contaminating surface water bodies and drinking water wells. To control the spread of underground waste, a system of monitoring wells is in operation at this site. The monitoring wells installed in Aquifer IV, within the injection areas, do not show trends in hydraulic heads over time related to the injection regime, while the wells installed in Aquifer III are affected by this regime (Fig. 12.1). The goal of this study is to measure the temporal pressure responses in Aquifer IV, overlying Aquitard D, on temporal termination of injection in Aquifer III. Specially designed high-resolution hydraulic sensors, connected to digital recording hardware, were used. These
ELSE_TSANG_CH012.qxd
152
8/19/2005
6:47 PM
Page 152
Experimental Study of Injection-Interval Hydraulic Isolation from Overlying Formation
Fig. 12.1. Groundwater head fluctuations in monitoring wells A-29 and A-45 located within injection Area 18. Screen of well A-29 captures Aquifer III, and screen of well A-45 captures Aquifer IV.
sensors allow the detection of pressure changes in absolute values that are four orders-ofmagnitude less than the pressure changes in the injection aquifer. Preliminary calculation shows that registration of such a response can be considered as evidence of leakage associated with preferential flow paths in the aquitard. 12.2 MONITORING EQUIPMENT AND MEASUREMENTS High-accuracy groundwater head measurements were performed over one month, in September 2002, during temporal termination of waste injection. Special equipment named “Uroven-1M,” developed by the design firm “Geophyspribor” of the Russian Academy of Sciences for automatic measurement and recording of monitoring parameters, was installed within Injection Area 18. This equipment was provided with sensors for water level and temperature measurement in monitoring wells, and sensors for atmospheric pressure recording. The accuracy of water level measurements was 0.5 mm. Water-level sensors were installed in two wells, T-8 and T-22, that captured Aquifer IV. Spatially, well T-22 was located about 10 m from the nearest injection well, N-13, and 500 m from the monitoring well, A44, that captured injection Aquifer III. In the vertical cross section, the screening intervals of wells T-22 and T-8 were located 30–40 m above the screening interval of injection and monitoring wells screened for injection to Aquifer III. This interval between Aquifer III and
ELSE_TSANG_CH012.qxd
8/19/2005
6:48 PM
Page 153
12.3 Monitoring Data Processing
153
Aquifer IV is composed of clay deposits with sand lenses, and is marked as Aquitard D. Water level, temperature, and atmospheric pressure were recorded digitally, once every 5 min, during 1 month of injection-temporal termination.
12.3 MONITORING DATA PROCESSING Preliminary numerical simulation of one month of injection termination, performed before sensor installation, indicated typical groundwater-head drawdown of about 10 m in Aquifer III, and that the drawdown of head in Aquifer IV could be in the range of 0.1–10 mm. This simulation used the 3-D high-spatial-resolution flow model of Area 18 described in Pozdniakov et al. (2005). Simulation was performed for a typical range of permeability and elastic storage values. The simulated drawdown of hydraulic head in Aquifer IV depends on the values of hydraulic conductivity, and elastic storage of clay and sand units, composing semipermeable layer D. The maximum modeling drawdowns were obtained in the simulation run by applying the maximum hydraulic conductivities acceptable for this site along with minimal storage. Thus, the accuracy of the “Uroven-1M” sensor allows monitoring the hydraulic response of Aquifer IV, on termination of the injection, for certain (unfavorable) combinations of hydraulic parameters in units composing semipermeable layer D. The measured change of hydraulic head in monitoring well A-44 (see Fig. 12.2) indicates that, during injection termination, the drawdown of hydraulic head in Aquifer III was about 6.5 m.
Fig. 12.2. Change of groundwater head in the monitoring well A-44 that captured the injection Aquifer III.
ELSE_TSANG_CH012.qxd
154
8/19/2005
6:48 PM
Page 154
Experimental Study of Injection-Interval Hydraulic Isolation from Overlying Formation
Processing of measurement results included decomposition of hydrographs in the wells by estimating water-level response to atmospheric pressure time-variation (see Fig. 12.3) and periodic response to Earth tides. After elimination of these effects, trend analysis of the water level was performed to estimate its directed change (drawdown) resulting from hydraulic response to termination of the injection (Fig. 12.4). Comparing Figures 12.2 and 12.4, one can see that the temporal changes in groundwater level in well T-22 differ from those changes in well A-44.
12.4 RESULTS The results of high-accuracy measurement processing show: During all measurement periods, water levels fluctuate synchronously in both monitoring wells with amplitude of 1 cm. Note that monitoring well A-44, which was closest to the sensor wells, captured an injection interval where the maximum observed drawdown of groundwater head was 6.5 m. The standard deviation of fluctuation, about 12 mm, mostly relates to variation of atmospheric pressure (see Fig. 12.3). After elimination of atmospheric response, the residual standard deviation became 3.9 mm. Earth tides with periods of 12 and 24 hours are recognized in level fluctuations, but their maximum amplitude is <1 mm. 2. After the elimination of atmospheric pressure and Earth tidal responses, residual water level fluctuations in both observation wells do not show a downward trend (see Fig. 12.4). This result is additional proof of good isolation of the injected interval from the overlying formation at the disposal site. 1.
Depth to groundwater level in Well T-22, mm
-1300 Fit Linear: Equation Y = -0.136 * X + 30.07 Coef of determination, R-squared = 0.83 -1320
-1340
-1360
-1380 9900
10000 10100 10200 10300 Atmospheric pressure, mm of water
10400
Fig. 12.3. Hydraulic response of well T-22 on atmospheric pressure fluctuation.
ELSE_TSANG_CH012.qxd
8/19/2005
6:48 PM
Page 155
References
155 10
Residual water level fluctuation in Well T-22, mm
Start of injection 5
0
-5
-10
-15 0
10 20 30 Time from termination of injection, day
40
Fig. 12.4. Residual fluctuation of groundwater level fitted by polynomial trend (dashed line).
ACKNOWLEDGMENTS This work was supported by the Civilian and Development Research Foundation through contract RG2-2395-MO-02. REFERENCES Pozdniakov, S.P., Bakshevskay, V.A., Zubkov, A.A., Danilov, V.V., Rybalchenko, A.I. and Tsang, C.-F., 2005. 3-D modeling of injected waste transport in sandy clay formation. In: J.A. Apps and C.-F. Tsang (Eds), Underground Injection Science and Technology. Elsevier Press, Amsterdam. Rybalchenko A.I., Pimenov, M.K., Kostin, P.P. and Kurochkin, V.M., 1996. Scientific and practical results of deep-injection disposal of liquid radioactive waste in Russia. In: C.-F. Tsang and J. Apps (Eds), Deep-Injection Disposal of Liquid Radioactive Wastes. Academic Press, NY. Rybalchenko, A.I., Pimenov, M.K., Kostin, P.P., Balukova, V.D., Nosuckhin, A.V., Mikerin, E.I., Egorov, N.N., Kaimin, E.P., Kosareva, I.M. and Kurochkin, V.M., 1998. Deep Injection of Liquid Radioactive Waste in Russia. Battelle Press, Columbus Richland, WA. Shestakov, V.M., Kuvaev, A.A., Lekhov, A.V., Pozdniakov, S.P., Rybalchenko, A.I., Zubkov, A.V., Davis, P.A. and Kalinina, E.A., 2002. Flow and transport modeling of liquid radioactive waste injection using data from Siberian Chemical Plant Injection Site. Environ. Geol., 42(2–3): 214–221.
This page intentionally left blank
156
ELSE_TSANG_CH013.qxd
8/22/2005
7:04 AM
Page 157
Chapter 13
GULF COAST BOREHOLE-CLOSURE-TEST WELL NEAR ORANGE, TEXAS J.E. Clarka, D.K. Bonurab, P.W. Papadeasc, and R.R. McGowend a
E.I. du Pont de Nemours & Co., Inc., Beaumont, TX, USA Bonura Geological Consulting Inc., Beaumont, TX, USA c Sandia Technologies, Houston, TX, USA d Terra Dynamics Inc., Austin, TX, USA b
13.1 INTRODUCTION AND BACKGROUND A borehole-closure (BHC)-test well protocol for a Gulf Coast plant near Orangefield, Texas, was developed by DuPont. This protocol addressed EPA’s concerns about movement of injected fluids toward the Orange salt dome as part of a “no-migration” demonstration petition. The BHC well (Orange Petroleum #35 Hagar) was located between the eastern flank of Orange salt dome (Fig. 13.1) and the plant site about 5 miles east and within the potential path of the 10,000-year modeled plume. Oil and gas have been heavily produced from the Orange salt dome area since 1919, and regulators were concerned about the possibility of “unknown” APs that might be conduits for fluid migration. Given the worst-case scenario that injected fluids might migrate across faults at Orange salt dome over a period of several thousand years, this test was to see if there was potential for migration of hazardous constituents from the injection zone upward through unidentified APs. Chemical fate transformation of low pH wastes to nonhazardous levels was also demonstrated through chemical reaction modeling, adding another margin of safety to the underground injection. Previous studies reported qualitatively that wells drilled in unconsolidated (soft) rock experience natural BHC (Johnston and Green, 1979; Davis, 1986; Johnston and Knape, 1986; Clark et al., 1987). This test was a quantitative analysis of natural BHC based on EPA recommendations. The worst-case scenario included: (1) a test interval within the injection zone consisting of a thin injection sand overlain by a thick, sand-free shale; (2) an open borehole with a maximum diameter equal to the largest abandoned well diameter (11 in.); (3) a mud program in accordance with drilling practices in general use during 1919; and (4) actual testing with 9.0 lb/gal brine, a worst case of an abandoned well without plugging records. Natural BHC would be proven if (1) a water flow or oxygen activation (OA) log, run at stations above the injection sand interval, showed no upward fluid channeling; and (2) a pressure transducer located outside the tubing and inside the casing showed no pressure buildup when a 100 psi pressure increase was applied. The 100 psi increase is conservative, because the maximum calculated value for potential pressure was 80 psi, which included the possible density sources of pressure increase: (1) a maximum specific gravity/density contrast between natural formation fluid and the injected waste of 0.092; and (2) a worst-case density drive if a plume potentially containing low-density waste extended from the plant wells to Orange salt dome for a maximum formation height of 2000 ft.
ELSE_TSANG_CH013.qxd
158
8/22/2005
7:04 AM
Page 158
Gulf Coast Borehole-Closure-Test Well near Orange, Texas
Fig. 13.1. Geographic location of plant site and BHC-test well.
More likely, the long-term effect of buoyancy occurs when the head of the plume has drifted from the plant to the dome and the waste plume levels off and fills the available anticlinal space at the dome. In this case, the available thickness of formation with potential lowdensity waste is considerably less, only 300 ft, and the pressure due to the buoyancy would be <12 psi (300 ft × 0.092 × 0.433 psi/ft). If the formation had been perfectly horizontal, then the resulting pressure increase equals one-half the thickness (ft) of the formation, times the density difference between the waste and formation fluid (Miller et al., 1986). In an abandoned AP, the casing would have corroded, based on conservative data from Orange salt dome AP well data and National Association of Corrosion Engineers data (Graver, 1985). With a maximum casing wall thickness of 0.557 in. for 8 5兾8 in. casing and a conservative corrosion rate of 20 mils per year, casing would corrode in 28 years—long before waste could reach the Orange salt dome. This time is consistent with the casing corrosion data available from producing wells in the Orangefield area. The geologic formations from 2000 to 8000 ft consist mainly of middle to upper Miocene sands, with lower Oligocene Anahuac Shale and Frio sands below. Tertiary sands and shales were deposited in a series of stacked, progradational wedges that dip and thicken toward the Gulf of Mexico. The lower Miocene Lagarto and middle Miocene Oakville Formation are very thick, fine- to very-fine-grained sands, silts, and shales deposited in fluvial and deltaic environments. The regional geologic structural setting is one of salt tectonics, with salt dome intrusions, minor salt ridges, and deep synclines. Orange salt dome is a piercement type salt dome, with the top of the salt at approximately 7000 ft and numerous radial faults. 13.2 TEST INTERVAL SELECTION Criteria for the test interval called for a thin, clean, injection sand overlain by a thick, sand-free shale within the correlatable injection zone of the site. Following evaluation and analysis of the mudlog, lithology samples, openhole logs, and visual examination of sidewall cores, the section chosen for the test was 36 net ft clean sand (2926–2962 ft; all depths are
ELSE_TSANG_CH013.qxd
8/22/2005
7:04 AM
Page 159
13.3 Borehole-Closure Testing
159
from kelly bushing) overlain by 88 net ft clean shale (2838–2926 ft). Casing was set at 2838 ft (Fig. 13.2). Porosity in the test sand ranged from 29.6 to 31.8%, based on sidewall cores and 29–31%, based on neutron-density log. Permeabilities for the test sand sidewall cores ranged from 900 to 1400 mD. X-ray diffraction of the test interval shales indicated >80% expandable smectite clays. Median grain size of 0.0046 in. was determined from sidewall cores in the sand. To best fit the particle size for a natural completion, we chose a screen assembly with 0.006 in. slotted, wire-wrapped reinforced tubing with 120 holes/ft at 3/8 in. diameter per hole. This screen construction minimized friction losses in the screen assembly. To satisfy another EPA worst-case condition, DuPont evaluated electric logs of representative APs in the modeled 10,000-year waste plume for continuity of the shale overlying the test sand. This shale was continuous and correlatable across the highest point of Orange salt dome. Also, the well location was updip (shallower) from the plant site, minimizing the geologic overburden that helps shale creep into open boreholes. 13.3 BOREHOLE-CLOSURE TESTING The well was drilled during the spring of 1991, to a depth of 5024 ft. Standard oilfield equipment was used for drilling and testing. Two cement plugs were set with the top of the
Fig. 13.2. BHC-test interval.
ELSE_TSANG_CH013.qxd
160
8/22/2005
7:04 AM
Page 160
Gulf Coast Borehole-Closure-Test Well near Orange, Texas
higher plug at 3027 ft, below the interval chosen for testing. After setting a 7 in. casing at 2938 ft and cementing, the open borehole was under reamed to 11 in. and prepared for testing. Natural BHC testing occupied 14 days. 13.3.1 Conditioning Hole, Mud Displacement, and Transducer Tests The open borehole was conditioned with 9.7 lb/gal mud, and then the drill string was pulled into the casing above the casing shoe. Drilling mud inside the casing was displaced with a fresh-water buffer followed by 9.1 lb/gal filtered brine (to limit mud invasion of the well screen), and the drill string was tripped out and laid down. The lower transducer was placed inside the well screen, approximately 5 ft from the top of the screen openings; the upper transducer was attached to the outside of the 2 7兾8 in. tubing approximately 120 ft above the lower transducer. The screen and tubing assembly with transducers was temporarily placed near the bottom of the casing shoe at 2838 ft. Each transducer was then tested under static and dynamic conditions to ensure that the electrical equipment was functioning properly. The lower transducer (temporary placement at 2758 ft) recorded a static pressure reading of 1305 psi. Correct operation was verified by a hydrostatic pressure measurement of 9.1 lb/gal of brine (0.052 × 2758 ft × 9.1 lb/gal = 1305 psi). The upper transducer (temporary placement at 2638 ft) was also operating properly (0.052 × 2638 ft × 9.1 lb/gal = 1248 psi). Another method of verifying that the transducers were recording accurately was to measure the distance by which the transducers were separated, i.e., (1305 – 1248 psi)/(0.052 × 9.1 lb/gal) = 120 ft. Filtered brine was pumped at various rates up to a maximum of 8.5 bbl/min to determine the friction loss in the screen assembly beside the lower transducer. The dynamic test conducted near the casing shoe revealed that the friction loss would be less than 12 psi for a 2 bbl/min flow rate in the screen assembly. The upper transducer reflected a 10 psi buildup for this same time, indicating that the 12 psi loss could not be entirely attributed to friction loss inside the screen. The BHC injection test itself was conducted at less than 0.5 bbl/min, much less than the 8.5 bbl/min during the dynamic pretest. 13.3.2 Mud Displacement and Shut In After the screen was run to the desired depth, the 9.7 lb/gal mud in the open borehole was immediately displaced with 9.1 lb/gal filtered brine. A total of 401 bbl of 9.1 lb/gal filtered brine was pumped through the injection tubing with returns to the surface (Fig. 13.3). Mud displacement caused an increase in the pressure of the upper transducer until the mud was displaced from the wellbore (Fig. 13.4). The discharge rate was gradually reduced to prevent sudden well surges that might cause the well screen to fill with sand. Final brine returns were clean, with only minute traces of gumbo shale. Following mud displacement, the wellbore was shut in and pressures were recorded. Recovery data for both transducers (Fig. 13.5) showed a slow pressure decline. Within 3–5 days, each of the transducers had reached static pressure. Data from the lower transducer (screened interval) showed stabilized pressure in less than 4 days, indicating that natural BHC had occurred. The lower transducer reflects static formation pressure of 1314 psi during the shut-in, whereas the upper transducer (inside the well casing) indicated a pressure–time slope change during shut-in. Only minor pressure changes occurred after this time period for the upper transducer, which would be expected because the brine could react with the shale below the casing shoe. Figures 13.4 and 13.6 show the pressures recorded from mud displacement to the
ELSE_TSANG_CH013.qxd
8/22/2005
7:04 AM
Page 161
13.3 Borehole-Closure Testing
161
Fig. 13.3. Well schematic after drilling mud was displaced, and well shut in and pressures were recorded.
Fig. 13.4. Data from upper transducer (2821 ft) from mud displacement to the end of test.
ELSE_TSANG_CH013.qxd
8/22/2005
162
7:04 AM
Page 162
Gulf Coast Borehole-Closure-Test Well near Orange, Texas
Fig. 13.5. BHC injection test showing upper and lower transducer pressures.
end of testing for the upper (2821 ft) and lower (2941 ft) transducers, respectively. Calculation of different fluid levels from the upper and lower transducers showed isolation of the two zones. 13.3.3 Injection Testing The purposes of the preinjection slug test (Fig. 13.6) were to determine: (1) if the screen was open and operating properly; (2) if the volume of water was sufficient to conduct a pressure buildup in the formation; and (3) if there was a pressure response in the upper transducer. Two series of five 2.5 gal brine slugs at 9.1 lb/gal were performed 1 week after shut in. Falloff curves in the lower transducer indicated that the screen was open. There was no pressure response in the upper transducer during or after the slug testing, indicating that the two transducers were indeed isolated and the well was holding pressure. The pre-injection slug test revealed that a pump truck would be required to control the low flow rate of 9.1 lb/gal brine injection. Because the low flow rates could be lower than a truck could pump (<20 gpm), a valve was installed to regulate even lower flow rates. Early testing data showed that the lower transducer was recording a pressure buildup, whereas the upper transducer had no pressure increase. The flow rate was increased from 16 to 22 gpm to obtain a 40 psi buildup. Before reaching 50 psi of formation buildup and before pressure testing with brine injection, Schlumberger was called in to run a water flow log (OA log) to check for upward fluid channeling (Figs. 13.7 and 13.8). While waiting on Schlumberger, the flow rate was reduced to 16 gpm to conserve brine. The upper transducer continued to show no pressure change from injection, except for minor early temperature anomalies associated with the cooling effect of injection fluids.
ELSE_TSANG_CH013.qxd
8/22/2005
7:04 AM
Page 163
13.3 Borehole-Closure Testing
163
Fig. 13.6. Data from lower transducer (2941 ft) from mud displacement to the end of test.
Transducer pressure recordings and flow rates are shown in Figure 13.9. OA logging was conducted in the tubing under pressure control conditions at 90 psi above static formation pressure at three different levels—25, 50, and 75 ft—above the test sand (depths of 2900, 2875, and 2850 ft, respectively). Neither vertical fluid flow nor channeling through the shale section above the sand was recorded. Following the pressure test at 90 psi, the formation was pressured to 110 psi above static, which was greater than EPA’s 100 psi requirement, so this test was more conservative than required. The OA log was rerun at the same stations described previously with the same results: no vertical flow was detected. The borehole was then pressured to 140 psi over static, or 40 psi above EPA’s requirement. Again, the OA log was run at the same three stations and with the same results: no vertical flow was detected. For these tests, flow rates were reduced in an attempt to maintain constant formation pressure during each OA log run. Flow rates were increased to obtain the next formation pressure OA log run; however, the formation pressure continued to increase, and the flow rates were further reduced to, maintain a consistent formation pressure increase over static. Both the upper transducer and the OA logging indicate no upward channeling of fluid. The final run of the OA showed no upward movement of fluids even as shallow as 25 ft above the injection sand, the minimum depth above the injection sand tested. Following testing, pressures on both transducers continued to be monitored, with pressure on the lower transducer returning to static within 2–3 hours. DuPont conducted a postinjection slug test to verify that the lower transducer was still working and that the upper transducer would respond to fluid placed in the annulus prior to cutting the transducer lines and pulling the tubing and screen assembly (Fig. 13.10). The upper transducer was working, and there was no bleed off of pressure into the lower transducer. When the annulus was
ELSE_TSANG_CH013.qxd
164
8/22/2005
7:04 AM
Page 164
Gulf Coast Borehole-Closure-Test Well near Orange, Texas
Fig. 13.7. Well schematic during water flow log.
Fig. 13.8. Recorded pressures from upper and lower transducers and brine temperature during OA logging.
ELSE_TSANG_CH013.qxd
8/22/2005
7:05 AM
Page 165
13.3 Borehole-Closure Testing
165
Fig. 13.9. Recorded pressure from upper and lower transducers and brine temperature compared to flow rate during OA logging.
Fig. 13.10. Post-injection transducer test results for upper and lower transducers.
filled to the surface with fluid, there was a pressure increase recorded for the upper transducer, as expected. However, there was no pressure increase recorded for the lower transducer, because the lower transducer was isolated from the annulus by the shale closing around the tubing assembly. This post-injection test verified natural BHC—the sealing of
ELSE_TSANG_CH013.qxd
8/22/2005
166
7:05 AM
Page 166
Gulf Coast Borehole-Closure-Test Well near Orange, Texas
the wellbore by the shale (expandable smectite clay layers) between the injection sand and the tubing.
13.4 SUMMARY During this test, natural BHC was demonstrated in Gulf Coast sediments and a rate of BHC was quantified. The BHC-test was designed and constructed for a worst-case scenario that: (1) hazardous injectate might migrate across a nonsealing fault; (2) the hazardous injectate encounters an AP; (3) the AP has the maximum diameter; and (4) the AP is filled with 9.0 lb/gal brine. The test interval was thin sand overlain by thick, sand-free shale. The BHC well was tested to the specified pressure and higher, with no upward fluid flow or channeling detected during OA logging, even with a minimum of 25 ft of overlying shale. The pressure differences recorded by the two transducers proved that no channeling of fluid had occurred, because there was no path of communication: the shale had closed in around the tubing in the open-hole section. The rate of BHC was determined to occur rapidly over a period of few days. Results of the test provided conclusive evidence that in this geological environment, a borehole closes naturally within a few days, even under a worst-case scenario. This test provided a significant additional margin of confidence that there would be no migration of hazardous constituents from the injection zone through an improperly constructed or unplugged AP. REFERENCES Clark, J.E., Howard, M.R. and Sparks, D.K., 1987. Factors that can cause abandoned wells to leak as verified by case histories from class II injection. Texas Railroad Commission files. Proceedings of the International Symposium on Subsurface Injection of Oilfield Brines, Underground Injection Practices Council, Oklahoma City, OK, pp. 166–223. Davis, K.E., 1986. Factors affecting the area of review for hazardous waste disposal wells. Proceedings of the International Symposium on Subsurface Injection of Liquid Wastes, March 3–5, New Orleans, LA. Water Well Journal Publishing Co., Dublin, OH, pp. 148–194. Graver, D.L. (Ed.), 1985. Corrosion Data Survey-Metal Section, 6th edn. National Association of Corrosion Engineers, Houston, p. 7. Jackson, M.P.A. and Galloway, W.E., 1984. Structural and depositional styles of Gulf Coast Tertiary continental margins: Application to hydrocarbon exploration. American Association of Petroleum Geologists Continuing Education Course Note Series No. 25. Johnston, O. and Green, C.J., 1979. Investigation of artificial penetrations in the vicinity of subsurface disposal wells. Texas Department of Water Resources. Johnston, O.C. and Knape, B.K., 1986. Pressure effects of the static mud column in abandoned wells. Texas Water Commission LP86-06. Miller, D., Fisher II, T.A., Clark, J.E., Porter, W., Hales, C.H. and Tilton, J.R., 1986. Flow and containment of injected waste. Proceedings of the International Symposium on Subsurface Injection of Liquid Wastes, March 3–5, New Orleans, LA. Water Well Journal Publishing Co., Dublin, OH, pp. 520–559.
ELSE_TSANG_CH014.qxd
8/20/2005
9:26 PM
Page 167
Chapter 14
INTERPRETATION OF TRANSIENT PERMEABILITY TESTS TO ANALYZE THE EVOLUTION OF A BRINE-FILLED SALT CAVERN Aron Behr Freiberg University of Mining and Technology, Freiberg, Germany
14.1 INTRODUCTION The chapter presents an approach developed to evaluate transient permeability tests on spherical salt samples. In the course of inverse modeling, the complete step-by-step history of the flow under study is reproduced by a purpose-built computer code. The results obtained under various load conditions were systemized to derive the relationship between the permeability and stress state in rock salt around caverns. A long-term behavior analysis of deep underground salt caverns assumes particular importance in view of their potential for disposing of different hazardous wastes. In this connection, the Solution Mining Research Institute (SMRI), USA, supported transient salt permeability tests (performed by Ecole Polytechnique, France) to investigate fluid permeation through salt cavern walls, and the change in the permeability as a function of stress (Berest et al., 2000). A brine-filled cavern was imitated by a cavity leached out in a salt sphere. Different multistage test scenarios were realized to mimic how stress would develop in an actual salt formation, whereby the test parameters responsible for the stress state in a sample would be changed in a prearranged way. The sophisticated design of the experiments, as well as the amount and quality of source information contained in the tests, required a special interpretation approach. This chapter describes an evaluation method that was developed and a number of important results obtained by the author within the framework of the SMRI research project. Permeabilities, estimated at acceptable confidence intervals, were drawn on to find, or to verify, some patterns in relation to the stress state. The evaluation of five tests was performed by an in-house software tool specifically designed for high-resolution estimation of the hydrodynamic parameters in low-permeability rock. The computer code included a numerical simulator with an automatic history-matching capability (including generating the useful statistics) that allowed it to identify permeability development by reproducing the total history of the flow, without exception for the relatively short stress buildup and stabilization stages.
14.2 EXPERIMENTAL DESIGN A salt sphere (24 cm diameter) was chosen as an experimental sample in the permeability tests (Fig. 14.1). The reason for this was that when using cylindrical cores, the leakage of the
ELSE_TSANG_CH014.qxd
168
8/20/2005
9:26 PM
Page 168
Interpretation of Transient Permeability Tests to Analyze the Evolution
Fig. 14.1. Illustration of the LMS permeability test (from Ratigan, 2003).
fluid around a sample is comparable to flow through it. In the center of a sample, a small cavity (around 4 cm diameter) was leached out. The cavity was connected with the injected system through a thin metallic tube, and the salt sphere was placed inside the tight jacket. The small space between jacket and salt was filled with draining glass fibers that transmitted the external load to the external face of the sample thus mimicking the overburden pressure in a salt formation. This pressure is different from brine pressure in the space, regarded as the outer pore pressure. Thus, three pressures were independently controlled during the tests: (1) internal cavity pressure, (2) external pore pressure, and (3) confining pressure. In addition, both the injected volume (into the cavity) and the withdrawn volume were measured. However, the difference between the confining pressure and outer brine pressure caused shrinkage of the outer space and, in turn, misinterpretation of the data on the withdrawn flow. So, these measurements were unsuitable for test evaluation. All tests were performed following the multistep scenarios: the controlled pressures were changed step by step with time (or kept constant) in a predetermined way. In doing so, more emphasis was focused on the gap between confining pressure and cavern pressure. At the beginning of each test, the pressures were gradually raised to the specified level during the so-called stress build-up period. This was followed by a relatively long stabilization period, over which the pressures were maintained constant.
14.3 CODE PATE FOR TEST EVALUATION An in-house numerical simulator with an automatic inverse modeling capability (called “PaTe”, short for Packer Test) was originally developed for the high-resolution evaluation of in situ tests in low-permeability rocks (Behr et al., 1998, 2002) and was adopted here to the
ELSE_TSANG_CH014.qxd
8/20/2005
9:26 PM
Page 169
14.3 Code PaTe for Test Evaluation
169
design of the laboratory experiments under consideration. The underlying mathematical model describes a transient spatial (axial symmetric) single-phase gas/liquid flow in rock induced by the pressure changes in gas/liquid-filled chambers within the porous media. The mass conservation equation, with consideration for Darcy’s law in cylindrical coordinates r and z, is k ∂p k ∂p ∂ ∂(φρ) 1 ∂ ᎏ ᎏ rρ ᎏ ᎏ ⫹ ᎏ ρ ᎏ ᎏ ⫽ ᎏ µ ∂r µ ∂z ∂z ∂t r ∂r
冢
冣
冢
冣
(14.1)
where p is the pressure [M L–1 T–2], t the time [T], β the fluid density [M L–3], k the permeability [L2], ϕ the porosity [–], and µ the viscosity [M L–1 T–1]. For slightly compressible fluids, which are relevant to the case at hand, it is assumed that d(ϕ ρ) ⫽ β ϕ ρ d(p), with β being the total compressibility of fluid in saturated porous media [M–1 L T2]. Both the spatial-distribution heterogeneity of permeability and porosity and their temporal changes are considered. As an extra possibility, the stress (pore pressure), and time dependence of the permeability and porosity may be taken into account as a function of a prearranged form, including (optionally) some of the parameters to be estimated. Three types of boundary conditions are foreseen: 1. The first-type boundary condition on the surface with a given pressure. 2. No flow boundary condition on the impermeable boundary. 3. The boundary condition on the interface between rock and chamber where the change of the pressure pc is monitored or managed (thereby, chamber pressure is coupled with the filtration process in rock). qc ⫹
( p )ρ (p )] ᎏᎏ , 冕 ρ ᎏµk grad (p) dA ⫽ d[V dt c
c
c
c
(14.2)
〈c
where Ac is the interface area (L2), Vc the chamber volume (L3), and qc the additional source/sink term (M T–1). The problem outlined is solved numerically with a simulation code based on the finite-difference (control-volume) method with fully implicit formulation by the block-center geometry. The domain under study is divided into cells, or gridblocks, by concentric cylindrical surfaces and planes, which are normal to z-axis. All the gridblocks are classified as either passive or active elements. The passive cells reflect the no-flow boundary or impermeable inner regions, while the active blocks reflect permeable formations, inner regions that are open to the atmosphere, outer boundaries with known pressure, and the chambers. Note that the concept “chamber” extends to any set of “hollow” gridblocks, which are connected in a closed system. One volume, one pressure, and one compressibility value should be attributed to each chamber specified in this way. The schematic of the model being considered is illustrated in Figure 14.2. The inner chamber presents the totality of all the gridblocks forming the cavity. The outer chamber is the space outside the salt sphere enclosed by impermeable boundaries of the simulation domain. The compressibility and volume of the total injection system (cavity + injector + transmitter + tube) are attributed to the inner chamber. The cells of the tube are denoted as passive elements. The approach to inverse modeling consists of minimizing the differences between the observed and simulated pressure responses by using the least-squares objective function. A variety of optimization algorithms has been implemented in the PaTe and can be chosen for the inverse procedure. These are the Gauss–Newton method, the Powell method, the Gradient method, and our own modification of the Levenberg–Marquardt method. In addition, the simulator was integrated with the universal inverse program UCODE (Poeter and Hill, 1998). The main advantage of this PaTe-version is that UCODE can generate the useful statistics.
ELSE_TSANG_CH014.qxd
170
8/20/2005
9:26 PM
Page 170
Interpretation of Transient Permeability Tests to Analyze the Evolution
Fig. 14.2. Model for test evaluation by PaTe.
14.4 TEST EVALUATION PROCEDURE The general principle of the evaluation procedure is based on the belief that the total history of flow development can be reproduced by PaTe, and that the unsteady-state effects are not ignored by fitting the simulated pressure to be measured. In this context, a simulation period covers all test phases, also including pressure buildup and stabilization. The inversing was performed in such a manner that the controlled cavity pressure was considered a response for matching, while the measured injected brine volume was used as a known input (source term related to the cavity-chamber). For lack of reliable data on the withdrawn flow, the outer chamber lost its importance for history matching, and the brine pressure at the external sample space should provide a needed boundary condition. In accordance with step-by-step test performance, the parameters to be estimated were introduced as averaged stage values. In case of fast-changing permeability, and also in view of the piecewise constant temporal distribution of sample parameters, stages were divided into multiple substages. Note that some uncertainty exists in the relevant model parameters, in addition to the permeability. First, this is true for the porosity and compressibility of the injection system. Porosity could be extracted from the time in which the elastic wave induced by the increase of pressure in the cavity reaches the external surface of sample (Behr et al., 1998, 2002). However, because the outflow measurement failed, this time was not detectable. Also, the sensitivity analysis showed a poor response in the cavity pressure to change in the porosity, in comparison to the permeability or compressibility of the injection system. As a result, the porosity remained unchanged and common to all calculations (rough value of the porosity ϕ ⫽ 0.1 was given, as by Berest et al., 2000). The compressibility of the injection system βinj [M–1 L T2] is a parameter with a high sensitivity for unsteady-state flow, especially at the beginning of the pressure buildup. It was necessary for each test to identify the extra “early” compressibility βinj0 by the first measurement points. After that, it became possible to obtain a good fit to the observations over the entire stress buildup period by applying the prearranged, experimentally estimated, compressibility for the rest
ELSE_TSANG_CH014.qxd
8/20/2005
9:26 PM
Page 171
14.5 Evaluation Results
171
of the time and by matching only one permeability value. Figure 14.3 illustrates this through one of the studied tests (Test No. 16).1 For each subsequent stage, one permeability value and one compressibility value for the injection system are identified. The compressibility is corrected to match the highly transient flow with stepwise changing cavity pressure, thereby providing the ultimate in fit for the two jump measurement points (the last point of one stage and the first point of the next stage) required for a progressive simulation throughout all stages. Note that in all the tests, the deviation of the estimated compressibilities from the experimental values was not more than 10% (except of course for the “early” values at the beginning of the pressure buildup). This reinforces the reliability of measurements and the modeling approach.
14.5 EVALUATION RESULTS The results for three selected permeability tests are presented in Figures 14.4 – 14.7. The development of permeability can be tracked from the stress-buildup stage to the end of the experiment (except for a few periods where there are too many uncertainties). The measured and simulated cavity pressures are entered on the graphs to show the fit, which is achieved by inverse modeling. Confining pressures are also shown to give an indication of the stress state in the salt sample. 14.5.1
Test No. 7 (Fig. 14.4)
During the test, the cavity pressure was first increased by 1 MPa incremental steps to the level of the confining pressure and then kept constant, while the confining pressure was decreased step-by-step. External fluid pressure remained unchanged (10 MPa). The test shows very
20
Cavity pressure (MPa)
k = 1.1.10-18 m2 15 inj0 = -1 . -3 10 2.4 10 MPa
inj = 0.5 . 10-3MPa-1
5
Observations Solution
0 0
0.1
0.2
0.3
0.4
0.5
Time (hours)
Fig. 14.3. Interpretation of the stress buildup stage in Test No. 16.
1
From here on, test numbering is consistent with that in the test report (Berest et al., 2000).
0.6
ELSE_TSANG_CH014.qxd
8/20/2005
172
9:26 PM
Page 172
Interpretation of Transient Permeability Tests to Analyze the Evolution 1.E-17
22 20
Pressure (MPa)
18 1.E-19 16
Cavity (observations) Cavity (simulation) Confinement Permeability
14
1.E-20 1.E-21
12
Permeability (m2)
1.E-18
1.E-22
10 u. z.
1.E-23
8 -2
18
38
58
78 98 Time (hours)
118
138
158
20
1.E-17
18
1.E-18
16
1.E-19
14
1.E-20
Cavity (observations) Cavity (simulation) Confinement Permeability
12 -2
38
78
118 Time (hours)
158
198
Permeability (m2)
Pressure (MPa)
Fig. 14.4. Evolution of the permeability in Test No. 7.
1.E-21 238
Fig. 14.5. Evolution of the permeability in Test No. 13.
clearly the impact of the pressure gap between confining and cavity pressures on permeability, which varied over three orders of magnitude. It gradually decreased during the first half of the test. At the fourth and fifth stages (from hour 97 to hour 127), we observed a considerable slowing in the growth of the injected volume, and even a slight decrease. This effect can be attributed to the shrinkage of the cavity, which makes itself evident in the range where the cavity and confining pressures come close together. As a result, we determined that the “apparent” flow rates could lead to misinterpretation of the test with respect to permeability, and therefore assigned this test interval to the uncertainty zone (u.z.). At subsequent stages,
ELSE_TSANG_CH014.qxd
8/20/2005
9:26 PM
Page 173
14.5 Evaluation Results
173 1.E-11
20
1.E-14
18 1.E-17
u. z.
uncertainty zone
16
Permeability (m2)
Pressure (MPa)
Cavity (observations) Cavity (simulation) Confinement Permeability
1.E-20
1.E-23
14 -2
98
198
298
398
498
Time (hours)
Fig. 14.6. Evolution of the permeability in Test No. 17.
Fig. 14.7. Test No. 17. Development of the injected volume and cavity pressure after pressure reduction. Best possible match to the calculated cavity pressures.
the mechanical phenomena were not more dominant than the hydrodynamic process. A significant increase of the permeability was observed as the difference between the cavity and confining pressures increased. 14.5.2
Test No. 13 (Fig. 14.5)
Cavity pressure was increased stepwise, while the confining pressure was kept constant during the entire test. In contrast to Test No. 7, the greater part of the permeability decline occurred during the initial periods; thereafter, the permeability changed only slightly. This can be accounted for by the fact that in this case the outer pressure was controlled to be in parallel with
ELSE_TSANG_CH014.qxd
174
8/20/2005
9:26 PM
Page 174
Interpretation of Transient Permeability Tests to Analyze the Evolution
the cavity pressure, with a gap of 2 MPa. This pressure difference is responsible for the extent of heterogeneity in the permeability distribution along the radius. However, the test design is not suitable for inferring conclusions with respect to this phenomenon. At a later time, as the confining pressure was approached, a trend toward permeability increase was observed. Moreover, the test showed strong evidence for a rapid increase in permeability occurring when the internal pressure exceeded the confining pressure. 14.5.3
Test No. 17 (Figs. 14.6 and 14.7)
The test scenario during hours 0–220 was very similar to that of Test No. 13, except that the cavity pressure came up to the immediate vicinity of the confining pressure just after the completion of the stabilization period, and increased in steps with smaller pressure amplitude. After 220 hours, conditions returned to the initial state (i.e., immediately after the stress-buildup phase) for 100 hours. Then, the same test schedule employed for the first 220 hours was used. The plot clearly shows the characteristic periods of the system evolution: (1) a decrease in permeability until the cavity pressure reaches the level of the confining pressure, (2) a time interval within which cavity shrinkage predominantly influenced the internal pressure, and (3) a steep increase in permeability. The situation following the sharp pressure reduction after 220 test hours permitted a more sophisticated treatment of experimental data. Figure 14.7 shows the downward jump in the curve of cumulative injected volume at the time of the pressure drop. We would expect that the further fall in the volume is balanced out by the flow from the sample into the cavity, which was caused by the quick change of the pressure gradient sign in the vicinity of the cavity surface. Extended study of pressure distribution in the spherical sample, performed by PaTe, shows that the period of contrary flow was very short. Even at a very low permeability (10–23 m2), the opposite sign in the pressure gradient disappeared in a period of a few minutes. This means that the long-term decrease in cumulative volume can be explained only by a change in the cavity volume. In this case, the simulated cavity pressures approached the observed pressures as the permeability changed to extremely small values. Nevertheless, the visible gap between the two curves remained (Fig. 14.7). In such a situation, this gap acted as a measure of cavern creep intensity. As observed in Figure 14.7, the shrinkage effect relaxed during the two subsequent stages of the test.
14.6 RELATIONSHIP BETWEEN PERMEABILITY AND STRESS STATE The data concerning the evolution of permeability under a variety of conditions can be considered sufficiently reliable to find, or to verify, some patterns in the relationship between the permeability and the stress state of the salt sample. At the same time, however, the design of the tests, as well as the amount and quality of source material enclosed in the five tests, do not allow definite conclusions to be drawn that support the most modern theories. The effective stress is equal to the compressive total (salt matrix) stress minus the brine pressure. When the cavity brine pressure exceeds the overburden pressure, salt matrix destruction (i.e., formation of microfractures) begins. Berest et al. (2000) observed a slow increase in the permeability just before this balance point. But the dramatic increase does not take place until the effective stress reaches a certain threshold value. We have attempted to bring these descriptive conclusions to an applicable and unified model. The results for all five tests are presented in Figure 14.8 by variables, making generalization possible. The abscissa
ELSE_TSANG_CH014.qxd
8/20/2005
9:26 PM
Page 175
14.6 Relationship between Permeability and Stress State
175
1.4
Normalized logarithm of permeability ratio
1.2 Test No. 3 1 Test No. 7
0.8
Test No. 13
0.6
Test No.16
0.4
Test No. 17
0.2 0 -1.5
-1
-0.5
0
0.5
1
1.5
Normalized (cavity pressure-confining pressure) Fig. 14.8. Normalized relationship between the permeability and stress state.
is the gap between cavity pressure and confining pressure normalized to the difference between cavity and external fluid pressures: Cavity pressure⫺Confining pressure 苶p苶 ⫽ ᎏᎏᎏᎏᎏ . ∆ Cavity pressure⫺External pore pressure
(14.3)
Note that introducing the fluid pressure difference as a scale resulted in reducing the scatter of points. The ordinate is the normalized logarithm of the permeability ratio: ⫽ ᎏᎏ . 冢 冣冒ln冢 ᎏ k 冣 ln k ⫺ln k
k l苶n苶 k苶 ⫽ ln ᎏ kmin
k0
ln k⫺ln kmin
min
0
(14.4)
min
Here, the permeability of the test stage is divided by the smallest (during the test) value kmin, which is close to the permeability of the virgin sample. Bearing in mind the possible exponential form of the relationship between permeability and effective stress (presented, for instance, by Berest et al., 2000), the logarithmic scale is used. An additional normalizing takes into account that, in some cases, the permeability ratio is not much more than one during a typical period; whereas in other cases, it may vary over wide limits (103–104). In Equation (14.4), k0 is the initial value of the permeability, obtained during the stress-buildup period, and is therefore close to (or coincides with) the maximum permeability over the test. Thus, the permeability span stretching from kmin to k0 is a characteristic scale of the permeability change. The data in Figure 14.8 are reasonably fit by a piecewise linear function, with a break point slightly to the left of the origin of coordinates: ∗ ∗ l苶n苶k苶 ⫽ ␣ ∆p ⫺∆p , ∆p ⫽⫺0.05…⫺0.1 . ∗ ∗ if ∆p ⬎ ∆p , α ⫽ 1.5, if ∆p ⬍∆p , α ⫽ ⫺0.5 (14.5)
冢
冣
It must be emphasized that this function is no more than a result of handling limited data obtained in the five tests. In the range within which the cavity pressure exceeds the confining
ELSE_TSANG_CH014.qxd
8/20/2005
176
9:26 PM
Page 176
Interpretation of Transient Permeability Tests to Analyze the Evolution
pressure (“right wing” in Fig. 14.8), Equation (14.5) reflects the actual nature of the fracturing phenomenon. As for the “left wing,” it is difficult to estimate to what extent the time dependence of the fissure closure/creation processes under compressive stresses affects the results. If the typical time of such processes is relatively large, we can assume that another scenario of the test performance would cause a sufficiently different correlation between permeability and pressure-gap values, where the cavity pressure is smaller than the confining pressure.
14.7 CONCLUSIONS Permeability evolution during multistep transient tests was simulated by means of a method based on inverse numerical modeling. Three main characteristic periods were observed: 1. A decrease in permeability until the internal pressure approaches the immediate vicinity of the confining pressure. 2. A time interval within which cavity shrinkage has a predominant influence on the internal pressure. A slight increase in permeability is observed, or can be expected, when system behavior is not reproducible by the hydrodynamic model with realistic parameters and, correspondingly, the uncertainty in the estimated permeability value remains very high. In such a situation, the slope of the curve for the injected volume may reverse the sign, and the change of volume cannot be balanced out by the flow from the sample. 3. A rapid increase in permeability when the cavity pressure noticeably exceeds the level of the confining pressure. Other features inferred by analyzing the evaluation results do not have a general nature, and hence need further investigation or statistical verification. At the same time, the concept for estimating cavern creep intensity can be considered a useful and applicable recommendation. The results of all five tests were unified in terms of the normalized logarithm for permeability ratio versus normalized difference between cavity pressure and confining pressure. This relationship was approximated as a piecewise linear function, with a break point slightly to the left of the origin of coordinates.
REFERENCES Behr, A., Förster, S., Häfner, F., and Pohl, A., 1998. In situ messungen kleinster permeabilitäten im festgestein (In situ measurements of extremely low permeabilities in solid rock). Freiberger Forschungshefte, A 849: 274–284, Freiberg University. Behr, A., Voigt, H.D., Häfner, F., and Belohlavek, K.-U., 2002. An advanced well test with automatic model calibration for soil and rocks. In: Proceedings of ModelCARE 2002 (International Conference on Calibration and Reliability in Groundwater Modeling), Prague, Czech Republic, 17–20 June 2002, Acta Universitatis Carolinae-Geologica, 46(2/3): 274–277. Berest, P., Brouard, B. and De Greef, V., 2000. Salt Permeability Testing—The Influence of Permeability and Stress on Spherical Hollow Salt Samples, SMRI Project No. 2001-1. Laboratoire de Mecanique des Solides, Ecole Polytechnique, Paris, France. Poeter, E.P., and Hill, M.C., 1998. Documentation of UCODE, a computer code for Universal Inverse Modeling. U.S. Geological Survey Water-Resources Investigations Report 98-4080. Ratigan, J.L., 2003. Summary Report—The Solution Mining Research Institute Cavern Sealing and Abandonment Program, 1996 through 2002. Research Report 2002-3-SMRI.
ELSE_TSANG_CH015.qxd
8/20/2005
9:31 PM
Page 177
Chapter 15
POTENTIAL CORROSION AND MICROBIOLOGICAL MECHANISMS AND DETECTION TECHNIQUES IN SOLUTION MINING AND HYDROCARBON STORAGE WELLS Ken E. Davis and Larry K. McDonald Subsurface Technology, Inc., Houston, TX, USA
15.1 INTRODUCTION The diversity of operational problems associated with solution mining and hydrocarbon storage in salt caverns covers a wide range of technologies and operator experience. In some cases, corrosion has been a contributor to the overall problems. Portions of this article are excerpts from a research report funded by the Solution Mining Research Institute (SMRI) titled State of the Art Review of the Understanding, Mitigation and Monitoring of Corrosion in Brine and Water Piping, Tubing and Casing at Brine Production and Hydrocarbon Storage Cavern Facilities. This report and practical experience confirm that corrosion in brine solutions is a major cause of equipment degradation, pipeline failure, and downhole tubular failure in the brine mining and hydrocarbon storage industries. No matter how many billions of dollars corrosion prevention, maintenance, and repair might cost the operators of salt caverns and other types of hydrocarbon storage facilities, this cost would never equal the possible catastrophic effects of a corrosion-related failure. Over the past few years, the mining industry has placed increasing emphasis on minimizing corrosion failures by mitigating corrosion through planned maintenance. New technologies to prevent corrosion continue to be developed, and cost-based corrosion management techniques are available to lower corrosion mitigation costs. For example, we must: ● Increase awareness of the large corrosion costs and potential savings. ● Change the misconception that nothing can be done about corrosion. ● Change policies, regulations, standards, and management practices to increase corrosion cost savings. ● Improve education and training of operators in recognizing corrosion control methods. ● Advance design practices for better corrosion management. ● Develop prediction and performance assessment methods. This chapter presents a general description of corrosion processes, various mitigation measures, and monitoring techniques applicable to brine production and hydrocarbon storage in solution-mined salt caverns and depleted gas reservoirs. It also addresses the effects of bacteria, which can live in saturated brine solutions. However, research to date has not identified the halophilic bacteria known to be corrosive. Further research is needed to address this issue.
ELSE_TSANG_CH015.qxd
178
8/20/2005
9:31 PM
Page 178
Potential Corrosion and Microbiological Mechanisms and Detection Techniques
15.2 THE CORROSION PROCESS Corrosion in brine and mining water is an important issue in solution mining, because it is the primary degradation mechanism within steel pipe and well tubulars. It is also important in oil/gas well drilling, which uses salt-saturated muds. In the storage industry, concentrated salt brines are used to displace hydrocarbons from salt caverns. All corrosion in brine is electrochemical or galvanic corrosion. For galvanic corrosion to occur, four conditions must all be satisfied: 1. There must be anodes (positive/oxidizing) and cathodes (negative/reducing). 2. There must be a potential difference between the anodes and cathodes. 3. There must be a metallic path between the anodes and cathodes. 4. There must be an electrolyte. A schematic of the corrosion process for iron is shown in Figure 15.1. Note that Figure 15.1 is merely an illustration. Metal atoms do not necessarily dissolve at a single point on a metal surface, nor are cathode areas restricted to one area on the surface. In the case of localized corrosion, such as pitting, these processes are limited to localized areas. However, in the case of general corrosion, the reactions occur randomly over the metal surface (Patton, 1981). A galvanic cell acts just like a battery, and all the necessary conditions are met in brine piping systems. The voltage difference between the anodes and cathodes (thermodynamic driving force for the reaction) is one of the main determining parameters for the reaction rate. This voltage difference is proportional to the free energy change for the reaction. The total circuit resistance of the galvanic cell determines the amount of corrosion current, which flows in an Ohm’s law circuit. The corrosion current determines the corrosion rate. For example, 1 amp of current flowing for 1 year can dissolve approximately 20 lbs of iron. 15.2.1 The Oxygen Corrosion Process In almost all cases, corrosion in brine is caused by oxygen (Smart and van Oostendorp, 1999). The chemical reaction is simple. Under aerated neutral pH conditions in brine, oxygen reacts
Fig. 15.1. A schematic representation of a corrosion cell (after Patton, 1981).
ELSE_TSANG_CH015.qxd
8/20/2005
9:31 PM
Page 179
15.2 The Corrosion Process
179
with iron and water by the following equation: Fe → Fe2 2e−
Anode half-cell reaction
(15.1a)
1 O 2 2
H2O 2e− → 2(OH)−
Cathode half-cell reaction
(15.1b)
1 O 2 2
H2O Fe → Fe(OH)2
Complete reaction
(15.1c)
Corrosion rates are controlled by: ● The concentration of oxygen in the brine and the diffusion of dissolved oxygen to the metal at the cathode. ● Flow velocity, which controls the thickness of the boundary layer at the surface, affecting oxygen diffusion rates. ● The conductivity of the solution, which controls ionic charge transfer in the electrolyte. Corrosion of carbon steels resulting from oxygen attack occurs because this dissolved gas can enter water in many ways. Dissolved oxygen occurs naturally or can be manmade in the system. Fresh water from a natural source (e.g., river or lake water) can be saturated with oxygen. The oxygen content, water pH, and temperature govern how corrosive the water will be. The aggressive corrosive effects of oxygen in fresh water at various temperatures are shown in the literature. For example, in tap water having a temperature of 70°F, a pH of 7, and dissolved oxygen content of 2.0 ppm, corrosion rates of over 200 mL/yr have been published in the literature. Oxygen can also be introduced into the water by aeration, thus putting oxygen into formerly oxygen-scarce waters. Manmade practices could prove difficult to control, because each source would need to be isolated and monitored. The best practice would be to identify and eliminate each source. Oxygen can enter a system in several ways, for example: ● Drawing water from a pond too close to the surface and occasionally sucking air. ● Defective or worn out water pump seals (pond discharge) on the suction side. ● Saturating return water or return brines with air at the pond location by allowing the return to fall freely into the brine pit. Oxygen is corrosive, even at low concentrations, because of its participation in creating differential aeration cells beneath scale deposits on metal surfaces. With oxygen present, deposition of scales on the metal surface is rarely uniform and never fully protective. Oxygen can be easily transported through these porous scales, which allows the active corrosion process to continue. The resultant corrosion form caused by oxygen is pitting. In many cases, this pitting becomes severe in highly localized areas, resulting in sporadic or spot perforations of well strings. Figure 15.2 depicts a deep pit in the first joint of the pipe body of a hanging string. Oxygen corrosion can be minimized by using corrosion-resistant alloys and nonmetallic materials, and by removing the oxygen from the system with chemical injection (Patton, 1981). Typical examples of the latter prevention method include the use of oxygen scavengers such as: ● Sodium sulfite 2Na2SO3 O2 → 2NaSO4 ●
Sodium bisulfite 2NaHSO3 O2 → Na2SO4 H2SO4
ELSE_TSANG_CH015.qxd
8/20/2005
180
9:31 PM
Page 180
Potential Corrosion and Microbiological Mechanisms and Detection Techniques
Fig. 15.2. Deep pit in first joint of pipe body hanging string (Davis and McDonald, 1998). ●
Ammonium bisulfite 2NH4HSO3 O2 → (NH4)2SO4 H2SO4
●
Sulfur dioxide 2SO2 2H2O O2 → 2H2SO4
●
Hydrazine N2H4 O2 → N2 2H2O
15.2.2 Effect of Salt on Corrosion The first principle in brine corrosion that needs to be recognized is that salt by itself does not cause corrosion and does not enter into the reaction. However, salt does affect the reaction rate due to its effects on brine conductivity and oxygen solubility. Corrosion in brine is caused mainly by dissolved gases, especially oxygen. Corrosion can also be caused by carbon dioxide and hydrogen sulfide, if present, chlorine, and possibly bacteria. 15.2.3 Oxygen Corrosion Rates in Brine The corrosion rate of steel in brine is directly proportional to the oxygen concentration (ASM Metals Handbook, 1987), as is shown in Figure 15.3. This direct proportionality holds to a concentration as low as 50 parts per billion (0.05 ppm). At oxygen concentrations below 50 ppb, other mechanisms become rate controlling. Therefore, one way to control corrosion in a brine system is to remove the oxygen to this level. 15.2.4 Oxygen Solubility in Brine The maximum concentration of oxygen in brine, as a function of temperature and salinity, is given by the following equation (ASM Metals Handbook, 1987): ln[O2]173.4292 249.6339 (100/T) 143.3483 ln(T/100)21.8492 (T/100) S[0.033096 0.014259 (T/100)0.0017000 (T/100)2]
(15.2)
where oxygen concentration O2 is expressed in mL/L; salinity, S in parts per thousand; and temperature T in K.
ELSE_TSANG_CH015.qxd
8/20/2005
9:31 PM
15.2 The Corrosion Process
Page 181
181
Fig. 15.3. Effect of oxygen concentration on the corrosion of low carbon steel in slowly moving water containing 165 ppm CaCl2. The 48-hour test was conducted at 25°C (76°F) (ASM Metals Handbook, 1987).
The most common method of characterizing the strength of a brine is its salinity, S, given in parts per thousand (‰). Salinity is defined as the total weight of salts in 1 kg of brine when all bromides and iodides have been replaced by an equivalent quantity of chlorides, and all carbonates are replaced by an equivalent quantity of oxides. Salinity is measured either by measuring the chlorinity or conductivity of the brine. Chlorinity is related to salinity by S 1.80666 Cl, where salinity and chlorinity are measured in parts per thousand. The combination of oxygen concentration and brine content results in a corrosion rate for oxygen-saturated brine, as shown in Figure 15.4. At first, the addition of salt increases the corrosion rate, but at about 3% brine, the reduced oxygen solubility results in a lowering of the corrosion rate as salinity increases until saturation. The quiescent uninhibited corrosion rate of steel in saturated brine, also saturated with oxygen, is about 10 mL/yr at 70°F, with the oxygen content of the saturated brine at 0.80 mL/L. In the present case, where the oxygen concentration is estimated to be 0.35 mL/L, the uninhibited oxygen corrosion rate is then estimated at 4.4 mL/yr on the basis of its oxygen concentration. 15.2.5 Oxygen Concentration Cell An oxygen concentration cell is a galvanic cell that is created when two areas on a piece of pipe have different oxygen concentrations. The difference in oxygen concentration causes a potential difference to develop between the two areas. An oxygen concentration cell will develop beneath sediment or scale deposits, beneath bacteria colonies, and under gas bubbles. The electrolyte under the deposit typically becomes depleted in oxygen, as a result of oxygen consumption by some mechanism such as corrosion or bacterial activity.
ELSE_TSANG_CH015.qxd
182
8/20/2005
9:31 PM
Page 182
Potential Corrosion and Microbiological Mechanisms and Detection Techniques
Fig. 15.4. Corrosion rate of steel in oxygen-saturated brine at 70°F (Uhlig and Revie, 1985).
Looking at the two oxygen-corrosion half-cell reactions of steel: 1 O2 2
H2O 2e− → 2(OH)−
Fe → Fe2+ 2e−
Cathode half-cell reaction Anode half-cell reaction
(15.3)
indicates that oxygen is involved only in the cathode half-cell reaction. Therefore, an electrolyte with a high concentration of oxygen will stabilize the cathode reaction. The area with less oxygen then becomes the anode and corrodes. As a result, corrosion occurs beneath deposits where oxygen is depleted. It also occurs in dead-leg areas where flow is stagnant and oxygen can also become depleted. This typically results in corrosion along the bottom of the pipe if sediments are the cause of the galvanic cell, and in a randomly scattered pattern all over the pipe if bacteria are the cause. The strength of an oxygen concentration cell is proportional to the difference in oxygen concentration between the bulk solution and under the deposits. In saturated brines, therefore, the driving force will be relatively small compared to fresh water. The high conductivity of the brine will counteract this, however, and most likely will give an oxygen concentration cell corrosion rate proportional to the normal ratio of oxygen corrosion in brine, as compared to fresh water. Oxygen concentration cells are not as active in saturated brines as in more dilute brines, since the solubility of oxygen is less. Figures 15.5(a) and (b) show the pitting type of corrosion caused by oxygenated brine. 15.2.6
Carbon Dioxide Corrosion
Carbon dioxide can also cause corrosion in brine, but is not expected to be significant in the case of mined brine. In oxygen-free 5% brine, deWaard and Lotz (1993) have developed a correlation for the initial corrosion rate, before the development of protective iron carbonate corrosion-product films. Their algorithm— log CR (mpy) 8.78 - 2320/T°K - 0.00555 T°C 0.67 log (PPco2, psi), (15.4) where log CR (mpy) is the log corrosion rate (mL/yr) T is the temperature (K) T is the temperature (°C) PPco2, psi is the carbon dioxide partial pressure (psi)—
ELSE_TSANG_CH015.qxd
8/20/2005
9:31 PM
Page 183
15.2 The Corrosion Process
183
Fig. 15.5. Examples of pitting corrosion.
has found wide acceptance in analysis of wells containing carbon dioxide. The actual mechanism of carbon dioxide attack is that of acid attack after dissociation of the carbon dioxide to carbonic acid. Naturally occurring acids or buffering ions in the water can have a large effect on the corrosion rate. This is often the case in corrosive gas wells, where acetic acid is often produced along with the gas, a product of bacterial degradation of methane over geological time. Production of formation water, on the other hand, can sometimes buffer the acidity from carbon dioxide and reduce corrosion rates. 15.2.7 Hydrogen Sulfide Corrosion Corrosion from hydrogen sulfide in brine is similar to carbon dioxide, as dissolved hydrogen sulfide also dissociates into a weak acid (Patton, 1981). Hydrogen sulfide corrosion is
ELSE_TSANG_CH015.qxd
184
8/20/2005
9:31 PM
Page 184
Potential Corrosion and Microbiological Mechanisms and Detection Techniques
usually not as rapid as carbon dioxide corrosion; however, corroded iron is almost immediately precipitated as iron sulfide on the surface of the steel. As a result, metal loss from hydrogen sulfide corrosion is usually low. The serious aspect of hydrogen sulfide corrosion is that hydrogen sulfide corrosion can embrittle steel by catalyzing the entry of atomic hydrogen into the steel. In steels with a hardness of Rockwell C-22 or less, the hydrogen can precipitate at internal microdefects and recombine to form hydrogen gas. Continued collection of the gas can build up to extraordinary pressures, measured to be above 300,000 psi. This very high pressure causes blisters to form in the steel and can cause hydrogen-induced cracking oriented in the direction of stress. Cracks form and grow, sometimes connecting together, and can cause failure of a pipe. In high-strength steels, with yield strengths above 80,000 psi, the hydrogen ions can precipitate along metal crystal planes and interfere with crystal plane bonding. This results in embrittlement of the steel. The phenomenon is known as sulfide stress cracking or hydrogen embrittlement. Steels can fail in a brittle manner, fracturing just like glass and resulting in catastrophic failure. This problem must be addressed in the design of the equipment by specifying low-strength grades of steel, or specialty alloys or coatings. Figure 15.6 shows the effect of hydrogen sulfide corrosion. 15.2.8 Effect of Flow Velocity Flow velocity can have a large effect on brine corrosion, but is often ignored. Velocity can have four important effects: 1. Stagnant or low velocity usually gives a low corrosion rate, but pitting is likely. 2. Low velocity allows sediment to accumulate in a pipeline. As a rule of thumb, sediments can accumulate at flow velocities less than about 5 ft/s. 3. Velocity increases the corrosion rate of brine by a factor of 7/8 (V7/8) (approximate), but there is an upper limit for the effect of velocity on the corrosion rate of steel in seawater, as shown in Figure 15.7.
Fig. 15.6. Corrosion due to hydrogen sulfide (SMRI, 2003).
ELSE_TSANG_CH015.qxd
8/20/2005
9:31 PM
Page 185
15.2 The Corrosion Process
185
Fig. 15.7. The effect of velocity on the corrosion rate of carbon steel in seawater (Uhlig and Revie, 1985).
4. High velocity in brine can cause erosion corrosion due to wall shear stress on passivating corrosion-product films and also by carrying particulates. The mechanism of erosion corrosion is the impingement of particulates on the wall of a pipe, especially at elbows and other points of directional change. The velocity required for erosion corrosion in steel pipe is about 10–12 ft/s. Other studies (Annand et al., 1977) have shown that the effect of velocity on the corrosion rate of carbon steel pipe by seawater can be given by Corrosion rate constant (V 7/8) quiescent corrosion rate.
(15.5)
The 7/8-power relationship was reported to be valid even when a thick rust film had developed, when corrosion inhibitors were added, and when the pH was changed. The slope and intercept would change, but the 7/8-power law remained valid. For quiescent conditions, when freshly exposed steel was exposed to deaerated seawater containing 0.05 ppm oxygen, the corrosion rate is given by Corrosion rate (mpy) 0.057V 7/8 0.4.
(15.6)
This equation indicates corrosion rates of 0.5 mpy at 1 ft/s, 0.6 mpy at 5 ft/s, and 0.8 mpy at 10 ft/s. These rates are generally considered satisfactory for corrosion control and justify the deaertion of seawater down to 0.05 ppm oxygen for corrosion control in seawater injection projects. For fully aerated seawater, and for erosive conditions in which a protective film cannot build up to a thickness of 1/8 in. or more, Annand et al. (1977) have described the corrosion rate as Corrosion rate (mpy) 11.4V 7/8 63.
(15.7)
This equation indicates corrosion rates of 74 mpy at 1 ft/s, 110 mpy at 5 ft/s, and 148 mpy at 10 ft/s. Any of these corrosion rates would be unacceptable for most piping systems, and illustrates why untreated seawater piping must be either coated, made of a corrosionresistant alloy, or frequently replaced. Corrosion rates for deaerated and aerated seawater are shown in Figure 15.8. Figure 15.9 depicts erosion/corrosion in the near-surface connection of a hanging string that was removed from a hydrocarbon storage well.
ELSE_TSANG_CH015.qxd
186
8/20/2005
9:31 PM
Page 186
Potential Corrosion and Microbiological Mechanisms and Detection Techniques
Fig. 15.8. The velocity effect on steel pipe wall corrosion due to air-saturated seawater containing 5 ppm ( ) and 10 ( )ppm oxygen (by Annand et al., 1977).
Fig. 15.9. Erosion/corrosion in near-surface connection (Davis and McDonald, 1998).
15.3 MICROBIOLOGICALLY INFLUENCED CORROSION (MIC) The MIC of carbon steels has been attributed to sulfate-reducing bacteria (SRB). Of perhaps even more importance to corrosion are acid-producing bacteria (APB), which decompose nonliving organic material by oxidizing it in a series of steps, first to organic acids and eventually to carbon dioxide and water. These bacteria come from both freshwater sources, such as natural water (i.e., well, river, lake, and seawater) used for cavern leaching or for hydrocarbon displacement, and from bacteria that already exist in the formation. These water sources are in recent years being used more frequently than well water for injection into brine cavern mining and/or hydrocarbon storage facilities.
ELSE_TSANG_CH015.qxd
8/20/2005
9:31 PM
Page 187
15.3 Microbiologically Influenced Corrosion (MIC)
187
15.3.1 The Microbiological Environment Bacteria are microscopic single-cell organisms that prefer to attach to metal surfaces in the form of discrete colonies. These colonies are covered by a biofilm, a protective layer of polysaccharides emitted by the bacteria that acts to protect the bacteria underneath it (see Fig. 15.10). Colonies contain many different species of bacteria that are synergistic with each other—i.e., they live better with each other than alone. For instance, APB deplete the water of any oxygen underneath the biofilm, enabling strict anaerobes such as SRB to grow. APB then partially oxidize carbon sources to form organic acids, with the acids serving as nutrient for SRB. Corrosion products, such as iron oxide and iron sulfide, will form deposits by being trapped in the biofilm. Other bacteria can oxidize ferrous iron to ferric iron, precipitating ferric hydroxide and manganese compounds from solution. Under the biofilms, any oxygen in the bulk fluid can be used by aerobic APB and even anaerobic APB to develop the anaerobic conditions required for SRB to thrive. For the most part, SRB decompose the organic acids given off by the APB, taking oxygen from sulfate ions for this purpose and giving off hydrogen sulfide. In other words, these anaerobic bacteria remove oxygen from sulfate ions and produce hydrogen sulfide. In time, these deposits form a hard crust (containing iron sulfide) on the metal surface that essentially protects the colony. Corrosion is caused by an oxygen concentration cell established under the bacteria colony by both depletion of oxygen and the presence of iron sulfide. MIC has a characteristic appearance of randomly scattered phat corkscrews and tend to pit. The resultant pitting can also be described as an aerial view of an open pit mine in which tiny channels or grooves are being cut as one goes deeper into the mine. Figure 15.11 shows corrosion resulting from SRB that has completely penetrated the body of the pipe. When active, bacteria can be identified by the appearance of black, red, or green slime on the wall of the pipe or vessel. Another indication of MIC can be smell. A freshly pulled well string or opened pipeline may have the classic “rotten egg” odor if MIC is present. Once an active MIC pit has developed, bacteria may no longer be necessary in the corrosion process, because the pit may become self-sustaining and continue to corrode. MIC is often found in low-salinity waters such as mining water or seawater. Whether MIC is an active mechanism in brine pipelines is not known at the present time. Most bacteria cannot survive in a saturated brine environment, and those that can have not been
Fig. 15.10. Schematic of a sessile bacteria colony under a biofilm on the surface of a metal (SMRI, 2003).
ELSE_TSANG_CH015.qxd
188
8/20/2005
9:31 PM
Page 188
Potential Corrosion and Microbiological Mechanisms and Detection Techniques
Fig. 15.11. Sulfate-reducing bacteria-influenced corrosion (SMRI, 2003).
shown to cause corrosion. A similar but nonbacterial type of corrosion can occur in brine pipelines caused by oxygen concentration cells forming beneath sediment deposits in the bottom of a pipeline (Fig. 15.12). Salt-tolerant organisms are widely distributed among the bacteria, fungi, yeasts, algae, and protozoa (Kushner, 1978). Bacteria have been around for over a billion years. They are highly adaptive and are great survivors, with spore-forming bacteria able to survive for thousands of years under hostile conditions, only to thrive when conditions improve. When considering bacteria, one should think of them as surviving or thriving, not living or dying, and also consider the almost certain recontamination of a system by bacteria if the system is exposed to outside water or air. To give an idea of the numbers of bacteria in natural waters, 1 mL of clean seawater typically has between 100,000 and 1,000,000 bacteria. Contaminated waters can have over 100,000,000 bacteria/mL. Experience with well water and seawater used in salt dome mining has shown that bacteria are typically present in the mining water at levels of 10,000–100,000 bacteria/mL in freshwater and 100,000–1,000,000 bacteria/mL in seawater (Confidential Communications). Just because bacteria can exist in a particular environment, however, does not mean that they can cause corrosion. Most bacterial activity is controlled by the availability of nutrients to the bacteria, nutrients such as hydrocarbons and the organic content of natural waters. 15.3.2 Halophilic Bacteria Halophilic bacteria are bacteria that exist in the presence of high salt concentration in water, or, looked at in the opposite manner, in a low-solvent environment. Interest in halophilic bacteria originated from the observation that halophilic bacteria could spoil food preserved in salt or sugar. Initial interest in red halophiles resulted from their ability to spoil salted fish and hides, although their presence had been obvious for centuries due to the red color they imparted to salterns, used to prepare solar sea salt from seawater. NASA conveniently obtained red halophilic bacteria from salterns in the Moffett Field area near San Francisco Bay. Natural salt lakes provide other good sources of halophilic bacteria. The Dead Sea has supplied some strains for study by Israeli scientists.
ELSE_TSANG_CH015.qxd
8/20/2005
9:31 PM
Page 189
15.3 Microbiologically Influenced Corrosion (MIC)
189
Fig. 15.12. Pit in wall due to oxygen concentration cell (courtesy of Baker Petrolite).
For solution mining purposes, the most distinctive organisms in this group are the extremely halophilic bacteria of the genera Halobacterium and Halococcus, all of which can grow in saturated NaCl solutions. They are critically characterized by the lower limit of NaCl (12–15%) required for growth. Several food spoilage organisms have been classed as moderate halophiles, being able to grow in the molarity range of 0.5–3.5 M NaCl (3–20%). This definition also fits many marine bacteria, many of which require about 0.5 M (3%) NaCl for growth and have been found to be able to withstand 20%, 25%, or even 30% NaCl. The definition of the lower salt concentration for growth of moderately halophilic bacteria must also include the specification of temperature. For instance, at 20°C, the bacterium Planococcus halophilis can grow in NaCl solutions up to 4.0 M (23%), and can also grow in the virtual absence of Na ions. However, if the temperature is raised to 25°C, at least 0.5 M NaCl is needed for growth, and NaCl cannot be substituted by KCl or nonionic solutes such as sugar. Thus, this organism would be classified as a highly salt-tolerant organism at 20°C, whereas at 25°C, it must be considered a moderate halophile. Halophilic bacteria in mined brine and storage brine While halophilic and salt-tolerant bacteria may exist in saturated brines, there have been no reports in the literature on bacteria of the kind known to cause corrosion. It may be that this type has never been investigated. Two things to remember about dealing with bacteria: they are tremendously adaptive and great survivors, so it is certainly possible that corrosive bacteria may already be tolerant of saturated mined brine. This should be tested. However, there is a long history of using salt as a preservative, which would indicate that bacteria probably do not commonly exist in saturated brine. Further, there have been no reports of bacteria corrosion problems in high salt, nonstarch drilling muds, or very saline brines produced from oil and gas wells. In all cases of corrosive salt drilling muds, corrosion has been controlled by deaeration. One obstacle that bacteria must overcome in all environments, including solution mining halite, is to obtain a nutrient source. Hydrocarbons are readily available in hydrocarbon storage operations, and there may be enough organic material in well water and seawater to sustain bacteria growth in brine halite mining operations.
ELSE_TSANG_CH015.qxd
8/20/2005
190
9:31 PM
Page 190
Potential Corrosion and Microbiological Mechanisms and Detection Techniques
Testing for halophilic bacteria Testing for halophilic or salt-tolerant bacteria should be done using a combination of techniques. Probably the most reliable test is to filter some water and microscopically examine the filter for bacteria. Another is to culture, in the usual nutrient broth, those cultures made with the same salinity as the mined brine. Field testing should also include examination of freshly removed steel from the brine, such as when casing is pulled or if corrosion coupons are exposed. Bacteria will be easily observed as slimy black, green, red, or brown growths on the surface of the steel, or underneath corrosion products. Table 15.1 shows the salt tolerance of different microorganisms. 15.3.3 Corrosive Bacteria MIC is normally characterized by the presence of several specific types of bacteria for which testing procedures are available and are convenient. From experience, when there are large numbers of these bacteria in the system, one can characterize the system as potentially corrosive or not. However, the mere presence of bacteria in a system does not necessarily mean that MIC will be significant. The types of bacteria most often used to characterize nonhighly saline systems as corrosive or not are: ● Acid-producing bacteria and fungi (APB) ● Sulfate-reducing bacteria (SRB) ● Iron- (and manganese-) precipitating bacteria (IPB) ● Sulfur-oxidizing bacteria (SOB).
Table 15.1. Salt tolerance of different microorganisms Category
Reaction
Examples
Nonhalophile
Grow best in media containing <1% salt1 Grow best in media containing 1–2% salt Grow best in media containing 2–8% salt. Organisms able to grow in <0.4% salt are considered facultative halophiles Grow best in media containing 5–14% salt Grow best in media containing 7.5–26% salt (saturated) Non halophile that can tolerate salt. If the growth range extends above 2.5 M salt, may be tolerant yeasts and fungi. Extremely halotolerant
Most normal cubacteria and most freshwater microorganisms Many marine microorganisms
Slight halophile Moderate halophile
Borderline extreme halophile Extreme halophile Halotolerant
1
Bacteria and some algae
Ectohiorhodospira halophila2 Actinopolyspora halophila The “red halophiles,” halobacte ria, and halococci Stapholococcus aureus and other staphylococci, solute-tolerant yeasts and fungi; Staphylococcus elabens,3 Staphylococcus epidermidis4
Salt is usually NaCl, but can be other salts in addition to a minimum amount of NaCl. A blue-green alga unable to grow in <1.5 M NaCl. 3 A blue-green alga which can grow in 25% NaCl (4.3 M). 4 Staphylococcus epidermidis can grow in 4.0 M NaCl. 2
ELSE_TSANG_CH015.qxd
8/20/2005
9:31 PM
Page 191
15.4 Conclusions and Recommendations
191
These types of bacteria are well known for their effects in other environments such as soils and oil pipelines, many of which carry highly saline-produced waters. It is not known if these same bacteria are salt-tolerant and can cause corrosion in saturated brines. Tests run on mined brines in a U.S. Gulf Coast salt dome mining plant, using standard 1% saline culture media, showed between 1000 and 100,000 bacteria/mL in the mining water, but only very small bacteria concentrations in these produced brines, e.g., 0–1 bacteria/mL. It is not known if these bacteria were salt-tolerant and had survived in the brine, were contaminating the brine from the air above the brine storage tank, or were actually able to grow in it. SRB are a well-known type of bacteria (J.R. Postgate, 1984). They are usually blamed for corrosion found in hydrocarbon pipelines, although they are just one of many species of bacteria living in synergistic communities and have difficulty in utilizing hydrocarbons directly as nutrients. SRB are strict anaerobes; i.e., they cannot live in the presence of oxygen. However, they can and do thrive in aerobic systems when they can live in sediment deposits that have become depleted in oxygen, or associated with colonies of aerobic APB. Corrosion under sediments is frequently found in hydrocarbon pipelines, and large numbers of SRB are usually associated with these deposits. The result is a channel or series of pits in the bottom of the pipeline similar in appearance to that experienced in brine pipelines. Recent studies have determined that several SRB species are actually archaea and capable of living under extreme conditions, including high salt concentration.
15.4 CONCLUSIONS AND RECOMMENDATIONS Because of the lack of previous research on halophilic bacteria, a bacteria-testing program would be required to establish that halophilic bacteria are causing corrosion in brine pipelines. This would not be a straightforward process, since culturing techniques may have to be developed for some of these bacteria. This would require some modifications to the normal bacteria-testing procedures for pipelines, as given by NACE International and previously by American Petroleum Institute (API). Sampling for bacteria should be done when pigs are run, as well as for random bulk samples. Bacteria that cause corrosion live on the wall of the pipeline and are dislodged by pigging. A second modification would be to use sessile bacteria monitoring techniques to culture bacteria in situ, to be able to detect very low bacteria concentrations. Finally, bacteria can be detected by filtering brine through a 0.45-µm filter and examining any bacteria under the microscope, or by using readily available DNA analysis techniques. It may be necessary to stop chlorine injection into the line to be able to grow bacteria in it. To ensure the integrity of a pipeline and to reduce risk, a corrosion monitoring program to determine actual corrosion rates should be implemented using the following, or a combination of the following: ● Coupons ● Linear polarization resistance probes ● Electrical resistance probes ● Galvanic probes. Downhole tubulars should also be regularly inspected using one or more of the following: ● Digital caliper tools ● Electromagnetic magnetic flux leakage tools ● Electromagnetic phase-shift tools ● Ultrasonic casing inspection tools ● A downhole video camera.
ELSE_TSANG_CH015.qxd
8/20/2005
192
9:31 PM
Page 192
Potential Corrosion and Microbiological Mechanisms and Detection Techniques
REFERENCES Annand, R.R., Hilliard, H.M., and Tait, W.S., 1977. Factors in the corrosivity of seawater used for secondary petroleum recovery. In: C.C. Wright, D. Cross, A.G. Ostroff and J.R. Stanford (Eds), Oilfield Subsurface Injection of Water. ASTM Special Technical Publication No. 641, ASTM, Philadelphia, PA. ASM Metals Handbook, 9th edn, Vol. 13, Corrosion, p. 903, 18. ASTM, Special Technical Publication 641, 1977. Davis, K. and McDonald, L.K., 1998. An overview of potential corrosion mechanisms and detection techniques in solution mining and hydrocarbon storage wells. Presented at the SMRI Technical Meeting, Rome, Italy. deWaard and Lotz, 1993. Prediction of CO2 corrosion of carbon steel. NACE, New Orleans, LA, Corrosion 93(69). Kushner, D.J. (Ed), 1978. Chapter 8: Life in high salt and solute concentrations: halophilic bacteria. Microbial Life in Extreme Environments. Academic Press, New York, pp. 317–368. Patton, C.C., 1981. Oilfield Water Systems. Campbell Petroleum Series. Postgate, J.R., 1984. The Sulfate Reducing Bacteria, 2nd edn. Cambridge University Press, Cambridge, UK. Smart, J. and Van Oostendorp, D., 1999. Corrosion mechanisms in brine. Solution Mining Research Institute Meeting Paper, Las Vegas. Solution Mining Research Institute (SMRI), 2003. State of the Art Review of the Understanding, Mitigation and Monitoring of Corrosion in Brine and Water Piping, Tubing and Casing at Brine Production and Hydrocarbon Storage Cavern Facilities. SMRI Report. Uhlig, H.H. and Revie, R.W., 1985. Corrosion and Corrosion Control. 3rd edn. John Wiley, New York, p. 106.
ELSE_TSANG_CH016.qxd
8/20/2005
9:33 PM
Page 193
Chapter 16
CHARACTERIZATION OF SUBSURFACE HETEROGENEITY: INTEGRATION OF SOFT AND HARD INFORMATION USING MULTIDIMENSIONAL COUPLED MARKOV CHAIN APPROACH Eungyu Parka,*, Amro Elfekib, and Michel Dekkingb a b
Environmental Sciences Division, Oak Ridge National Laboratory, Oak Ridge, TN, USA Faculty of Civil Engineering and Geosciences, Delft University of Technology, Delft, The Netherlands
16.1 INTRODUCTION Characterization of the subsurface is an important step for subsurface flow and transport simulations, which may reveal useful information about the existing heterogeneity of fields. The subsurface is inherently not fully accessible with existing techniques, and consequently there exists substantial uncertainty that needs to be resolved to produce a plausible picture of the ground. Geostatistical simulations are currently the most promising solutions to these uncertainty problems, and their effectiveness has been proven in many case studies. Indicator-based simulation methods are well suited for many geological characterization cases (Carle and Fogg, 1996). They have been intensively used in many fields such as soil, stratigraphy, hydrogeology, and sedimentology. In most of these simulations, indicator variograms have been widely applied. Recently, Carle and Fogg (1997) and Carle et al. (1998) developed a new type of sequential indicator simulation (SIS) algorithm that uses Markovian transition probabilities instead of the indicator variograms. The Markovian property states that the conditional distribution of any future state, given the past and present states, is independent of the past history and depends only on the present state (Ross, 2000). The approaches of Carle et al. (1998) brought many improvements in terms of asymmetry, which could not be modeled by conventional geostatistical approaches. By using the transition probability approach, many soft and hard data in an early stage of the model can be utilized. In the approach presented by Carle and Fogg (1997), the transition probabilities are built in a Markovian framework in the estimation phase. However, in the simulation phase, they use the conventional simulation methods (i.e., SIS, followed by a simulated quenching technique). Therefore, their approach loses some of the benefits of the Markovian transition probabilities. Elfeki and Dekking (2001) developed a two-dimensional (2-D) conditional indicator simulation algorithm using Markovian transition probabilities under a Markovian framework. This model is directly branched from the onedimensional (1-D) Markov chain model developed by Krumbein (1967). The coupled Markov chain (CMC) model is very convenient to implement for stochastic simulation, because the CMC model does not require parametric fitting of a semivariogram model nor cumbersome indicator co-kriging techniques. Also, conditioning is simple, through the use of explicit formulae. The main objective of this study is to extend the CMC to three-dimensional (3-D) in order to better suit the model for many practical problems. * Currently at Department of Geology, Kyungbuk National University, Daegu, Korea.
ELSE_TSANG_CH016.qxd
194
8/20/2005
9:33 PM
Page 194
Characterization of Subsurface Heterogeneity: Integration of Soft and Hard Information
16.2 THEORETICAL BACKGROUND In the Markovian framework, the conditional distribution of any future state is independent of the past history if the present state is given. In formulas, let the discrete stochastic process {Zi, i = 0, 1, 2, …} be a sequence of random variables taking values in the state space {S1, S2, …, SM}. The sequence is a Markov chain or a Markov process, if Pr(Zi ⫽ Sk冷Zi⫺1 ⫽ Sl, Zi⫺2 ⫽ Sn, Zi⫺3 ⫽ Sr, … , Z0 ⫽ Sp)
(16.1)
⫽ Pr(Zi ⫽ Sk冷Zi⫺1 ⫽ Sl ) ⫽ plkⴢ In 1-D problems, a Markov chain is described by a single transition-probability matrix. Transition probabilities correspond to relative frequencies of transitions from certain states to other states. These transition probabilities can be arranged in a square matrix form:
冤
p1l ⴢ p⫽ ⴢ ⴢ pM1
ⴢ ⴢ ⴢ ⴢ ⴢ
ⴢ ⴢ plk ⴢ ⴢ
ⴢ ⴢ ⴢ ⴢ ⴢ
冥
p1M ⴢ ⴢ , ⴢ pMM
(16.2)
where plk denotes the probability of transition from state Sl to Sk, and M is the number of states in the system. Thus, the probabilities of a transition from S1 to S1, S2, …, SM are given by p1l, p12,…, p1M where l = 1, 2, …, M and so on. The matrix p has to fulfill specific properties: (1) its elements are nonnegative, plk ⱖ 0; (2) the elements of each row sum up to 1. The transition probabilities considered in Equation (16.2) are called one-step transitions. We also consider N-step transitions, which means that transitions from one state to another take place in N steps. The N-step transition probabilities can be obtained by multiplying the single-step transition-probability matrix by itself N times. Under some mild conditions on the transition matrix (aperiodicity and irreducibility), the successive multiplications lead to identical rows (w1, w2 … , wM). Thus, the wk (k = 1, 2, … , M) are given by lim plkN ⫽ wk.
N→∞
(16.3)
These are called marginal probabilities, where wk is no longer dependent on the initial state Sl. The aforementioned method is a discrete-time Markov chain model. The methods for building continuous-time Markov chains are also available and thoroughly reviewed by Carle et al. (1998). Note that using the continuous-lag Markov chains, we can build Markovian transition probabilities even out of sparsely distributed data, because the continuous-lag formulation can handle irregularly spaced data using the transition rate and the exponential decaying lag-transition model. This method is potentially helpful when we build the horizontal transition matrix using sparsely distributed lithological data (Carle et al., 1998).
16.3 THE 3-D COUPLED MARKOV CHAIN MODEL The coupled transition probability plma,bfh on the state space {S1, S2, ... , SM}×{S1, S2, ... , SM}×{S1, S2, ..., SM}is given by Figure 16.1: hy v plma, bfh ⫽ phlbx ⴢ pmf ⴢ pah
(16.4)
ELSE_TSANG_CH016.qxd
8/20/2005
9:33 PM
Page 195
16.3 The 3-D Coupled Markov Chain Model
195
These transition probabilities form a stochastic M3×M3 matrix. For the 3-D generalization of the 2-D CMC model, a triple coupling of 1-D Markov chains (Xi), (Yj), and (Zk) is used for constructing a 3-D spatial stochastic process (Γijk) on a cubic lattice, as shown in Figure 16.1. Each cell has a layer number k, a row number j, and a column number i. Using the analogy applied for developing the 2-D CMC model (Elfeki and Dekking, 2001), we can write the three coupled chains on the cubic lattice by forcing these three independent chains to have the same outcome as plmn, o ⫽ Pr(Γi,j,k ⫽ So冷Γi⫺1,j,k ⫽ Sl, Γi,j⫺1,k ⫽ Sm, Γi,j,k⫺1 ⫽ Sn ) hy v plohxⴢpmo ⴢpno ⫽ ᎏᎏ , hx hy Σf plf ⴢpmf ⴢpnfv
(16.5)
o ⫽ 1, … , Μ.
We can also write the conditional probability of cell (i, j, k) to be in state So, given the past [cell (i−1, j, k) is in state Sl, cell (i, j−1, k) is in state Sm, and cell (i, j, k−1) is in state Sn], and the future [cell (Nx, j, k) is state Sp, and cell (i, Ny, k) is in state Sq] as plmn,oⱍp, q ⫽ Pr(Γi,j,k ⫽ So冷Γi⫺1,j,k ⫽ Sl, Γi, j⫺1,k ⫽ Sm, Γi,j,k⫺1 ⫽ Sn, ΓNx, j,k ⫽ Sp, Γi,Ny, k ⫽ Sq)
(16.6)
hx (Nx⫺i) hy hy (Ny⫺j) plohx pop pmo poq ᎏᎏ ⫽ C ᎏᎏ pv , hy (Ny⫺j⫹1) no plphy (Nx⫺i⫹1) pmq
where C is a normalizing constant given by
冢冱
C⫽
r
hy p p x (Nx⫺i) hy(Ny⫺j) phlrx pmr prq nr rp ᎏᎏᎏ ⫺i⫹1) (N (N hy y⫺j⫹1) phlpx x pmq v
h
冣
⫺1
(16.7)
Also, in the derivation of Equation (16.6), we assume that the conditioning data can only be found along the horizontal plane and that the conditioning scheme goes from top to bottom layer by layer. In practical calculation of the 3-D CMC model, all 1-D, 2-D, and 3-D CMC equations must be applied together (Park et al., 2002). The application of each CMC equation depends on its location in the 3-D lattice. For the periphery of the top layer, the 1-D CMC equation
Fig. 16.1. Computational cubic lattice of a 3-D CMC model.
ELSE_TSANG_CH016.qxd
196
8/20/2005
9:33 PM
Page 196
Characterization of Subsurface Heterogeneity: Integration of Soft and Hard Information
is used. For the remaining part of the top layer and the four side layers, the 2-D CMC equation is applied. For the rest of the domain, the 3-D CMC equation—(Equation (16.6))—is applied. 16.3.1 Algorithm of 3-D CMC Model The developed algorithm for determining the lithology of each cell in a single realization using 3-D CMC is as follows: Step 1: The 3-D domain is discretized using proper sampling intervals: for the vertical direction, the thinnest lithology observed will be used as a discretizing unit; for the vertical direction, Walther’s law is adopted for deciding the horizontal discretizing unit from the available information. Step 2: The borehole data are saved in their proper locations for conditioning. Step 3: The Markovian vertical transition probabilities are calculated from the tally matrix of the transition lithology, using borehole data. Horizontal transition probabilities are inferred from the vertical transition probabilities using Walther’s law. In most of the presented simulation, the horizontal transition probabilities are used for the averages of the upward- (i.e., from bottom to top) and downward- (i.e., from bottom to top) transition probabilities to avoid directional bias. Step 4: The 1-D and 2-D CMC probabilities are applied on the periphery of the top layer and the periphery of the cube, respectively. The lithology for each corresponding cell is filled in according to its corresponding conditional transition-probability distribution. Step 5: The 2-D and 3-D CMC probabilities are applied on the top layer and inside the cube, respectively. The lithology for each corresponding cell is filled in according to its corresponding conditional transition-probability distribution. Step 6: The procedure stops after having visited all the unassigned cells in the domain. The Monte Carlo simulation is also possible by repeating the above-mentioned algorithm until desirable numbers of realizations are generated. After generation of enough single realizations, the final decision is made by the dominancy of the lithology in the corresponding cells. Calculation of the uncertainty is also possible through these multiple realizations.
16.4 APPLICATION OF THE 2-D AND 3-D CMC MODEL 16.4.1 2-D Application As an illustration of the applicability of the CMC model, we consider an example based on data from the Alabama MADE Test Site on unconsolidated coastal plain sediments. Data from the site were available from 16 borehole measurements along a transect (276 × 12.2 m – horizontal × vertical). The developed 2-D CMC model is applied to the data (Fig. 16.2a). The input is prepared from the lithological data from the 16 boreholes located at the site. Data are discretized using the thickness of the finest lithology observed along the vertical direction, which is 0.1 m. The transition probabilities along the vertical direction are calculated from the 16 boreholes using a descending sequence of lithologies in the boreholes. However, we use horizontal transition probabilities calculated by averaging the vertical ones in both descending and ascending sequences. This approach takes into account the various possibilities of the regional geologic setting in the horizontal direction at the site, thereby avoiding directional bias.
ELSE_TSANG_CH016.qxd
8/20/2005
9:33 PM
Page 197
16.4 Application of the 2-D and 3-D CMC Model
197
Fig. 16.2. (a) Hydraulic conductivity distribution along section A–A. Tick marks denote measurements of hydraulic conductivity using borehole flowmeter (from Adams and Gelhar, 1992); (b) single realization by 2-D CMC model; (c) reduced information down to 1.94% from simulated image; (d) resimulated image (conditioned on the 1.94% known data) recovering of 68.58% of the simulated image.
ELSE_TSANG_CH016.qxd
198
8/20/2005
9:33 PM
Page 198
Characterization of Subsurface Heterogeneity: Integration of Soft and Hard Information
Using our developed 2-D CMC model, we generate a single realization image conditioned on 16 boreholes (Fig. 16.2b). Unaided visual comparison confirms the similarities between the hand-drawn geologic map (Fig. 16.2a) and the simulation result (Fig. 16.2b). We also performed a test using this simulated image by reducing the data down to 1.94% from the original image (Fig. 16.2d) as an extreme case, and resimulate with this data (Fig. 16.2d) to see the effectiveness of the newly improved 2-D CMC model. The single realization generated by this procedure reproduces 68.58% of the original image (Fig. 16.2b). From the structural point of view, the CMC model mimics very closely the structure of the original one conditioned on the given sparse data. 16.4.2 3-D Application Based on the developed 3-D CMC model, a single realization can be generated, as shown in Figure 16.3. To illustrate the 3-D CMC model, we “bent” the previous transect data (at the MADE site) to form a cube. Therefore, the input data used for in the 3-D CMC model are somewhat hypothetical at borehole locations along the periphery of the domain (Fig. 16.3). On each side, four hypothetical boreholes are located where two boreholes are shared by neighboring sides. The domain size is 40 × 40 × 12.2 m (hor. x × hor. y × vert. z), discretized into 40 cells along the x- and y-directions, and 122 cells along the z-direction, with uniform intervals. The outside and the inside of the simulation results (single realization) by 3-D CMC are shown in Figure 16.3a and b, respectively. Each face of the block shows a geologically plausible 2-D simulation profile as shown in the 2-D CMC model. From the slice map (fence diagram) of the inside of the block, no unrealistic geologic changes are observed. Because the simulation is based on hypothetical data, our simulation results cannot be compared with real data. However, we can compare a number of realizations to check the stability of the model. In other words, if two realizations are totally different from each other, we can consider the model to be unstable. The instabilities, if any, can mostly be observed from inside the 3-D block. Through our Monte Carlo analysis, we confirmed that our model is stable: the generated images vary only slightly from each other, and those variations occur near the lithology boundaries (Fig. 16.4). Through our Monte Carlo analysis, we generated a statistical map, in 3-D space, of the ensemble indicator function (EIF) of each lithology. As a typical example, Figure 16.4a shows the statistical structure in 3-D space of the EIF of lithologies 3 and 2. In the figure, the EIF value with the dark red color is 1; dark blue is 0. If an EIF equals 0, there is almost no probability of the corresponding lithology appearing, whereas an EIF equal to 1 ensures that the corresponding lithology is present at that location in space. Figure 16.4b is an isoEIF of lithology 3 for EIF values of 0.1, 0.5, and 0.9, respectively. These values represent a 10% (yellow), 50% (green), and 90% (blue) degree of certainty for finding the corresponding lithology when the exploration reaches the inside of the EIF shell. The generated 3-D statistical mapping can be used for engineering purposes. Based on the EIF for each lithology, we extract an ensemble hypothetical geological map by assigning a lithology with the maximum EIF at a given cell (Fig. 16.5). There is no significant difference when we compare the final geological image to a single realization (Fig. 16.3), except for some local variations at interfaces between lithologies that are smoothed out by the ensemble calculation. From this, we can conclude that the 3-D CMC model is a stable simulator, honoring all the given conditioning data.
ELSE_TSANG_CH016.qxd
8/20/2005
9:33 PM
Page 199
16.5 Summary and Conclusions
199
Fig. 16.3. Simulated single realization using the 3-D CMC model: (a) outer slices; (b) inner slices.
16.5 SUMMARY AND CONCLUSIONS We developed a 3-D CMC model and corresponding software called CMC3D. It is an extension of previous work by Elfeki and Dekking (2001). (The software is available directly from the authors.) A brief algorithm has been presented here. We also provided improved schemes for the CMC 2-D conditioning. Monte Carlo simulations show that our
ELSE_TSANG_CH016.qxd
200
8/20/2005
9:33 PM
Page 200
Characterization of Subsurface Heterogeneity: Integration of Soft and Hard Information
Fig. 16.4. Statistical mapping of the EIF in 3-D space: (a) EIF of lithology 3 where red indicates high probability (~1.0) and black indicates low probability (~0.0) of occurrence of the lithology; (b) superposition of the iso-EIF of encountering lithology 2 with a probability = 0.1 (light grey), 0.5 (dark grey), and 0.9 (black) in 3-D space. [red and blue refer to the color version of the figure, which can be found in the CD enclosed].
model is stable for most cases. The 2-D application on the MADE site shows that the model has promise for delineating the complex geological structure of an aquifer, even with only sparse data.
ELSE_TSANG_CH016.qxd
8/20/2005
9:33 PM
Page 201
Acknowledgments
201
Fig. 16.5. Final image generated from 30 realizations: (a) outer slices; (b) inner slices.
ACKNOWLEDGMENTS This work is funded by NWO Project No. 812.02.003, and performed during a stay of the first author at Delft University of Technology, The Netherlands. This work was also supported in part by the U.S. Department of Defense Strategic Environmental Research and Development Program (SERDP) Cleanup Program, directed by Dr. Andrea Leeson under the project “Integrated Protocol for Assessment of Long-Term Sustainability of MNA for Chlorinated Solvent Plumes.”
ELSE_TSANG_CH016.qxd
202
8/20/2005
9:33 PM
Page 202
Characterization of Subsurface Heterogeneity: Integration of Soft and Hard Information
REFERENCES Adams, E.E. and Gelhar, L.W., 1992. Field study of dispersion in a heterogeneous aquifer, 2 spatial moment analysis. Water Resour. Res., 28(12): 3293–3307. Carle, S.F. and Fogg, G.E., 1996. Transition probability-based indicator geostatistics. Math. Geol. 28(4), 453–476. Carle, S.F. and Fogg, G.E., 1997. Modeling spatial variability with one- and multidimensional continuous Markov chains. Math. Geol. 29(7), 891–917. Carle, S.F., LaBolle, E.M., Weissmann G.S., VanBrocklin, D. and Fogg, G.E., 1998. Conditional simulation of hydrofacies architecture: a transition probability/Markov approach. In: G.S. Fraser and J.M. Davis (Eds), Hydrogeologic Models of Sedimentary Aquifers, SEPM Concepts in Hydrological Environmental Geology, vol. 1, Society for Sedimentary Geology, Tulsa, OK, pp. 147–170. Elfeki, A.M.M. and Dekking, M., 2001. A Markov chain model for subsurface characterization: theory and applications. Math. Geol. 33(5), 569–589. Krumbein, W.C., 1967. FORTRAN IV computer program for Markov chain experiments in geology. Computer contribution 13, Kansas Geological Survey, Lawrence, KS. Park, E., Elfeki, A.M.M. and Dekking, F.M., 2002. Characterization of subsurface heterogeneity: integration of soft and hard information using multi-dimensional Coupled Markov chain approach. Annual Progress Report, Civil Engineering and Geosciences, Delft University of Technology, Delft, The Netherlands. Ross, S., 2000. Introduction to Probability Models. 7th edn. Academic, San Diego, CA.
ELSE_TSANG_CH017.qxd
8/20/2005
10:19 PM
Page 203
Chapter 17
MODELING OF WASTE INJECTION IN HETEROGENEOUS SANDY CLAY FORMATIONS S.P. Pozdniakova, V.A. Bakshevskaya, A.A. Zubkovb, V.V. Danilovb, A.I. Rybalchenkoc, and C.-F. Tsangd a
Faculty of Geology, Moscow State University, Moscow, Russia Siberian Chemical Combine, Seversk, Russia c All-Russia Designing and Research Institute of Production Engineering (VNIPIPT), Moscow, Russia d Lawrence Berkeley National Laboratory, Berkeley, CA, USA b
17.1 INTRODUCTION Since 1963, radioactive waste has been injected into deep artesian aquifers of Cretaceous terrigenous deposits at the Siberian Chemical Combine in Western Siberia, near Tomsk, (Fig. 17.1). A detailed hydrogeological description of the disposal site and injection setting is given by Rybalchenko et al. (1996, 1998) and Foley et al. (1995). The total volume of waste injected within the two disposal areas (Areas 18 and 18a, Fig. 17.1) into heterogeneous aquifers 300–400 m below land surface, is about 4 ⫻ 107 m3. Most of the waste injected at Area 18 is low-level, unprocessed waste with a total specific activity ranging from 10−8 up to 10−6 Ci/L (which, according to Russian Federation regulations, exceeds the maximum allowed activity). The waste disposed at Area 18 contains nitrate and sulfate ions as well as detergent. Total dissolved solids do not exceed 20 g/L and are typically less than 10 g/L, which explains why the physical properties (viscosity and density) of injected fluid are similar to the properties of natural water in aquifers. The current scenario for disposal site operation is that injection will be terminated within the next decade. The future long-term subsurface migration of wastes injected up to the date of injection termination is a subject of intensive flow and transport modeling predictions (Rybalchenko et al., 1998; Shestakov et al., 2002). These models are based on a simplified layered geospatial description of natural heterogeneity within the injection formation. The layered geospatial model represents the vertical heterogeneity of the formation as a chain of seven aquifers, marked from bottom to top as I, II, III, IV, IVa, V, and VI. These aquifers are divided by six semipermeable units marked from bottom to top as A, B, C, D, E, F (Rybalchenko et al., 1998). Aquifers II and III are used for waste injection. The overlaid Aquifer III semipermeable layer (Layer D) is considered to be the main protective layer against the vertical migration of waste. Lithological study of well cores shows that each aquifer and each semipermeable layer marked within the studied area has a complex internal architecture (Fig. 17.1), consisting of alternating relatively high and relatively low permeable units. In this chapter, we describe the development of a 3-D high-resolution model of Injection Area 18, to study the impact of heterogeneity on the subsurface spreading of waste.
ELSE_TSANG_CH017.qxd
204
8/20/2005
10:19 PM
Page 204
Modeling of Waste Injection in Heterogeneous Sandy Clay Formations
Fig. 17.1. The site location and main geological features of the injection formation.
17.2 DEVELOPMENT OF A 3-D MODEL FOR HETEROGENEITY The geologic formation that includes Aquifers II and III and the overlaying intervals up to the E layer was formed during the Cretaceous Era in a continental near-sea border environment. The terrigenous deposits of this formation are represented by clay, sandy clay loam, sandy loam, sand, and other fluvial soil types. To take into account these essential features of heterogeneity, we processed lithological data from 212 wells placed irregularly within an area approximately 4 ⫻ 4 km (Fig. 17.2) within Area 18. This dataset includes about 46 km of lithological logs obtained from characterization, monitoring, and injection wells (which were drilled over the past 40 years within the site). The vertical resolution of logs is 0.5 m. Table 17.1 summarizes the distribution of soil types found in the logs of these wells. We can see from this table that within captured soil types, the largest fraction (40%) is sand, and the second largest fraction (33.5%) is clay. Thus, within the internal architecture of the formation of interest, high-permeable (i.e., sand) and low-permeable (i.e., clay) facies form about 80% of the overall formation volume. The other soils presented in Table 17.1 can also
ELSE_TSANG_CH017.qxd
8/20/2005
10:19 PM
Page 205
17.2 Development of a 3-D Model for Heterogeneity
205
Fig. 17.2. Plan view of the modeled square area. Circles are the places of wells with lithological logs. Table 17.1. Distribution of soil types found in lithological logs Soil type
Number of layers
Total thickness (m)
Fraction (%)
Sand Clay Kaoline clay Clay Breccia Sandy loam Shale Sandy clay loam Conglomerate Other soils
3,493 3,019 265 584 663 72 370 50 8
18,609.0 15,583.5 3,620.5 2,484.5 2,440.5 1,343.0 1,302.5 670.5 119.5
40.0 33.5 7.8 5.3 5.3 2.9 2.8 1.4 0.9
Total
8,524
46,173.5
100.0
Indicator 1 0 0 0 1 0 0 1 1 Mean 0.476
be divided into relatively high-permeable (the shadowed names in the first column of Table 17.1) and relatively low-permeable soils. Thus, as a very simple model of the medium, the binary sequences of facies can be applied for spatial heterogeneity modeling. For this, in Table 17.1, all relatively high-permeable soils are joined in one facies (called “sand”) and are assigned an indicator-function value of 1. The low-permeable soils (all other soil types) are joined in another facies (called “clay”) and are assigned an indicator-function value of 0. The estimated volumetric fraction of both facies in the modeled medium is close to half of the volume—the calculated mean value of the indicator function is 0.476. The magnitude
ELSE_TSANG_CH017.qxd
8/20/2005
206
10:19 PM
Page 206
Modeling of Waste Injection in Heterogeneous Sandy Clay Formations
of expected permeability contrast for this binary is about four orders, as derived from fieldtest results indicating that the hydraulic conductivity of aquifers at this site is 0.3⫺1 m/day, whereas the hydraulic conductivity of semipermeable layers is about 10⫺4 m/day. To describe and model lithological heterogeneity, we use the model of two facies’ spatial succession, based on a transition-probabilities approach (Carle and Fogg, 1997). For two facies, four transition probabilities describe stochastic properties of heterogeneity. The evolution of these probabilities in a given direction as a function of separation h is found by the system solutions of Kolmogorov’s equations for Markov’s processes (Rozanov, 1971):
冢
冣
(17.1a)
h pcc(h) ⫽ (1 ⫺ pc) exp ⫺ ᎏ ⫹ pc, λef
冢
冣
(17.1b)
psc(h) ⫽ 1 ⫺ pss(h),
(17.1c)
pcs(h) ⫽ 1 ⫺ pcc(h),
(17.1d)
h pss(h) ⫽ (1 ⫺ ps) exp ⫺ ᎏ ⫹ ps, λef
where pss is the probability of sand-to-sand transition, pcc the probability of clay-to-clay transition, psc the probability of sand-to-clay transition, pcs the probability of clay-to-sand transition, ps ⫽ Ls/(Ls ⫹ Lc) the marginal probability of sand, pc ⫽ 1 ⫺ ps the marginal probability of clay, λef ⫽ LsLc/(Ls ⫹ Lc) the effective correlation scale. Ls the characteristic length of sand facies, and Lc the characteristic length of clay facies. Thus, the evolution of transition probabilities depends on the characteristic lengths of facies, which is the parameter of probabilistic exponential distribution of facies length in a given direction (Rozanov, 1971). The more general, closed form of the Kolmogorov’s system-solution equation for the three facies case can be found in Lu and Zhang (2002). The Markov’s chain of transition probabilities was computed with Transition Probability Simulator (TSIM) code (Carle, 1998) for horizontal and vertical directions. These probabilities were fitted by a theoretical model, Equations 17.1(a–d). The results of calculations and fitting are shown in Figures 17.3a and b. Computed characteristic lengths and vertical/horizontal correlation scales for sand and clay facies are shown in Table 17.2. Figure 17.3 (and in particular 17.3a) and Table 17.2 demonstrates the essential spatial anisotropy of this system: the horizontal correlation scale exceeds the vertical scale by more than 60 times. Using fitted transition probabilities and the TSIM code (Carle, 1998), which simulated lithological heterogeneity by co-kriging with a transition probabilities algorithm, we created a 3-D lithological heterogeneity model of Injection Area 18. The planar size of this lithological model is 4200 ⫻ 4200 m, with a step of 33 m, and a vertical length of 250 m, with a step of 1 m. Thus, the total number of blocks representing spatial heterogeneity is about 4 million. The used interval of absolute elevation in the vertical direction, from ⫺50 to ⫺300 m, covers (according to modern hydrogeological stratification adopted at this site) Aquifers II, III, and IV, as well as semipermeable layers C and D (Rybalcenko et al., 1998). Figure 17.4 shows the sand and clay units’ distribution in a vertical cross section along the line AB, and Figure 17.5 gives a 3-D view of the site heterogeneity. From these figures, it is
ELSE_TSANG_CH017.qxd
8/20/2005
10:19 PM
Page 207
17.2 Development of a 3-D Model for Heterogeneity
207
Table 17.2. Parameter of transition probabilities in vertical and horizontal directions Parameter
Sand length Ls (m)
Clay length L c (m)
Effective correlation scale λef (m)
Sand fraction (ps )
Vertical direction Horizontal direction
5.1 350.0
5.53 380.00
2.65 182.00
0.48
Fig. 17.3a. Vertical transition probabilities. Circles are calculated using well data, lines are fitted to the theoretical model.
impossible to mark by sight two horizontal continuous layers (i.e., C and D) predominantly consisting of clay units divided by three aquifers composed of sand. Figure 17.6 shows the vertical distribution of the clay fraction averaged over the simulated domain area. The pattern of alternating chain aquifer/semipermeable layers would imply the appearance of the clay fractions every five intervals, close to 1 and 0, whereas the simulated results show disordered noise-like changes in clay fractions from 0.3 to 0.8. These simulation results verify that the internal structure of this formation is more complex than a simply layered system. Figure 17.6
ELSE_TSANG_CH017.qxd
208
8/20/2005
10:19 PM
Page 208
Modeling of Waste Injection in Heterogeneous Sandy Clay Formations
Fig. 17.3b. Horizontal transition probabilities. Circles are calculated using well data, lines are fitted to the theoretical model.
Fig. 17.4. Vertical cross section of sand–clay distribution along the line AB obtained with TSIM. Dark areas are clay units and light areas are sand units.
also compares the simulated distribution of clay fractions in the vertical direction with clay fractions obtained by averaging logs over all well data. This figure indicates good agreement between simulated and measured vertical distribution of clay fractions within the studied domain.
ELSE_TSANG_CH017.qxd
8/20/2005
10:19 PM
Page 209
17.2 Development of a 3-D Model for Heterogeneity
209
Fig. 17.5. Three-dimensional view of sand–clay distribution in modeled domain.
Fig. 17.6. (1) Comparison of averages over all wells; (2) vertical distribution of clay fraction with modeling result.
ELSE_TSANG_CH017.qxd
210
8/20/2005
10:19 PM
Page 210
Modeling of Waste Injection in Heterogeneous Sandy Clay Formations
17.3 FLOW AND TRANSPORT MODEL CALIBRATION Our 3-D flow and transport model of the injection area was developed by MODFLOW2000 (Harbaugh et al., 2000) and MT3DMS (Zheng and Wang, 1998) codes, using a 3-D spatial lithological heterogeneity model. To optimize computer time and the numerical convergence of proven solution methods, the individual steps of the numerical grid was increased by five times in the vertical direction. For each enlarged gridblock, the gridblock horizontal conductivity was calculated by arithmetic averaging over five gridblocks, and the gridblock vertical conductivity was calculated by harmonic averaging. Thus, the total number of flow/transport gridblocks was reduced to about 800,000. Horizontal Kh and vertical Kv hydraulic conductivities of sand, as well as the isotropic hydraulic conductivity of clay and elastic storages of clay, were selected for flow calibration. It is assumed that the elastic storage of sand is a much less uncertain parameter than clay storage and can be defined using published data. That is why the specific elastic storage of sand was taken as a function of the depth, according to a recommendation by Shestakov (2002). Selection of an anisotropic model of sand conductivity was made after analysis of the available 2000 sand core samples taken from the injection interval during site characterization. A calculation of sand hydraulic conductivity, using the Kozeni–Carman equation and core-sample grain-size distribution data, shows that the conductivity changes from 10⫺3 to up to 10 m/day, with a geometric mean of about 0.5 m/day. This means that sand facies is heterogeneous hydraulically in the lithological model. Taking this into account, the anisotropic model of hydraulic conductivity for sand was selected. This approach assumes that sand bodies (layers) composed of sublayers with different conductivities can be modeled as effective anisotropic media with horizontal and vertical conductivities. Calibration of the flow model was achieved by modeling the injection cycle over 1 year. Data about groundwater heads in monitoring wells, screened injection intervals, and the overlie formation were used for the calibration procedure (Fig. 17.7). For verification of the calibrated model, the monitoring data from other years were used, and changes in hydraulic heads (delta H) from the beginning of this year’s injection was modeled over the calibration period. Because the hydraulic influence of a 1-year injection cycle extends more in the horizontal direction by distance than in the model, a general head boundary package (Harbaugh et al., 2000) at the model external boundary was used during calibration. The Parameter Estimation (PEST) code (Doherty, 2001), combined with MODFLOW 2000, was used for automatic calibration of the transient flow model. During the calibration, we took into account that, in some gridblocks, as a result of vertical averaging, the block conductivity differs from the conductivity of clay or sand. This combination of the PEST and MODFLOW 2000 can determine, through inversion, the multiplication coefficient for initially assigned parameter values in a certain zone. Thus, the mixed gridblocks (i.e., the blocks with averaged conductivity) were designated “sand zone,” if the fraction of sand in these blocks was more than 0.5 or “clay zone,” if the fraction of sand in these blocks was less than 0.5. After inversion, the initial conductivity values in the mixed blocks were recalculated using the new values for sand and clay conductivities; these new values were also assigned as pure sand or clay blocks. The same approach was applied to the clay storage. Next, PEST performed a new calibration iteration with the set of starting parameters obtained in the previous iteration. By the third calibration iteration, the multiplication coefficient remained equal to 1, and the parameters obtained from the second
ELSE_TSANG_CH017.qxd
8/20/2005
10:19 PM
Page 211
17.3 Flow and Transport Model Calibration
211
Fig. 17.7. Spatial distributions of injection and monitoring wells used for 1-year injection cycle calibration modeling.
iteration were accepted as the calibration results. The results of calibration and verification indicate that the model closely reproduces the changes in hydraulic heads during injection (Fig. 17.8). For calibration of sand porosity and dispersivity, a 1-month experiment involving waste injection into Injection Well D2, described in Rybalchenko et al. (1998), was followed. An injection well and monitoring well placed 58 m from this well were installed within Area 18A (Fig. 17.1), outside of our modeled domain. The structure of the geological formation in Area 18A (Rybalchenko et al., 1998) is the same as the structure in Area 18, which is why the results obtained in Area 18A are acceptable for Area 18. A support-scale transport model between Well D2 and the observation well was created, and the breakthrough curve for a neutral nitrate tracer in the observation well was adopted to calibrate sand porosity and local-scale dispersivity. The results of transient-flow-model calibration and local-scale transport-model calibration are summarized in Table 17.3. Obtained values fall within acceptable ranges for hydraulic parameters of these lithological units within the studied region (Rybalchenko et al., 1998; Shestakov et al., 2002).
ELSE_TSANG_CH017.qxd
8/20/2005
212
10:19 PM
Page 212
Modeling of Waste Injection in Heterogeneous Sandy Clay Formations
Fig. 17.8. Comparison of modeled (continuous lines) and observed (discontinuous lines) changes of hydraulic heads in the monitoring wells during 1 year of injection, with temporal variations in injection rates within the injection wells. Table 17.3. Hydraulic and transport parameters obtained during calibration Parameters
Kh (m/day)
Kv (m/day)
Elastic storage (m–1)
Porosity
Longitudinal dispersivity (m)
Sand Clay
0.86 4 ⫻ 10⫺5
0.009 4 ⫻ 10⫺5
— 8.4 ⫻ 10⫺7
0.14 —
0.5 —
17.4 ANALYSIS OF EFFECTIVE HYDRAULIC AND TRANSPORT PROPERTIES Numerical estimation of effective hydraulic properties was performed for the modeled domain. For this estimation, 250 m of the modeled formation were divided into five layers, with each layer 50 m thick. For each layer, the effective horizontal and effective vertical conductivities were estimated by applying horizontal and vertical flows, respectively, using a vertical grid size of 1 m. The results show the domain’s essential hydraulic anisotropy of medium effective conductivity: the vertical effective conductivity is about three orders of magnitude less than the effective horizontal hydraulic conductivity. This phenomenon is in accord with the anisotropy of effective correlation scales and should prevent vertical migration of injected waste. Vertical effective conductivities for all five layers fall within one order of magnitude (3–6 ⫻ 10⫺3 m/day). The same result was obtained for the effective horizontal hydraulic conductivity that varies from 0.3 to 0.5 m/day for different layers. It conflicts with
ELSE_TSANG_CH017.qxd
8/20/2005
10:19 PM
Page 213
17.4 Analysis of Effective Hydraulic and Transport Properties
213
modern hydrogeologic stratification, i.e., the layered system of sequences for Aquifers I–VI and semipermeable layers A–F (Rybalchenko et al., 1998). Within the simulated interval of elevations (from ⫺50 to ⫺300 m), the studied formation includes two semipermeable (C and D) layers and three divided aquifers (II–IV). Because the effective conductivity for five layers does not seriously differ from layer to layer, the conceptual model of anisotropic formation (instead of a layered system) could be applied in predictive modeling of flow and transport for this site. Long-term transport of waste from an instant small source placed within the modeled medium was simulated to study the possibility of using a macrodispersive approach for this medium. Two sets of simulation runs with different mean flow directions (horizontal and vertical) were performed. In the first (horizontal) case, the simulated situation represented natural groundwater flow from the injection site to the discharge zone, with flow direction coinciding with the largest correlation scale of sand and clay bodies. In the second (vertical) case, we simulated upward flow within the discharge zone, with flow direction along the smallest correlation scale of lithological units. The main modeled processes include advection, diffusion with a characteristic coefficient value of 10⫺5 m2/day, and local-scale dispersion with longitudinal dispersivity characteristic of sand (from Table 17.3), and a longitudinal dispersivity for clay of 0.05 m. For both facies, the porosity value was assigned to be 0.14. Figure 17.9 shows the distributions of longitudinal flow velocities, scaled by dividing by the mean flow velocity calculated for each case and plotted in log scale for both cases. As mentioned in a number of publications (e.g., Desbarats, 1990; Dagan and Lessoff, 2001), the obtained flow velocity field is characterized by bimodal distribution which, at least for relatively short time periods, leads to non-Fickian transport. In both cases, the temporal changes in the plume over 3000 years of spatial second moments was studied. These moments correspond to the main diagonal terms of effective macrodispersivity tensors, and according to macrodispersion theory, macrodispersitvity is a constant parameter—i.e., transport is a Fickian process) if the second spatial moment grows linearly with time (Gelhar, 1993). The results from these numerical experiments for the horizontal mean flow are fitted by a power regression line with power exponents more than 1. It shows that the effective longitudinal macrodispersivity grows throughout the modeled transport process (Fig. 17.10). In contrast, the transversal terms are more stable, because the power exponent for the vertical transversal term is 0.91 and the exponent for the horizontal transversal term is 1.16. For vertical flow (Fig. 17.11), the second spatial moments grow more nonlinearly than in the horizontal flow case. These results confirm that the effective macrodispersivity tensor in such a heterogeneous system depends very strongly on flow direction (Dagan and Lessoff, 2001). For the horizontal flow case, horizontal and vertical transversal dispersion terms are much less than the longitudinal term, whereas for the vertical flow case, the transversal terms exceed the longitudinal terms by two orders of magnitude. Figure 17.12 shows the result of two additional experiments that simulate vertical and horizontal flow and transport. In these experiments, over the entire flow area of the studied medium, contamination from the initial source was only found in sand. For the vertical flow case, at all grid nodes associated with sand facies in the lowest model layer, the initial value of relative concentration was assigned a value of 1, while clay nodes in this layer (and all nodes within other layers) were assigned a value of 0. For the horizontal flow case, the same initial concentration distribution was assigned for the left vertical section of the grid. Distribution of arriving concentration time to the oppositive boundary versus dimensionless time, τ ⫽ Vt/nL, is shown in Figure 17.12. Here, V is Darcy’s velocity, n the porosity, and L the transport distance. This figure shows the typical long-tail distribution of a heterogeneous
ELSE_TSANG_CH017.qxd
214
8/20/2005
10:19 PM
Page 214
Modeling of Waste Injection in Heterogeneous Sandy Clay Formations
Fig. 17.9. (a) Distribution of longitudinal flow velocity for horizontal mean flow direction; and (b) vertical mean flow direction. The horizontal axis is a decimal logarithm, scaled by dividing by the mean value velocity.
system. The interesting property of this system is that peak dimensionless arriving time, according to Figure 17.12, is approximately the same for both runs, about 0.32, while the dimensionless advective time, i.e., the time of advective transport through sand, should be equal to the sand fraction in the overall modeled volume, ps ⫽ 0.48. This means that, because of the connectivity of certain sand bodies, transport in this system is faster than in a perfectly layered medium. According to these runs, in the simulated medium only about 67% of the sand bodies are perfectly connected, while in a perfectly layered medium, 100% of the sand bodies are connected and thus overall advective transport is slower. The vertical transport result thus looks more dispersive than the horizontal transport, possibly because, within permeable facies, there is a wider distribution of flow velocity for vertical flow (Fig. 17.9). Physically, this accords with the idea that for horizontal flow, the
ELSE_TSANG_CH017.qxd
8/20/2005
10:19 PM
Page 215
17.5 Modeling of the Injection History
215
Fig. 17.10. Evolution of second spatial moments of waste transport over time for the case of horizontal flow. The lines are an approximation of numerical data by power fitting.
effective extent of heterogeneity that leads to spatial fluctuation of a flow path within permeable bodies is proportional to the vertical size of the clay bodies. For the vertical case, spatial fluctuations of flow lines in permeable bodies are proportional to the horizontal size of the clay bodies. This relationship leads to the more dispersive behavior of vertical breakthrough curves. In addition, the long-tail distribution of arriving times indicates that a dualporosity model could be studied as a possible model for regional-scale transport. In general, these numerical experiments show that, at this site, the problem of selecting the best model for predicting long-term and regional-scale waste transport still exists. This is because high-resolution 3-D models cannot be applied to regional-scale transport modeling: upscaling to parameters of simplified models should be done. The results from modeling indicate that the dispersion parameters of effective macrodispersive models are time and space dependent, and that, for sites such as the one modeled here, dual-porosity models need to be further investigated.
17.5 MODELING OF THE INJECTION HISTORY The last step of our study was the simulation of 40 years of injection history, up to the present, to obtain 3-D distributions of injected neutral species (nitrate) of the wastes within
ELSE_TSANG_CH017.qxd
216
8/20/2005
10:19 PM
Page 216
Modeling of Waste Injection in Heterogeneous Sandy Clay Formations
Fig. 17.11. Evolution of second spatial moments of waste transport over time for the case of vertical flow. The lines are an approximation of numerical data by power fitting.
the studied formation. For this, we used a 3-D flow and transport model with estimated hydraulic parameters and local-scale mixing and diffusion. Within this model, a case history was simulated, taking into account temporal changes of injection rate and injection well location, and an actual screening interval for each well. The time-averaging of injection rates in the model was 1 year. A third-order Total Variation Diminishing (TVD) scheme with an implicit solution of dispersive terms was modeled by the MT3DMS code (Zheng and Wang, 1998). At this site, the length of the screening interval for an injection well is typically 30–50 m. Thus, each injection interval crosses 6–8 model layers. For accurate simulation of injection along such screening intervals, the following approach was used: the total well injection rate for a given time step was assigned to the upper sand block crossed by the well screen. Vertical hydraulic-conductivity values along screening intervals were increased by two orders of magnitude over their intrinsic values. Thus, the vertical redistribution of waste within gridblocks with well screens was simulated. The drawback of this approach was the very small grid Peclet number for such gridblocks, which seriously decreased transport time steps during simulation. The resulting complex, irregular shape of the waste body was a function of the changes in injection well locations over time, their injection rates, and the influence of 3-D topology on clay and sand facies. Analysis of modeled waste distribution shows that at present, the main mass of waste predominantly distributes within sand, with the filling of clay bodies by the waste, caused by diffusion and advection, still playing only a small role in overall waste
ELSE_TSANG_CH017.qxd
8/20/2005
17.6 Conclusions
10:19 PM
Page 217
217
Fig. 17.12. Probability distribution function (PDF) of arriving concentration time. Dimensionless time τ ⫽ Vt/nL.
transport. The more important issue, within the highly permeable unit of the formation under study, is how 3-D clay/sand topology affects the velocity field. Despite our model not having the continuous semipermeable layers (like Layers C and D used in previous models) (Shestakov et al., 2002), our modeling results did not show any significant upward waste movement through preferential flow paths of this heterogeneous system.
17.6 CONCLUSIONS In this chapter, a 3-D high-spatial-resolution model of Injected Area 18, a disposal site at the Siberian Chemical Complex, was developed. The new feature of this model is the ability to describe in detail the site formation’s internal 3-D architecture, using a Markov’s chain for modeling binary lithological heterogeneity. Results from modeling lithological fielddata parameters show the essential flow and transport elements of this site: the volume fraction for high- and low-permeable facies are approximately equal within the studied formation. Moreover, the anisotropy of this site is demonstrated by the fact that its horizontal correlation scale exceeds its vertical scale by more than 60 times. Its vertical effective
ELSE_TSANG_CH017.qxd
8/20/2005
218
10:19 PM
Page 218
Modeling of Waste Injection in Heterogeneous Sandy Clay Formations
conductivity is three orders of magnitude less than its horizontal effective hydraulic conductivity, a phenomenon that accords with the anisotropy of heterogeneity correlation scales. This result indicates that a conceptual model of anisotropic formation could be applied in predictive modeling of flow and transport at this site, instead of the previously utilized layered system. Numerical studies of transport at this site show that the problem of selecting the best model for predicting long-term regional-scale waste transport at this site still exists. The effective macrodispersion model and dual-porosity model are competitive approaches for such predications. Effective macrodispersive model are limited by the fact that dispersion parameters are time and space dependent, whereas the dual-porosity model needs to be further investigated to specify the relationship between dual-medium parameters and heterogeneity. The simulation of 40 years of injection history, up to the present, to obtain 3-D distributions of injected wastes within the studied formation, shows that today, the main mass of waste is predominantly distributed within the sand. The filling of clay bodies through diffusion and advection still plays only a small role in overall waste transport. The more important issue, within the highly permeable unit of the formation under study, is how 3-D clay/sand topology affects the velocity field. The last important practical note is that modeling results did not show any significant upward waste movement during 40 years of injection through preferential flow paths that could be formed in such a heterogeneous system. This fact indicates that the anisotropy of effective correlation scales, which leads to the hydraulic anisotropy of the system, prevents vertical migration of injected waste.
ACKNOWLEDGMENTS This work was supported by the Civilian and Development Research Foundation through Contract RG2-2395-MO-02.
REFERENCES Carle, S.F., 1998. T-PROGS: Transition Probability Geostatistical Software, V. 2.0. Hydrologic Science graduate group, University of California, Davis, CA. Carle, S.F. and Fogg, G.E., 1997. Modeling spatial variability with one- and multi-dimensional continuous Markov chains. Math. Geol., 29(7): 891–918. Dagan, G. and Lessoff, S.C., 2001. Solute transport in heterogeneous formations of bimodal conductivity distribution. Water Resour. Res., 37(3): 465–642. Desbarats, A.J., 1990. Macrodispersion in sand-shale sequences. Water Resour. Res., 26(1): 153–163. Doherty, J., 2001. MODFLOW-ASP—Using MODFLOW-2000 with PEST-ASP. Watermark Computing, Australia. Foley, M.G., Bradley, D.G., Cole, C.R., Hanson, J.P., Hoover, K.A., Perkins, W.A. and Williams, M.D., 1995. Hydrogeology of West Siberian Basin and Tomsk Region. PNL10585, Pacific Northwest Laboratory, Richland, WA. Gelhar, L.W., 1993. Stochastic Subsurface Hydrology. Prentice-Hall, Englewood Cliffs, NJ.
ELSE_TSANG_CH017.qxd
References
8/20/2005
10:19 PM
Page 219
219
Harbaugh, A.W., Banta, E.R., Hill, M.C. and McDonald, M.G., 2000. MODFLOW-2000, The U.S. Geological Survey Modular Ground-Water Model—User Guide to Modularization Concepts and the Ground-Water Flow Process. U.S. Geological Survey Open-File Report 00–92, Reston, VA. Lu, Z. and Zhang, D., 2002. On stochastic modeling of flow in multimodal heterogeneous formations. Water Resour. Res., 38(10): 1190. Rozanov, Yi. A., 1971. Stochastic Processes. Nauka, Moscow. Rybalchenko, A.I., Pimenov, M.K., Kostin, P.P., Balukova, V.D., Nosuckhin, A.V., Mikerin, E.I., Egorov, N.N., Kaimin, E.P., Kosareva, I.M. and Kurochkin, V.M., 1998. Deep Injection of Liquid Radioactive Waste in Russia. Battelle Press, Columbus Richland, Washington, DC. Rybalchenko, A.I., Pimenov, M.K., Kostin, P.P. and Kurochkin, V.M., 1996. Scientific and practical results of deep injection disposal of liquid radioactive waste in Russia. In: J.A. Apps and C.-F. Tsang (Eds). Deep Injection Disposal of Hazardous and Industrial Waste. Academic Press, London. Shestakov, V.M., 2002. Development of relationship between specific storage and depth of sand and clay formation. Environ. Geol., 42(2–3): 127–129. Shestakov, V.M., Kuvaev, A.A., Lekhov, A.V., Pozdniakov, S.P., Rybalchenko, A.I., Zubkov, A.V., Davis, P.A. and Kalinina, E.A., 2002. Flow and transport modeling of liquid radioactive waste injection using data from Siberian chemical plan injection site. Environ. Geol., 42(2–3): 214–221. Zheng, C. and Wang, P.P., 1998. MT3DMS, A Modular Three-Dimensional Multispecies Transport Model for Simulation of Advection, Dispersion and Chemical Reactions of Contaminants in Groundwater Systems, Documentation and User’s Guide. Departments of Geology and Mathematics, University of Alabama.
This page intentionally left blank
220
ELSE_TSANG_CH018.qxd
8/20/2005
9:40 PM
Page 221
Chapter 18
NON-DARCY FLOW BEHAVIOR NEAR HIGH-FLUX INJECTION WELLS IN POROUS AND FRACTURED FORMATIONS Y.-S. Wu Earth Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, CA, USA
18.1 INTRODUCTION Darcy’s law has been used exclusively in studies of porous-medium phenomena. However, high-velocity non-Darcy flow occurs in many cases involving subsurface flow systems, such as in the flow near high-flux wells during oil or gas production, water pumping, and liquid waste injection. Theoretical, field, and experimental studies on non-Darcy flow in porous media have been performed, most of which have been focused on singlephase flow conditions that pertain to the oil and gas industry (e.g., Tek et al., 1962; Swift and Kiel, 1962; Lee et al., 1987). Other researchers have devoted their efforts to finding and validating correlations of non-Darcy flow coefficients (e.g., Liu et al., 1995). In studies of non-Darcy flow through porous media, the Forchheimer equation has been generally used to describe single-phase non-Darcy flow. Several studies in the literature extend the Forchheimer equation to multiphase flow and provide equations for correlating non-Darcy flow coefficients under multiphase conditions (Evans et al., 1987; Evans and Evans, 1988; Liu et al., 1995; Wu, 2001). A general numerical method has recently been developed in modeling single-phase and multiphase non-Darcy flow in multidimensional porous and fractured reservoirs (Wu, 2002a). This numerical model incorporates the extended Forchheimer equation using an integral finite-difference or a control-volume numerical discretization scheme, and implements an extended dual-continuum approach (such as the double- or multiple-porosity, or the dual-permeability method) for simulating non-Darcy fracture-matrix flow in a fractured medium. Relative permeability and capillary pressure functions required for modeling two- or three-phase flow are borrowed directly from the multiphase Darcy flow theory; i.e., they are treated as functions of saturations only. In addition, an approximate analytical solution is obtained for analyzing non-Darcy wellflow behavior through fractured reservoirs (Wu, 2002b). The objectives of this study are (1) to obtain insights into non-Darcy flow as it relates to transient pressure behavior through porous and fractured reservoirs, and (2) to provide type curves for well test analyses of non-Darcy flow wells. The type curves generated include cases of drawdown, injection, and buildup tests with non-Darcy flow occurring in porous and fractured reservoirs. In addition, non-Darcy flow into partially penetrating wells is also considered. The transient pressure type curves for flow in fractured reservoirs are based on the double-porosity model. The type curves provided in this work will find applications in interpreting well tests of non-Darcy flow in porous and fractured reservoirs, using a typecurve matching technique.
ELSE_TSANG_CH018.qxd
8/20/2005
9:40 PM
Page 222
222
Non-Darcy Flow Behavior Near High-flux Injection Wells
18.2 MATHEMATICAL MODEL AND NUMERICAL FORMULATION A multiphase system in a porous or fractured aquifer is assumed to be composed of three phases: nonaqueous phase liquid (NAPL) (oil), gas (air), and water. For simplicity, the three fluid components (water, NAPL, and gas) are assumed to be present only in their associated phases, with single-phase flow regarded as a special case of multiphase flow. According to the multiphase extension of the Forchheimer equation for non-Darcy flow, each phase flow, in response to pressure, gravitational, and capillary forces, is governed by its mass conservation:
⭸ ᎏ (φ Sf f)⫽⫺ⵜ•( fvf)⫹qf , ⭸t
(18.1)
where f is the density of fluid f (f = w for water, f = n for NAPL or oil, and f = g for gas); is the Darcy (or volumetric) velocity of fluid f; Sf is the saturation of fluid f; φ is the effective porosity of formation; t is time; and qf is the sink/source term of phase (component) f per unit volume of formation, representing mass exchange through injection/production wells or resulting from fracture–matrix interactions. Volumetric flow rate (namely, Darcy velocity in the conventional Darcy flow case) for non-Darcy flow of each fluid may be described using the multiphase extension of the Forchheimer equation (Evans and Evans, 1988; Liu et al., 1995; Katz and Lee, 1990):
f ⫺(ⵜPf⫺fg)⫽ ᎏ vf⫹ffvf|vf|, kkrf
(18.2)
where Pf is the pressure of phase f; g is the gravitational constant vector; f is the dynamic viscosity of fluid f, k is the absolute/intrinsic permeability of the formation; krf is relative permeability to phase f and is regarded as a function of saturation; and f is the effective non-Darcy flow coefficient, with a unit m−1 for fluid f under multiphase flow conditions (Evans and Evans, 1988). Equation (18.1), the mass-balance governing equation for three phases, needs to be supplemented with constitutive equations, which express all the secondary variables and parameters as functions of a set of primary thermodynamic variables of interest. Here, we borrow relative permeability and capillary pressure relations, and other correlations such as density and viscosity, from the multiphase Darcy flow model to complete the problem description. The additional nonlinearity introduced by non-Darcy flow to the governing equations makes it in general necessary to use a numerical approach. Equation (18.1) can be discretized in space using an integral finite-difference or control-volume finite-element scheme for a porous and/or fractured medium (Wu, 2002a). The time discretization is carried out with a backward, first-order, finite-difference scheme. The discrete nonlinear equations for water, NAPL, and gas flow at node i are written as follows:
冦( S )
冧 ᎏ⌬t ⫽冱(F )
n⫹1 n f f i ⫺( Sff) i
Vi
n⫹1 n+1 f ij ⫹Qfi ,
(18.3)
j⑀i
where n denotes the previous time level; n+1 is the current time level; Vi is the volume of element i (porous or fractured block); ∆t is the time step size; and ηi contains the set of neighboring elements (j), porous or fractured block, to which element i is directly
ELSE_TSANG_CH018.qxd
8/20/2005
9:40 PM
Page 223
18.2 Mathematical Model and Numerical Formulation
223
connected. Ff is a mass-flow term between elements i and j, defined (when Equation (18.2) is used) as
冦
冤冢 ᎏ 冣 ⫺␥ (ψ ⫺ψ )冥 冧
Aij 1 Ff⫽ ᎏᎏ ⫺ᎏ⫹ 2(kf)ij⫹1/2 f
1
2
1/2
ij
fj
fi
(18.4)
f
where subscript ij+1/2 denotes a proper averaging of properties at the interface between the two elements and Aij is the common interface area between connected elements i and j. The mobility of phase f is defined as krf f⫽ ᎏ , f
(18.5)
and the flow potential term is ψfi⫽Pfi⫺ij+1/ 2gDi,
(18.6)
where Di is the depth to the center of element i. The mass sink/source term at element i, Qfi for phase f, is defined as Qfi⫽qfiVi .
(18.7)
In Equation (18.4), the transmissivity of flow terms is defined (if the integral finite-difference scheme is used) as 4(k2ff)ij⫹1/2 , ␥ij⫽ ᎏᎏ di⫹dj
(18.8)
where di is the distance from the center of element i to the interface between elements i and j. In the model formulation, absolute permeability, relative permeability, and the effective non-Darcy flow coefficient are all considered to be flow properties of the porous media and need to be averaged between connected elements in calculating the mass-flow terms. In general, absolute permeability is harmonically weighted along the connection between elements i and j, and relative permeability and non-Darcy flow coefficients are both upstream weighted. Then, the nonlinear, discrete Equation (18.3) is solved using a Newton/Raphson iteration. The technique used for handling non-Darcy flow through fractured rock follows the dualcontinuum methodology (Barenblatt et al., 1960; Warren and Root, 1963; Pruess and Narasimhan, 1985; Wu, 2002a). The method treats fracture–matrix interactions with a multicontinuum numerical approach, including the double- or multiporosity method, the dual-permeability method, and the more general “multiple interacting continua” (MINC) method. The non-Darcy flow formulation, Equation (18.1) or (18.3), is applicable to both single-continuum and multicontinuum media. Using the dual-continuum concept, Equation (18.1) or (18.3) can be used to describe single-phase and multiphase flow, respectively, both in fractures and inside matrix blocks, when dealing with fractured reservoirs. Special attention needs to be paid to treating fracture–matrix flow terms with Equations (18.3) and (18.4) for estimating mass exchange at fracture–matrix interfaces using a double-porosity approach. In particular, Wu (2002a) has shown that for the double-porosity or nested discretizations, the characteristic length of non-Darcy flow distance between fractures and matrix crossing the
ELSE_TSANG_CH018.qxd
8/20/2005
9:40 PM
224
Page 224
Non-Darcy Flow Behavior Near High-flux Injection Wells
interface may be approximated using the results for Darcy flow (Warren and Root, 1965; Pruess, 1983). The flow between fractures and the matrix is still evaluated using Equation (18.4); however, the transmissivity for the fracture–matrix flow is then given by 4(k2mff)ij+1/2 ␥ij⫽ ᎏᎏ , lfm
(18.9)
where km is matrix permeability, and lfm is the characteristic distance for flow crossing fracture–matrix interfaces (Table 4.1; Wu, 2002a). In modeling flow through a fractured rock using the numerical formulation of this work, a modeler needs to essentially figure out how to generate a grid that represents both the fracture and matrix systems. Several fracture–matrix subgridding schemes exist for designing different meshes for different fracture–matrix conceptual models (Pruess, 1983). Once a proper mesh of a fracture–matrix system is generated, fracture and matrix blocks are specified to represent fracture or matrix domains, separately. Formally, they are treated in exactly the same way in the solution of the discretized model. However, physically consistent fracture and matrix properties and modeling conditions must be appropriately specified for fracture and matrix systems, respectively. 18.3 DIMENSIONLESS VARIABLES AND ANALYTICAL SOLUTIONS Let us define the following group of dimensionless variables (Earlougher, 1977; Warren and Root, 1963): the dimensionless radius is r rD⫽ ᎏ , rw
(18.10)
with r being the radial coordinate or distance and rw a well radius. The dimensionless time is kt tD⫽ ᎏ rw2 C
(18.11a)
for a porous formation, and kf t tD⫽ ᎏᎏ rw2 (mCm⫹fCf)
(18.11b)
for a fractured formation. In Equation (18.11), kf is fracture absolute permeability in a fractured formation, C is compressibility, and subscripts m and f denote matrix and fracture, respectively. The dimensionless non-Darcy flow coefficient is kqm D⫽ ᎏ 2 rwh
(18.12a)
for a porous formation, and kf qmf D⫽ ᎏ 2rwh
(18.12b)
ELSE_TSANG_CH018.qxd
8/20/2005
9:40 PM
Page 225
18.4 Type Curves of Non-Darcy Flow
225
for a fractured formation. The dimensionless fracture pressure is Pi⫺Pw(t) PD⫽ ᎏᎏ , q Ⲑ2kfh
(18.13)
where Pi is the initial pressure of the formation, a constant, and Pw(t) is the well pressure, a function of time. Note that in Equation (18.12), qm is a mass production or injection rate, treated as a constant. In addition, Warren and Root define two more dimensionless parameters to characterize double-porosity flow behavior. The first one is the ratio of fracture porosity–compressibility to the total system porosity–compressibility product as
φfCf ⫽ ᎏᎏ , φmCm⫹φfCf
(18.14)
and the second is the interporosity flow parameter:
␣rw2 km ⫽ ᎏ , kf
(18.15)
with α being a shape factor of rock matrix blocks. An approximate analytical solution for transient non-Darcy flow in a fractured medium is derived (Wu, 2002b) as a superposition of the Warren-Root solution (1963) and the nonDarcy flow coefficient: tD 1 tD PD(rD⫽1, tD)⫽D⫹ ᎏ 1n tD⫹0.80907⫹Ei ⫺ ᎏ ⫺Ei ⫺ᎏ 2 (1⫺) (1⫺)
冤
冢
冣 冢
冣冥.
(18.16)
Note that the non-Darcy flow coefficient is the extra flow resistance introduced by nonDarcy flow under steady-state flow conditions in the fracture. The reason for this approximate solution, Equation (18.16), is that for transient non-Darcy flow in a double-porosity reservoir, flow may be quick to approach a quasi-steady state within a region near the well, because of the low storage capacity and high permeability of the fracture continuum. Pressure drops caused by non-Darcy flow are approximated by a summation of steady-state flow resistance and the transient Darcy flow term—the second term on the right-hand side of Equation (18.16). This approximate solution has been shown to be very accurate for nonDarcy flow through normal double-porosity fractured media (Wu, 2002b).
18.4 TYPE CURVES OF NON-DARCY FLOW In this section, we present several applications and discuss single-phase, non-Darcy flow behavior. In addition, we provide several commonly used dimensionless pressures or type curves for non-Darcy-flow well-test analyses, including: 1. Pressure drawdown and buildup analyses 2. Effects of finite boundaries for reservoirs 3. Pressure drawdown in fractured reservoirs 4. Pressure responses in partially penetrating wells of porous and fractured reservoirs
ELSE_TSANG_CH018.qxd
8/20/2005
226
9:40 PM
Page 226
Non-Darcy Flow Behavior Near High-flux Injection Wells
These application examples deal with single-phase, slightly compressible fluid transient flow. In addition, type curves of non-Darcy flow through a single well are generated using numerical solutions for single-phase, slightly compressible non-Darcy fluid flow in infiniteor finite-acting reservoirs. 18.4.1
Pressure Drawdown and Buildup
This example deals with non-Darcy flow through an infinite-acting reservoir. The flow is approximated by a one-dimensional, radially symmetrical formation in the numerical model, with an outer boundary radius of 5 × 106 (m), discretized into a one-dimensional grid of 3100 gridblocks in logarithmic scale. Initially, the system is undisturbed and at constant pressure. A fully penetrating production well, represented by a well element, starts pumping at t = 0, specified at a constant water-pumping rate. Input parameters for this problem are presented in Table 18.1. Figure 18.1 shows a set of type curves for pressure drawdown, calculated by the numerical model in terms of dimensionless pressure versus dimensionless time. As shown here, the non-Darcy flow coefficient is a very important and sensitive parameter for pressure drawdown plots. As a result, the figure indicates that the non-Darcy flow coefficient can be effectively estimated using the type curves with the traditional type-curve matching approach, owing to its sensitivity. Figure 18.2 presents simulated pressure drawdown and buildup curves, in which the well is pumped for 1 day only and then shut off. Pressure variations in the well during the entire pumping and shut-in period, as shown in Figure 18.2, indicate that pressure buildup is insensitive to the values of non-Darcy flow coefficients, compared to drawdown in pumping periods. This insensitivity results from the rapid reduction in flow velocity near the well after a well is shut off and non-Darcy flow effects become ineligible. Many additional modeling investigations have verified this observation. From these investigations as well as our own study, we conclude that pressure-buildup tests are not suitable for estimating non-Darcy flow coefficients. On the other hand, the pressure-buildup method following non-Darcy flow pumping tests will be a good way to determine permeability values, since there is no significant non-Darcy flow effect.
Table 18.1. Parameters for the pressure drawdown and buildup analysis Parameter
Value
Unit
Initial pressure Initial porosity Reference fluid density Formation thickness Fluid viscosity Fluid compressibility Rock compressibility Permeability Water pumping rate Wellbore radius Outer boundary radius Dimensionless non-Darcy Flow coefficient
Pi = 10 φi = 0.20 ρi = 1000 h = 10 µ = 1 × 10−3 Cf = 5 × 10−10 Cr = 5 × 10−9 k = 9.869 ×10−13 qv = 0.1 rw = 0.1 re= ∞ ≈5 × 106 βD = 1 × 10−3 , 1, 10, 100, 1 × 103, 1 × 104, 1 × 105
bar kg/m3 m Pa s Pa−1 Pa−1 m2 m3/d m m
ELSE_TSANG_CH018.qxd
8/20/2005
9:40 PM
Page 227
18.4 Type Curves of Non-Darcy Flow
227
104
Dimensionless Pressure (PD)
103
102
101 Theis Solution βD=10
-3
βD=100 βD=10+1
100
βD=10
+2
βD=10
+3
βD=10
+4
βD=10+5
10-1 100
101
102
103
104
105
106
107
108
109
Dimensionless Time (tD)
Fig. 18.1. Type curves for dimensionless pressures of non-Darcy flow in an infinite system without wellbore storage and skin effects. 20
Dimensionless Pressure (PD)
βD=0 βD=1 βD=10 15
10
5
100
101
102
103
104
105
106
107
108
109
Dimensionless Time (tD)
Fig. 18.2. Dimensionless pressures for 1-day pumping, followed by pressure buildup, of non-Darcy flow in an infinite system without wellbore storage and skin effects.
ELSE_TSANG_CH018.qxd
228
8/20/2005
9:40 PM
Page 228
Non-Darcy Flow Behavior Near High-flux Injection Wells
18.4.2 Effects of Finite Reservoir Boundaries Boundary effects or well interference in finite, developed reservoirs will show up in well tests sooner or later. Two types of boundary conditions, closed and constant pressure, are commonly used to approximate the effects of finite reservoir/well boundaries. A finite flow system and its parameters are similar to those above. Only two finite radial systems with outer boundary radii, re = 1000 and 10,000 m, are considered. Figures 18.3 and 18.4 show dimensionless pressure drawdown curves, for closed and constant-pressure boundaries, as well as the two radii. For a smaller, finite formation system with re = 1000 m, Figure 18.3 shows that significant boundary effects occur at about dimensionless time tD = 108 (1 day in real time), at which the well pressure responses deviate from the infinite-acting solution (say, the Theis solution for small non-Darcy flow coefficients). For the larger system with re = 10,000 m, boundary effects are very similar, but show up much later (Fig. 18.4). 18.4.3 Non-Darcy Flow in Fractured Media This problem exemplifies non-Darcy flow through a fractured reservoir. The fracture–matrix formation is described using the Warren and Root double-porosity model (Wu, 2002a). The physical flow model is that of a typical transient flow toward a well that fully penetrates a radially infinite, horizontal, uniform, fractured reservoir. In numerical modeling, for comparison, a radially finite reservoir (re = 5 × 106 m) is used and discretized into a one-dimensional (primary) grid. The r-distance of 5 × 106 m is subdivided into 3100 intervals in a logarithmic scale. A double-porosity mesh is generated from the primary grid, in which a three-dimensional fracture network and cubic matrix blocks are used. The uniform matrix block size is 1 × 1 × 1 m, and fracture permeability and aperture are correlated by the cubic law. Model input parameters are given in Table 18.2. Figure 18.5 shows a comparison of the numerical modeling results and the approximate analytical solution (3.7) with different dimensionless non-Darcy flow coefficients. The two characteristic parameters for these cases are ␥⫽6 ⫻ 10⫺5 and ⫽2 ⫻ 10⫺3, from the parameters used as listed in Table 18.2. Note that the values of these two parameters are within a typical range of double-porosity flow behavior, as discussed by Warren and Root. Figure 18.5 shows excellent agreement between the analytical (circled-symbol curves, labeled as PD,WR + βD) and numerical (solid-line curves) solutions, except at earlier times (tD < 100) or for large non-Darcy flow coefficients (βD > 10). For non-Darcy flow into a well from an infinite fractured system, well-pressure type curves are shown in semi-log plots of Figure 18.5. The type curves in the figure show that well (fracture) pressures are extremely sensitive to the value of non-Darcy flow coefficients; therefore, well pumping tests will help to determine this constant in a fractured reservoir. Furthermore, Figure 18.5 indicates that the effects of non-Darcy flow on early transient pressure responses are very strong, such that the first semi-log straight lines may not develop when non-Darcy flow is involved in a fracture reservoir. 18.4.4 Non-Darcy Flow with Partial Penetrating Wells Here, non-Darcy flow is considered as occurring in a partially penetrating well from an infinite-acting, homogeneous, isotropic, porous, or fractured reservoir. Flow near a partially penetrating production well is three-dimensional toward the wellbore, and it can be handled mathematically using a 2D, axially symmetrical (r–z) grid. In the numerical model, the
ELSE_TSANG_CH018.qxd
8/20/2005
9:40 PM
Page 229
18.4 Type Curves of Non-Darcy Flow
229
103 The is Solution βD=0.1, Closed Boundary βD=0.1, Constant Pressure βD=1.0, Closed Boundary
Dimensionless Pressure (PD)
βD=1.0, Constant Pressure βD=10, Closed Boundary
10
βD=10, Constant Pressure
2
101
100 103
104
105
106
107
108
109
1010
1011
1012
Dimensionless Time (tD)
Fig. 18.3. Type curves for dimensionless pressures of non-Darcy flow in a finite system with an outer boundary radius of 1000 m. 103 βD=0.1, Closed Boundary βD=0.1, Constant Pressure βD=1.0, Closed Boundary βD=1.0, Constant Pressure
Dimensionless Pressure (PD)
βD=10, Closed Boundary βD=10, Constant Pressure
102
101
100 105
106
107
108
109
1010
1011
1012
1013
Dimensionless Time (tD)
Fig. 18.4. Type curves for dimensionless pressures of non-Darcy flow in a finite system with an outer boundary radius of 10,000 m.
ELSE_TSANG_CH018.qxd
8/20/2005
9:41 PM
230
Page 230
Non-Darcy Flow Behavior Near High-flux Injection Wells Table 18.2. Parameters for the single-phase, fractured-medium flow problem
Parameter
Value
Matrix porosity Fracture porosity Reference water density Water-phase viscosity Matrix permeability Fracture permeability Water production rate Rock compressibility Water compressibility Dimensionless non-Darcy Flow coefficient for fracture Dimensionless non-Darcy Flow coefficient for matrix Wellbore radius
φm = 0.30 φf = 0.0006 ρw = 1000 µw = 1 × 10−3 km = 1.0 × 10−16 kf = 9.869 × 10–13 qm = 0.1 Cr = 1.0 × 10–9 Cw = 5.0 × 10–10 βD, f = 1 × 10−4, 1, 5, and 10 βD, m= 1 × 10–3, 10, 50, and 100 rw = 0.1
35
Unit
kg/m3 Pa s m2 m2 kg/s 1/Pa 1/Pa
m
Warren-Root Solution βD=1 βD=5 βD=10
30
βD=15 βD=20
Dimensionless Pressure (PD)
PD,WR + βD
25
20
15
10
5
100
102
104
106
108
1010
Dimensionless Time (tD)
Fig. 18.5. Type curves for dimensionless pressures of non-Darcy flow in an infinite fractured system, with comparisons to the approximate analytical solution.
infinite-acting reservoir is approximated by a 2-D, radially symmetrical reservoir with an outer boundary radius (r = 1 × 107 m) and a thickness of 10 m in the vertical, z-direction. The system is discretized into a 2D grid of 1000 divisions in the r-direction, using a logarithmic
ELSE_TSANG_CH018.qxd
8/20/2005
9:41 PM
Page 231
18.4 Type Curves of Non-Darcy Flow
231
scale and five uniform grid layers in the z-direction for the porous reservoir. For the fractured flow example, the single-porosity, porous reservoir grid is further processed into a doubleporosity grid using the MINC technology. Initially, the two single-phase systems are both at vertical-gravity equilibrium. Partially penetrating wells with a percentage of wellbore completion are represented by single-well elements, and the results are compared. The parameters for the porous reservoir are those given in Table 18.1, and the fracturedreservoir properties are given in Table 18.2, with the fractured reservoir handled using the double-porosity model. Two type curves for pressure drawdown, calculated in terms of dimensionless pressure versus dimensionless time, are shown in Figures 18.6 and 18.7, respectively, for the porous and fractured reservoirs. Figures 18.6 and 18.7 show the significant impact of well-penetration percentage on well pressure behavior in both the porous medium and fractured reservoirs. As completed well screen lengths decrease (i.e., as wellbore penetration or open-screen length becomes smaller), the flow resistance and pressure drops at the well increase significantly to maintain the same production rates. We could expect a larger impact of well partial penetration on the non-Darcy flow regime near a well than on Darcy flow, because of higher flow rates or large non-Darcy flow effects near the wellbore. However, comparison of the straight lines developed in the type curves at late times (Figs. 18.6 and 18.7) indicates that the same pseudo-skin concept (Earlougher, 1977) may also be applicable to analyzing the partial penetration effects of non-Darcy flow at wells.
Dimensionless Pressure (PD)
103
102
101 100% Penetration 80% Penetration 60% Penetration 40% Penetration 20% Penetration
100
100
101
102
103
104
105
106
107
108
109
Dimensionless Time (tD)
Fig. 18.6. Type curves for dimensionless pressures of non-Darcy flow at partially penetrating wells in an infinite porous reservoir (βD = 10), with different degrees of well penetration.
ELSE_TSANG_CH018.qxd
8/20/2005
9:41 PM
232
Page 232
Non-Darcy Flow Behavior Near High-flux Injection Wells 50 45 40 Dimensionless Pressure (PD)
100% Penetration
35
80% Penetration 60% Penetration 40% Penetration
30
20% Penetration
25 20 15 10 5 0
100
101
102
103
104
105
106
107
108
Dimensionless Time (tD)
Fig. 18.7. Type curves for dimensionless pressures of non-Darcy flow at partially penetrating wells in an infinite fractured reservoir (βD = 1) with different degrees of well penetration.
18.5 SUMMARY AND CONCLUSIONS This chapter presents a theoretical study of non-Darcy flow behavior through porous and fractured rock, which may occur near pumping or production wells during high-flux injection of waste fluids into underground formations. In this study, both numerical and analytical solutions are used, with non-Darcy flow simulated using the Forchheimer equation and a three-dimensional, multiphase flow reservoir simulator. Non-Darcy flow through a fractured reservoir is handled using a general dual-continuum concept, covering such commonly used conceptual models as double-porosity, dual-permeability, and explicit-fracture models. Using numerical simulation results, we analyzed fundamental non-Darcy flow behavior for transient pressures in porous and fractured reservoirs. In particular, we provide a number of dimensionless type curves for well-test analyses of non-Darcy flow wells. The type curves generated include various types of drawdown, injection, and buildup tests with nonDarcy flow occurring in porous and fractured reservoirs. In addition, non-Darcy flow into partially penetrating wells is also considered. The type curves provided in this work for nonDarcy flow in porous and fractured reservoirs can be used in well-test interpretations using a type-curve matching technique.
ELSE_TSANG_CH018.qxd
8/20/2005
References
9:41 PM
Page 233
233
ACKNOWLEDGMENTS The author is indebted to Guoxiang Zhang and Dan Hawkes for their careful and critical review of this manuscript. This work was supported in part by the Assistant Secretary for Energy Efficiency and Renewable Energy, Office of Geothermal and Wind Technologies of the U.S. Department of Energy. The support is provided to Berkeley Lab through U.S. Department of Energy Contract No. DE-AC03-76SF00098.
REFERENCES Barenblatt, G.I., Zheltov, I.P. and Kochina, I.N., 1960. Basic concepts in the theory of seepage of homogeneous liquids in fissured rocks, PMM Sov. Appl. Math. Mech., 24(5): 852–864. Earlougher, R.C., Jr., 1977. Advances in Well Test Analysis. SPE Monograph, Vol. 5. SPE of AIME, Dallas. Evans, E.V. and Evans, R.D., 1988. Influence of an immobile or mobile saturation on nonDarcy compressible flow of real gases in propped fractures. J. Petrol. Technol. 40(10): 1343–1351. Evans, R.D., Hudson, C.S. and Greenlee, J.E., 1987. The effect of an immobile liquid saturation on the non-Darcy flow coefficient in porous media. J. SPE Prod. Eng. Trans. AIME, 283, 331–338. Katz, D.L. and Lee, R.L., 1990. Natural Gas Engineering, Production and Storage, Chemical Engineering Series. McGraw-Hill, New York. Lee, R.L., Logan, R.W. and Tek, M.R., 1987. Effects of turbulence on transient flow of real gas through porous media. SPE Formation Eval., 108–120. Liu, X., Civan, F. and Evans, R.D., 1995. Correlations of the non-Darcy flow coefficient. J. Can. Petrol. Technol. 34(10): 50–54. Pruess, K., 1983. GMINC—A mesh generator for flow simulations in fractured reservoirs. Report LBL-15227, Lawrence Berkeley National Laboratory, Berkeley, CA. Pruess, K. and Narasimhan, T.N., 1985. A practical method for modeling fluid and heat flow in fractured porous media. Soc. Petrol. Eng. J., 25: 14–26. Swift, G.W. and Kiel, O.G., 1962. The prediction of gas-well performance including the effects of non-Darcy flow. J. Petrol. Technol. Trans. AIME, 222: 791–798. Tek, M.R., Coats, K.H. and Katz, D.L., 1962. The effects of turbulence on flow of natural gas through porous reservoirs. J. Petrol. Technol., Trans. AIME, 222: 799–806. Warren, J.E. and Root, P.J., 1963. The behavior of naturally fractured reservoirs. Soc. Petrol. Eng. J., Trans. AIME, 228: 245–255. Wu, Y.S., 2001. Non-Darcy displacement of immiscible fluids in porous media. Water Resour. Res., 37(12): 2943–2950. Wu, Y.S., 2002a. Numerical simulation of single-phase and multiphase non-Darcy flow in porous and fractured reservoirs. Trans. Porous Media, 49(2): 209–240. Wu, Y.S., 2002b. An approximate analytical solution for non-Darcy flow in fractured media, LBNL-48197. Water Resources Res. 38(3): 5-1–5-7.
This page intentionally left blank
234
ELSE_TSANG_CH019.qxd
8/22/2005
7:06 AM
Page 235
Chapter 19
MODELING DENSITY CHANGES IN HAZARDOUS DISPOSAL WELL PLUMES R.G. Larkina and J.E. Clarkb a
R.G. Larkin Consulting, Austin, TX, USA E.I. du Pont de Nemours & Co., Inc., Beaumont, TX, USA
b
19.1 INTRODUCTION Class I hazardous disposal well operators must submit a “No Migration” petition to the U.S. Environmental Protection Agency (EPA) to apply for an exemption from the federal land ban restrictions on hazardous waste disposal. In many Class I disposal well operations, the injectate density can vary daily and usually has a different density from the native formation brine. In most cases, the influence of the density contrast on the fluid potential must be included in a petition modeling demonstration. After the model well(s) is shut-in, the density contrast results in a net updip (lower-density injectate) or downdip (higher-density injectate) component of flow near the center of mass of the calculated plume (neglecting regional flow). In the past, two model runs were required by EPA Region VI: one using the lowest and one using the highest anticipated injectate density for the entire operational period. The results were then combined into one large composite plume. This was intended to provide plume delineation over the injectate-density range, and implies that buoyancy effects on the plume position are not controlled by the average density. At the time when this policy was implemented, no studies had been done to determine if injectate-density variations could produce a split or “dumbbell-shaped” plume with heavier wastes segregating downdip and lighter wastes segreglating updip. A series of modelingh studies designed to address this issue—Larkin and Clark (1994a), Larkin et al. (1994b, 1996), and Fahy et al. (1992)—demonstrated with site-specific and generic modeling that dumbbell-shaped, lateral-density segregation is not predicted by modeling results. The 10,000-year plume location was found to be controlled by the operational period average of injectate density, and not by shorter-term density fluctuations. In contaminant transport models of deep-well injection, Fahy et al. (1992) used an injectate-density input that varied sinusoidally with time. The authors reported that injectate-density variations in their models were strongly damped by dispersion. At a distance of 600 ft from the model well, density variations were not significantly different from the average injectate density. The conclusion of this study was that periodic injectate-density variations about the mean will not yield a 10,000 year plume location that is significantly different from the plume location calculated by using the average injectate density. Larkin and Clark (1994a), and Larkin et al. (1994b, 1996), used the Sandia WasteIsolation Flow and Transport code (SWIFT) in generic and site-specific two-dimensional models to examine buoyancy effects in disposal-well plumes. The models used constant and
ELSE_TSANG_CH019.qxd
236
8/22/2005
7:06 AM
Page 236
Modeling Density Changes in Hazardous Disposal Well Plumes
variable injectate-density input values. In the generic cases, model runs using random daily injectate-density fluctuations about the mean were compared with a model that used the mean-density-input value for the entire 20-year operational period. Temporal sensitivity was tested in other generic runs; the time period for which a random density was selected (called the density-averaging period) was increased to 30, 180, 365, and 3650 days. Once selected, the injectate density was held constant for the specified time period. Significant deviations from the mean results were only calculated in the unlikely case of a 3650-day averaging period. Differences between the calculated plume location and extent for all other cases could not be distinguished in contour plots of 10,000-year simulations. Larkin et al. (1996) used daily injectate density and flow-rate data from an operating Class I injection well facility in the Gulf Coast, in a 10,000 year model. The model plumes were compared with the average case. The site-specific results agreed with the generic modeling results. The extent, location, and maximum concentrations of the modeled plumes at 250 and 10,000 years postinjection were so similar that significant differences could not be distinguished in contour plots. The current study encompasses generic modeling designed to be complementary to the earlier density modeling. The goal of the study is to provide modeling results to support the use of a range of quarterly averages as opposed to a range of extreme injectate-density values in future petition modeling. As in earlier studies, the SWIFT model (Reeves et al., 1986; Ward et al., 1984) was used. Model input parameters were chosen to be representative of values for the Gulf Coast. In two runs, the goal of the modeling was to determine the relative importance of changes in flowrate schedule. In other runs, the injectate density was assigned the value of the formationfluid density for one quarter, and the original injectate density for the next quarter over the entire operational period, to simulate the effects of extreme injectate-density changes. These runs were designed to provide a “worst case” result to determine the impact of implementing a quarterly average injectate-density modeling policy in EPA Region VI. In reality, injectate density is more likely to vary on a daily basis due to rainfall events and process variations. This means that the quarterly modeling performed here is extremely conservative.
19.2 GENERIC MODEL DESCRIPTION Generic Gulf Coast models were used to investigate the effects of variable-density injection. The models are generic in the sense that the input parameters are representative of hydrogeologic conditions in deep brine aquifers of the Gulf Coast. Most input parameters were chosen to be representative of extreme values for the Gulf Coast that would generate the largest plume. This was done in order to include conditions representative of the subsurface for a large number of Class I hazardous injection well operators. For the purposes of the modeling, the extent of the plume was determined by the location of the relative concentration (C/Co) contour equal to 1 ⫻ 10–12. At this extent, the concentration of the fullstrength model waste has been reduced by 12 orders of magnitude. This would be a reduction sufficient to render virtually any injected waste nonhazardous. The generic groundwater flow and contaminant transport model was designed using the SWIFT code (Reeves et al., 1986; Ward et al., 1984). The same basic generic model was used in the different model cases described below. Any density input changes and other parameter variations are given in the descriptions below. SWIFT is a three-dimensional finite difference code that can be used to simulate groundwater flow, contaminant transport,
ELSE_TSANG_CH019.qxd
8/22/2005
7:06 AM
Page 237
19.4 Modeling Results—Variation of Rate Schedule, Runs 1 and 2
237
and heat transport in single or dual porosity media. In SWIFT, the equations governing groundwater flow and solute transport are coupled through: (1) the pore fluid velocity; (2) the dependence of the fluid density on pressure, solute mass fraction, and temperature; and (3) the dependence of fluid viscosity on solute mass fraction and temperature.
19.3 GENERIC MODEL INPUTS The model is homogenous, isotropic, and two-dimensional with 381 variably spaced nodes in the x-direction (aligned along the hydraulic gradient), and 66 variably spaced nodes in the y-direction. The model thickness in the z-direction is constant at 40 ft (12.2 m). The model was designed to be two-dimensional lateral with no-flow boundaries along the sides. The model also took advantage of symmetry about the y-axis to reduce the needed number of nodes and represents one-half of the model region perpendicular to the direction of groundwater flow. One injection well was located along the axis of symmetry at node x = 122, y = 1 or at x = 222, y = 1 (in low-average injectate-density cases). The model well was operated at a constant injection rate of 400 gallons per minute (gpm) (0.0252 m3/s) for 20 years in one case; all other cases used a rate of 1000 gpm (0.0631 m3/s) for 8 years. The modeled injection interval dips at a constant angle of 2.0° in all cases. In the models, the direction of dip corresponds to the direction of increasing magnitude along the x-axis. Constant pressure boundary conditions were assigned at the updip and downdip ends of the grids to impose an average linear groundwater velocity of 0.0 or 2.0 ft/yr (0.61 m/yr). The average linear velocity is the product of the hydraulic conductivity and the hydraulic gradient divided by the effective porosity. The initial pressure and temperature at the well location were 2325 psi (16.03 ⫻ 106 Pa) and 155°F (68.3°C) corresponding to a depth of 5000 ft (1524 m). The other input parameters used in the generic model are given in Table 19.1. Unless noted, the parameters given in the table are common to all models.
19.4 MODELING RESULTS—VARIATION OF RATE SCHEDULE, RUNS 1 AND 2 The generic model discussed above was used to model two different scenarios for direct comparison. The goal of the model Runs 1 and 2 was to determine the relative importance of changes in the model rate schedule on the 10,000-year plume configuration. The model scenario and results are summarized for each run below. All other inputs are given in Table 19.1. In Run 1, the injectate density was assigned to be 71.32 lb/ft3 (1142.44 kg/m3) while the formation-fluid density was 65.52 lb/ft3 (1049.53 kg/m3). One well injected at 1000 gpm (0.0631 m3/s) for 8 years. The hydraulic conductivity was 11.22 ft/day (3.96 × 10–5 m/s), the dip was 2°, and the regional velocity was 2.0 ft/yr (0.61 m/yr). In Run 2, one well injected at 400 gpm (0.0252 m3/s) for 20 years. All other inputs were the same. In this manner, the same volume was maintained between the two runs. The model-calculated relative concentrations at the end of 10,000 years are contoured in Figures 19.1 and 19.2. From these two runs, it was concluded that the injection schedule would have no bearing on the model-calculated plume as long as the volume remained constant. Because no difference was observed between calculated plumes in Runs 1 and 2, the shorter operational period was chosen for the remaining runs.
ELSE_TSANG_CH019.qxd
8/22/2005
7:06 AM
238
Page 238
Modeling Density Changes in Hazardous Disposal Well Plumes Table 19.1. Generic model input parameters
Input Parameter
Value
Permeability
2 darcies (1.974 ⫻ 10
Hydraulic conductivity Formation-fluid density
Variable with run (see text) Variable with run (see text) 66.14 lb/ft3 (1059.46 kg/m3) at 155°F (68.3°C) 65.52 lb/ft3 (1049.53 kg/m3) at 155°F (68.3°C) Variable with run (see text) 71.32 lb/ft3 (1142.44 kg/m3) at 155°F (68.3°C) 62.4 lb/ft3 (1000 kg/m3) at 155°F (68.3°C) 64.27 lb/ft3 (1029.51 kg/m3) at 155°F (68.3°C) 0.30 Range ⫽ 1.136 cP (0.001136 Pa s) at 75°F (23.9°C) to 0.447 cP (4.47 ⫻ 10⫺4 Pa s) at 175°F (79.4°C) Range ⫽ 1.78 cP (0.00178 Pa s) at 75°F (23.9°C) to 0.377 cP (3.77 ⫻ 10⫺4 Pa s) at 175°F (79.4°C) 0.0 or 2.0 ft/yr (0.62 m/yr) 2.0° 40 ft (12.2 m) 3.3 ⫻ 10⫺6 psi⫺1 (4.79 ⫻ 10−10 Pa⫺1) 3.3 ⫻ 10⫺6 psi⫺1 (4.79 ⫻ 10−10 Pa⫺1) 200 ft (61 m)
Injectate-fluid density
Porosity Formation-fluid viscosity
Injectate-fluid viscosity
Regional flow rate Formation Dip Injection interval thickness Fluid compressibility Formation compressibility Longitudinal dispersivity Transverse dispersivity Effective molecular diffusivity Injection rate and duration Datum depth and temperature
Comments ⫺12
2
m)
20 ft (6.1 m) 2.14 ⫻ 10−3 ft2/day (2.3 ⫻ 10−9 m2/s) 1 well at 400 gpm (0.0252 m3/s) for 20 years or 1000 gpm (0.0631 m3/s) for 8 years 5000 ft (1524 m) at 155°F (68.3°C)
Upper range for Gulf Coast sand 117,800 mg/l NaC1 (1.06 sg) 1.05 sg 26% NaC1 solution 22,300 mg/l NaC1 (1.0 sg) 69,200 mg/l NaC1 (1.03 sg) Typical Gulf Coast sand Basis: 1.05 – 1.14 sg NaC1 Solution at 155°F (68.3°C)
Basis: 1.0 – 1.06 sg NaC1 Solution at 155°F (68.3°C)
Range for Gulf Coast sand Upper range Gulf Coast sand Worst case for Gulf Coast sand Typical for Gulf Coast sand Typical for Gulf Coast sand Appropriate for model scale (Adams and Gelhar, 1992) Assumed 0.1 ⫻ 200 ft (61 m) Worst case for Gulf Coast sand Typical upper range
Representative of Gulf Coast
19.5 MODELING RESULTS—VARIATION OF INJECTATE DENSITY Injectate density and formation-fluid density were alternated quarterly and annually for an 8 year operational period to simulate the effects of extreme-density changes. The generic model discussed above was used to model four different scenarios (Model Runs 3–6). In all cases, a dip of 2° (in the direction of increasing distance along the x-axis) and a regional velocity of 0.0 were assigned to maximize updip plume movement. A symmetric relationship between the injectate density and the formation-fluid density was used. The specific gravity
ELSE_TSANG_CH019.qxd
8/22/2005
7:06 AM
19.7 ResultsModel Run 4
Page 239
239
Fig. 19.1. Run 1 C/Co at 10,000 years. 8 years injection at 1000 gpm. k = 2000 md, v = 2.0 ft/yr.
Fig. 19.2. Run 2 C/Co at 10,000 years. 20 years injection at 400 gpm. k = 2000 md, v = 2.0 ft/yr.
(sg) of the injectate was assigned to be either equal to 1.0 or 1.06 alternately for 91.25 days (Runs 4 and 6) or annually (Run 5) to yield an average injectate sg of 1.03. The model scenario and results are summarized for each run below. All other inputs are given in Table 19.1. 19.6 RESULTS—MODEL RUN 3 Run 3 was designed to be a base-case run for comparison with future runs. The injectate density was assigned to be 64.27 lb/ft3 (1029.51 kg/m3) while the formation-fluid density was 66.14 lb/ft3 (1059.46 kg/m3). One well injected at 1,000 gpm (0.0631 m3/s) for 8 years. The hydraulic conductivity was changed from the earlier runs to 10.91 ft/day (3.86 × 10−5 m/s) to account for the change in formation-fluid density. The dip was 2.0°, and the regional velocity was 0.0 ft/yr. The model-calculated relative concentrations are contoured in Figure 19.3 for the end of 10,000 years. In the figure, the extent of the plume was determined by the location of the relative concentration (C/Co) contour equal to 1 × 10–12. After 10,000 years, the center of mass moved updip, as expected, for a distance of 43,060 ft (13,125 m). The extent of the plume in Run 3 was 70,530 ft (21,498 m). 19.7 RESULTS—MODEL RUN 4 Run 4 was designed to be a variable-density run for comparison with Run 3. The injectate density was assigned to be 62.4 lb/ft3 (1000 kg/m3) while the formation-fluid density was 66.14 lb/ft3 (1059.46 kg/m3). The low (62.4 lb/ft3 or 1000 kg/m3) and high (66.14 lb/ft3 or 1059.46 kg/m3) injectate densities were alternately injected for the period of 91.25 days at 1000 gpm (0.0631 m3/s) for 8 years. The low-density fluid was injected first. In this manner, the average injectate density was 64.27 lb/ft3 (1029.51 kg/m3). All other inputs were unchanged from Run 3. After 10,000 years, the center of mass moved updip, for a distance of 41,060 ft (12,515 m) or 95% of the distance in Run 3. The model-calculated relative concentrations are contoured in Figure 19.4 at the end of 10,000 years. The extent of the plume for a 12 order-of-magnitude reduction in the initial concentration was 68,830 ft (20,979 m).
ELSE_TSANG_CH019.qxd
240
8/22/2005
7:06 AM
Page 240
Modeling Density Changes in Hazardous Disposal Well Plumes
Fig. 19.3. Run 3 C/Co at 10,000 years. Base variable-density run. Average density = 64.27 lb/ft3.
Fig. 19.4. Run 4 C/Co at 10,000 years. Injectate density alternates quarterly for 8 yr: 62.40 – 66.14 lb/ft3. Low density injected first. Average density = 64.27 lb/ft3.
This is 1700 ft (518 m) less than the Run 3 plume (98% of the Run 3 extent). The shift in the location of the center of mass in Run 4, relative to Run 3, is small and not considered significant given the fact that the Run 4 plume extent is actually less than that calculated in the average run.
19.8 RESULTS—MODEL RUN 5 Run 5 was designed to be a variable-density run for comparison with Run 3 and Run 4. As in Run 4, the injectate density was assigned to be 62.4 lb/ft3 (1000 kg/m3) and the formation-fluid density was 66.14 lb/ft3 (1059.46 kg/m3). The low and high densities were alternately injected for the period of 365 days at 1000 gpm (0.0631 m3/s) for 8 years. The low-density fluid was injected first. In this manner, the average injectate density was 64.27 lb/ft3 (1029.51 kg/m3). All other inputs were unchanged from Run 4. After 10,000 years, the center of mass moved updip for a distance of 39,860 ft (12,149 m), equivalent to 93% of the distance in Run 3. The model-calculated relative concentrations are contoured in Figure 19.5 at the end of 10,000 years. The extent of the plume for a 12 orderof-magnitude reduction in the initial concentration was 67,280 ft (20,507 m). This is 3250 ft (991 m) less than the Run 3 plume, or 95% of the Run 3 extent. The shift in the location of the center of mass in Run 5, relative to Run 3, is small and again is not considered significant given the fact that the Run 5 plume extent is less than that calculated in the average run.
19.9 RESULTS—MODEL RUN 6 Run 6 was designed to be a variable-density run for comparison with Run 4. As in Run 4 and Run 5, the injectate density was assigned to be 62.4 lb/ft3 (1000 kg/m3) while the formation-fluid density was 66.14 lb/ft3 (1059.46 kg/m3). The low and high densities were alternately injected for the period of 91.25 days at 1000 gpm for 8 years. As in previous runs, this yielded an average injectate density of 64.27 lb/ft3 (1029.51 kg/m3). In Run 6, the
ELSE_TSANG_CH019.qxd
8/22/2005
Conclusion
7:06 AM
Page 241
241
high-density fluid was injected first to determine the impact on the model-calculated plume. All other inputs were unchanged from Run 4. After 10,000 years, the center of mass moved updip a distance of 42,660 ft or 13,003 m (99% of the distance in Run 3). The model-calculated relative concentrations are contoured in Figure 19.6 at the end of 10,000 years. The extent of the plume for a 12 order-of-magnitude reduction in the initial concentration was 69,730 ft (21,254 m). This is 800 ft (244 m) less than the Run 3 plume (99% of the Run 3 extent). The shift in the location of the center of mass in Run 6, relative to Run 3, is small and again is not considered significant given the fact that the Run 6 plume extent is less than that calculated in the average run. In comparison with Run 5, the density of the fluid injected during the last period has the greatest impact on the location of the center of mass. For example, in Run 6, the light-end injectate was injected last, and the center of mass moved farther updip than in Run 4, where the heavy fluid was injected last. Similar results were found by Larkin and Clark (1994a), and Larkin et al. (1996).
CONCLUSION The results of this study supported the findings of the earlier modeling studies. The plume extent and location were controlled by the average injectate density. The model-calculated plumes resulting from alternating the extremes of the injectate density, ranging over 91.25 days or over 1 year, were smaller than the average injectate-density plume. This means that if operators model a quarterly average injectate-density plume, it will provide the most conservative case. The results of this study demonstrate that injectate-density fluctuations, inherent in chemical processing, do not contribute measurably to the long-term calculated plume extent. The density of the fluid injected during the last period impacts the location of the center of mass. The modeling results demonstrate that “No Migration” Petition models, using a quarterly average injectate density, are justifiable at Class I injection well facilities where conditions are similar to the ones modeled here.
Fig. 19.5. Run 5 C/Co at 10,000 years. Injectate density alternates annually for 8 yr: 62.40 – 66.14 lb/ft3. Low density injected first. Average density = 64.27 lb/ft3.
Fig. 19.6. Run 6 C/Co at 10,000 years. Injectate density alternates quarterly for 8 yr: 62.40 – 66.14 lb/ft3. High density injected first. Average density = 64.27 lb/ft3.
ELSE_TSANG_CH019.qxd
8/22/2005
242
7:06 AM
Page 242
Modeling Density Changes in Hazardous Disposal Well Plumes
Based on the results of this study and earlier modeling work, EPA Region VI has altered its policy and will accept the use of a range of quarterly average injectate values as opposed to a range of extreme injectate-density values in future petition modeling. Under the new method, operators will keep a daily record of injectate density. The facility would retain its density range but the end members would be changed, from instantaneous restrictions to running 3-month-weighted average restrictions. The running average is the average over the most recent 3 months of the daily injectate measurements weighted by the daily injected volume. This greatly reduces the necessary model injectate-density range. As long as the operator’s quarterly average density remains within the modeled quarterly range, the petition conditions would be satisfied.
REFERENCES Adams, E.E. and Gelhar, L.W., 1992. Field study of dispersion in a heterogeneous aquifer, 2. Spatial moments analysis. Water Resour. Res., 28(12): 3293–3307. Fahy, E., Miller, C., Golod, I., and Hales, C., 1992. Subsurface dispersion of waste density variations (abs.), San Diego, CA. Ground Water Protection Council, Oklahoma City, OK. Larkin, R.G. and Clark, J.E., 1994a. Effect of Density Averaging Period Duration on Modeled Disposal Well Plumes. Texas Natural Resources Conservation Commission Clean Texas 2000, UIC Workshop Proceedings, Austin, TX, pp. 149–162. Larkin, R.G., Clark, J.E. and Papadeas, P.W., 1994b. Comparison of modeled disposal well plumes using average and variable injectate densities. Ground Water, 32: 35–40. Larkin, R.G., Clark, J.E. and Papadeas, P.W., 1996. Modeling the Effect of Injectate Density Changes on Disposal Well Plumes. In: J.A. Apps, and C.F. Tsang (Eds), Deep Injection Disposal of Hazardous and Industrial Wastes. Academic Press, San Diego, CA, pp. 381–402. Reeves, M., Ward, D.S. and Johns, N.D., 1986. Theory and implementation for SWIFT II, The Sandia Waste-Isolation Flow and Transport Model for Fractured Media, Release 4.84. Sandia National Laboratories, Albuquerque, NM, NUREG/CR-3328, SAND831159. Ward, D.S., Reeves, M. and Duda, L.E., 1984. Verification and field comparison of the Sandia Waste-Isolation Flow and Transport Model (SWIFT). Sandia National Laboratories, Albuquerque, NM, NUREG/CR-3316, SAND83-1154.
ELSE_TSANG_CH020.qxd
8/20/2005
9:43 PM
Page 243
Chapter 20
LEAKOFF MODELING OF FLUID INJECTED IN GAS RESERVOIR AT FRACTURE STIMULATION A. Behr and G. Mtchedlishvili Freiberg University of Mining and Technology, Freiberg, Germany
20.1 INTRODUCTION The objective of this chapter is to develop a practical method by which to identify a leakoff coefficient for fluid loss in hydraulic fracturing within tight gas reservoirs, and to determine its dependence on formation permeability. The schedule of the slurry injection in the fracture treatment, as well as the data of the fluid backflow and gas production during the cleanup period, provides the input information for the algorithm. A special model (in the form of system of algebraic recurrence equations) was derived to translate the monitored fracturing scenario into the relationships that imitate the step-wise fracture development and specify the exposure time of injected fluid to the reservoir throughout the fracture area. In the course of solving this model, the overall leakoff coefficient is iteratively adjusted. This average value is split when the leakoff coefficient is considered as a permeability-dependent power function. The unknown exponent is considered to be a subject of study by history matching. A hypothetical example and a case study were utilized to test the effectiveness of the cleanup production data in identifying this exponent. The production from tight gas formations may be substantially improved through hydraulic fracture treatment, which is accompanied by the injection of a special fracturing fluid. The penetration of the fluid into the matrix results in the formation of an “invaded zone” around the fracture. This invaded zone increases the resistivity of the rock and may, in the limiting case, completely block any gas inflow into the fracture. Therefore, the fluid distribution in the fracture environment is to a great extent responsible for the post-fracture production. A more precise way of determining the fluid loss is to solve the complete fracturing model, including equations for the developing fracture mechanics coupled with flow equations for the fractures and reservoir (Settari, 1980; Nghiem et al., 1984). In such a model, the mass exchange between fractures and the formation can be described on the basis of Darcy’s law—that is, by the general theory of flow in porous media. However, the peculiar conditions on the fracture surface (e.g., deposition of the filter cake, enormous pressure drop) and the specific features of the fracture fluid properties require a special model for flow into a reservoir. In the classical leakoff theory, the flow rate varies inversely with the square root of the time. The factor of proportionality, the leakoff coefficient, accounts for the combined resistance effect of the filter cake, the zone invaded by the fracturing fluid, and the uninvaded part with the original compressible reservoir fluid. Correspondingly, the leakoff coefficient is essentially a characteristic of the formation, the fracturing fluid, and the reservoir fluid. Its canonical form is derived from the conventional
ELSE_TSANG_CH020.qxd
244
8/20/2005
9:43 PM
Page 244
Leakoff Modeling of Fluid Injected in Gas Reservoir at Fracture Stimulation
hydraulic equations, but includes several experimental parameters. On these semi-phenomenological grounds, it was made possible to solve the fracture propagation model independently of the reservoir equations (Howard and Fast, 1970; Williams, 1970). However, such approaches are not aimed at providing an accurate solution for placement of the invaded fluid in the near-fracture region and, therefore, allow the use of the overall leakoff coefficient. Its value can be estimated in a laboratory or in situ by a special microfrac test. On the basis of his own generalized leakoff model, Settari (1987) has suggested a partially decoupled model of fracturing processes whereby processes in the formation are taken into account, but in a manner that does not require a numerical solution of the reservoir model. The introduced spatial and time-dependent leakoff coefficient reflects variation of several local parameters, governing the process of fluid invasion, over the fracture area and during the treatment. Note that the gain in accuracy can be achieved by additional experimental work to obtain the specific empirical information involved in the model. The problem of estimating fluid loss implies a radically different way of handling the modeling—when the background for this issue is not the fracturing design calculation but the modeling of the post-fracture well performance. In this case, it is assumed that the geometry parameters of the stationary fracture are available from a fracture computer package or from geophysical measurements. With this starting point, the challenge is to allocate the given amount of invaded fluid (simply calculated from the material balance) in the fracture vicinity as if it were a consequence of the leakoff process during the fracturing period. In other words, it ceases to be meaningful to solve the complete fracture model and thereby to apply the description of the already-performed fracturing as fitting data. But the advantage of such existing information is that it provides a framework for other methods of initializing the reservoir model for the post-fracture simulation. Typically in such an approach, we model a hypothetical injection of a known amount of water (total for the fracture and invaded zone) into the stationary “dry” fracture (Robinson et al., 1992). As it takes place in reality, the resulting depth of invasion depends on porosity, net thickness, and reservoir water saturation, and decreases away from the well. At the same time, replacing the fracturing with a fictive process can be mistaken at many points. For example, the time taken for the water front to arrive at a specified location of the fracture (in other words, the reduction of the maximum exposure time of water to the formation) depends heavily on fracture conductivity and well pressure. But a major criticism of this approach is that it does not account for the leakoff theory. In the present chapter, we show another way to estimate fluid loss after the fracturing treatment. The idea is to: (1) reconstruct the prehistory using available treatment scenarios and a simplified fracture propagation model; (2) extract from its solution the overall leakoff coefficient and the distribution of the exposure time for the fracturing fluid to formation throughout the stationary fracture; (3) expand the leakoff coefficient about its average value in terms of dependence on rock permeability and net-to-gross thickness ratio, and on this basis; and (4) calculate the inflow into the formation over the fracture plane. The model retains the simplicity of the classical leakoff theory, although it is more comprehensive and potentially more accurate than other methods estimating fluid loss, which do not resort to a fracture simulator. The approach of splitting the leakoff coefficient is of importance for highly layered or heterogeneous formations. It implies a permeability dependence in the form of a power function. It is thought that the leakoff coefficient grows in proportion to the square root of the permeability, if the fluid loss is mainly controlled by reservoir properties. However, this relationship was derived assuming the validity of Darcy’s law for fluid loss velocity. In actuality,
ELSE_TSANG_CH020.qxd
8/20/2005
9:43 PM
Page 245
20.2 Reconstruction of Fracture Propagation
245
the flow is more complicated due to, among other reasons, the very high pressure at the fracture face and multiphase effects. In addition, the proportionality breaks down as the filter cake becomes nonignorable. There is a reason to think that the exponent in the generalized permeability dependence of leakoff coefficient is different from 0.5. We suggest estimating its value by history matching of the production data over the cleanup period. The sensitivities needed for inversion are proved by hypothetical and real examples.
20.2 RECONSTRUCTION OF FRACTURE PROPAGATION The proposed fracture propagation model provides the first step of the fluid loss estimation procedure. Here, the purpose is to use the available data about the fracture treatment and fracture plane geometry to gain a perception of how the fracture developed over space and time, and to translate this information into the exposure-time distribution. The characteristic periods of the fracturing process are routinely recorded in the operation protocol. Slurry volumes VS injected into the fracture by the end of these intervals specify the treatment schedule. Taking into account the fluid loss volume Vl, the material balance principle can be applied for each ith period to yield the fracture volume VF: VFi ⫽ Vsi ⫺ Vli
(20.1)
With the leakoff equation Vl ⫽ Cl At 1/2,
(20.2)
we can close the model by the reasonable assumption that the fracture grows proportionally in all directions, i.e., L A 1/2 VFi 1/3 ᎏi ⫽ ᎏi ᎏ ⫽ ⫽ αi (20.3) L A VF
冢 冣
冢 冣
with a proportionality factor αi for linear dimensions; Li, L, Ai, A, VFi, VF are the linear size, the area of fracture (both faces), and the fracture volume at the end of the ith period and after closure, respectively. Based on Equations (20.1)–(20.3), a system of algebraic recurrence equations was derived: VFα 3i ⫹ Cl A
∆ti
ᎏ α ⫽r, 冪莦 2 2 i
i
i ⫽ 1,..., N,
(20.4)
with the term on the right-hand side as follows:
冤
ri ⫽ Vsi ⫺ Cl
i⫺1
j⫽1
⫺1 ⌬tj ⌬ t ⫹ ⫺ ᎏ 冱∆Aj 冱 k 2 j⫽1 k⫽j⫹1
冪莦莦莦莦莦
冱∆Aj
i
i
∆ti
ᎏ 冥. 冪莦 2
(20.4a)
Here, N is the number of characteristic periods in the fracturing schedule, ∆ti is length of the ith period, ∆Aj is an area increment in jth period, which can be expressed from Equation (20.3) as ∆Αj ⫽ Αj⫺Αj⫺1 ⫽ A(α 2j⫺α 2j⫺1).
(20.4b)
Equations (20.4), (20.4a), and (20.4b) are the recurrence cubic equations with respect to variable αi, which can be solved successively to get the fracture expansion factor at different time points. The variation of the factor α over time reflects the fracture propagation. At the same time, from the similarity equation, Equation (20.3), α-value can be assigned to each point of the
ELSE_TSANG_CH020.qxd
8/20/2005
246
9:43 PM
Page 246
Leakoff Modeling of Fluid Injected in Gas Reservoir at Fracture Stimulation
stationary fracture surface. So, the corresponding initial point of exposure period is defined simply by interpolation between nearest neighbors in the discrete function (αi, ti).
20.3 ESTIMATION OF THE LEAKOFF COEFFICIENT Obviously, for the last solution of Equation (20.4), i ⫽ N, the condition
αN ⫽ 1
(20.5)
must be satisfied. Hence, Equation (20.5) serves as a basis for improving the predetermined overall leakoff coefficient. The correction can be performed by an iteration procedure. In the case of pronounced heterogeneity in the surrounding fracture region, using only one average leakoff coefficient over the whole fracture surface may lead to substantial errors. As a corrective measure, a leakoff coefficient dependence on formation properties can be involved in the approach. Let us propose this as a relationship with permeability k and net-to-gross thickness ratio η in the form of Cl ⫽ C ∗l η k␥,
(20.6)
where factor C ∗l , which can be referred to as “specific leakoff coefficient,” has a different dimensionality from the original leakoff coefficient. Its overall value was derived from the condition that the total fluid loss remains unchanged:
冕
t1/2 dA
A C*l ⫽ C l pr ᎏᎏ
冕
(20.7)
t1/2 ηkγ dA
A
Here, Cl pr is the primary average leakoff coefficient, obtained by solving the fracture propagation model, Equations (20.4) and (20.5). Summarizing, Equations (20.6) and (20.7) signify a spatial distribution of the leakoff coefficient, while its modification is invariant. The ideas above were incorporated into a tool designed specifically to prepare the reservoir model for the post-fracture simulation (Behr et al., 2003). To initialize the gridded model with respect to fluid saturation in the fracture vicinity, the exposure times were computed for the centers of all the gridblocks containing the stationary fracture. After that, the classical leakoff coefficients recalculated by Equations (20.6) and (20.7), with the local permeabilities and net-to-gross thickness ratios, were distributed over all elements of the fracture faces. The saturation profiles around the fractures were worked out using Buckley–Leverett equations for two-phase nonmiscible displacement. The leakoff equation, Equation (20.2), provides the boundary condition on the fracture face, which allows constructing a self-similar solution in the form of a universal saturation profile, dependent on a dimensionless combined time–distance variable. 20.4 IDENTIFICATION OF EXPONENT γ It should be expected that the exponent in the permeability-related dependence of the leakoff coefficient ␥, Equation (20.6), is different from the value of 0.5 derived from the conventional
ELSE_TSANG_CH020.qxd
8/20/2005
9:43 PM
Page 247
20.4 Identification of Exponent ␥
247
equations for flow in porous media. This difference results from the deposition of the filter cake and the obvious deviation from Darcy’s law for the process of fluid penetration into the matrix, in view of the specific conditions at the fracture face during the fracturing. It is not possible to produce such conditions in the laboratory. Therefore, in actual practice, only field data can provide the information for extracting the leakoff schematic. 20.4.1 Hypothetical Study We demonstrate the possibility of γ-parameter identification by a hypothetical but representative example of post-fracture performance in a tight gas reservoir. The model in the study presents a vertical, hydraulically fractured well placed in the formation of two layers having different properties: permeability of 0.01 and 0.001 mD, and porosity of 0.1 and 0.05, respectively. The elliptical fracture with semiaxes 100 m (horizontal) and 50 m (vertical) is evenly divided between the layers. Dimensionless fracture conductivity can be estimated as 10. The pressure drawdown increases gradually (with an exponential asymptote) up to 250 bar over a typical time scale of about 5 days. Such a pressure progression is typical for the relatively short-term cleanup period (recovery of the water injected during fracturing) when well fluid dominance is changing from water to gas. The capillary pressure and relative permeability characteristics were determined by Brooks–Corey relationships in conformity with data presented by Ward and Morrow (1987) for tight sandstones. The automatic model preparation for commercial reservoir simulator, including the calculation of the initial water distribution in the fracture environment with consideration of leakoff schematic outlined above, was performed by our own support-tool (Behr et al., 2003). Figure 20.1 exhibits the development of the gas production rate and cumulative recovered water computed for the cleanup phase at various values of γ in the range from 0 to 1. (The monitored production data presented in such a manner are usually used for history matching, in view of the importance of the achieved level of gas production rate and fraction of the injected fracturing fluid produced back to the surface.) The curves for the water backflow show a poor response to the change in γ, in contrast to those for the gas production. Parameter γ=0 γ = 0.25 γ = 0.5 γ = 0.75 γ=1
50000
600 Gas 500
40000
400
Water
30000
300
20000
200
10000
100
0
Cumulative water recovery, m3
Gas production rate, m3/day
60000
0 0
2
4
6 Time, day
8
10
12
Fig. 20.1. Hypothetical study: simulated gas production rate and cumulative water recovery.
ELSE_TSANG_CH020.qxd
8/20/2005
248
9:43 PM
Page 248
Leakoff Modeling of Fluid Injected in Gas Reservoir at Fracture Stimulation
γ considerably affects the gas breakthrough as well as the characteristics of gas-rate development at early times. This statement is especially true for γ ⬎ 0.5. Figures 20.2–20.4 introduce γ-functions for characteristics of the gas-rate curves, which may be used as matching criteria in the identification of γ by inversing. All the plotted characteristics are presented in the normalized form, by dividing them by their values obtained for γ ⫽ 0.5. Referring to Figure 20.2, the maximum gas rate decreases monotonically with γ. However, only half the γ-interval presents the sensitivity that may be accepted as sufficient for identification. Figure 20.3, with the slope calculated between points of 20 and 80% levels of the maximum gas rate, shows the same basic behavior we observed in Figure 20.2. The curve for the gas breakthrough time, shown in Figure 20.4, has a distinct extremum. To explain this phenomenon, in Figure 20.5 we compare the developments of water saturation profiles in both layers calculated for γ ⫽ 0, 0.5, and 1. In the first two cases, the breakthrough occurs in the more permeable layer. Hence, the deeper the fracturing fluid penetrates in this layer (i.e., as
Normalized maximum gas rate
1.3
1.2
1.1
1
0.9
0.8 0
0.2
0.4
0.6
0.8
1
γ
Fig. 20.2. The γ dependence of normalized maximum gas rate.
Normalized gas rate slope
2.5
2
1.5
1
0.5
0 0
0.2
0.4
γ
0.6
0.8
Fig. 20.3. The γ dependence of normalized gas rate slope.
1
ELSE_TSANG_CH020.qxd
8/20/2005
9:43 PM
Page 249
Normalized breakthrough time
20.4 Identification of Exponent ␥
249
1 0.8 0.6 0.4 0.2 0 0
0.2
0.4
γ
0.6
0.8
1
Fig. 20.4. The γ dependence of normalized gas breakthrough time.
the γ increases), the later the breakthrough occurs. After γ exceeds a certain critical value (here it is about 0.5), the depth of the invaded zone in the less permeable layer becomes so small that the gas forces its way up to the fracture first in this layer. That reverses the trend of the γdependence of gas breakthrough time, making it unreliable for history matching. Here is the proper place to mention that the deviation of γ from 0.5 in the direction of its decrease is much more realistic than in the opposite direction. This is because the abovenamed reasons for such a deviation reduce the role of formation permeability in controlling fluid loss processes, rather than the reverse. Thus, γ can be searched for in the range that has a higher sensitivity. 20.4.2
Case Study
Here, we extend the above concept to an actual field situation. A hydraulically fractured well within a tight gas formation in northern Germany was considered. The fracture intersects eight layers with permeabilities varying from 0.0016 to 0.014 mD and porosities from 0.01 to 0.10. The fracture is characterized by a half-length of 80 m, maximum size in vertical direction of 110 m (net to gross thickness ratio η ⫽ 0.4), and dimensionless conductivity of about 100. The amount of water injected in the fracture treatment is 300 m3. During a cleanup period of 4 days, the pressure drawdown increases to approximately 500 bar. A distinctive feature of such a real case is the uncertainty in the two-phase properties of the formation near the fracture. We introduce the concept of effective relations for the capillary pressure and relative permeabilities of the heterogeneous system and consider them, as well as the exponent γ, as a subject of investigation by history matching. As before, in the hypothetical example, the model for simulating post-fracture performance (with a reservoir simulator) was prepared with an in-house tool. In doing so, we assumed that the fracture itself, as well as the fracture zones forming the basis for a pattern for a piecewise constant proppant (correspondingly, permeability and porosity) distribution, is elliptical in form. Figure 20.6 demonstrates the best matches between the measured and computed gas and water production data, achieved at fixed γ ⫽ 0 and 0.5. The corresponding inverse solutions for relative permeabilities are plotted in Figure 20.7. Because of a large pressure drawdown, the fitting was not sensitive to the capillary pressure. One result observed here is that the matched functions of water relative permeability are not dependent on γ. Also, the optimal curves of cumulative water recovery are practically the same. Thus, this type of data cannot be used for γ-identification.
ELSE_TSANG_CH020.qxd
8/20/2005
250
9:43 PM
Page 250
Leakoff Modeling of Fluid Injected in Gas Reservoir at Fracture Stimulation 0.9 γ=0
Water saturation
0.6 0.3
k = 0.01 mD, φ = 0.1
Distance from well, m
0 0 -0.3
0.5
1
1.5
2
2.5
3
k = 0.001 mD, φ = 0.05
-0.6 initial
after 5 days
-0.9 0.9 γ = 0.5
Water saturation
0.6 k = 0.01 mD, φ = 0.1
0.3
Distance from well, m
0 0
0.5
1
1.5
2
2.5
3
-0.3 k = 0.001 mD, φ = 0.05
-0.6 -0.9 0.9
γ=1 Water saturation
0.6 k = 0.01 mD, φ = 0.1
0.3
Distance from well, m
0 0
0.5
1
1.5
2
2.5
3
-0.3 -0.6
k = 0.001 mD, φ = 0.05
-0.9
Fig. 20.5. Development of water backflow from the invaded zone.
In contrast, the matched curves for gas relative permeability differ noticeably. Though the corresponding simulated-gas-rate production graphs agree very closely for much of the cleanup period, they do exhibit differences in the gas breakthrough point and slope in the adjacent region. Unfortunately, in this example, the indicated distinctions are insignificant, and there is a lack of data monitored during the transition period. This excludes the possibility of extracting (to some extent) a reliable exponent γ for leakoff calculation. At the same time, the availability of the relationship for gas relative permeability opens the way for estimating the parameter γ on the basis of production data during cleanup. The sensitivity of these data to γ is confirmed by Figure 20.8, which shows a considerable deviation from the
ELSE_TSANG_CH020.qxd
8/20/2005
9:43 PM
Page 251
251
75000
250
60000
200
45000
150
30000
100
15000
50
0
Cumulative water recovery, m3
Gas production rate, m3/day
20.5 Conclusions
0 0
1
2 3 Cleanup time, day
Observed gas rate Simulated gas rate, y = 0 Simulated water recovery, y = 0,5
4
Simulated gas rate, y = 0.5 Observed water recovery Simulated water recovery, y= 0
Fig. 20.6. Development of gas flow rate and water recovery at the bottomhole during the cleanup period.
0.35
=0
Relative permeability
0.3
= 0.5
0.25 0.2 Gas 0.15 0.1 Water 0.05 0 0.3
0.4
0.5 0.6 Water saturation
0.7
0.8
Fig. 20.7. Relative permeability curves matched in the case study at fixed γ ⫽ 0 and 0.5.
curves matched for γ ⫽ 0.5 when changing γ to 0 and maintaining the relative permeabilities. The behavior of the dash curve, which is a solution for γ ⫽ 1, is consistent with conclusions inferred from the hypothetical example.
20.5 CONCLUSIONS A special algorithm was developed to determine fluid loss into the matrix during fracturing of a tight gas reservoir. This algorithm uses the classical leakoff theory and is based
ELSE_TSANG_CH020.qxd
8/20/2005
Page 252
75000
250
60000
200
45000
150
30000
100
15000
50
0
Cumulative water recovery, m3
Leakoff Modeling of Fluid Injected in Gas Reservoir at Fracture Stimulation
Gas production rate, m3/day
252
9:43 PM
0 0
1
2 Cleanup time, day
Observed gas rate Simulated gas rate, γ = 0.5 Observed water recovery Simulated water recovery, γ = 0.5
3
4
Simulated gas rate, γ = 1 Simulated gas rate, γ = 0 Simulated water recovery, γ = 1 Simulated water recovery, γ = 0
Fig. 20.8. Comparison of production data simulated by the fixed relative permeabilities for different γ-exponents.
on the purpose-built fracture propagation model. The geometry of the stationary fracture and the fracture treatment schedule provide the input for calculation. The solution presents the spatial distribution for the exposure time of the fracture fluid to the formation, as well as the overall leakoff coefficient, i.e., the data needed to compute the saturation profiles around the fracture. This leakoff coefficient is corrected by relation to the formation permeability and net-togross thickness ratio. The exponent γ in the power permeability dependence is proposed to be estimated by history matching production data monitored during the cleanup phase. By hypothetical and real examples, the possibility of this estimation was confirmed, but for the range of the exponent less than about 0.5. However, this range is approximately equivalent to the interval of the most plausible values for γ. For the purpose of γ identification, the data for gas production rate are much more suitable than those for water recovery. The maximum of the curve for the gas production rate and its slope after the gas breakthrough can be used as matching criteria rather than the breakthrough time itself. Although the latter is sensitive to γ, it has a nonunique correlation to the invaded fluid distribution within the layers of a heterogeneous formation.
NOMENCLATURE Roman Letters A Cl C *l
fracture face area (m2) leakoff coefficient (m/s1/2) specific leakoff coefficient (m/s1/2 mDγ )
ELSE_TSANG_CH020.qxd
8/20/2005
9:43 PM
Page 253
References k L N r t V
253
permeability (mD) linear size (m) number of pumping periods (dimensionless) right-hand-side term (m3) time (s) volume (m3)
Greek Letters α ∆ γ η
fracture expansion factor (dimensionless) increment power index (dimensionless) net-to-gross-thickness ratio (dimensionless)
Indices F l pr s *
fracture leakoff primary slurry specific
ACKNOWLEDGMENTS The authors would like to thank the German Society for Petroleum and Coal Science and Technology (DGMK) for organizing and funding this work. We particularly appreciate the help from the participating companies: Preussag Energie GmbH (Lingen), Wintershall AG (Kassel), RWE DEA AG (Hamburg), Erdöl-Erdgas GmbH (Berlin), and EMPG (Hannover). REFERENCES Behr, A., Mtchedlishvili, G., Friedel, T. and Haefner, F., 2003. Consideration of damaged zone in tight gas reservoir model with hydraulically fractured well. Proceedings of the SPE European Formation Damage Conference, SPE 82298, The Hague, The Netherlands. Howard, G.C. and Fast, C.R., 1970. Hydraulic Fracturing, Vol. 2 SPE Monograph Series. Dallas, TX. Nghiem, L.X., Forsyth, P.A. and Behie, A., 1984. A fully implicit hydraulic fracture model. Journal of Petroleum Technology, 36(7): 1191–1198. Robinson, B.M., Holditch, S.A., Whitehead, W.S. and Peterson, R.E., 1992. Hydraulic fracturing research in East Texas: Third GRI staged field experiment. Journal of Petroleum Technology, 44(1): 78–87. Settari, A., 1980. Simulation of hydraulic fracturing processes. Proceedings of the SPE Fifth Symposium on Reservoir Simulation, SPE 7693, Denver, TX. Settari, A., 1987. Partially decoupled modeling of hydraulic fracturing process. Proceedings of the SPE Reservoir Simulation Symposium, SPE 16031, San Antonio, TX. Ward, J.S., and Morrow, N.R., 1987. Capillary pressures and gas relative permeabilities of low permeability sandstone. SPE Formation Evaluation, September, 345–356. Williams, B.B., 1970. Fluid loss from hydraulically induced fractures. Journal of Petroleum Technology, 22(7): 882–888.
This page intentionally left blank
254
ELSE_TSANG_CH021.qxd
8/22/2005
7:02 AM
Page 255
III. GEOCHEMISTRY
255
This page intentionally left blank
256
ELSE_TSANG_CH021.qxd
8/22/2005
7:02 AM
Page 257
Chapter 21
PREDICTING TRACE METAL FATE IN AQUEOUS SYSTEMS USING A COUPLED EQUILIBRIUM-SURFACE-COMPLEXATION DYNAMIC-SIMULATION MODEL J.A. Dyera, N.C. Scrivnera, B.C. Fritzlera, D.L. Sparksb, S.J. Sandersc, and P. Trivedid a
du Pont Engineering Research and Technology, Wilmington, DE, USA Department of Plant and Soil Sciences, University of Delaware, Newark, DE, USA c Process Systems Enterprise, Denville, NJ, USA d Department of Civil and Environmental Engineering, University of Alaska, Fairbanks, AK, USA b
21.1 INTRODUCTION Trace metal discharges from industrial manufacturing processes are being increasingly scrutinized and regulated, with new regulations pushing metal effluent limits from parts-permillion (ppm) to parts-per-billion and lower levels. The increased regulatory focus on metals means that operators of Class I underground injection wells must fully understand the geochemical fate of trace metals in the subsurface, aqueous environment. Trace metals are often present in wastewater discharges, groundwater, and surface water at concentrations well below their respective solubility limits. As a result, a large fraction of the discharged metals is often complexed with dissolved organic matter and/or sorbed onto biological and inorganic suspended solids, including hydrous iron, aluminum, and manganese oxides. Moreover, in the subsurface environment, clay, carbonate, and oxide minerals, such as goethite, ferrihydrite, dolomite, and calcite, play a major role in controlling the fate and mobility of trace metals. Equilibrium surface-complexation models (SCMs) have been developed and extensively studied over several decades to help scientists and engineers better understand the geochemical fate of trace metals in aqueous systems containing reactive solid surfaces. Still, the ability to predict trace-metal partitioning in complex aqueous systems involving solution- and surface-complexation, precipitation, and dissolution reactions is evolving. Few studies have combined molecular- and macroscopic-scale investigations with surface complexation modeling to predict trace-metal speciation and partitioning in aqueous systems over a broad range of environmental conditions. Even fewer have demonstrated the practical application of SCMs, calibrated with fundamental macroscopic and spectroscopic metal sorption data, in helping to address industrial trace-metal emission problems. Over the past 5 years, DuPont Engineering Research and Technology, in partnership with the University of Delaware and OLI Systems, Inc., have been jointly developing the methodology, tools, and thermodynamic data necessary to better understand and predict the fate and mobility of trace metals in aqueous systems. This has led to new capabilities in simulating the equilibrium aqueous speciation and partitioning of trace metals in both steady-state and dynamic systems using the OLI Software (OLI Systems, Inc., Morris Plains, NJ).
ELSE_TSANG_CH021.qxd
258
8/22/2005
7:02 AM
Page 258
Predicting Trace Metal Fate in Aqueous Systems Using a Coupled Equilibrium
In this chapter, an equilibrium SCM within the OLI Software Environmental Simulation Program was calibrated against fundamental macroscopic and spectroscopic data for predicting single-solute lead (Pb) and zinc (Zn) sorption onto ferrihydrite over a broad range of concentration, pH, and ionic strength conditions. The resulting equilibrium SCM was then coupled with OLI Systems’ dynamic simulation tool (DynaChem) to predict the geochemical fate of acidic lead- and zinc-containing aqueous waste streams following injection into a subsurface sandstone formation containing calcite, dolomite, and iron oxyhydroxides. The impact of metal type and pH on trace-metal breakthrough time was investigated.
21.2 OLI SOFTWARE The OLI Software is a commercial simulation package that models aqueous electrolyte equilibria, including chemical speciation and redox reactions, equilibria between aqueous, vapor, organic liquid, and multiple solid phases, biochemical and inorganic reaction kinetics, and ion exchange, adsorption, and coprecipitation phenomena. The system is built around the OLI Engine, which is the foundation for the Environmental Simulation Program (ESP), Corrosion Simulation Program (CSP), ElectroChem, and DynaChem. At the heart of the OLI Engine are the databank, the thermodynamic framework, and the equation solvers (Rafal, et al., 1994b). The OLI databank contains the thermodynamic and physical properties of over 8300 inorganic and organic species. The thermodynamic variables (Gibbs free energy, enthalpy, entropy, and heat capacity) in the databank are from six main sources—Chase et al. (1985), Daubert and Danner (1989), Glushko et al. (1981), Gurvich et al. (1989), and Oelkers et al. (1995), Wagman et al. (1982). Each of these sources is an extensive, carefully evaluated, and well-referenced compilation of thermodynamic data. The state-of-the-art thermodynamic framework uses generalized correlations based upon regressed experimental data in the databank to predict the required thermodynamic and physical properties for the chemical system of interest (i.e., infinite-dilution, standard-state values for Gibbs free energy, enthalpy, entropy, heat capacity, and volume, liquid-phase activity coefficients for systems at finite concentrations, and so on). The theoretical basis for the thermodynamic framework is explained elsewhere (Rafal et al., 1994a; Zemaitis et al., 1986). Shock and Helgeson (1988) describe the semi-theoretical basis for temperature and pressure extrapolation of standard-state properties, along with procedures for estimating missing parameters. The aqueous activity coefficient model is an extension of the work of Bromley and Meissner (i.e., the Bromley–Zemaitis framework) (Anderko et al., 1997; Zemaitis et al., 1986). Details on the equation solvers used to solve the nonlinear equilibrium, electroneutrality, and mass-balance equations are reported by Rafal et al. (1994b). Permissible operating ranges for the software are given by Scrivner et al. (1996).
21.3 DYNACHEM MODEL DynaChem is a dynamic simulation code that is based upon the discrete, modular computation of process units, such as a tank, a section of pipe, or a schedule of inflows to a process (i.e., mass and energy flows as a function of time). In this case, “process” refers to any chemical system, which can be a typical chemical manufacturing process or a geochemical system being altered by the flow of contaminated wastewater. In isolating a
ELSE_TSANG_CH021.qxd
8/22/2005
7:02 AM
Page 259
21.3 DynaChem Model
259
process unit as a discrete computation, homogeneity and chemical and thermodynamic equilibrium can be assumed; however, limits to homogeneity and equilibrium can also be imposed through imperfect mixing and the addition of reaction kinetics. Once a process has been defined as a series of discrete units, a two-tier stepping technique is used to simulate the process dynamics. The inner tier steps through the units in a predefined order for the specified time increment, which addresses the movement of mass and energy through the process during a small, but finite, time increment. The outer tier utilizes the final state of the process as defined by the inner tier as the beginning state for the next time increment. The outer tier, therefore, steps through time, with each step resulting in one complete pass through the inner tier. During a defined time increment, “packets” of mass and energy are introduced into a unit. The mass and energy are combined with the mass and energy already present in the unit, and the equilibrium condition is calculated. Based upon defined unit parameters, packets of outgoing mass and energy are calculated and placed at collection points called nodes. The transmission of mass and energy from one unit to another during a discrete time increment, then, occurs by accepting packets of mass and energy from upstream nodes and depositing packets of mass and energy at downstream nodes. In this study, the DynaChem process units can be conceptualized as concentric, annular rings of porous sandstone with an increasing radius (r) and a fixed thickness (∆r) and height (h). These variable-volume annular rings surround an inner cylindrical borehole of fixed radius (i.e., the injection well) where contaminated wastewater is injected and flows radially outward from the well, reacting with minerals in the annular rings of sandstone. This is depicted schematically in Figure 21.1. Ideally, the thickness of the annular rings (and, hence, the volume of the DynaChem process units) would be infinitesimally small to approximate continuous flow through the subsurface sandstone. As ∆r approaches zero, the predicted concentration profiles will become more realistic; however, computation time will increase significantly. As a result, a compromise was made between accuracy (i.e., process
Native Brine and Sandstone ∆r Process Unit 10
ll
r on cti
We
e
Inj
h Injected Wastewater
Fig. 21.1. Conceptual model of underground injection zone showing 10 process units or annular rings of reactive sandstone surrounding the injection well.
ELSE_TSANG_CH021.qxd
260
8/22/2005
7:02 AM
Page 260
Predicting Trace Metal Fate in Aqueous Systems Using a Coupled Equilibrium
unit size) and computation time. The use of discrete, finite-sized process units leads to artifacts (i.e., discontinuities or spikes) in the model-predicted concentration profiles as shown in Figures 21.4–21.6 below.
21.4 SURFACE COMPLEXATION MODEL CALIBRATION Over the years, a variety of SCMs have been developed and utilized for predicting tracemetal sorption onto mineral oxides. These include the nonelectrostatic model (NEM), the constant capacitance model (CCM), the diffuse-layer model (DLM), the generalized twolayer model (GTLM), and the modified triple-layer model (TLM). SCMs attempt to explicitly account for the reaction processes occurring at the solid–water interface; they assume that metal ions form complexes with surface functional groups in a manner similar to metal–ligand complexation reactions in solution (Hayes and Katz, 1996). They are thermodynamic models that differ in their physical description of the solid–water interfacial region (i.e., the location of sorbed species with respect to the surface as well as the description of surface charge–potential relationships across the interfacial region) and in their assumptions regarding number of site types and the structure and composition of the sorbed species. Hayes and Katz (1996) provide more background on SCMs. Four different SCMs are available for use in DynaChem simulationsthe NEM, CCM, GTLM, and modified TLM. In this study, the modified TLM was employed to predict Pb and Zn equilibrium partitioning between the aqueous phase and the amorphous iron oxyhydroxide (ferrihydrite) phase as described below. The thermodynamic framework for the modified TLM is described by Sahai and Sverjensky (1998). Calibration of the modified TLM was based on macroscopic and spectroscopic data for single-solute Pb(II) and Zn(II) sorption onto 2-line ferrihydrite (N2 atmosphere, room temperature, and 4 hours equilibration time) that are reported and discussed by Dyer et al. (2003, 2004) and Trivedi et al. (2003, 2004). Trivedi et al. (2003) describe the ferrihydrite preparation method and present potentiometric titration data for 2-line ferrihydrite in 0.001, 0.01, and 0.1 M NaNO3 solutions (N2 atmosphere and room temperature). Analysis of the potentiometric titration data is described by Dyer et al. (2003). Raw tabulated potentiometric titration data as well as Pb(II) and Zn(II) sorption data are reported by Dyer (2002). The modified TLM was chosen over other SCMs available in the OLI code, because it provided excellent fits of the potentiometric titration, constant-pH isotherm, and pH edge data for Pb(II) and Zn(II) sorption onto ferrihydrite. Figures 21.2 and 21.3 show optimized TLM fits of the single-solute Pb(II) and Zn(II) constant-pH isotherm data. See Dyer et al. (2003, 2004) for more details.
21.5 DEFINITION OF CASE STUDIES DynaChem simulations were based on the conditions summarized in Table 21.1. TLM parameters used in the simulations are reported by Dyer et al. (2003, 2004). In addition, a homogeneous sandstone formation, constant wastewater composition and flow rate, constant porosity, and equilibrium conditions were assumed. Equilibrium conditions were justified for the following reasons. First, one is ultimately interested in evaluating chemical fate in underground injection wells over a long timeframe (e.g., 10,000 years). For this reason, conditions in the leading portion of the reaction front (i.e., far from the injection well where r is large)
ELSE_TSANG_CH021.qxd
8/22/2005
7:02 AM
Page 261
21.5 Definition of Case Studies
261
1.E+00
Moles Pb/Mole Fe
1.E-01
1.E-02
1.E-03
pH 4.5 Data pH 5.5 Data pH 6.5 Data pH 4.5 Model pH 5.5 Model pH 6.5 Model
1.E-04
1.E-05
1.E-06 1E-05 0.0001 0.001
0.01
0.1
1
10
100
1000 10000
Pb in Solution (ppm) Fig. 21.2. Optimized triple-layer model fits of constant-pH equilibrium isotherm data for single-solute Pb(II) sorption onto 2-line ferrihydrite (0.1 and 1.0 g ferrihydrite/l in 0.01 M NaNO3; Pb added as Pb(NO3)2; 4 hours equilibration time; room temperature; N2 glovebox).
1.E+00
Moles Zn/Mole Fe
1.E-01
1.E-02 pH 4.5 Data pH 5.5 Data pH 6.5 Data pH 7.5 Data pH 4.5 Model pH 5.5 Model pH 6.5 Model pH 7.5 Model
1.E-03
1.E-04
1.E-05 0.0001
0.001
0.01
0.1
1
10
100
1000
10000
Zn in Solution (ppm) Fig. 21.3. Optimized triple-layer model fits of constant-pH equilibrium isotherm data for single-solute Zn(II) sorption onto 2-line ferrihydrite (0.1 and 1.0 g ferrihydrite/l in 0.01 M NaNO3; Zn added as Zn(NO3)2; 4 hours equilibration time; room temperature; N2 glovebox).
are particularly important. Far from the injection point, the radial flow velocity will be very low and, hence, the residence time for reaction will be long. For example, for the conditions assumed in these case studies, the average interstitial pore velocity at 300 m (r) from the
ELSE_TSANG_CH021.qxd
262
8/22/2005
7:02 AM
Page 262
Predicting Trace Metal Fate in Aqueous Systems Using a Coupled Equilibrium Table 21.1. Conditions for DynaChem simulations Case study
Parameter
1
2
3
4
Wastewater pH Pb feed conc. (ppm) Zn feed conc. (ppm) HNO3 Feed Conc. (molal) Simulation time (hour)
2 100 — 0.011 500
2 — 32 0.011 500
1 100 — 0.110 100
1 — 32 0.110 100
All Case Studies: • • • • • • • • • • • •
10 process units (annular rings) surrounding injection well. Each process unit is completely mixed. 150 l/hour contaminated water flow at 25°C with radial flow outward from center. Contaminated water contains HNO3 and Pb and Zn as nitrate salts. Wastewater feed pipe in center [0.075 m radius (r) × 1 m high (h)]. Concentric annular rings [0.0375 m thick (∆r) × 1 m high (h)]. 0.022, 0.031, 0.040, 0.049, 0.057, 0.066, 0.075, 0.084, 0.093, and 0.102 m3 unit volumes for annular rings 1 through 10, respectively. 75 vol% sandstone, 25 vol% aqueous brine at 45°C and pH ∼6.5 initially. Sandstone density of 2.75 kg l−1. 93 wt% quartz, 3 wt% dolomite, 2 wt% iron, 1 wt% calcite. Iron is 10% ferrihydrite/90% goethite; effective sorption capacity for Pb and Zn is based on ~18% Fe as “sorbing” ferrihydrite. 0.1 hour time increment; 150 atm total pressure.
injection well is only 3 m/year. Second, the neutralization, dissolution, precipitation, and surface complexation reactions involving calcite, dolomite, and iron oxyhydroxide are fast relative to those for clay minerals and silicon oxides. The impact of a low radial velocity and relatively fast reaction rates is that the pore water will approach equilibrium with the reactive constituents in the sandstone. Third, the equilibrium assumption allows one to evaluate the best-case scenario. In other words, what is the best that one can expect? The authors recognize that there are limits to the equilibrium assumption. For example, close to the injection well, interstitial flow velocities will be higher (e.g., 1.3 m/hour close to the well screen) and, hence, dissolution kinetics may in fact be limiting. In addition, the potential exists for flow channeling due to heterogeneities (e.g., fractures) in the sandstone formation and/or pore clogging caused by solids in the injected wastewater and/or secondary precipitation reactions. All of these effects will tend to decrease trace-metal breakthrough times and retardation factors (i.e., earlier arrival of the trace-metal front). The effect of permeability changes on sorption due to dissolution of carbonate and iron oxyhydroxide minerals will be small (5–10%) when compared to the uncertainties introduced to the tracemetal breakthrough curves from the other simplifying assumptions discussed above.
21.6 RESULTS AND DISCUSSION Results for case studies 1 and 2 are summarized in Figures 21.4–21.6. Figure 21.4 shows the concentration/pH time profiles for DynaChem process unit 10 (i.e., the 10th annular ring or equilibrium stage from the injection well). Figures 21.5 and 21.6 present the trace-metal breakthrough curves for process units 1 through 10 for Pb and Zn, respectively. While the simulation
ELSE_TSANG_CH021.qxd
8/22/2005
7:02 AM
Page 263
21.6 Results and Discussion
263
1.E+04 Artifact of using discrete, finite-sized DynaChem process units
1.E+03 1.E+02
Pb(aq)
Zn(aq)
Dolomite
1.E+01
Variables
Pb (aq) (ppm)
pH
pH
1.E+00
pH
Zn (aq) (ppm)
PbCO3 (s)
1.E-01
CaCO3 (s) (mol)
ZnCO3 (s)
1.E-02 1.E-03
Dolomite (s) (mol)
CaCO3 (s)
1.E-04 1.E-05
Pb(aq)
Zn(aq)
PbCO3 (s) (mol)
Dolomite
ZnCO3 (s) (mol)
1.E-06 1.E-07 1.E-08 0
50
100
150
200
250
300
350
400
450
500
Time (hr)
Fig. 21.4. Concentration/pH profiles for process unit 10 for case studies 1 and 2.
10000 1 2
3
4
5
6
7
8
9
10
Total Soluble Pb (ppm)
1000 100 10
Unit 1 Unit 2 Unit 3 Unit 4 Unit 5 Unit 6 Unit 7 Unit 8 Unit 9
1 0.1 0.01
Unit 10
0.001 0
50
100
150
200
250 300 Time (hr)
350
400
450
500
Fig. 21.5. Pb(II) breakthrough curves for case study 1.
timeframe is relatively short (500 hours), the behavior, shape, and magnitude of the parameters and breakthrough curves are essentially the same at much longer times (i.e., much farther from the injection well). The only differences will be in the duration (i.e., width) of the dissolution, precipitation profiles, and Pb and Zn desorption profiles. In other words, for a fixed ∆r, the mass
ELSE_TSANG_CH021.qxd
8/22/2005
264
7:03 AM
Page 264
Predicting Trace Metal Fate in Aqueous Systems Using a Coupled Equilibrium 1000
1 2
3
4
5
6
7
8
9
10
Total Soluble Zn (ppm)
100
10
Unit 1 Unit 2 Unit 3
1
Unit 4 Unit 5
Artifact of using discrete, finitesized DynaChem process units
0.1
Unit 6 Unit 7 Unit 8
0.01
Unit 9 Unit 10
0.001 0
50
100
150
200
250
300
350
400
450
500
Time (hr)
Fig. 21.6. Zn(II) breakthrough curves for case study 2.
of dolomite, calcite, and sorbing ferrihydrite in each annular ring increases with increasing (r). Figures 21.4–21.6 illustrate a number of important points about case studies 1 and 2. First, upon reaction with the acidic wastewater, calcite and dolomite in the sandstone formation buffer the pH at about 6.5; therefore, sorption of Pb and Zn onto ferrihydrite is very favorable. However, once acidity in the wastewater consumes the calcite and dolomite, the pH drops sharply to 2.0, releasing the accumulated Pb and Zn back into solution. This results in a concentration wave for each metal that will move slowly outward from the injection point with time; the amplitude of these waves (or spikes) will exceed the wastewater feed concentration by an order of magnitude as illustrated in Figures 21.5 and 21.6. The duration (or width) of the spikes is a function of the mass of metal that has accumulated within each process unit or annular ring prior to the breakthrough time. For Pb, the breakthrough curves look like step waves of increasing duration (or width) because of Pb’s favorable sorption characteristics. For Zn, less metal has accumulated in each reaction zone prior to breakthrough; therefore, the breakthrough curves look more like short-duration spikes than longer-duration step waves. The short-duration, downward-pointing spikes in Pb and Zn concentration that occur before the breakthrough time are artifacts of the DynaChem model simulation. As mentioned earlier, the thickness of the annular rings (and, hence, the volume of the DynaChem process units) would ideally be infinitesimally small (i.e., ∆r in Fig. 21.1 is close to zero) to approximate continuous flow through the subsurface sandstone. The use of discrete, finite-sized process units, however, leads to artifacts (i.e., discontinuities or spikes) in the model-predicted concentration profiles. Second, the mobility of both Pb and Zn is significantly retarded relative to the movement of the water itself. As shown in Table 21.2, the ratios of the unit 10 breakthrough times for Pb and Zn to the hydraulic residence time are 375 and 183, respectively. The significant retardation of Pb and Zn migration (as well as the rapid desorption that occurs when the pH drops sharply to 2 upon complete dissolution of dolomite and calcite) is also illustrated in Figures 21.7 and 21.8, respectively. These two figures show the magnitude and variability
ELSE_TSANG_CH021.qxd
8/22/2005
7:03 AM
Page 265
21.6 Results and Discussion
265 Table 21.2. Case study results
Case study
Buffered pH of sandstone formation during sorption
Unit 10 breakthrough time 1 (hour)
tb (Pb/Zn)/ tHydraulic2
Breakthrough time1 relative to case study 1
1 2 3 4
~6.5 ~6.5 ~5.5 ~5.5
387 188 53 37
375 183 51 36
— 1/2 1/7 1/10
100 ppm Pb, pH 2 32 ppm Zn, pH 2 100 ppm Pb, pH 1 32 ppm Zn, pH 1
1
Breakthrough time is defined in this study as the time at which the Pb or Zn concentration reaches the inlet concentration for the first time.
2 Ratio of breakthrough time for Pb/Zn to the hydraulic residence time (1.03 hour for 10 units). Hydraulic residence time is based on a porosity of 25%.
1.E+07 1 2
3
4
5
6
7
8
9
10
1.E+06
Unit 1 Unit 2 Unit 3
Kd,eff (ml/g)
1.E+05
Unit 4 Unit 5
1.E+04
Unit 6 Unit 7 Unit 8
1.E+03
Unit 9 Unit 10
1.E+02
1.E+01 0
100
200
300
400
500
Time (hr) Fig. 21.7. Effective Pb(II) partition coefficients (Kd,eff) for units 1 to 10 in case study 1.
in effective Kd values (i.e., linear partition coefficients) for each metal. The effective Kd value (Kd,eff) includes the contributions of both sorption onto ferrihydrite as well as precipitation as the lead or zinc carbonate. Third, there are two sinks for both metalssorption onto the iron oxyhydroxides and precipitation as the metal carbonate. The dominant mechanism for each metal differs, however. For Pb, PbCO3 is the predominant sink, while for Zn, sorption onto ferrihydrite predominates. This is illustrated in Figures 21.9 and 21.10, respectively. Finally, the effective retardation of Zn relative to Pb at the same pH is only about 1/2 as shown in Table 21.2. Results for case studies 3 and 4 are summarized in Figures 21.11–21.13. Figure 21.11 shows the concentration/pH time profiles for process unit 10 for both case studies 3 and 4. Figures 21.12 and 21.13 present the trace-metal breakthrough curves for the ten process units for Pb and Zn, respectively. Notice the significant impact of a drop in the wastewater
ELSE_TSANG_CH021.qxd
8/22/2005
266
7:03 AM
Page 266
Predicting Trace Metal Fate in Aqueous Systems Using a Coupled Equilibrium 1.E+04 Unit 1 Unit 2 Unit 3 Unit 4
1.E+03 Kd,eff (ml/g)
Unit 5 Unit 6 Unit 7 Unit 8 Unit 9
1.E+02
Unit 10
1
2
3
4
5
6
7
8
9
10
1.E+01 0
100
200
300
400
500
Time (hr)
Fig. 21.8. Effective Zn(II) partition coefficients (Kd,eff) for units 1 to 10 in case study 2.
1000000 Kd PbCO3 100000
Kd sorption
Kd,eff (ml/g)
10000
1000
100
121.4
115.6
109.7
103.8
97.9
92.0
86.2
80.3
74.4
68.5
62.6
56.8
50.9
45.0
39.1
33.2
27.4
21.5
1
15.6
10
Time (hr)
Fig. 21.9. Contributions to Kd,eff from Pb(II) sorption onto ferrihydrite vs. precipitation as PbCO3 for unit 5 in case study 1.
feed pH from 2 to 1, which results in the buffered sandstone formation pH being lowered from 6.5 to 5.5. A drop of only 1 pH unit has a two-pronged effect. First, Pb and Zn sorption onto ferrihydrite is less favorable, while Pb and Zn carbonate solubilities increase. Second, the pH buffering agents, calcite and dolomite, are consumed much more rapidly. The net effect is that Pb and Zn breakthrough times are 1/7 and 1/5, respectively, of what
ELSE_TSANG_CH021.qxd
8/22/2005
7:03 AM
Page 267
21.6 Results and Discussion
267
10000 Kd ZnCO3 Kd sorption
Kd,eff (ml/g)
1000
100
131.5
124.2
116.9
109.6
102.3
95.0
87.7
80.4
73.1
65.8
58.5
51.2
43.9
36.6
29.3
22.0
14.7
7.4
1
0.1
10
Time (hr)
Fig. 21.10. Contributions to Kd,eff from Zn(II) sorption onto ferrihydrite vs. precipitation as ZnCO3 for unit 5 in case study 2.
1.E+04 1.E+03
Zn(aq) Dolomite
1.E+02
Pb(aq)
pH
1.E+01
Pb (aq) (ppm)
pH
1.E+00 Variables
pH 1.E-01
Zn (aq) (ppm)
ZnCO3 (s)
CaCO3 (s) (mol)
1.E-02
CaCO3 (s)
1.E-03
PbCO3 (s)
Dolomite (s) (mol)
Zn(aq)
1.E-04
PbCO3 (s ) (mol)
1.E-05
Pb(aq)
Dolomite
1.E-06
ZnCO3 (s) (mol)
1.E-07 1.E-08 0
10
20
30
40
50
60
70
80
Time (hr)
Fig. 21.11. Concentration/pH profiles for process unit 10 for case studies 3 and 4.
they were at a buffered pH of 6.5 (case studies 1 and 2). Table 21.2 summarizes the impact of both pH and metal type on trace-metal breakthrough time. Interestingly, the difference in metal retardation for 100 ppm Pb at pH 6.5 versus 32 ppm Zn at pH 5.5 is about one order of magnitude.
ELSE_TSANG_CH021.qxd
8/22/2005
268
7:03 AM
Page 268
Predicting Trace Metal Fate in Aqueous Systems Using a Coupled Equilibrium 10000 1 2
Total Soluble Pb (ppm)
1000
3
4
5
6
7
8
9
10
100 10
Unit 1 Unit 2 Unit 3 Unit 4 Unit 5 Unit 6 Unit 7 Unit 8 Unit 9 Unit 10
1 0.1 0.01 0.001 0
10
20
30
40
50
60
70
80
Time (hr)
Fig. 21.12. Pb(II) breakthrough curves for case study 3.
10000 1 2
3
4
5
6
7
8
9
10
Total Soluble Zn (ppm)
1000 100 10
Unit 1 Unit 2 Unit 3 Unit 4 Unit 5 Unit 6 Unit 7 Unit 8 Unit 9 Unit 10
1 0.1 0.01 0.001 0
10
20
30
40
50
60
70
80
Time (hr)
Fig. 21.13. Zn(II) breakthrough curves for case study 4.
In summary, small changes in environmental parameters, such as pH and trace-metal type, can have a significant impact on the mobility of trace metals in subsurface aqueous systems. This is reflected by the significant variability in trace-metal partition coefficients, which are often assumed to be constant in contaminant fate and transport models. In reality, the Kd values are far from constant in these highly reactive subsurface systems. In addition, while there
ELSE_TSANG_CH021.qxd
8/22/2005
References
7:03 AM
Page 269
269
are limitations with the equilibrium assumption close to the injection well where flow velocities are higher, these results highlight the best that one can expect to achieve for an extended time period far from the injection point. Finally, this study highlights the importance of coupling fundamental microscopic and spectroscopic studies of trace-metal sorption with surface complexation modeling to help predict the complex aqueous inorganic chemistry occurring over extended time periods in the dynamic subsurface environment. REFERENCES Anderko, A., Sanders, S.J. and Young, R.D., 1997. Real-solution stability diagrams: A thermodynamic tool for modeling corrosion in wide temperature and concentration ranges. Corrosion, 53: 43–53. Chase, M.W., Jr., Davies, C.A., Downey J.R. Jr., Frurip, D.J., McDonald R.A. and Syverud A.N., 1985. JANAF thermochemical tables. J. Phys. Chem. Ref. Data, Suppl. No. 1, 3rd ed., 14: 1–1856. Daubert, T.E. and Danner R.P., 1989. Physical and Thermodynamic Properties of Pure Chemicals: DIPPR Data Compilation. Hemisphere Publishing Corp., New York. Dyer, J.A., 2002. Advanced Approaches for Modeling Trace Metal Sorption in Aqueous Systems. Ph.D. Dissertation, University of Delaware, Newark, DE. Dyer, J.A., Trivedi, P., Scrivner, N.C. and Sparks, D.L., 2003. Lead Sorption onto Ferrihydrite. 2. Surface Complexation Modeling. Environ. Sci. Technol., 37(5): 915–922. Dyer, J.A., Trivedi, P., Scrivner, N.C. and Sparks, D.L., 2004. Surface complexation modeling of Zinc sorption onto Ferrihydrite. J. Colloid. Interface Sci. 270(1): 56–65. Glushko, V.P., Medvedev, V.A., Bergman, G.A., Vasil’ev, B.P., Kolesov, W.P., Gurvich, L.V., Yungmand, V.S., Khodakovskii, I.L., Resnitskii, L. A., Smirnova, N.L., Gal’chenko, G.L., Alekseev, V.I., Vorob’ev, A.F., Baibuz, V.F., Kostryukov, B.N. and Biryokov, B.P., 1965–1981. Thermo Constants of Compounds. Academy of Sciences, Moscow, USSR, Vols. 1–10. Gurvich, L.V., Veyts, I.V., Medvedev, V.A., Khachkuruzov, G.A., Yungman, V.S., Bergman, G.A., Iorish, V.S., Yurkov, G.N., Gorbov, S.I., Kuratova, L.F., Trishcheva, N.P., Przheval’skiy, I.N., Leonidov, V.Ya., Ezhov, Yu.S., Tomberg, S.E., Nazarenko, I.I., Rogatskiy, A.L., Dorofeyeva, O.V. and Demidova, M.S., 1989. Thermodynamic Properties of Individual Substances, 4th ed., Vols. 1–5, USSR Academy of Sciences, Institute for High Temperatures and State Institute of Applied Chemistry, Hemisphere Publishing Corp., New York. Hayes, K.F. and Katz, L.E., 1996. Application of X-ray Absorption Spectroscopy for Surface Complexation Modeling of Metal Ion Sorption. In: P.V. Brady (Ed.), Physics and Chemistry of Mineral Surfaces. CRC Press, Boca Raton, FL, pp. 147–223. Oelkers, E.H., Helgeson, H.C., Shock, E.L., Sverjensky, D.A., Johnson, J.W. and Pokrovskii V.A., 1995. Summary of the apparent standard partial molal Gibbs free energies of aqueous species, minerals, and gases at pressures 1 to 5000 bars and temperatures 25 to 1000 °C. J. Phys. Chem. Ref. Data, 24: 1401–1560. Rafal, M., Berthold, J.W., Scrivner, N.C., and Grise, S.L., 1994a. Models for Electrolyte Solutions. In: S.I. Sandler (Ed.), Models for Thermodynamic and Phase Equilibria Calculations. Marcel Dekker, Inc., New York, pp. 601–670. Rafal, M., Black, P., Sanders, S.J., Tolmach, P.I. and Young, R.D., 1994b. Development of a comprehensive environmental simulation program. AIChE Spring National Meeting, Atlanta, GA, April 17–21.
ELSE_TSANG_CH021.qxd
270
8/22/2005
7:03 AM
Page 270
Predicting Trace Metal Fate in Aqueous Systems Using a Coupled Equilibrium
Sahai, N. and Sverjensky, D.A., 1998. GEOSURF: A computer program for modeling adsorption on mineral surfaces from aqueous solution. Comput. Geosci. 24: 853–873. Scrivner, N.C., Butler, P.B. and Karmazyn, J., 1996. Modeling: An excellent solution for remediation, 69th Annual Conference and Exposition of the Water Environment Federation, Session No. 5, Dallas, TX. Shock, E.L. and Helgeson, H.C., 1988. Calculation of the thermodynamics and transport properties of aqueous species at high pressures and temperatures: Correlation algorithms for ionic species and equation of state predictions to 5 kbar and 1000°C. Geochim. Cosmochim. Acta, 52: 2009–2036. Trivedi, P., Dyer, J.A. and Sparks, D.L., 2003. Lead sorption onto Ferrihydrite. 1. A Macroscopic and Spectroscopic Assessment. Environ. Sci. Technol., 37(5): 908–914. Trivedi, P., Dyer, J.A. and Sparks, D.L., 2004. Mechanistic and Thermodynamic Interpretations of Zinc sorption onto Ferrihydrite. J. Colloid. Interface Sci. 270(1): 77–85. Wagman, D.D., Evans, W.H., Parker, V.B., Schumm, R.H., Halow, I., Bailey, S.M., Churney, K.L. and Nuttall R.L., 1982. The NBS tables of chemical thermodynamic properties. Selected values for inorganics and C1 and C2 organic substances in SI units. J. Phys. Chem. Ref. Data, Suppl. No. 2, 11: 1–392. Zemaitis, J.F., Jr., Clark, D. M., Rafal, M. and Scrivner, N.C., 1986. Handbook of Aqueous Electrolyte Thermodynamics: Theory and Application. Design Institute for Physical Property Data, American Institute of Chemical Engineers, Inc., New York.
ELSE_TSANG_CH022.qxd
8/21/2005
9:12 AM
Page 271
Chapter 22
REVIEW OF THE STUDIES OF RADIONUCLIDE ADSORPTION/DESORPTION WITH APPLICATION TO RADIOACTIVE WASTE DISPOSAL SITES IN THE RUSSIAN FEDERATION V.G. Rumynina, L.N. Sindalovskiya, P.K. Konosavskya, A.V. Mironovaa, E.V. Zakharovab, E.P. Kaiminb, E.B. Pankinac, and A.A. Zubkovd a
Institute of Environmental Geology of the RAS, St. Petersburg Division, St. Petersburg, Russia b Institute of Physical Chemistry of the RAS, Moscow, Russia c A.P. Alexandrov Technical Research Institute, Sosnovyi Bor, Russia d Siberian Chemical Combine, Seversk, Russia
22.1 INTRODUCTION Recent research findings strongly suggest that subsurface radionuclide transport is often accompanied by nonideal (anomalous) phenomena caused by (a) adsorption hysteresis facilitated, in particular, by irreversible uptake of radionuclides by a specific group of minerals in the rock matrix or by secondary geochemical alteration of the minerals, (b) deterministic and stochastic heterogeneity of the geological strata resulting in preferential flow paths of the radioactive component, and (c) colloid-facilitated transport. In particular, nonideal radionuclide transport behavior was observed during investigations that have been conducted in Russia at several sites associated with near-surface and subsurface radioactive solid and liquid waste disposal (Fig. 22.1). Two such disposal sites are located within the Northwestern Center of Nuclear Energy (NWCNE), near St. Petersburg: (a) a solid radioactive waste disposal site (the so-called Radon site), which is used as the northwestern regional surface repository; and (b) an Engineered or Designed Underground Repository (EUR site) in the Cambrian clay (a regional aquitard) for spent nuclear fuel and high-level radioactive waste storage and isolation. Two other sites are associated with deep-well injection repositories, operated by the Siberian Chemical Plant and the Siberian Mining-and-Chemical Plant—the Tomsk-7 site and Krasnoyarsk-26 site, located in Western and Eastern Siberia, respectively. The last site is associated with one of the surface reservoirs, Lake Karachai, which has been used over a long period for the disposal of liquid radioactive waste by the Mayak Production Association (the Lake Karachai site), South Ural. Radioactive wastes contain various long-lived decay products. The most dangerous radionuclides are fission products (90Sr and 137Cs) and actinides (239Pu and 241Am). To evaluate radionuclide behavior in the subsurface environment, sorption and desorption kinetics and equilibrium were measured in batch, diffusion, and column (dynamic) experiments. Rock samples both of sedimentary (unconsolidated and consolidated) and crystalline/fractured types were selected from the radioactively contaminated site (Table 22.1). Core samples of the sandy sediment (the Radon Site) were also taken from boreholes to study spatial variability of the adsorption and desorption constants based on a variogram analysis.
ELSE_TSANG_CH022.qxd
272
8/21/2005
9:12 AM
Page 272
Review of the Studies of Radionuclide Adsorption/Desorption with Application
Fig. 22.1. A map showing locations of waste disposal sites. Table 22.1. Summary of the experimental setup: samples, solutions, temperature/pressure, and experimental conditions Site
Type of rock (core samples)
Radionuclides
Solutions
Temperature/ pressure conditions
Sample conditioning
Radon
Sedimentary: sand
90
SGW*
Conditioned
EUR
Sedimentary: clay
90
SGW, SrCl2
Tomsk-7
Sedimentary: clayey sand
90
SGW, NaNO3
Krasnoyarsk-26
Sedimentary: clayey sand
90
NaOHNaNO3
Crystalline/ fractured: tuffs, tuffand-lava
90
Room (T20°C, P0.1 MPa) Room (T20°C, P0.1 MPa) Room (T70°C, P3 MPa) Room (T70°C, P3 MPa) Room (T20°C, P0.1 MPa)
Lake Karachai
Sr, 137Cs
Sr, 36Cl
Sr, 137Cs
Sr, 137Cs, Pu, 241Am
239
Sr
SGW, NaNO3
Conditioned
Conditioned, unconditioned Conditioned, unconditioned Unconditioned
*SGW—synthesized groundwater.
The purpose of this study is to quantitatively evaluate the adsorption-related reactions with respect to natural attenuation, sorption, and desorption kinetics and equilibrium under different geochemical, temperature, and pressure conditions (Table 22.1). Earlier developed analytical and numerical models were applied that take into account the microscopic heterogeneity of multimineral geosorbents and variations of an external chemical potential resulting from the rate-limited transformations of the mineral phase.
ELSE_TSANG_CH022.qxd
8/21/2005
9:12 AM
Page 273
22.2 Radon Site
273
22.2 RADON SITE 22.2.1 Groundwater Contamination Specifics Radioactive solutions originated as a result of seepage of atmospheric precipitation (rainfall and snow) through solid radioactive wastes, which were placed in surface impoundments with imperfectly designed waterproofing properties. During a period of around 10 years (1980–1990) a leachate percolated down to the Cambrian (Lomonosovsky) sand aquifer (Fig. 22.2) and migrated with the natural gradient flow toward the Finland Gulf (Rumynin et al., 2003). The distribution of dissolved radionuclides in the aquifer shows the temporal variations in the leachate leakage rate from different impoundments. Unfortunately, radionuclide migration monitoring does not include an efficient leachate flow control system, and consequently, there is a degree of uncertainty in the solute transport analysis. The average rate of the radioactive solution release was estimated to be about 450 kBq day1 (90Sr) and 24,000 kBq day1 (3H). In 1990, the radioactive waste storage impoundments were reconstructed, and the leakage rate was significantly reduced. Monitoring data on radionuclide concentrations versus time (Fig. 22.3) indicate that only natural attenuation processes are currently taking place. It is clear that the efficiency of the site rehabilitation is controlled by the ability of the rock matrix to release the earlier adsorbed radionuclides, and therefore the study of desorption plays an important role in prediction of groundwater contamination. 22.2.2 Experimental Tasks Experiments focused on evaluating adsorption and desorption parameters that control plume spreading and natural attenuation. The first step of the study was concerned with a detailed
NW
Izhorskoe Plateu
EUR site Radon site
100 50
100 1 -2
50
Finland Gulfs 1 -2
0 -50 1
Vkt1
-100 -150
lm
1
0
ln
-50 -100 -150
Vkt2
-200 -250
0
5
10
-200
AR-PR 1-2
15 km
-250
-300
-300 -350
-350 H, m
1
2
3
4
5
6
7
8
Fig. 22.2. A schematic hydrogeological cross section of the Radon and DUR sites: 1—limestone; 2—sand; 3—clay; 4—sandy clay; 5—clayey sand; 6—crystalline rock; 7—GW table of Ordovician (O1-2) aquifer; 8—GW table of Lomonosovsky (ε1 ln) aquifer.
ELSE_TSANG_CH022.qxd
8/21/2005
274
9:12 AM
Page 274
Review of the Studies of Radionuclide Adsorption/Desorption with Application 7.0E-07
3.5E-04 3.0E-04
6.0E-07
H-3 Sr-90
Ci.L-1 90Sr,
0.0E+00 20 1/ 1/
20 1/ 1/
20 1/ 1/
19 1/ 1/
19 1/ 1/
19 1/ 1/
19 1/ 1/
19 1/ 1/
19 1/ 1/
19 1/ 1/
04
0.0E+00 02
1.0E-07
00
5.0E-05
98
2.0E-07
96
1.0E-04
94
3.0E-07
92
1.5E-04
90
4.0E-07
88
2.0E-04
86
Ci.L-1
5.0E-07
3H,
2.5E-04
Fig. 22.3. Variance of radionuclide concentration as a function of time (observation well # 7, Radon site).
investigation of the adsorption process based on laboratory and column experiments with three reference samples, which represented typical lithologic units. A special field and laboratory program was also conducted in order to study the spatial variability of adsorption and desorption constants. 22.2.3
Batch and Column Experiments
Table 22.2 shows a distribution of minerals between silt and clay fractions in the basic samples. In particular, it is seen that samples differ in their lepidocrocite—a ferro-oxide mineral that might control radionuclide uptake—content. Before conducting the sorption measurements, the sand samples were conditioned with synthesized groundwater (SGW) (mg l1/mg-eq l1: Na—69/3; Ca2—20/1; Mg2—12/1; Cl—71/2; HCO3—183/3; Sr2— 0.1/0.0023) to ensure that the system was in steady state. Observations for the process kinetics showed that a period of 2 weeks was sufficient to reach equilibrium. The best fit for the log-transformed experimental data characterizing adsorption equilibrium is a straight line (Fig. 22.4). It means that adsorption isotherms are of the Freundlich type: N s KFs C ns,
(22.1)
where C is the equilibrium concentration in the solution, N s is the concentration on the solid phase at the adsorption stage of the experiment, KFs , and ns are the apparent linear and exponential Freundlich constants. On the other hand, the laboratory tests (Fig. 22.4), which included a stage of step-by-step desorption (by SGW) of previously adsorbed 90Sr, clearly indicate that adsorption exhibits irreversibility and hence hysteresis. The overall set of experimental data, represented by both adsorption and desorption points in plots (Fig. 22.4), can be fitted to a nonlinear dual-site model (Rumynin et al., 2002a). We postulated that the rock is chemically heterogeneous, and there are two types of reaction sites (1 and 2) within the mineral phases of the rock, which differ in their abilities
ELSE_TSANG_CH022.qxd
8/21/2005
9:12 AM
Page 275
22.2 Radon Site
275
to take up and release radionuclides (Pickens et al., 1981). The partial linear (Ki) and exponential (nis) Freundlich phase distribution coefficients are responsible for the adsorption/desorption process; thus the total adsorption/desorption isotherm is as follows:
N
冦
ns
s
ns
N f K1C 1 (1f )K2C 2 , ns
d
(22.2)
ns
N f K1C 1 (1f )K2C02,
where N sd is the adsorbed concentration of a component at the adsorption (s) and desorption (d) stages, C is the concentration in the liquid phase, Ki and nsi are the partial linear and exponential constants, and f is the fraction of Site 1. At the desorption stage, the concentration at Table 22.2. Mineralogy (w%) for three reference sand samples (#1n, #1c, #2b) from the Radon site Minerals
Illite Kaolin Quartz Gypsum Lepidocrocite Halite
Fraction size, mm 0.05–0.01 1n 1c 2b
0.01–0.002 1n 1c
32 56 12 1 1
23 44 6 2 25
—
31 54 10 5 1
23 45 4 5 26
—
—
4
—
—
100000
13 21 20 42
2b 13 22 13 35
0.002 1n 1c
2b
—
30 58 2 8 3
19 34 1 6 42
14 20 11 41
—
20
—
—
14
a
N=19.9C0.89
N, Bq cm-3
10000
1000
N0=25.3C0.57
100
10 0.01
1.00
100.00
10000.00
C, Bq cm-3
Fig. 22.4. Adsorption/desorption isotherms for three basic samples: measured concentration points at the adsorption and multistep desorption stages for three sets of experiments (a—number of samples, correspondingly 1c): black-filled circles—experimental adsorption measurements; triangles— experimental desorption measurements; crosses and solid lines—modeling results; open circles and thin lines—N0 vs. C.
ELSE_TSANG_CH022.qxd
8/21/2005
276
9:12 AM
Page 276
Review of the Studies of Radionuclide Adsorption/Desorption with Application 100000
b
N=13.0C0.8
N, Bq cm-3
10000
1000
100
N0=26.9C0.45
10 0.01
1.00
100.00
10000.00
C, Bq cm-3 10000
c N=11.0C0.72
N, Bq cm-3
1000
100 N0=9.1C0.54
10 0.01
1.00
100.00 C, Bq
10000.00
cm-3
Fig. 22.4. (continued). number of samples: b, c—correspondingly 1n and 2b.
Site 2 remains constant and is equal to the value of N2d N 0, which is reached at the end of ns the adsorption stage (when C C0): N0 ⬅ N d2 (1 f )K2C0 2 Fitted parameters Ki and nis for three lithological types of the sediment are presented in Table 22.3. They were found using a numerical solution of a full system of equations describing the experimental setup (Appendix A). The fitting curves satisfactorily describe the experiments with 90Sr (Fig. 22.4). The fraction coefficient ( f) was assigned to be 0.95, which corresponds to the quartz content in the mineral phase of the sediment (the remaining
ELSE_TSANG_CH022.qxd
8/21/2005
9:12 AM
Page 277
22.2 Radon Site
277
Table 22.3. Fitting equilibrium constants controlling radiostrontium adsorption/desorption n2s n 1s Effective (overall) isotherm (1 f)K2 fK1 of the Freundlich Type† K 2* K1*
Sample No.
1c
7.6 8.0
25.3 506
1.0
0.57
N 19.9 C 0.89
1n
3.8 4.0
26.9 538
1.0
0.45
N 13.0 C 0.85
2b
4.75 5.0
9.1 182
0.8
0.54
N 11.0 C 0.75
* For f 0.95 † N KFs C n s—linearization of the modeling curves (K Fs and ns are the apparent/effective Freundlich coefficients); s s units for K1 and K2 are (Bq cm3)n11 and (Bq cm3)n21, respectively.
5%—clay minerals and Fe/Al/Mn oxide coatings on the sand grains—are assumed to be responsible for the specific adsorption). The kinetic constants (Appendix A) were assumed to be α 1s α s2 50 day1; α d1 10 day1, and α d2 0.01 day1. In addition to the batch tests, package-wise dynamic (column) experiments with the same samples of the sand and radioactive solutions were carried out. A simplified balance approach was used to determine the adsorption irreversibility. The results are consistent with data from the batch tests. The performed analysis indicates that the modeling of 90Sr transport in the studied aquifer must account for the irreversible adsorption. 22.2.4 Spatial Variability of the Freundlich Adsorption and Desorption Constants Taking into account the actual aquifer heterogeneity, we can expect that the relevant adsorption constants of the Freundlich isotherm might vary over wide ranges, depending on the spatial variability of physical and physicochemical properties of the geosorbent. Therefore, we describe below the experimental investigations aimed at a geostatistical estimation of the hydrogeological system for bridging the gap between field/laboratory measurements and numerical aquifer modeling. Description of experimental work Thirty-seven shallow boreholes were drilled to study the spatial distribution of sand properties (Fig. 22.5). Two core samples of the sandy sediment were taken from each of the boreholes from intervals at depths of 0.5 m and 1.0 m, respectively. Thus, 74 rock samples were available for further laboratory investigations (Appendix B). Each core sample was examined in batch laboratory adsorption tests. To evaluate the parameters that are responsible for the adsorption process, and to assess the relations between adsorption and physical parameters of the aquifer material, the grain-size distribution and hydraulic conductivity were also measured. Isotherms and estimated parameters The major features of 90Sr adsorption are illustrated by Figure 22.6, a standard isothermal presentation of experimental data, using a bilogarithmic system of coordinates. This indicates that the Freundlich type of the adsorption isotherm dominates. However, in some experiments, the concentration points, related to the individual isotherms Ns–C (Fig. 22.6), did not perfectly fit the Freundlich model. The scatter of the concentration points around the
ELSE_TSANG_CH022.qxd
8/21/2005
278
9:12 AM
Page 278
Review of the Studies of Radionuclide Adsorption/Desorption with Application
Fig. 22.5. Sampling pattern.
(a) 100000
Samples #1-1 through #9-2
Ns, Nd, Bq cm-3
10000
1000 Adsorption Desorption 1 Desorption 2 100 10
100
1000
10000
C, Bq cm-3
Fig. 22.6. Adsorption and desorption isotherms (C is the equilibrium concentration in the solution; Ns and Nd are the concentrations in the solid phase for adsorption [s] and desorption [d]; Ns,d are normalized to the specific volume of the rock; Ns,d ρ bN, ρ b is the bulk density of the rock; and N is the amount/activity of the radionuclide adsorbed per unit weight of the rock).
ELSE_TSANG_CH022.qxd
8/21/2005
9:12 AM
Page 279
22.2 Radon Site
279
(b) 100000 Samples #10-1 through #18-2
Ns, Nd, Bq cm-3
10000
1000
100 10
100
1000
10000
C, Bq cm-3 (c) 100000 Samples #19-1 through #27-2
Ns, Nd, Bq cm-3
10000
1000
100 10
100
1000 C, Bq cm-3
Fig. 22.6 (continued).
10000
ELSE_TSANG_CH022.qxd
280
8/21/2005
9:12 AM
Page 280
Review of the Studies of Radionuclide Adsorption/Desorption with Application (d) 100000 Samples #28-1 through #37-2
Ns, Nd, Bq cm-3
10000
1000
100 10
100
1000
10000
C, Bq cm-3
Fig. 22.6 (continued).
fitting lines may be due to experimental errors or actual deviation of the adsorption process from the Freundlich type. Such a situation occurs for about 20% of the experimental data. Along with assessment of the Freundlich constants, an effective distribution coefficient Kd (dimensionless) was calculated (Appendix B) for a given value of concentration (C), as follows: s
s ∆N N s Kd ≡ KFs (C)(n 1) C ∆C
(22.3)
Analysis of the experimental data indicates that the values of KFs and Kd vary over a wide range (Appendix B). The adsorption process is noticeably nonlinear: the majority of the isotherms are characterized by values of the constant, ns, which are essentially less than 1. Thus, the medium can be considered highly heterogeneous with respect to variations of its chemical properties. Analysis of the grain-size distribution shows that the sediment is primarily represented by quartz grains (up to 93–97%) of sand size (0.05 mm dparticle 0.5 mm). The fines content (silt- and clay-type grains) varies on average between 2% and 5% (3.1% is the arithmetic mean). Thus, the sediments can be classified as pure quartz sand with a negligible amount of fines; it should be noted that the fines content is close to the sensitivity of the hydrometer test. The uniformity coefficient, Cu (d60/d10 ratio), is less than 4 (with few exceptions), and varies within a rather narrow range. This means that the sand is well sorted and,
ELSE_TSANG_CH022.qxd
8/21/2005
22.2 Radon Site
9:12 AM
Page 281
281
at the micro level, can be described as a homogeneous porous medium. Despite the low content of silt and clay particles, one can suppose that the small-size grains can play a major role in radionuclide adsorption. The sandy material is rather permeable: in the majority of the flow-column experiments, hydraulic conductivity was determined to fall in the range of 1.5–7.7 m d1 (the mean value is about 2.7 m d1). Statistical distribution of the parameters (histograms) A standard geostatistical analysis was applied to establish statistical variations and distributions of the adsorption constants (Fig. 22.7 and Table 22.4). It was found that there are two types of two-dimensional histograms for the studied parameters: (1) the exponential Freundlich constant (ns) shows a normal frequency distribution, and (2) the linear Freundlich constant (KFs) and the effective distribution coefficient (Kd) (for a particular case, C 10,000 Bq cm3, as defined by Equation 22.3) show lognormal frequency distributions. Analysis of the experimental data indicates that there is an inverse linear relationship between constants KFs and ns. At the same time, the statistical analysis shows that the chemical constants are correlated neither with the studied grain-size distribution characteristics nor with the hydraulic conductivity. Variations of secondary mineral species (hydroxides and oxides of Fe/Mn/Al) may be responsible for variations in the chemical properties of the aquifer material (Jackson and Inch, 1989). Spatial variability of the adsorption constants The relevant characteristics of spatial variability have been studied through variogram analysis. Two variogram subroutines from GSLIB (Deutsch and Journel, 1992) for irregularly spaced data in two and three dimensions were used. To handle two- and three-dimensional (a)
Fig. 22.7. Histograms of the basic adsorption constants of the Cambrian sandy aquifer (bars are experimental data; solid lines are theoretical distributions).
ELSE_TSANG_CH022.qxd
8/21/2005
282
9:12 AM
Page 282
Review of the Studies of Radionuclide Adsorption/Desorption with Application (b)
No of obs
Histogram of Kfs y = 74 * 18.358333 * lognorm (x; 3.152037; 1.160958) 42 39 36 33 30 27 24 21 18 15 12 9 6 3 0 1.7
20.0
38.4
56.8
75.0
93.5 112.0 130.0 148.6 167.0 185.3 203.6 222.0 Kfs
(c) Histogram of Kd y = 74 * 2.0500805 * lognorm (x; 1.137088; 0.7670445) 36 33 30 27 No of obs
24 21 18 15 12 9 6 3 0 0.3
2.3
4.4
6.4
8.5
10.5
12.6 Kd
14.6
16.7
18.7
20.8
22.8
24.9
Fig. 22.7. (continued).
data, the empirical semivariograms Var(h) were calculated for four parameters—(Y(k)), Y(1) ns, Y(2) ln(KFs), and Y(3) ln(Kd) (Fig. 22.8)—using Equation (4): 1 N(h) Var(h) 冱 [Y(k)i Y(k)j]2, 2N(h) 1
(22.4)
where N(h) is the number of pairs, Y(k)i the value at the start (or tail) of the pair i, Y(k)j the corresponding end (or head) value, and h the separation vector, which is specified with some direction and distance (lag) tolerance.
ELSE_TSANG_CH022.qxd
8/21/2005
9:12 AM
Page 283
22.2 Radon Site
283
Table 22.4. Statistics for adsorption and physical properties of the Cambrian (Lomonosovsky) sandy aquifer
ln(Kfs)† ns* ln(Kd)† ln(K)† ln(d60/d10)† ln(SCPC)† ln(GSV)†
ymin
ymax
ymean
σy2
0.53 0.29 1.24 1.20 0.41 0.51 0.55
5.4 1.1 3.21 2.05 2.30 2.3 1.7
3.1 0.78 1.14 0.98 0.86 0.94 1.28
1.31 0.028 0.59 0.42 0.076 0.32 0.12
*Normal distribution. s † Lognormal distribution; Kfs is the linear Freundlich constant, (cm3/Bq)n 1; ns is the exponential Freundlich constant; Kd is the effective distribution coefficient; K is the hydraulic conductivity, md1; d60/d10 is the uniformity coefficient; SCPC is the silty/clay particle content (%), GSV is the grain-size variance.
(a) 2
1.6
Variogram
1.2
0.8
Semivariograms for ln(KFs) 2D: Overall (z=0.5 m) 2D: Overall (z=1.0 m) 3D: Overall (tolerance=90 degrees) 3D: X-direction (tolerance=45 degrees) 3D: Y-direction (tolerance=45 degrees) Fit for 2D-var (z=0.5 m): var=0.16+1.54(1-exp(-h/12.0)) Fit for 2D-var (z=1.0 m): var=0.045+1.5(1-exp(-h/5.0)) Fit for 3D-var: var=0.34+1.3(1-exp(-h/10.8))
0.4
0 0
10
20
30
Distance, m
Fig. 22.8. Experimental semivariograms for KFs (upper row, left plot), ns (upper row, right plot), Kd, and fitted theoretical curves.
A two-dimensional analysis focused on construction of the variograms based on two sets of data obtained for each of the sampling intervals (z 0.5 and 1.0 m). For three-dimensional variogram-analysis-based data, the two intervals were considered to belong to one data set.
ELSE_TSANG_CH022.qxd
8/21/2005
284
9:12 AM
Page 284
Review of the Studies of Radionuclide Adsorption/Desorption with Application
(b) 0.05
0.04
Variogram
0.03
0.02
Semivariograms for ns 2D: Overall (z=0.5 m) 2D: Overall (z=1.0 m) 3D: Overall (tolerance=90 degrees) 3D: X-direction (tolerance=45 degrees)
0.01
3D: Y-direction (tolerance=45 degrees) Fit for 3D-var: var=7e-7+0.03(1-exp(-h/0.43)) Fit for 2D-var (z=0.5 m): var=3e-4+0.02(1-exp(-h/3.4)) Fit for 2D-var (z=1.0 m): var=2e-5+0.03(1-exp(-h/1.9)) 0 0
10
20
30
Distance, m
Fig. 22.8. (continued).
The semivariograms were calculated in all directions, including two perpendicular directions. The corresponding assessments showed that there is essentially no anisotropy in the Y(1), Y(2), and Y(3) fields. Mean semivariograms (in all directions) were therefore calculated and approximately fitted with theoretical curves γ (h) of the standard exponential type:
冤
冢
h γ (h) c 1 exp a
冣冥.
(22.5)
The fitted constants a and c are presented in the plot legends of Figure 22.8. It appears that a certain number of pairs on the variograms was not sufficient to perfectly fit the exponential model. Variograms for the different sampling intervals were fitted with the exponential model using different ranges and sill values. The range of ln(KFs) is comparable with the size of the studied site. The range of ns is about one-third of the size of the tested site. The variograms show a nugget effect.
ELSE_TSANG_CH022.qxd
8/21/2005
9:12 AM
Page 285
22.2 Radon Site
285
(c)
0.8
Variogram
0.6
0.4
Semivariogram for ln(Kd) 2D: Overall (z=0.5 m) 2D: Overall (z=1.0 m) 3D: Overall (tolerance=90 degrees)
0.2
3D: X-direction (tolerance=45 degrees) 3D: Y-direction (tolerance=45 degrees) Fit for 3D-var: var=0.18+0.54(1-exp(-h/9.5)) Fit for 2D-var (z=0.5 m): var=0.0+0.6(1-exp(-h/8.0)) Fit for 2D-var (z=1.0 m): var=3e-3+0.62(1-exp(-h/3.0)) 0 0
10
20
30
Distance, m
Fig. 22.8. (continued).
22.2.9 Geostatistical Interpolation of the Adsorption Constants The ordinary kriging of Y(k) (Deutsch and Journel, 1992) was used to map the expected Y(k) values. The kriging interpolation was applied to construct two-dimensional fields of Y(k) for each of the elevation levels (z 0.5 and 1.0 m). The resulting space distributions were transformed to get fields of the actual constants (Fig. 22.9). 22.2.10 Variability of the Desorption Constants Measured concentration points at the adsorption and multistep desorption stages for all sets of experiments are shown in Figure 22.6. The partial adsorption/desorption constants have been estimated (Table 22.5), based on the dual-site model (2). The fitting curves satisfactorily describe about 40% of the experiments with 90Sr. The remaining 60% of the experiments could not be properly interpreted in the framework of the model, probably due to experimental errors
ELSE_TSANG_CH022.qxd
8/21/2005
286
9:12 AM
Page 286
Review of the Studies of Radionuclide Adsorption/Desorption with Application 30
1.7
5.2
7.7
30
4.9
17.5
3.6 KFs z=1.0 m
34.5
29.3
4.6
3.2
18.4 20
6.3
88.9
11.5
85.2
41.2 15
5.5
29.0
16.2
103.0
3.9
46.2
54.7
70.8
21.2
34.4
16.3 5
10.4
25.1
43.9
43.2
205.6
22.7 0 0
43.0
29.7
29.2
52.9
54.8
5
10
15
20
25
10
3.5
25
28.7
20.9
12.0
7.0
7.6
19.4
200.0
16.2
14.3
28.7
3.8
9.5
10.9
125.5
119.0
81.0
18.9
6.6
23.1
35.1
9.5
30.7
73.8
144.6
17.3
63.9
34.8
51.8
211.6
19.1
0
5
10
15
20
25
16.6
222.0 20
Y, m
Y, m
KFs z=0.5 m 70.3 25
15
94.6
10
5
0
X, m 2
10 1.10 30
20
25
30 40 0.92
X, m 200 0.92
0.91 30
0.66
0.93
1.02
0.85 25
0.75
0.77
0.92
0.90
0.31 20
0.87
0.78
0.29
0.80
0.57 15
0.88
0.77
0.91
0.88
0.94
0.59
0.53
0.64
0.81
0.92
ns z=0.5 m
ns z=1.0 m
0.55 25
0.68
0.76
0.93
1.06
0.78 20
0.92
0.48
0.77
0.59
0.79 15
0.95
0.80
0.73
0.45
0.84
0.77
0.67
0.70
0.83
0.82
0.89 5
1.00
0.81
0.82
0.74
0.60
0.83 5
0.78
0.96
0.83
0.65
0.59
1.01 0 0
0.81
0.77
0.79
0.75
0.69
0.72
0.84
0.65
0.38
0.90
5
10
15
20
25
0.97 0 0
5
10
15
20
25
10
X, m 0.45 0.7 0.75 0.79 0.83 0.9
Y, m
Y, m
2.7
10
1.02
X, m 1.1
Fig. 22.9. Estimated values of the adsorption constants KFs , ns, and Kd.
in the detection of low 90Sr concentrations in solution and to the heterogeneity of the rock samples (two desorption branches belonging to one isotherm actually reflect the solution’s chemical interaction with two subsamples, which may differ at the microscopic level). Therefore, the available information is rather restricted in its use for a comprehensive geostatistical analysis and generalization. Meanwhile it should be noted that (1) variability of the calculated constants (Table 22.5) is significant; (2) partial isotherms N si f(Ki, n si) can be both concave and convex;
ELSE_TSANG_CH022.qxd
8/21/2005
9:13 AM
Page 287
22.2 Radon Site
287
4.3
2.5
3.7
2.1 30
30
0.8
1.9
Kd z=0.5 m 1.8
3.2
2.4
5.6
2.4
3.0
0.7
1.4
2.0
6.0
3.5
4.6
1.3
0.6
2.2
5.6
2.6
4.5
4.4
6.6
5.9
10.4
4.4
8.4
3.9
5.2
24.9 0 0
7.5
3.6
4.2
5.3
3.2
5
10
15
20
25
20
15
10
5
4.2
Y, m
1.1
25
Y, m
Kd z=1.0 m 4.2 25
2.9
2.5
5.7
2.8
0.4 20
2.3
2.6
0.3
2.6
1.8 15
4.7
3.5
1.7
3.1
6.3
2.9
1.6
2.9
3.3
3.2
4.8 5
4.6
6.6
6.4
2.9
3.3
13.1 0 0
4.8
8.0
2.1
0.7
7.6
5
10
15
20
25
10
X, m 0.6
2
3
4
5
6
3.2
X, m 20
Fig. 22.9 (continued). Table 22.5. Fitted parameters from interpretation of the adsorption/desorption experiments Borehole interval
fK1
ns1
(1 f )K2
ns2
KsF
ns
1-1 1-2 2-1 2-2 3-1 3-2 4-2 6-1 6-2 8-1 10-1 10-2 11-1 12-1 13-1 14-2 15-1 16-1 16-2 22-2 23-2 25-1 29-2 34-2 35-1 36-1
4.2 5.9 8.3 18.5 0.2 4.6 1.7 0.2 0.05 2.8 2.6 20.8 0.3 3.7 0.11 0.17 25.5 6.9 31.7 13.6 17.3 8.5 4.9 19 29.7 0.51
1.05 0.8 0.81 0.67 1.18 0.85 0.87 1.23 1.51 1.05 1.11 0.84 1.28 1.05 1.42 1.25 0.69 0.91 0.64 0.74 0.78 0.79 0.94 0.73 0.74 1.18
23 21.4 38.4 50.6 37.6 23.1 138 61.1 25.5 21.5 87.9 24.5 197 541 46.4 574 114 0.14 0.18 1.3 3.9 0.31 37.2
0.97 0.93 0.8 0.72 0.71 0.88 0.49 0.62 0.81 0.84 0.63 0.68 0.44 0.36 0.72 0.22 0.58 1.33 1.31 1.05 0.99 1.21 0.62
28.3
0.49
22.7 17.3 43 63.9 29.7 34.8 51.8 54.8 19.1 10.4 43.9 30.7 43.2 205.6 46.2 119 70.8 21.2 19 9.5 10.9 6.3 16.6 2.7 1.7 5.2
1.01 0.97 0.81 0.72 0.77 0.84 0.65 0.69 0.9 1.0 0.82 0.83 0.74 0.6 0.77 0.53 0.7 0.83 0.81 0.88 0.94 0.92 0.85 1.02 1.1 0.92
ELSE_TSANG_CH022.qxd
8/21/2005
288
9:13 AM
Page 288
Review of the Studies of Radionuclide Adsorption/Desorption with Application
(3) in comparison with the relationship between apparent constants KFs and ns, there is a similar correlation between the linear (Ki) and exponential (nsi) partial Freundlich constants.
22.3 EUR SITE 22.3.1 Geological Environment The planned underground repository in the Cambrian clay, the EUR site (located 25 km away from the Radon site), is potentially suitable for disposal of radioactive waste and spent nuclear fuel (Rumynin et al., 2003). The clay formation is underlain by the Cambrian (Lomonosovsky series) aquifer and overlain by the Ordovician aquifer (Fig. 22.2). Downward leakage of groundwater through the clay from the upper to lower aquifer, and diffusion of the radionuclides, can potentially result in contamination of the subsurface environment. 22.3.2 Results of Diffusion Experiments To study barrier properties of the rock, in particular to determine the rate of radiostrontium (90Sr) release from the designed repository, diffusion/adsorption experiments have been carried out under room temperature and pressure conditions. Before testing, the clay samples were conditioned through multistep saturation by 0.1 and 3.0 N solutions of SrCl2. This allows us to assume that the composition of pore water is identical to the composition of the radioactive solutions. The basic experimental scheme consists of a through-diffusion test followed by an outdiffusion test using a diffusion cell. The core material was placed in the columnar cell, which was in direct contact with a container filled with a radioactive solution of 0.1 or 3.0 N SrCl2 (the through-diffusion test) or radionuclide-free solutions with the same concentrations of the principal components (the out-diffusion test). Along with 90Sr, the diffusive solutions contained 36Cl, which is considered to be a conservative tracer. Radionuclide concentrations were measured as functions of time in the liquid phase (in the chamber as shown in Figs. 22.10a and 22.10b). Additionally, the mass of the radionuclides taken up by the clay was controlled by means of activity measurements of the core material (solid phase) in different cross sections (the core was cut into slices after completion of each of experiments). Thus, the spatial distribution of the radionuclides remaining in the column was also measured (Fig. 22.10c). Experimental data were interpreted using analytical (Mironenko and Rumynin, 1998) and numerical solutions of the one-dimensional diffusion equation: C 2C 0, neff DM t x2
(22.6)
where neff n Kd ρ b, n the porosity, Kd the distribution coefficient, ρb the bulk density of the rock, and DM the coefficient of molecular diffusion. Table 22.6 summarizes the fitted parameters. As one can see, the values of the molecular diffusion coefficients of chloride and strontium ions are similar, and ion diffusivity tends to decrease when a solution’s ionic strength is increased. Values of the effective porosity
ELSE_TSANG_CH022.qxd
8/21/2005
9:13 AM
Page 289
22.3 Eur Site
289 (a)
C, Ci/L
5.0E-05
4.0E-05
3.0E-05
2.0E-05
1.0E-05
t, d
0.0E+00 0 0.1Cl-e
10 0.1Cl-m
3.0Cl-e
20 3.0Cl-m
30 0.1Sr-e
40
0.1Sr-m
3.0Sr-e
3.0Sr-m
(b)
C, Ci/L
4.0E-06
3.0E-06
2.0E-06
1.0E-06
0.0E+00 5
0 0.1Cl-e
0.1Cl-m
3.0Cl-e
10 3.0Cl-m
15 0.1Sr-e
0.1Sr-m
20 3.0Sr-e
t, d
3.0Sr-m
Fig. 22.10. Experimental and modeling (fitting) curves: a—through-diffusion test, Cin(t); b—outdiffusion test, Cout(t). In legend: “e” is experimental curves; “m” is fitting (modeling) curves.
controlling chloride migration are close to the total porosity of the clay (neff ≈ n0 0.5). At the same time, 90Sr is found to be a reactive component that is actively adsorbed by the rock. The Kd values depend on the solution’s ionic strength and the type of diffusion experiment. Thus, in the through-diffusion experiments, an increase in ionic strength from 0.1 to 3.0 N leads to a decrease in Kd ⬅ Kdin, constant from 4.2 to 0.7 cm3 g1. These values differ significantly from Kdout values that were obtained in the out-diffusion tests, respectively, 11.4 and 2.4 cm3 g1. Thus, the adsorption process is nonsingular.
ELSE_TSANG_CH022.qxd
8/21/2005
290
9:13 AM
Page 290
Review of the Studies of Radionuclide Adsorption/Desorption with Application 6.0E-06 C, Ci/L
(c)
4.0E-06
2.0E-06
0.0E+00
x, cm 0
0.1Cl- e
10
5 0.1Cl-m
3.0Cl-e
3.0Cl-m
15 0.1Sr-e
20 0.1Sr-m
25 3.0Sr-e
3.0Sr-m
Fig. 22.10. (continued). c—distribution of the radionuclide concentrations in porous water, C(x)
Table 22.6. Molecular diffusion and adsorption/desorption coefficients from diffusion experiments (the EUR site) Parameters 2
DM neff (m /d)* DM (m2/d) neffin neffout Kdin (cm3/g) Kdout (cm3/g)
36
90
Cl 0.1 N SrCl2 3.11 10 5.66 105 0.55 0.55 ≈0.03 ≈0.03 5
3.0 N SrCl2
Sr 0.1 N SrCl2
3.0 N SrCl2
2.10 10 4.20 105 0.5 0.5 0.0 0.0
5.56 10 7.41 105 7.5 19.5 4.19 11.38
7.25 105 4.27 105 1.7 4.5 0.71 2.38
5
4
* neff nineff
22.4 TOMSK-7 SITE 22.4.1 General Information The site selected for this study is the repository near the city of Tomsk, operated by the Siberian Chemical Plant. The site is associated with two systems of deep injection wells for low-level (LLW) and intermediate-level (ILW) radioactive waste disposal in a deep geological formation. Sand horizons are used as the collector-layers, which are isolated from the surface and shallow waters by layers of low-permeability clay (Rybalchenko, 1998). The injection-flow (discharge) rate varies from 300 (ILW) to 800 (LLW) m3 day1, and the total activity of the waste varies from 3.7 104 to 3.7 108 Bq l1 (Rybalchenko, 1998). The injection of the ILW in the deep geological formation at the site is accompanied by heat evolution. The temperature of the formation rises to more than 100ºC. For this reason,
ELSE_TSANG_CH022.qxd
8/21/2005
9:13 AM
Page 291
22.4 Tomsk-7 Site
291
the behavior of radionuclides in the subsurface environment was studied with batch laboratory experiments, which were conducted under elevated temperature (T 70°C) and pressure (P 3 MPa) conditions. The pressure corresponds to an overburden of about 300 m. The experimental results were compared with data previously obtained under room conditions (T 20°C and P 0.1 MPa). During the natural attenuation period, which is expected to start when the injection of the waste is terminated and the contaminants are assumed to be flushed out of the pores by freshwater, the desorption process will play a dominating role. Thus, an experimental study of desorption is of special interest. 22.4.2 Laboratory Series Sorption and desorption kinetics and equilibrium were measured in batch experiments conducted with a radioactive acid (pH ⬇ 3) solution of NaNO3 (10 g l–1) and SGW (Table 22.7) under different temperature (T 20 and 70°C)/pressure (P PATM and P 3 MPa) conditions. Laboratory series indexes are: Sr(Cs)-GW-GW – adsorption and desorption of 90Sr (137Cs) in the groundwater; Sr(Cs)-Na-Na – adsorption of 90Sr (137Cs) in sodium nitrate solution and desorption under the same condition (environment); Sr(Cs)-Na-GW – adsorption of 90Sr (137Cs) in sodium nitrate solution and desorption under the groundwater condition. (GW and Na mean groundwater and sodium nitrate solution.) Water samples were analyzed by atomic adsorption spectrometry for cations that might compete with radionuclides for adsorption sites (Fig. 22.11). 22.4.3 Experiments under Room Conditions The hysteresis phenomenon is not strongly exhibited during the displacement of 90Sr, which was previously adsorbed in the formation water by the SGW of the same composition (Fig. 22.12a). Adsorption and desorption of 90Sr under a NaNO3 environment (Sr-Na-Na series) or a change of the electrolyte for the SGW (Sr-Na-GW series) results in strong hysteresis anomalies (Figs. 22.12b and c). In the experimental series Sr-Na-Na and Sr-Na-GW, adsorption and multistep desorption proceeded under different hydrogeochemical conditions (Fig. 22.11), leading to hysteresis in adsorption (Figs. 22.12b and c). Indeed, the chemical equilibrium at the first desorption step is characterized by the presence of a noticeably smaller amount of displaced cations in solution, compared with the adsorption stage (Fig. 22.11). Therefore, the equilibrium condition differs from the condition occurring at the adsorption stage. The same tendency remains also at subsequent desorption steps, which are characterized by a gradual decrease of cations of alkali earth metals from the system. Although the desorption equilibrium is described by the same type of isotherm, the values of KFd and nd Freundlich parameters change from one step to the next (see Fig. 22.12 caption). Table 22.7. Chemical composition of groundwater (a major horizon for the radioactive waste disposal, the Tomsk-7 site) Anions (mg l1)
Cations (mg l1)
CO2 3
HCO3
NO3
SO42
Cl
1.49
234
0.6
2.47
6.43
2
Na
Ca
111.2
3.7
pH Mg 0.6
2
7.6–8.0
ELSE_TSANG_CH022.qxd
8/21/2005
292
9:13 AM
Page 292
Review of the Studies of Radionuclide Adsorption/Desorption with Application 1000
a
1
3 2 Fe
100
Ci (mg/L), pH
Ca
Mg 10
pH Sr Ca
pH 1
Fe Mg
0 0
1
1
10000
2 3 N, number of incubation period
4
5
b 2
Ci (mg/L), pH
1000
3
Na
Ca
100
Na
Fe
pH
10
Ca
pH Sr
1
Mg
K Fe Mg
K
0 0
1
2 3 N, number of incubation period
4
5
Fig. 22.11. A change in the major component composition of solution, which was measured in the multistep flashing of the rock by a solution of NaNO3 (a) and groundwater (b); dotted lines show the initial composition of groundwater; 1—displacing solution (NaNO3, before the adsorption step); 2—equilibrium at the adsorption step; 3—equilibrium at the desorption steps.
ELSE_TSANG_CH022.qxd
8/21/2005
9:13 AM
Page 293
22.4 Tomsk-7 Site
293 1x10 6
1x10 9
Sr-Na-Na
Sr-GW-GW
Ns, Nd, Bq cm-3
Ns, Nd, Bq cm-3
1x10 8
1x10 7
1x10 6
1x10 5
1x10
4 1 2 3 4 5
1x10 5 1 2
1x10 4 1x10 2
(a)
3
1x10 3
1x10 4
1x10 5
1x10 6
1x10 1 1x10
1x10 7
C, Bq cm-3
1x10
2
1x10
3
1x10
4
1x10
5
1x10
6
C, Bq cm-3
(b) 6
1x10
Sr-Na-GW
Ns, Nd, Bq cm-3
5
1x10
4
1x10
1 2 3 4 5
3
1x10 1 1x10
(c)
2
1x10
3
1x10
4
1x10
5
1x10
6
1x10
C, Bq cm-3
Fig. 22.12. Isotherms of adsorption and desorption of Sr-90 (double-log scale) for experimental series: (a) Sr-GW-GW (1—adsorption, fitting curve: N s 140.7 C1.03, 2—desorption, fitting curve: N d 183.5 C1.04 ); (b) Sr-Na-Na (1—adsorption: N s 24.6 C 0.83, 2—desorption, step 1: N d 83.4 C0.86, 3—sorption, step 2: N d 136.9 C1.03, 4—desorption, step 3: N d 34.4 C1.57, 5—desorption, step 4: N d 1.0 C2.26); (c) Sr-Na-GW (1—adsorption: N s 24.6 C0.83, 2—desorption, step 1: N d 598.2 C0.88, 3—desorption, step 2: N d 77.0 C1.39, 4—desorption, step 3: N d 33.0 C1.57, 5—desorption, step 4: N d 72.6 C1.36).
The sorption ability of 137Cs is higher than that of 90Sr: KFs (series Cs-Na) KFs (series Sr-Na) (see Fig. 22.13 caption). In contrast to the above results, a decrease in the saturation of the solution by cations of alkali earth metals at the desorption steps in the experimental series Cs-Na-Na in Figure 22.13 does not lead to a retention of this radionuclide in the solid phase. Very likely, equilibrium in the system is controlled exclusively by the ionic strength of the solution, which changes slightly from one desorption step to the next. A step-by-step washout of the contaminated rock with the synthesized groundwater (series Cs-Na-GW) leads to a displacement of the desorption isotherms along the concentration axis C with respect to the adsorption line (Fig. 22.13). In contrast to the previous experiments, however, the desorption isotherms remain approximately parallel to the adsorption isotherm up to the fourth desorption step: the linear coefficient KFd increases considerably,
ELSE_TSANG_CH022.qxd
294
8/21/2005
9:13 AM
Page 294
Review of the Studies of Radionuclide Adsorption/Desorption with Application 1x106
Cs-Na-Na, Cs-Na-GW
C=0 Nd= 327080 5
-3
Ns, Nd Bq cm-3
1x10
4
1x10
1 2 3 4 5 6 7
C=0 Nd=43 78 3
1x10 0 -2 1x10
-1
1x10
0
1x10
1
1x10
2
1x10
3
1x10
4
1x10
5
1x10
C, Bq cm-3
Fig. 22.13. Isotherms of Cs-137 adsorption and desorption (double-log scale): experimental series Cs-Na-Na and Cs-Na-GW: 1—adsorption (NaNO3 environment: N s 90.0 C 0.89 ); 2—desorption in NaNO3: N d 133.0 C 0.86; 3—desorption in GW: 1 step, N d 642.3 C 0.81; 4—desorption in SGW: 2 step, N d 1417.9 C 0.80; 5—desorption in SGW: 3 step, N d 1706.0 C 0.88; 6—desorption in SGW: 4 step, N d 4982.2 C 0.77; 7—desorption in GW: 5 step.
while the exponential index nd changes only slightly. Flushing the rock samples, which were contaminated at the adsorption stage, with freshwater did not result in a release of a detectable amount of 137Cs from the mineral phase on the last (fifth) step (Fig. 22.13). A kinetic dual-site model (Appendix A, Rumynin et al., 2002) satisfactorily described experiments with 90Sr in the groundwater system and 137Cs in NaNO3 solution. From this, the following conclusions can be reached: (a) in comparison with the kinetic adsorption constants, the desorption constants have lower values, so that a complete chemical equilibrium may not be reached on the desorption steps; (b) desorption from reactive sites of Type 2 (clay and hydro-micaceous minerals) is characterized by the reduced kinetic constants in comparison with the desorption from the reactive sites of Type 1 (quartz and carbonate minerals); and (c) upon a change in composition of the solution at the desorption step in the SGW, the equilibrium linear constant of adsorption on the sites of Type 2 rises sharply. 22.4.4 Experiments at Elevated Temperature and Pressure The first set of experimental series with radionuclides was carried out at 70°C and under normal atmospheric pressure (P Patm). Another set of experiments was carried out at the same temperature, but at a pressure of 3 MPa.
ELSE_TSANG_CH022.qxd
8/21/2005
9:13 AM
Page 295
22.4 Tomsk-7 Site
295
T 70°C and P Patm: At the sorption step of the experimental series with 90Sr, rather slow changes of the concentration function C (Figs. 22.14a–c) were noted. This suggests that the process is affected by a dissolution of carbonate minerals and/or a hydrolysis of alumina silicates. In the desorption kinetic curves, plots C versus t, the extremes that were noted under room conditions could not be observed. Adsorption curves obtained for 137Cs with hot and cold acid solutions (Figs. 22.14d–f) did not differ much from each other. T 70°C and P 3 MPa: Breakthrough curves obtained at the adsorption stage of the experiments with NaNO3 solution have shown a nonmonotonic nature (Fig. 22.14). During the first days of the experiments, a drop in radionuclide concentration in the NaNO3 solution was observed. At later times, however, an increase in the radionuclide content in the solution was noted. This effect is most pronounced for 90Sr: its concentration at the end of the sorption experiment (after 14 days) reaches the initial values (Figs. 22.14a–c). This means that 90Sr was adsorbed at first by the rock in large quantities; then, it was completely desorbed into the solution, so that the rock was clean of the contaminants. The calculated apparent constant of the sorption distribution is assumed to be equal to zero (i.e., a release of Sr-90 is about 100%). A release of 137Cs reaches 15–20% (Figs. 22.14d–f). The phenomenon established experimentally can be explained if we take into account that adsorption equilibrium is controlled by dissolution kinetics of minerals in the solid phase. Indeed, adsorption and desorption are sufficiently rapid processes compared with mineral dissolution. Therefore, 90Sr is actively adsorbed at first, when the solution is undersaturated by competing bivalent cations (first of all, calcium). Later, however, the concentration of these cations gradually increases, due to dissolution of carbonate minerals; this leads to a release of 90Sr from the exchange complex of the rock. The degree of solution saturation by calcium ions is proportional to the concentration (partial pressure) of carbon dioxide, whose solubility rises with a pressure increase. Therefore, at the elevated pressure, calcium concentration can reach values that indicate that practically all 90Sr has been released to the solution. At elevated temperature and pressure, the acid solution of NaNO3 can also dissolve some silicate minerals, for example, hydromica. In this case, the solution must be enriched gradually in potassium ions that compete with 137Cs for the exchange positions on the matrix. Therefore, it can be supposed that an accumulation of potassium ions in the solution leads to the release of previously adsorbed 137Cs (Figs. 22.14d–f). A specific adsorption model (Appendix C) describes the mutual interference of adsorption kinetics and dissolution of carbonate minerals resulting in a nonmonotonic behavior of the concentration curves obtained at the adsorption stage:
冤 冕
t苶
冥
C1 C1/C1i eF(i ) 1 g(x)eF(x)d x , ¯
0
(22.7)
where
冕
t苶
F (t¯ ) f(x) dx, 0
1 f(x) 苶2i C 苶2), g(x) (C 苶2i C 苶2), a (C
C 苶2 C 苶2s (C 苶2s 1)eα苶 t苶 C1 and C2 are also the concentrations of the radionuclide and competitive cation; C 1i and C 2i are the initial concentrations; C s2 is the concentration of saturation; α1 and α2 are the kinetic constants of adsorption and dissolution; C 苶2 C2/C 2i , C 苶2s C s2/C 2i , and C 苶2i KsC 2i ; Ks is the selectivity coefficient; a Vρ b/ms; V is the volume of the solution; ms is the weight of the rock sample; ρn is the bulk density; and t苶 α1t, α 苶 α2/α1.
ELSE_TSANG_CH022.qxd
8/21/2005
296
9:13 AM
Page 296
Review of the Studies of Radionuclide Adsorption/Desorption with Application (a)
3800
38000
3600
36000
Please Reduce this figure by 6% or Please resize this figure by the height of 37p10.022
T = 70°C, P = Patm
3200
C, Bq cm-3
3400 C, Bq cm-3
(b)
T = 70°C, P = Patm
T = 70°C, P = 3MPa
3000
34000 32000 30000
2800
28000
T = 70°C, P = 3MPa
Ci = 3700 Bq cm-3
Ci = 37000 Bq cm-3
2600
26000 0
4
8
0
12
4
400 380000
8
12
t, d
t, d
(d)
(c) T = 70°C, P = Patm
360000
300 T = 70°C, P = 3MPa C, Bq cm-3
C, Bq cm-3
340000 320000
200
T = 70°C, P = Patm
300000 T = 70°C, P = 3MPa
100
T = 20°C, P = Patm
280000 Ci = 370000 Bq cm-3
Ci = 370 Bq cm-3 0
260000 0
4
8
0
12
4
t, d
(e)
4000
40000
12
16
(f)
30000
2000
T = 20°C, P = Patm
1000
T = 70°C, P = 3MPa
T = 70°C, P = Patm Ci = 3700 Bq cm-3
0 0
4
8 t, d
12
16
C, Bq cm-3
C, Bq cm-3
3000
8 t, d
T = 20°C, P = Patm T = 70°C, P = 3MPa
20000
10000
T = 70°C, P = Patm Ci = 37000 Bq cm-3
0 0
4
8 t, d
12
16
Fig. 22.14. Influence of the temperature and pressure on radiostrontium adsorption kinetics: a, b, c—Sr-90 adsorption (a—Ci 3700 Bq cm–3; b—Ci 3700 Bq cm3; c—Ci 37,000 Bq cm–3); d, e, f–Cs-137 adsorption (d—Ci 370 Bq cm3; e—Ci 3700 Bq cm3 f—Ci 37,000 Bq cm3).
The experimental adsorption curves of 90Sr at the elevated temperature and pressure are satisfactorily described by the solution of Equation 22.7. An example of the fit of Equation (22.7) to the experimental data is shown in Figure 22.15. As indicated in the graph, the agreement of the experimental and modeling results occurs with quite low values of the dis1 solution kinetic constant (α 苶 0.01 0.02, α2 0.15 0.30 d ).
ELSE_TSANG_CH022.qxd
8/21/2005
9:13 AM
Page 297
22.5 Krasnoyarsk-26 Site
297
1.0
ce y e g-
Sr-90: T=70oC, P=3 MPa _ 0 a=17, C2 =90, 1=15 d-1
_ C1
0.9
0.8
1 2 0.7
3 4 5
0.6 0
4
8
12
16
t, d
Fig. 22.15. Fitting curves for the adsorption kinetics of 90Sr (1—Ci 370,000 Bq cm3; 2—Ci 37,000 Bq cm3; 3—Ci 3700 Bq cm3; 4—C 2i 0.02, α2/α1 0.01; 5—C 苶 2i 0.004, α2/α1 0.02).
22.5 KRASNOYARSK-26 SITE 22.5.1 General Information The Krasnoyarsk-26 deep-well injection site is a part of the plutonium and enriched uranium production industrial complex (plant), which is located on the eastern side of the Yenisei River, about 40–50 km northeast of the city of Krasnoyarsk (the eastern edge of the West Siberia Basin). Deep-well injection has been ongoing at the Krasnoyarsk-26 site for more than 40 years. So far, about 6 million m3 of the waste has been injected into two confined aquifers at a depth of about 200 and 400 m below the surface. The injected solutions contain short-lived beta- and gamma-emitting radionuclides. According to the accepted safety strategy, the waste must remain isolated from the surface environment for at least 1000 years. HLW and ILW are injected into the lower aquifer (Horizon I), and LLW into the upper aquifer (Horizon II). The total present-day activity of the waste stored in both horizons is estimated to be about 450 million Curies (MCi). As in the Tomsk-7 site, radioactive decay of shortlived radionuclides results in the heating of the recharged horizons. 22.5.2 Experimental Setup and Analysis of the Major Results The major purpose of the experimental work was to assess how much the rock matrix’s mineral transformations influenced its capability to retain radioactive waste components.
ELSE_TSANG_CH022.qxd
8/21/2005
298
9:13 AM
Page 298
Review of the Studies of Radionuclide Adsorption/Desorption with Application
Samples of the clayey sand were taken from Horizon I. The mineral composition of the sand consists of quartz, feldspar, biotite, muscovite, chlorite, illite, and kaolinite. The synthesized alkaline waste (representative of the predominant type of the radioactive waste injected in the subsurface) consists of NaOH: 10.5 g/L; NaNO3: 44.5 g/L; Na2CO3: 2 g/L; and pH 13. Batch experiments were performed in autoclaves under different temperature and pressure conditions (Tables 22.8 and 22.9). The solid-to-liquid phase ratio (S:L) in different experimental series was also varied, with the weight of each sample equal to 1 g. Quenched solutions and sand samples were examined to determine changes in the composition of liquid and solid phases. Radionuclide concentrations were determined by the radiometric technique. Table 22.8 lists the observed results of the change in the composition of the waste after contact with rock samples at elevated temperature and pressure. It appears that exchangeable and structural cations (composing the minerals) were leached out of the rock. The observed concentrations depend on the S:L ratio rather than the duration of the experiments. Results of the batch experiments indicate that at elevated pressure and temperature, noticeable amounts of aluminum, silica, and (to a lesser extent) potassium were leached from the rock matrix. In contrast to the acid-waste experiments (samples from the Tomsk-7 site), calcium was observed in the trace-level concentrations. A similar set of experiments was performed under room conditions (T 20°C, P 0.1 MPa). It was found that a solution’s general tendency in component accumulation remained the same; however, the leaching process intensity was observed to be much lower. Table 22.8. A change in the composition of the solution in the batch experiments under different conditions (samples from the Krasnoyarsk-26 site) Dissolved solids (mg l1)
Experimental conditions S:L
T (°C)
P (Mpa)
t (h*)
Al 2%
Si 5%
K 5%
Ca 2%
1:10 1:20 1:30 1:10
70 70 70 70
3.0 3.0 3.0 3.0
100 200 300 500
1.27 7.41 18.52 1.38
2.67 8.17 9.99 2.80
0.25 1.51 2.20 0.24
0.003 0.008 0.002 0.003
* Duration of the experiment.
Table 22.9. Adsorption characteristics from the batch experiments (S:L1:10), the Krasnoyarsk-26 site Radio- Experimental nuclides conditions
Initial Equilibrium concentration concentration (Bq l1) (Bq l1)
Percent of Distribution sorption (%) coefficient (cm3 g1)
90
1.3 106 1.3 106 1.3 106 5.3 106 5.3 106 5.3 106 1.28 106 1.28 106 1.28 106 1.1 105 1.1 105 1.1 105
85.5–90.0 95.0–95.5 95.2–95.5 92.5–93.5 96.0–96.2 96.0–96.2 95.0–96.0 98.0–99.0 99.0–99.5 95.2–95.5 98.0–99.0 99.0–99.5
Sr
137
Cs
239
Pu
241
Am
20°C, 0.1 MPa, 100 h* 70°C, 3 MPa, 100 h 70°C, 3 MPa, 500 h 20°C, 0.1 MPa, 100 h 70°C, 3 MPa, 100 h 70°C, 3 MPa, 500 h 20°C, 0.1 MPa, 100 h 70°C, 3 MPa, 100 h 70°C, 3 MPa, 500 h 20°C, 0.1 MPa, 100 h 70°C, 3 MPa, 100 h 70°C, 3 MPa, 500 h
* Duration of the experiment.
1.9 105–1.5 105 6.5 104–5.9 104 6.8 104–5.9 104 4.0 105–3.4 105 2.1 105–2.0 105 2.1 105–2.0 105 6.4 104–5.0 104 2.5 104–1.3 104 1.3 104–6.4 103 5.2 103–5.0 103 2.2 103–1.1 103 1.1 103–5.0 102
60–80 190–210 200–210 120–145 240–250 240–255 200–240 500–970 1000–1300 200–210 500–950 980–1200
ELSE_TSANG_CH022.qxd
8/21/2005
22.5 Krasnoyarsk-26 Site
9:13 AM
Page 299
299
The tested (treated) samples of the sand were also subjected to a visual, microscopic examination to estimate the alteration in the mineral composition of the rock. It was revealed that when a solid material is dissolved by interacting with hot alkaline solution, new mineral associations precipitate under room conditions (Fig. 22.16). The predominant neogenic amorphous form is aluminum hydroxide coating mica minerals (in particular, biotite); some inclusions of aluminum hydroxide may be observed within the mica cleavage. Photos (Fig. 22.16) also illustrate that the hydroxide films may cement the mica particles together. In the batch experiments, the apparent phase distribution coefficients under different experimental conditions were evaluated (Table 22.9). Table 22.9 indicates that the higher the temperature and pressure, the greater the values of the adsorption distribution coefficient (Kd). The radionuclide adsorption also depends on the chemical properties of the elements, particularly their capability for hydrolysis in the alkaline solution. In our case, actinoids (such as 239Pu and 241Am) are most susceptible to hydrolysis. Therefore, the rise of the distribution coefficient values can be explained by actinide coprecipitation with hydroxyl aluminum in the porous matrix. Coprecipitation of 239Pu with silicate minerals is also possible.
Fig. 22.16. Photos showing a rock sample after interaction of the sand with alkaline solution (T 70°C, P 3 MPa). Neogenic aluminum oxide films are in a brown color. Left and right pictures differ in the power magnification, 20 and 40, respectively.
ELSE_TSANG_CH022.qxd
8/21/2005
300
9:13 AM
Page 300
Review of the Studies of Radionuclide Adsorption/Desorption with Application
Desorption of radionuclides in the process of washing (successive desorption technique) sand samples with different solutions was studied (Myasoedov et al., 1999): (1) water and a dilute solution of ammonium acetate (1 M, pH 4.8); (2) hydrochloric acid (1 M); and (3) hot (90°C) hydrochloric and nitric acids (6.0–7.5 M). The percentage of the radionuclides removed from the rock by washing the samples with the three types of solutions can be conditionally associated with quantities in the solid phase of (1) mobile, (2) intermediate mobile, and (3) low (poorly) mobile radionuclides. Results of the experiments are presented in Figure 22.17. From the results of the selective desorption, the following conclusions can be derived: A major fraction of 137Cs is incorporated into the mineral structure, and the radionuclides are basically leached from the rock when matrix dissolution occurs; 137Cs is partly absorbed by the hydroxide metals comprising the neogenic films. The percentage of 137Cs in the ionexchange complex is negligible. Also, with an increase in temperature and rock-waste contact time, the amount of the low mobility species increases. Strontium-90 is a relatively mobile and exchangeable species. Strontium-90 is partly absorbed by the neogenic films and by carbonate minerals. The percentage of the lowmobility species is low, but increases with rising temperature. In this respect, the behavior of 90Sr is the opposite of that in acid solutions (compared with experiments with sands from the Tomsk-7 site). Americum-241 can be considered a mobile, exchangeable species; it is absorbed by neogenic films and carbonate minerals in approximately equal parts. An increase in the rock-waste contact time results in a decrease in the quantity of 241Am absorbed by the films, probably due to a transformation in structure and a change in the properties of the films. The percentage of the low mobility species does not exceed 2%. In comparison with americium, 239Pu is less mobile; it is absorbed by films, clay, and sand grains. The low mobility fraction ranges up to 10–15% and increases with rising temperature. Only the water-soluble or ion-exchangeable radionuclides can be desorbed quickly from the rock under natural attenuation when groundwater flows through the contaminated rock. Based on the represented results, we established a mobility sequence for the radionuclides under consideration, the most mobile being 90Sr, followed by 241Am, 239Pu, and 137Cs. 100% 80% 60% 40% 20% 0% 1
2
3
1
Cs
2
3
1
Sr
2
3 Pu
1
2
3
Am
Experimental conditions: 1- 20 C, P=0.1 MPa, 100 h; 2- 70 C, P=3 MPa, 100 h; 3- 70 C, P=3 MPa, 500 h
mobile
intermediate mobile
low mobile
Fig. 22.17. The selective desorption of radionuclides.
ELSE_TSANG_CH022.qxd
8/21/2005
9:13 AM
Page 301
22.6 Lake Karachai Site
301
22.6 LAKE KARACHAI SITE 22.6.1 General Description of the Migration Process A surface waste reservoir (formerly Lake Karachai) has been used over 50 years for storage of intermediate-level radioactive liquid wastes. The leakage from the reservoir has resulted in a contaminant plume traveling through the aquifer composed of fractured metavolcanic rocks. Lake Karachai is still filled with the waste that is very dense and highly radioactive. The high density is due to the salt content (mainly, sodium nitrate up to 100 g l–1). Radioactivity is caused by the presence of a mixture of various long-lived decay products among which are 90 Sr (up to 4 10–3 Ci l–1) and 137Cs (up to 9 10–3 Ci l–1). The plume boundaries mapped using nitrate and radionuclide concentrations (Drozhko and Glagolenko, 1997) are moving largely southward and northward from the reservoir at a velocity of 70–80 m yr–1. Velocities of the dissolved radionuclide constituents are noticeably lower, although ultralow concentrations of 90Sr are observed at the advancing front of the plume. Subsurface radionuclide transport at the Lake Karachai site is accompanied by a range of physical and physicochemical processes, including (Rumynin et al., 1998, 2002b) advection, mechanical dispersion, chemical diffusion, radioactive decay, chemical reaction, and adsorption. Depending on the nature of the physical and physicochemical processes, the mass transport potential of the contaminants may be either enhanced or diminished. The mass-transport potential is determined mostly by the bulk salt flux, which is characterized by the current content of the total dissolved solids and solution density. The concentration of nitrate ions, which dominates the solution and correlates with its density, can serve as a major conservative component. Meanwhile, spatial and temporary changes in concentrations of the minor species (e.g., radionuclides) do not affect the overall transport potential and can be excluded from consideration. On the other hand, the behavior of radionuclides is governed by the concentration distribution of many other dissolved components and complexing species. Therefore, predicting the fate of the radionuclides requires the use of reactive solute transport models. A simplified analysis can be based on a “Kd approach,” taking into account the variability of coefficients of equilibrium distribution. 22.6.2 Samples and Experimental Setup A set of adsorption experiments has been carried out for studying 90Sr sorption onto fracture surfaces (Rumynin et al., 1998). Three major aspects of the problem have been evaluated: (1) types of adsorption/desorption isotherms; (2) influence of the NaNO3 concentration on the coefficient of equilibrium distribution (Ka) at adsorption and desorption stages; and (3) the irreversibility of the adsorption process. Core materials were taken from three different boreholes in uncontaminated areas, within separate depth intervals (Table 22.10). The samples are represented by welded tuffs, scoriaceous lavas, tuffs, and tuff breccias moderately metamorphosed into greenschist facies (Table 22.10). Core samples were cut into disc-shaped wafers (up to 90 mm in diameter and 2–4 mm thick). Sets of discs were placed into glass containers filled with NaNO3 solutions containing 90Sr. Three basic concentrations (Cs) of NaNO3 were selected for creating the hydrogeochemical environment related to the site under consideration: 1, 15, and 70 g l–1. This gradation corresponds to the typical hydrogeochemical zones of groundwater contamination. For the desorption step, the solutions were replaced by background (uncontaminated) water.
ELSE_TSANG_CH022.qxd
8/21/2005
302
9:13 AM
Page 302
Review of the Studies of Radionuclide Adsorption/Desorption with Application
Table 22.10. Characteristics of studied rock samples (from the Lake Karachai site) and calculated values of the equilibrium distribution coefficient Rock characteristic
Number Interval Number of m of boreholes experiments
Tuff-and-lava, tuff 1–96 Tuff-and-lava Lava breccias Tuff-and-lava
1–96 8002 1–96
Tuff 1–96 Lava breccias 211A
1 98–100 2 3 16 4 690 5 6 71 7 8 9 26 10 11 12 13 — 14
NaNO3 content (g l1)
Adsorption distribution coefficient Kas (cm)
Desorption distribution coefficient Kad (cm)
1.0 15.0 70.0 1.0 1.0 1.0 15.0 70.0 1.0 15.0 70.0 1.0 15.0 70.0
0.57 0.21* 0.18* 1.89 0.57 0.31 0.05 0.04 3.49 0.27 0.04 0.20 0.04 0.03
1.54 1.0 11.7 12.8 1.85 2.9 4.1 6.6 436 99 69 2.0 9.1 9.1
* Desorption under constant NaNO3 content.
After a special series of experiments, it was established that a time period of about 100 h is sufficient to attain chemical equilibrium between liquid and solid phases during adsorption and desorption stages. 22.6.3 Adsorption Parameters Figure 22.18 illustrates an example of adsorption isotherms obtained for samples taken from one of the boreholes. It has been observed that the equilibrium relationship between 90 Sr concentrations on the surfaces of the discs and in liquid phases are nearly linear for Cs 1 and 15 g l1. An increase in NaNO3 content in the solution leads to a shift of adsorption in a nonlinear field (ns 1). Unfortunately, experimental data for a comprehensive isotherm-based analysis of the adsorption process are rather restricted; therefore, for further analysis, data on the values of the apparent phase distribution coefficient Ka (Table 22.10) will be mostly used. The values of Ka depend strongly on the concentration of sodium nitrate: the more dilute the solution, the greater the value of Ka. In the majority of the experiments with highly concentrated solutions (70 g l1), values of Ka were found to be less than 0.04 cm. In the diluted solutions (1 g l1), values of Ka are at least 5–10 times greater. The relationship between Ka and NaNO3 content (Cs) can be approximated by the following equation: Ka Ka1eγ Cs Ka2,
(22.8)
where the sum of Ka1 and Ka2 corresponds to the coefficients of adsorption in diluted solutions, and Ka2 in concentrated solutions. The average value of γ is 0.2 g l1 (Fig. 22.19).
ELSE_TSANG_CH022.qxd
8/21/2005
9:13 AM
Page 303
22.6 Lake Karachai Site
303
1.0E+001
N.10-3, Bq cm-2
1.0E+000
1.0E-001
N=0.90 C0.98
N=0.19 C1.30
1.0E-002 N=0.38 C1.05 1.0E-003
Adsorption M = 1 g/L Adsorption M = 15 g/L Adsorption M = 70 g/L Desorption points
1.0E-004 1.0E-003
1.0E-002
1.0E+000 1.0E-001 C.10-3, Bq cm-3
1.0E+001
Fig. 22.18. Adsorption isotherms (well #8002, depth 938 m) and desorption points. 10.00
Ka, cm
1.00
(1,2,3) = 0.18 Lg-1
0.10
(9,10,11) = 0.18 Lg-1 (6,7,8) = 0.21 Lg-1
(12,13,14) = 0.21 Lg-1 0.01 0
20
40 NaNo3, gL-1
60
80
Fig. 22.19. Influence of NaNO3 content on the distribution coefficient (theoretical curve calculations are based on Equation 22.8), numbers of samples (Table 22.8) are shown in brackets.
ELSE_TSANG_CH022.qxd
304
8/21/2005
9:13 AM
Page 304
Review of the Studies of Radionuclide Adsorption/Desorption with Application
22.6.4 Hysteresis in Adsorption The phenomenon was studied in the process of step-by-step desorption of previously adsorbed radiostrontium; at the end of each desorption step, the radioactive solution was replaced by radioactivity-free water. Measured concentration points at the adsorption and two-step desorption stages of the laboratory experiments for two sets of experiments are depicted in Figure 22.18. Analysis of the experimental results shows that all desorption concentration points are located above the adsorption isotherms. Furthermore, it was established that the desorption coefficients Kad (calculated according to the traditional balance approach; Vandergraaf and Abry, 1982; Wels et al., 1997) significantly exceed adsorption coefficients Ka (Table 10): the greater the concentration of NaNO3 in the solution at the adsorption stage, the greater the calculated ratio Kad/Ka. This means that desorption of radionuclides initially adsorbed in the highly concentrated solutions is limited. Results of multistep desorption experiments showing differences between Ka and Kad can be explained in light of the previous analysis of experiments with unconditioned sands from the Tomsk-7 site. Thus, the variability of the hydrogeochemical environment at different stages of the tests is responsible for the nonsingular process.
22.7 CONCLUSIONS An analysis of the experiments with sand, clay, and metavolcanic rock core samples indicates that the adsorption isotherms for 90Sr and 137Cs are of the Freundlich type. All described batch and diffusion experiments revealed hysteresis in radionuclide adsorption. In experiments that proceeded under geochemical conditions that remained constant at adsorption and desorption stages, the hysteresis can be described within the framework of a dualsite model, which takes into account the irreversibility of the radioactive solution’s interaction with one of the components of the mineral phase. Alterations in the chemical composition of the solution and the secondary mineral phase of the rock transformation might also lead to hysteresis in adsorption. Evidence for the latter phenomenon was obtained when we observed and described an anomalous behavior of adsorption kinetic curves in experiments with sand samples from the Tomsk-7 site, under elevated temperature and pressure conditions. A rapid drop in radionuclide concentration in solution, which is caused by its sorptive uptake, is modified by a gradual increase in the corresponding concentrations. To explain this phenomenon qualitatively, one can take into account the accompanying dissolution of carbonate minerals and hydrolysis of aluminum silicates, which causes saturation of the solution by calcium, magnesium, stable strontium, and potassium ions. Experiments showed that temperature noticeably affects the intensity of radionuclide adsorption. Depending on the solution composition, heating the system leads either to a decrease in the rock’s adsorptive capacity (in experiments with acid waste, such as the Tomsk7 site) or increase (in experiments with alkaline waste, such as the Krasnoyarsk-26 site). Experimental results are important for the subsequent analysis of radionuclide migration. Thus, the setup of the experiments with samples from Tomsk-7 and Lake Karachai sites can be associated with simulation (“static” modeling) of the groundwater system response to the acid brine inflow into the aquifer. The behavior of the principal solution components in the experiments with unconditioned rock samples is similar to what is often observed at actual
ELSE_TSANG_CH022.qxd
8/21/2005
9:13 AM
Page 305
22.7 Conclusions
305
contaminated sites, when the advancing front of an inflowing (waste) solution displaces groundwater. In particular, such system behavior was monitored at the Lake Karachai site (Fig. 22.20a) as well as at a number of sites in Western Urals, Russia, where subsurface reservoirs of limestone and sandstone formations are contaminated by chloride brine (Fig. 22.20b). Saturation of the solution with dissolution products of the rock matrix’s soluble minerals and with ion-exchange reactions might be so high that radionuclide adsorption is essentially suppressed. In particular, under such conditions, Sr-90 begins to behave similarly to a chemically nonreactive component, and the migration ability of the strontium grows sharply. This conclusion is related to one of the most acute problems in environmental hydrogeology, namely, the “fast transport” of Sr-90 in subsurface environments. In contrast, the migration of hot alkaline solutions can be accompanied by the coprecipitation of radionuclides with hydroxides of aluminum and silicate minerals in the mobile geochemical and heat barriers, as shown in the analysis of laboratory experiments with Krasnoyarsk-26 sand samples. Spatial variability of the adsorption properties of the studied sand material (the Radon site) is significant. Thus, the Freundlich linear adsorption constant varies within a range of two magnitudes. Such variations dramatically exceed the variability of the hydraulic conductivity and some pore-scale structural properties of the sediment (such as the uniformity coefficient, the fines content, and the grain-size variance). The studied adsorption constants (KFs, ns, and Kd) are correlated neither with the groundwater flow nor with physical characteristics of the porous media. Variations of the secondary mineral species’ content probably a Well #41/77
63/68 10/68 3/68 9/68 176 50/79
209
900 Na+ 800 700
C, mg-eq L-1
600 NO3 500 400 300
Ca2+
200 Mg2+ 100 K+
0 0
1000
2000 X, m
3000
4000
Fig. 22.20. Principal component distribution along flowpaths at two brine-contaminated sites: a—Lake Karachai site.
ELSE_TSANG_CH022.qxd
8/21/2005
306
9:13 AM
Page 306
Review of the Studies of Radionuclide Adsorption/Desorption with Application b Well #125 122 800
121
127
124
181
Ci, mg-eq L-1
600
(Na+):5 (Cl-):5 400
Ca2+ 200 Mg2+
(K+):5 0 0
1000
2000
3000
4000
5000
X, m
Fig. 22.20. (continued). b-Berezniki 41 site.
influenced variations of the sand material’s chemical properties more pronouncedly than characteristics of the grain-size distribution and permeability of the medium. The results of these experiments could affect the direction of the theoretical study of subsurface radionuclide migration, modeling activity, and numerical code developments. ACKNOWLEDGMENTS This work was partly funded by the Russian Foundation for Basic Research (Grant No. 03-05-64231-a), the Swiss National Science Foundation (Grant No.7SUPJ062261), ISTC (Project No. 1565), and the Presidium of the Russian Academy of Sciences (Program “Natural Catastrophes”). REFERENCES Charbeneau, R.J., 1982. Calculation of pollutant removal during restoration with adsorption and ion exchange. Water Resour. Res., 18(4): 1117–1125. Deutsch, C. and Journel, A., 1992. GSLIB, Geostatistical software library and user’s guide, Oxford University Press, New York, Oxford. Drozhko, E.G. and Glagolenko, Y.U., 1997. Environmental problems at the Mayak Site. Joint Russian-American Hydrogeology Seminar, July 8–9, 1997, PUB-804, LBNL, Berkeley, CA, pp. 4–14.
ELSE_TSANG_CH022.qxd
References
8/21/2005
9:13 AM
Page 307
307
Golubev, V.S., 1981. The dynamics of the geochemical processes, Nedra, Moscow (in Russian). Jackson, R.E. and Inch, K.J., 1989. The in-situ adsorption of 90Sr in a sand aquifer at the Chalk River Nuclear Laboratories. J. Contamin. Hydrol., 4: 27–50. Mironenko, V.A. and Rumynin, V.G., 1998. Problems of Environmental Hydrogeology. MGGU Publ., Moscow, Vol. 1 (in Russian). Myasoedov, B.F., Novikov, A.P. and Pavlotskaya, F.I., 1996. Problems of analysis of different nature components with respect to detection of radionuclide concentrations and migration forms. J. Anal. Chem., 51(1): 124–130 (in Russian). Pickens, J.F., Jackson, R.E., Inch, K.J. and Merritt, W.F., 1981. Measurement of distribution coefficient using a radial injection dual-tracer test. Water Resour. Res., 17(3): 529–544. Rumynin, V.G., Konosavsky, P.K. and Hoehn, E., 2002a. Batch laboratory, analytical, and modeling study of subsurface transport of irreversibly adsorbing Sr-90. Proceedings of the 4th International Conference on Calibration and Reliability in Groundwater Modelling (ModelCare’02), Prague, Czech, June 17–20, 2002, pp. 429–432. Rumynin, V.G., Mironenko, V.A., Sindalovsky, L.N., Boronina, A.V., Konosavsky, P.K. and Pozdniakov, S.P., 1998. Evaluation of conceptual, mathematical, and physical-and-chemical models for describing subsurface radionuclide transport at the Lake Karachai waste disposal site. Lawrence Berkeley National Laboratory Report, LBNL-41974, Berkeley, CA. Rumynin, V.G., Pankina, E.B. and Yakushev, M.F., 2003. Assessment of the Actual and Potential Impact on Groundwater Quality Caused by the Northwestern Center for Atomic Energy, St. Petersburg State University Publ., St. Petersburg (in Russian). Rumynin, V.G., Sindalovskiy, L.N., Konosavsky, P.K., Boronina, A.V., Gallo, C. and Willem, J.Z., 2002b. Study of groundwater contamination by radioactive brine (Lake Karachai case). Environ. Geol., 42(2–3): 187–198. Rybalchenko, A.I., 1998. Deep-well injection of liquid radioactive waste in Russia: Present situation. In: M.J. Stenhous and V.I. Kirko (Eds), Defense Nuclear Waste Disposal in Russia: International Perspective, NATO ASI Series. Kluwer Academic Publishers, Disarmament Technologies, 18: 199–217. Vandergraaf, T.T. and Abry, D.R.M., 1982. Radionuclide sorption in drill core material from the Canadian Shield. Nucl. Technol., 57: 399–412. Wels, C.S.L., Smith, L. and Vandergraaf, T.T., 1997. Influence of specific surface area on transport of sorbing solutes in fractures: An experimental analysis. Water Resour. Res., 32(7): 1943–1954.
ELSE_TSANG_CH022.qxd
8/21/2005
308
9:13 AM
Page 308
Review of the Studies of Radionuclide Adsorption/Desorption with Application
APPENDIX A. A NON-EQUILIBRIUM MODEL OF DUAL-SITE, ONE-COMPONENT ADSORPTION If the adsorption process is under the conditions of a local equilibrium and obeys the Freundlich isotherm, then the adsorption kinetics of the two types of reactive sites can be described as follows: dN s f 1 α 1s (K sF1C n1 N1), dt dN2 s (1 f ) α s2(K sF2C n 2 N2), dt
(A1)
where C and N1, N2 are the concentrations of the radionuclide in solution and on the rock (on each of the reactive sites); K sF1 and K sF2 are the partial linear Freundlich constants; n1s and ns2 are the partial exponential constants; α 1s and α s2 are the kinetic adsorption constants; and f is the fraction of site 1. The corresponding equations for the desorption are as follows: dN d f 1 α 1d(K dF1C n 1 N1) dt dN2 d (1 f ) α d2 (K dF2C n2 N2). dt
(A2)
Equation (A3) describes the mass balance in a batch experiment: dMT 0, dt
M T VC N(ms/ρ b),
(A3)
where V is the volume of the solution, ms the weight of the rock sample, ρ the bulk density, b and N the total radionuclide concentration in the solid phase: N fN1 (1 f )N2.
(A4)
The system—Equations (A1)–(A4)—is closed when the initial conditions (t0) are assumed to be as follows: N1 N2 0, N1 Ni1,
C Ci N1 Ni2,
(at the adsorption stage) C0
(at the desorption stage),
(A5)
where Ci is the initial concentration of the adsorbing component in the solution; and Ni1 and Ni2 are the initial concentrations of the component on the sorption sites upon substitution of the solution. For solving the system of differential equations (A1)–(A3), and (A5), a numerical algorithm with finite differences was developed. Having varied by the kinetic constants, one can s simulate/model different mass-exchange regimes. So, assuming α 1,2 , α d1 → ∞, α d2, → 0 we arrive at an equilibrium model described above—Equation (2).
ELSE_TSANG_CH022.qxd
8/21/2005
9:13 AM
Page 309
Appendix B
309
APPENDIX B. SORPTION COEFFICIENTS (KFS, n S, AND Kd) FROM BATCH SORPTION EXPERIMENTS WITH LOMONOSOVSKY SAND (THE RADON SITE) Borehole interval
KFs
ns
R
Kd
Borehole interval
KFs
ns
R
Kd
1-1 1-2 2-1 2-2 3-1 3-2 4-1 4-2 5-1 5-2 6-1 6-2 7-1 7-2 8-1 8-2 9-1 9-2^ 10-1 10-2 11-1 11-2 12-1 12-2 13-1 13-2 14-1^ 14-2 15-1 15-2 16-1 16-2 17-1 17-2^ 18-1 18-2 19-1* 19-2
22.7 17.3 43.0 63.9 29.7 34.8 29.2 51.8 52.9 211.6 54.8 19.1 16.3 23.1 10.4 35.1 25.1 9.5 43.9 30.7 43.2 73.8 205.6 144.6 46.2 125.5 54.7 119 70.8 81 21.2 19.0 34.4 6.6 41.2 94.7 5.5 14.3
1.01 0.97 0.81 0.72 0.77 0.84 0.79 0.65 0.75 0.38 0.69 0.9 0.89 0.83 1.0 0.78 0.81 0.96 0.82 0.83 0.74 0.65 0.6 0.59 0.77 0.59 0.67 0.53 0.7 0.64 0.83 0.81 0.82 0.92 0.79 0.57 0.95 0.88
0.999775 0.998466 0.999916 0.999920 0.999905 0.999427 0.962572 0.984204 0.985277 0.984580 0.991129 0.996739 1.000000 0.984402 0.995325 0.963013 0.983422 0.945259 0.999653 0.998946 0.999513 0.997197 0.995770 0.998441 0.987573 0.991791 0.952407 0.999662 0.990189 1.000000 0.985985 0.991226 0.999627 0.951245 0.999844 1.000000 0.937561 0.984363
24.3 14.0 11.5 9.2 6.0 11.5 6.8 4.6 9.4 2.9 6.4 9.5 7.6 7.1 10.4 7.6 6.7 7.2 12.6 9.4 7.1 6.5 12.9 8.5 9.4 7.3 5.5 4.6 8.9 6.7 6.5 1.7 9.9 2.8 9.6 4.8 3.8 6.2
20-1* 20-2* 21-1 21-2* 22-1* 22-2 23-1 23-2 24-1 24-2^ 25-1 25-2^ 26-1* 26-2^ 27-1 27-2* 28-1 28-2 29-1* 29-2 30-1^ 30-2 31-1 31-2 32-1* 32-2 33-1 33-2 34-1 34-2 35-1 35-2* 36-1 36-2* 37-1* 37-2*
29 28.7 16.2 3.8 103 9.5 3.9 10.9 18.4 222 6.3 7.6 88.9 19.4 11.5 200 85.2 16.2 70.3 16.6 34.5 28.7 29.3 20.9 4.6 12 3.2 ND 3.5 2.7 1.7 4.9 5.2 17.5 7.7 3.6
0.8 0.77 0.73 0.91 0.45 0.88 0.94 0.94 0.78 0.31 0.92 0.87 0.48 0.78 0.77 0.29 0.59 0.8 0.55 0.85 0.68 0.75 0.76 0.77 0.93 0.92 1.06 ND 1.02 1.02 1.1 0.91 0.92 0.66 0.92 0.93
0.818613 0.765290 0.996520 0.904042 0.844161 0.999925 0.983184 0.999251 0.999640 0.959560 0.998418 0.941480 0.684417 0.946164 0.979953 0.556008 0.974415 0.999385 0.869456 0.999231 0.943679 0.999325 0.999455 1.000000 0.804627 0.979874 0.972187 ND 0.983600 0.999043 0.994962 0.920783 0.996663 0.885587 0.858407 0.788602
7.2 5.8 2.5 2.0 2.3 4.1 1.2 7.2 4.0 1.8 3.6 3.0 2.4 4.2 2.3 1.4 5.0 4.0 3.1 5.8 3.7 5.1 5.5 4.2 2.8 6.9 4.8 ND 4.0 3.1 3.3 2.6 2.9 1.6 4.4 2.2
Notes: KFs is the linear Freundlich constant, (cm3/Bq)n 1; ns is the exponential Freundlich constant; R is the coefficient of determination; R2; Kd is the effective distribution coefficient [dimensionless, Equation (22.3) for C 10,000 Bq cm–3]; interval number: “1” stands for upper (z 0.5) and “2” stands for lower (z 1.0 m). Comments: ^—isotherm slightly departs from the Freundlich type (0.94 R 0.96). *—isotherm departs (very appreciably) from the Freundlich type (R 0.94). ND—No data. s
ELSE_TSANG_CH022.qxd
8/21/2005
310
9:13 AM
Page 310
Review of the Studies of Radionuclide Adsorption/Desorption with Application
APPENDIX C. A KINETIC MODEL OF ADSORPTION WITH CONCOMITANT MINERAL DISSOLUTION In accordance with the law of mass action, the ion-exchange equilibrium in the two-component system is determined by Equation (C1) (Charbeneau, 1982):
冢 冣冢 冣
N K12 1 C1
z2
C2 z , N2
(C1)
where C1 and C2 are the concentrations of the displacing cation of the radioactive metal (1) and of the displaced cation of stable component in the solution (2), respectively; N1 and N2 are the concentrations of radionuclide and stable components on solid surfaces, respectively; K12 is the selectivity coefficient for the cation-exchange reaction between 1 and 2; and z1 and z2 are the corresponding charges of cations. Assuming Qv N1 + N2 and N1 N2 (i.e., considering that the concentration of radionuclides on the solid phase is significantly smaller than the concentration of the displaced cation), we can transform Equation (C1) to an isothermal form [Equation (C2)]: C1 , N1 KsC2z苶
(C2)
where z苶 z1/z2, Ks (K12Qvz1)1/z2. In this case, the equation describing the adsorption kinetics can be written in a form [Equation C3)]: dN1 α1(C1 KsC2z苶N1), dt
(C3)
where α1 is the kinetic constant. A mass balance equation for the construction of a model for batch experiment interpretation is as follows [Equation (C4)]: dM T 0, dt
M T VC1 N1(ms/ρb)
(C4)
or dC1 dN1 a 0, dt dt
(C5)
where V is the volume of the solution, ms the weight of the rock sample, and ρb the bulk density, a Vρ b /ms. The radionuclide concentration on the solid phase (N1) for two phases of the experiment (adsorption and desorption) can be presented as follows: N1
a(C10 C1) for adsorption N0 aC1 for desorption
冦
(C6) (C7)
where C10 is the initial (before the sorption step) radionuclide concentration in the solution; N0 a(C10 C *1) is the concentration of the radionuclide on the rock before desorption; and C *1 is the concentration measured at the end of the adsorption step.
ELSE_TSANG_CH022.qxd
8/21/2005
9:13 AM
Page 311
Appendix C
311
By inserting (C5), (C6), and (C7) into kinetic equation (C3), we obtain the following equations: dC1 dt
冦
冤 冢 冣 冥 N 1 α 冤K C 冢 K C 冣C 冥 a a
1 α1 KsC z2苶 C T1 KsCz苶2 C1 a 1
s
0
z苶 2
s
z苶 2
1
for adsorption,
(C8)
for desorption.
(C9)
Considering a dissolution reaction that proceeds in parallel to the sorption reaction and that is presented by a first-order rate equation (Golubev, 1981): C2 C2H (C2H C 02)eα2t,
(C10)
where C20 is the initial concentration of the cation in the solution, C2H the concentration at saturation, and α2 the kinetic dissolution constant of the mineral. Initial conditions (t 0) for the proposed experimental setup are: N1 0,
C1 C10,
N1 N0,
C1 0,
C2 C20
for the adsorption stage
C2
for the adsorption stage
C20
(C11)
Equations (C8) and (C9) can be rewritten in a generalized dimensionless form as dC 苶 1 1 (C 苶20 C 苶2)z苶 (C 苶 02C 苶2)z苶 C 苶1 , dt苶 a
冤
冢
冣 冥
(C12)
where for the adsorptive process, C1 C1/C10, and for the desorption process, C1 aC1/N0; also C2 C2H 苶2 苶 C2H (C 苶2H 1)eα苶 t苶 C 苶2 C 苶2H , C 0, C2 C20
冢
冢
冣
冣
1 C 苶20 Ksz2/z1C20 Ksz2/z1 , N0 a(C01 C *1), 苶t α1t, 1/z1 K12 Qv
α 苶 α2/α1.
Thus, the problem can be reduced to the solution of the following ordinary differential equation of the first order: y f( 苶t )y g( t苶 ),
(C13)
where y ⬅ y( 苶t ) 苶 C1, f( 苶t ) 1/a tial conditions (C14):
苶 N1 0 苶 C1 1 苶 C2 1 N 苶1 1 苶 C1 0 苶 C2 1
(C 苶20 C 苶2)z苶,
and g( 苶t )
(N 苶1 N1/N0).
(C 苶20 C 苶2) z苶,
with the transformed ini(C14)
The general solution of Equation (C12) is as follows:
冤 冕 冕 g(x)e
冥
t苶
C1 eF(t苶 ) 1 g(x)eF(x) dx , 苶 0
苶1 eF(t苶 ) C
t苶
F(x)
(C15)
dx
0
for the adsorption (C1 C1/C 01) and desorption (C 苶1 aC1/N0) steps of the process, respectively.
This page intentionally left blank
312
ELSE_TSANG_CH023.qxd
8/21/2005
9:21 AM
Page 313
Chapter 23
CHEMICAL INTERACTIONS BETWEEN WASTE FLUID, FORMATION WATER, AND HOST ROCK DURING DEEP-WELL INJECTION N.F. Spycher a and R.G. Larkinb a
Lawrence Berkeley National Laboratory, Berkeley, CA, USA R. G. Larkin Consulting, Austin, TEX, USA
b
23.1 INTRODUCTION A new deep-disposal well (NDW-1) was drilled in 1994 within 200 m of an injection well that had been in operation for 12 years (ODW-1). One of the objectives of the new drilling activities was to assess whether waste products migrated from the nearby injection well, ODW-1, to the location of the new well, NDW-1. Another objective was to evaluate the fate of injected waste products and their effects on formation water and minerals. These data would prove useful for the planning of future deep-well injection in the area. While drilling NDW-1, fluid samples were collected at two depth intervals correlating with the injection interval at ODW-1 (approximately 2 km below the ground surface). From previous modeling analyses, it was anticipated that the fluid samples recovered at NDW-1 would be representative of waste fluids injected at ODW-1. However, after analysis, it was found that the recovered fluid did not have the main chemical characteristics of the injected waste, although some waste degradation products were detected in small concentrations. To evaluate the origin of the fluid from NDW-1, a geochemical investigation was carried out in two phases. First, we compared chemical analyses of formation waters (prior to waste injection) with analyses of fluid recovered from NDW-1 (before injection started in this well), and with analyses of waste injected at nearby ODW-1. We then simulated the chemical interaction between injected waste fluid and the formation water and minerals, to assess whether precipitation or dissolution of mineral phases would affect the chemical composition of fluid recovered at NDW-1. Focus was given to inorganic compounds, mostly sulfate and ammonia, which were the main components of waste injected at ODW-1. 23.2 CHEMICAL CHARACTERIZATION OF FORMATION WATERS PRIOR TO WASTE INJECTION NDW-1 was completed in the Oakville Formation (Texas Gulf Coastal Plain), which consists mainly of poorly consolidated fine-grained sand with some shale intervals. Table 23.1 summarizes the general chemical composition of water samples from NDW-1 and other nearby wells in the Oakville Formation prior to waste injection (i.e., formation waters). Table 23.2 summarizes the composition of waste injected at ODW-1. In addition to common constituents shown in Table 23.1, one of the samples from NDW-1 was found to contain a low concentration of a degradation product of organic constituents injected into ODW-1. For
7.0 1.095 1.094 162818 86200 — — 110 134 — — — — 49000 — 3200 890 1.5 0.86 13 — — 22 — ⫺1.2
6800 1977 7.25 1.096 1.095 156074 95666 — — — 153 — — — — 54908 † 3966 1355 — — 15 — — 88 — 0.2
5952 1961 7.31 1.095 1.094 155234 95039 — — — 193 — — — — 54836 † 3878 1276 — — 16 — — 82 — 0.2
6062 1961
ODW-2
7.42 1.095 1.094 155482 95087 — — — 238 — — — — 55504 † 3530 1163 — — 27 — — 48 — 0.2
6292 1961 6.6 1.0952 1.094 149100 79300 19 ⬍1 87 53 — 8.41 16 0.23 45800 209 4610 1370 43 1.27 30 1.9 7.6 52 236 2.6
~4500 1981
ODW-3
6.8 1.1 1.099 213201 101777 235 — — 230 — — — — 51000 450 3500 1100 3 — — — — — — ⫺7.0
NA 1993
ODW-4
6.7 1.099 1.098 143961 87900 43 — 108 132 — — — — 52136 — 2920 830 10 — — — — — — 0.2
5858 1975
ODW-5
7.8 1.026 1.025 34580 19000 2700 — — 142 67 1.3 — 0.67 10500 390 410 1350 0.003 0.002 6.4 0.001 4.5 0.02 8 0.4
Seawater
314
8.2 1.0735 1.0726 107257 66499 633 ⬍1 858 523 ⬍11 6 17 37 28985 4047 2254 644 14 ⬍2 ⬍459 24 — — — ⫺12
7089 1994
ODW-1
8/21/2005 9:21 AM
(**) Total carbonate concentration assumed equal to total alkalinity as HCO3⫺ unless a different value was specifically reported. (†) The concentration of potassium is included in the reported sodium concentrations, and not reported independently. (—) Not analyzed or reported. (⬍) Less than concentration shown, which is the detection limit.
7.5 1.0686 1.0677 105699 62993 683 23 246 150 ⬍11 9 22 0.67 26715 3623 1923 620 ⬍11 ⬍2 ⬍457 ⬍21 — — — ⫺12
Parameter pH (25°C assumed) Specific gravity (15–25°C) Density Total dissolved solids Chloride Sulfate Sulfide (as HS) Alkalinity (total, as CaCO3) Total carbonates (as HCO3⫺)** Bromide Fluoride Ammonia (as N) Nitrate/nitrite (as N) Sodium Potassium Calcium Magnesium Iron Manganese Silica (SiO2) Aluminum Boron Barium Strontium Charge imbalance
Units pH None g/cm3 mg/L mg/L mg/L mg/L mg/L mg/L mg/L mg/L mg/L mg/L mg/L mg/L mg/L mg/L mg/L mg/L mg/L mg/L mg/L mg/L mg/L %
6860 1994
NDW-1
Sampling depth (ft) Sampling year
Well number
Table 23.1. Analyses of formation waters from well NDW-1 and other wells in the Oakville Formation. These data were used for Figures 23.1–23.3. Data compiled by R.G. Larkin Consulting. Seawater data from Hem (1989)
ELSE_TSANG_CH023.qxd Page 314
Chemical Interactions between Waste Fluid, Formation Water, and Host Rock
ELSE_TSANG_CH023.qxd
8/21/2005
9:21 AM
Page 315
23.2 Chemical Characterization of Formation Waters Prior to Waste Injection
315
Table 23.2. Composite waste stream analysis (injectate in ODW-1) Units
Value
General parameters Specific gravity (25°C) pH (25°C) Total dissolved solids Sulfate Chloride Ammonia (as total N) Total carbonate (as HCO3⫺)** Sodium Potassium Calcium Magnesium Iron Manganese Silica (as SiO2) Chemical oxygen demand Total organic carbon
none pH units mg/L mg/L mg/L mg/L mg/L mg/L mg/L mg/L mg/L mg/L mg/L mg/L mg/L mg/L
1.04 7.6 59000 32000 3200 11000 18 2110 31 32 15 2.93 0.07 11 20060 13880
Organic species Organic acids Cyanide (total) Methanol Other organics
mg/L mg/L mg/L mg/L
3000 60 800 900
** Total carbonate concentration was calculated by assuming equilibration of the waste fluid (excluding organics) with atmospheric carbon dioxide (PCO2 ⫽ 10 ⫺3.5 bar).
comparison, the composition of seawater (Hem, 1989) is also included in this table. Correlation plots (Fig. 23.1), Piper diagrams (Fig. 23.2), and Stiff diagrams (Fig. 23.3) were used with these data to illustrate compositional differences and possible relationships between samples. The waters from NDW-1 and ODW-4 have a significant charge imbalance (12% and 7%, respectively) (Table 23.1). By correcting their sodium concentrations to bring charge balance, these waters yield sodium-to-chloride ratios similar to those in other formation waters (Fig. 23.1). Compared with the samples at other wells, the water from NDW-1 appears diluted, but with otherwise similar proportions of major constituents (sodium, chloride, calcium, and magnesium) (Figs. 23.2 and 23.3). However, the water from NDW-1 exhibits an increased content of sulfate, potassium, and possibly bicarbonate compared with other formation waters for which these data are available (Table 23.1). This difference is not visible on the Piper and Stiff diagrams because concentrations of sulfate, potassium, and bicarbonate constitute a very small fraction of the total dissolved constituents in these waters. Note that potassium and sulfate data were not available for all samples, thus leaving some margin of uncertainty regarding the concentration trends of these constituents. Besides these differences, all formation water samples, including those from NDW-1, have similar chemical compositions. The overall dilution of chloride and major cations, and apparently higher sulfate concentrations at NDW-1, could be the result of mixing with waste. However, if all dilution resulted from waste, sulfate and ammonia concentrations at NDW-1 would be higher than those observed. Also, the elevated potassium concentration at NDW-1 cannot be explained by
ELSE_TSANG_CH023.qxd
316
8/21/2005
9:21 AM
Page 316
Chemical Interactions between Waste Fluid, Formation Water, and Host Rock
Fig. 23.1. Correlation between sodium and chloride concentrations in Oakville Formation waters (prior to injection), fluid from NDW-1 after injection in nearby well (Table 23.1), liquid waste, and seawater. The dashed line represents waters with identical sodium-to-chloride ratios but various degrees of evaporative concentration (or dilution) with respect to each other. Arrows show corrections to sodium concentrations needed to bring NDW-1 and OWD-4 analyses in charge balance. After such correction, these waters exhibit a sodium-to-chloride ratio similar to that in other formation waters, supporting a similar origin. Waters from NDW-1 are the most diluted compared with other formation waters.
mixing with waste. It could result from exchange or dissolution reactions involving minerals in the formation, such as potassium-bearing clays, which are predominant in shale beds observed in core near the sampling intervals at NDW-1. It is possible that the higher carbonate concentration at NDW-1, compared with ODW-1, reflects some degradation of organic waste constituents. Organic carbon oxidation with sulfate reduction has been reported in the literature (e.g., Kelly and Matisoff, 1985), and we have personally observed trends of sulfate concentration decrease together with carbonate concentration (alkalinity) increase beneath landfills where groundwater was impacted by leachate and/or landfill gas. The oxidation of organic waste constituents by sulfate reduction would result in the formation of sulfide, which was reported in one of the NDW-1 samples. However, more data would be needed to confirm this hypothesis.
23.3 SIMULATION OF CHEMICAL INTERACTION BETWEEN NATIVE FLUID, WASTE, AND HOST ROCK Simulations of the chemical reaction of waste, formation water, and formation minerals were performed to evaluate the extent of waste–water–rock interaction resulting from injection. One goal was to calculate changes in formation porosity to help assess long-term injection efficiency. Another goal was to assess whether mineral precipitation/dissolution reactions might explain the lower-than-expected sulfate and ammonia concentrations and increased potassium level at well NDW-1. This work was performed in two stages. First, simple speciation computations and multicomponent reaction-path simulations were run to help define the general chemical system, the types of controlling primary mineral phases, and any potential secondary minerals that
ELSE_TSANG_CH023.qxd
8/21/2005
9:21 AM
Page 317
23.3 Simulation of Chemical Interaction between Native Fluid, Waste, and Host Rock
317
Fig. 23.2. Piper diagram comparing the compositions of native formation waters prior to injection, fluid from NDW-1 after injection in nearby well, liquid waste, and seawater. All formation-water samples, including NDW-1, fall near each other and thus appear to be of similar origin.
could form as a result of reaction with waste constituents. Using this information, a more sophisticated multicomponent reactive transport simulation was then run to predict the evolution of the modeled system under hydrological and chemical conditions more representative of actual field conditions. These simulations were run using thermodynamic data from various literature sources. The main original database was the SOLTHERM database (Reed and Palandri, 1998), with references therein, including data for ammonium minerals from Daniels (1992) and many data derived using SUPCRT92 (Johnson et al., 1992) with modifications as follows. For silicate minerals, equilibrium constants were adjusted for consistency with the thermodynamic properties of aqueous silica reported by Gunnarsson and Arnorsson (2000). Consistent feldspar thermodynamic data were taken from Arnorsson and Stefansson (1999). Equilibrium constants for illite and smectites were recomputed (courtesy J. Apps, LBNL) using data from Kulik and Aja (1997). These are also consistent with the silica data reported by Gunnarsson and Arnorsson (2000). 23.3.1 Initial Speciation and Reaction-Path Simulations The composition of the initial formation water input into simulations was taken as the analysis of water originally recovered from well ODW-1 before waste injection began (Table 23.1). The
ELSE_TSANG_CH023.qxd
318
8/21/2005
9:21 AM
Page 318
Chemical Interactions between Waste Fluid, Formation Water, and Host Rock
Fig. 23.3. Stiff diagrams comparing the composition of native formation waters prior to injection, fluid from NDW-1 after injection in nearby well, liquid waste, and seawater. Figures depicting similar shapes, regardless of size, indicate waters with similar proportions of constituents. Samples from NDW-1 appear diluted compared with the other formation water samples, but have otherwise similar proportions of constituents (i.e., similar diagram shape).
Table 23.3. Composition of formation matrix assumed for simulations. Data provided by x-ray analysis and petrographic examination of recovered core from the Oakville Formation Formation minerals Quartz Feldspars Calcite Dolomite (Fe-rich) Kalolinite Chlorite Illite/mica Montmorillonite (smectite)
Weight % 65 25 1 2 1.02 0.54 0.96 3.48
composition of reactant liquid waste (Table 23.2) and the initial formation mineralogy (Table 23.3) were also available from measurements. Note that organic species were assumed to be nonreactive and were omitted from simulations. The presence of organic acids could affect the modeled system chemistry, particularly pH. These effects were tested in limited simulations and appear to be small, although a more significant effect cannot be ruled out at this time. The measured temperature within the injection interval is around 65°C, and waste is heated up to this temperature before being injected. Therefore, to perform realistic simulations, the formation water and waste (25°C analyses) were numerically heated from 25 to 65°C before reacting them together. The SOLVEQ code (Reed, 1982; Spycher and Reed, 1992) was used to compute the pH and chemical speciation of these fluids at 65°C, from the 25°C analyses (Reed
ELSE_TSANG_CH023.qxd
8/21/2005
9:21 AM
Page 319
23.3 Simulation of Chemical Interaction between Native Fluid, Waste, and Host Rock
319
and Spycher, 1984). At 65°C, the recalculated pH of formation water was 6.9, and that of the waste was 6.5, compared to values of 7.0 and 7.6, respectively, at 25°C (Tables 23.1 and 23.2). Because concentrations of aluminum, potassium, and sulfate in the formation water (native fluid in OWD-1) were unknown, these were calculated by assuming equilibrium with illite, potassium feldspar, and barite at 65°C, respectively. This yielded concentrations of 0.006 ppm Al, 421 ppm K, and 45 ppm SO4. Sulfate and potassium concentrations computed in this way are in line with other formation water analyses (Table 23.1). Because the solution was supersaturated with respect to calcite (likely as the result of CO2 loss during sampling; see Palandri and Reed, 2001), this mineral was brought to exact saturation with the native fluid at 65°C by adjusting the bicarbonate concentration to 40 ppm, within the expected typical large uncertainty of analytical values for this constitutent. The adjustment of the bicarbonate concentration brought iron dolomite close to exact saturation, and, for this reason, a slight adjustment to the measured iron concentration (from 1.5 to 1.8 ppm) was made to allow this mineral to saturate exactly. Note that for simplicity, iron dolomite was treated as a pure end-member (ferrodolomite). In reality, an ankerite phase is likely to prevail, with a composition intermediate between dolomite and ferrodolomite. Using these adjusted compositions, the computed equilibration temperatures for several other minerals observed in the formation are close to measured downhole temperatures, around 65°C (Fig. 23.4). Note, however, that smectites are undersaturated and chlorite supersaturated by approximately 1.5 log(Q/K) units at the temperature of interest (Fig. 23.4). This is likely the result of poorly constrained thermodynamic data for these minerals. Also note that quartz is undersaturated at the formation temperature, possibly the result of silica loss by removal of precipitates when filtering samples for analysis (e.g., Palandri and Reed, 2001). These authors have shown that corrections for such silica deficiency, for CO2 loss, and for downhole pressure (~220 bar) effects on thermodynamic data significantly improves the clustering of log(Q/K) curves near the formation temperature on plots such as Figure 23.4. However, because of the preliminary nature of the simulations presented here, these corrections were not attempted.
Fig. 23.4. Saturation indices for typical formation minerals and anhydrite using the water analysis from well OWD-1 (prior to injection) (Table 23.1) somewhat modified (see text) to bring equilibrium with illite, microcline, barite, iron-dolomite, and calcite at 65°C (the approximate downhole temperature). Anhydrite is shown for information because it is predicted to precipitate upon waste injection. Equilibration temperatures [defined by log(Q/K) = 0] for several other formation minerals are close to downhole temperatures (~65°C).
ELSE_TSANG_CH023.qxd
320
8/21/2005
9:21 AM
Page 320
Chemical Interactions between Waste Fluid, Formation Water, and Host Rock
The reaction of waste with the formation water and minerals was then carried out in two stages using the CHILLER code (Reed, 1982; Spycher and Reed, 1992). First, the formation mineral assemblage (Table 23.1) was numerically titrated into the formation water (ODW-1, Table 23.1) until the calculated proportion of water and minerals was consistent with the observed porosity of the formation (approximately 28%, by volume). Then, the resulting water/mineral assemblage was reacted with waste fluid by simulating infusion of liquid waste into the formation. This pseudo-flow-through simulation involved incrementally mixing and reacting a finite amount of liquid waste with the formation mineral/water assemblage, and subsequently fractionating (i.e., removing) the resulting fluid by an amount suitable to accommodate for additional infusion of waste fluid. For this simulation, the mixing increment of waste fluid was taken as 10% of the initial pore-fluid volume. After infusing a total waste fluid amount equal to 10 pore-fluid volumes, the mixing increment was increased to a full pore-fluid volume for the rest of the simulation. The model therefore simulates fresh liquid waste infusing through a given volume of formation (e.g., a “box”), and calculates resulting porosity and other chemical changes within that “box.” This modeling exercise represents a crude but very useful approximation of the real system. It is also limited, in our case, by the assumption of thermodynamic equilibrium. Nevertheless, it is an important step in the development of more sophisticated reactive-transport simulations such as those presented later. The assumption of equilibrium provides a useful limiting case and allows for evaluating the relative stability of a multitude of potentially important secondary minerals. The modeling simplifications also enable a quick evaluation of reactions within a large chemical system, the number of which can then be narrowed down for further evaluations using more computationally intensive reactive-transport simulations. Several important observations were made from this initial simplified simulation. Sulfate and total ammonia (here reported as ammonium) concentrations quickly rise, as expected, because of the elevated concentrations of these constituents in the injected waste (Fig. 23.5a). The concentrations of other aqueous species decrease as a result of dilution because the waste (Table 23.2) is more dilute than the formation water (Table 23.1). Chloride, which is unreactive in this instance, best illustrates this trend (Fig. 23.5a). Anhydrite (calcium sulfate) quickly forms because of the added sulfate from the waste (Fig. 23.5b), but redissolves shortly thereafter because of the dilution effect on calcium concentrations and additional calcium depletion caused by calcite (calcium carbonate) precipitation. Eventually, anhydrite precipitates again as calcite dissolves due to further mixing with mildly acidic waste (calculated pH of 6.5 at 65°C). Note that the difference between the measured formation water pH value of 7 at 25°C and the initial value of 6.4 in this simulation results from the reequilibration at 65°C with formation minerals prior to reaction with waste. Another interesting observation from this simulation is the precipitation of ammonium silicates upon reaction with injected waste. Ammonium illite and, to a lesser extent, buddingtonite (ammonium feldspar) are predicted to form (Fig. 23.5b), although not in amounts sufficient to significantly decrease the elevated concentration of ammonia in solution. Albite and kaolinite completely dissolve, and some annite (an iron mica) precipitates. Other formation minerals are not affected significantly by waste injection, and potassium concentrations remain below initial ambient concentrations. The overall effect of injection is a slight increase in porosity (Fig. 23.5d). Consequently, the formation water evolves from a composition similar to that of the original fluid (OWD-1) to a composition similar to waste, with some reactivity but not enough mineral precipitation to keep sulfate and ammonia concentrations from increasing.
ELSE_TSANG_CH023.qxd
8/21/2005
9:21 AM
Page 321
23.3 Simulation of Chemical Interaction between Native Fluid, Waste, and Host Rock
321
Fig. 23.5. Reaction-path simulation results: (a) dissolved species (total aqueous concentrations), (b) minerals, (c) pH, and (d) porosity. The interaction between waste fluid, formation water, and formation minerals is simulated as a “box” of initial formation minerals and fluid (1 kg fluid and near 28% porosity) infused with liquid waste in successive increments (see text). Local equilibrium is assumed. Precipitating secondary minerals include mostly anhydrite with some ammonium illite, ammonium feldspar (buddingtonite), and annite. Conceptually, the well head is to the right side of the graph.
23.3.2 Reactive-Transport Simulation The computations described above were supplemented with a simulation of coupled flow, transport, and reaction using the TOUGHREACT code (Xu and Pruess, 2001; Xu et al., 2001, 2004). This simulation was kept as simple as possible, with the goal of focusing primarily on the evaluation of waste-formation reactions around the disposal well, and less so on exact long-term transport predictions. Waste injection was modeled as a one-dimensional radial flow problem (i.e., injection from a single well into a horizontal homogenous and confined geologic formation). A variable mass flux was applied at the well head (averaging approximately 16 kg/s, with fluctuations between 2 and 19 kg/s), corresponding to the actual injection flow rates recorded
ELSE_TSANG_CH023.qxd
322
8/21/2005
9:21 AM
Page 322
Chemical Interactions between Waste Fluid, Formation Water, and Host Rock
during the investigated time period (1982–1994). The outer model boundary was set to a constant pressure representative of regional conditions (around 21.9 MPa), and was located at a distance far enough from the injection well (10 km) to avoid boundary effects in the area of interest (i.e., at distances ⬍1000 m). The temperature of the formation and of the injected fluid was maintained at 65°C, consistent with values measured in the field. The permeability of the formation (around 1.8 ⫻ 10⫺13 m2) was estimated from field pumping tests. Injection was simulated for a period of 12 years corresponding to the time between the start of injection activities at well ODW-1 and the drilling of well NDW-1. The same primary formation minerals were used as previously (albite, K-feldspar, quartz, chlorite, kaolinite, Fe-dolomite, calcite, illite, and smectites). Potential secondary minerals
Fig. 23.6. Reactive-transport simulation results after 2 years of injection at ODW-1: (a) dissolved species (total aqueous concentrations), (b) minerals, (c) pH, and (d) porosity. The interaction between waste fluid, formation water, and formation minerals is simulated as a 1-D radial flow problem, coupling the effects of flow, transport, and reaction. Silicate minerals are reacting under kinetic constraints. Anhydrite is the dominant secondary mineral, precipitating as a front moving away from the well (negligible porosity decrease).
ELSE_TSANG_CH023.qxd
8/21/2005
9:21 AM
Page 323
23.3 Simulation of Chemical Interaction between Native Fluid, Waste, and Host Rock
323
Fig. 23.7. Reactive-transport simulation results after 12 years of injection: (a) dissolved species (total aqueous concentrations), (b) minerals, (c) pH, and (d) porosity. The interaction between waste fluid, formation water, and formation minerals is simulated as a 1-D radial flow problem, coupling the effects of flow, transport, and reaction. Silicate minerals are reacting under kinetic constraints. Near the well head, ammonium illite and kaolinite form at the expense of feldspar dissolution, and all calcite initially present dissolves (overall porosity increase with time).
were selected on the basis of the results of the CHILLER reaction-path simulation described earlier, and included anhydrite, barite, ammonium illite, buddingtonite, and annite. Muscovite and amorphous silica were also considered but never formed. Anhydrite, calcite, and barite were set to react at equilibrium because the reaction rate of these minerals is typically quite fast. The other minerals were set to react under kinetic constraints using effective reaction rates calculated from published rate constants and geometric surface areas estimated from a grain size around 10 µm. The implemented rate law was that derived from transition-state theory (Lasaga, 1981) yielding an affinity-dependent formulation (e.g., Aagaard and Helgeson, 1982; Steefel and Lasaga, 1994). The same thermodynamic data were used as in the previous CHILLER simulation.
ELSE_TSANG_CH023.qxd
324
8/21/2005
9:21 AM
Page 324
Chemical Interactions between Waste Fluid, Formation Water, and Host Rock
Results of this TOUGHREACT simulation are shown as a function of distance from the modeled well head after 2 and 12 years of continuous injection (Figs. 23.6 and 23.7, respectively). The dilution front, as shown by chloride concentrations, is predicted to move approximately 200 m after 2 years (Fig. 23.6a) and 500 m after 12 years (Fig. 23.7a). The profiles of sulfate and ammonia concentrations show a similar rate of expansion (Figs. 23.6a and 23.7a). A zone of anhydrite precipitation and calcite dissolution is predicted to form and migrate away from the well head, to approximately 20 m after 2 years (Fig. 23.6b) and 80 m after 12 years (Fig. 23.7b). Anhydrite does not form close to the well head because calcium concentrations are too small (only 32 mg/L calcium in waste), and does not form far away from the well head where calcite precipitation competes for calcium. Ammonium illite and kaolinite progressively precipitate near the well head as feldspars dissolve (Figs. 23.6b and 23.7b). As predicted in the previous CHILLER simulation, the precipitation of anhydrite and ammonium illite does not significantly affect the concentrations of sulfate and ammonia in solution. The pH within the waste plume (Figs. 23.6c and 23.7c) remains lower than previously predicted (Fig. 23.4c) by up to around one pH unit, because the kinetic constraints in the present TOUGHREACT simulation reduce the buffering effect of formation minerals reacting with waste. The porosity near the well is predicted to increase from the initial value of 28% to 28.8% after 2 years, and 29.7% after 12 years (Figs. 23.6d and 23.7d). The porosity barely decreases in the area of anhydrite precipitation because of simultaneous calcite dissolution. In general, this TOUGHREACT simulation displays trends consistent with earlier CHILLER results, showing mild reactivity and porosity change (after 12 years at the location where NDW-1 was drilled, at approximately 200 m from the simulated injection well), and an increase in concentrations of sulfate, ammonia, and bicarbonate, but a decrease in the concentrations of all other constituents.
23.4 DISCUSSION AND CONCLUSIONS The composition of fluid from NDW-1 is similar to formation water unaffected by waste. However, the presence of small concentrations of a waste degradation product, slightly increased sulfate concentrations, apparent dilution, and possibly higher alkalinity compared with other formation waters, all suggest that some mixing with waste has occurred. The waste appears to be, at least theoretically, not very reactive with formation water and minerals. If significant mixing with waste had occurred, one would expect elevated sulfate and ammonia concentrations at NDW-1. Sulfate and ammonium minerals were predicted to form. However, the precipitation of these minerals was not predicted to significantly reduce the elevated concentrations of sulfate and ammonia in the waste plume. It could be that the precipitation of sulfate and ammonium minerals was underpredicted by the models. It cannot be ruled out, at this time, that other important sulfate and ammonium solid phases form in the real system and significantly deplete sulfate and ammonium in solution. Another possible explanation for the observed dilution, without elevated sulfate and ammonia concentrations, may be that these species were retarded by mechanisms other than mineral precipitation (e.g., sorption and cation exchange for NH4+, sulfate reduction to sulfide). It may also be that dilution is caused by mixing with other less saline waters, either naturally or by introduction through the drilling/sampling process. Given the fact that the sampled intervals at NDW-1 coincide closely with injection intervals at ODW-1, and judging from the reactive transport simulation (which shows that migration occured several hundred
ELSE_TSANG_CH023.qxd
8/21/2005
References
9:21 AM
Page 325
325
meters away from the injection well after 12 years), we would expect that NDW-1 was located within the bulk of the waste plume from ODW-1. It must be recognized that the modeling work presented in this study is preliminary and subject to many uncertainties. Most notably, the effect of reactions involving organic acids present in the injected waste warrants further investigation. The consideration of organic species was not within the scope of this study, although the modeling tools could accommodate these species, their degradation, and potential interactions between such species. Uncertainties regarding kinetic and thermodynamic data must also be acknowledged, because these data are crucial to the outcome of simulations such as those presented here. Also, the potential formation of other sulfate and ammonium minerals should be further investigated. Thus, the simulation results presented here should not be interpreted without keeping in mind these uncertainties. Finally, this study illustrates how traditional graphical methods and geochemical modeling can be applied to problems related to deep-well injection of waste. Simple graphical analyses provide useful characterization tools. Geochemical simulators are more complex, but are powerful tools in helping to solve a wide range of problems—such as the assessment of chemical reactions between waste and host rock, waste compatibility and mixing studies, formation damage and reduced porosity, and the precipitation of scale in injection wells.
ACKNOWLEDGMENTS The authors are grateful to T. Xu and P. Dobson for their valuable comments during preparation of this paper. We also thank J. Apps for supplying some of the thermodynamic data, and S. Salah and E. Sonnenthal for supplying kinetic data used in this study. Analytical data were provided by a private firm that wishes to remain anonymous.
REFERENCES Aagaard, P. and Helgeson, H.C., 1982. Thermodynamic and kinetic constraints on reaction rates among minerals and aqueous solutions. I. Theoretical considerations. Am. J. Sci., 282: 237–285. Arnorsson, S. and Stefansson, A., 1999. Assessment of feldspar solubility constants in water in the range 0° to 350°C at vapor saturation pressures. Am. J. Sci., 299(3): 173–209. Daniels, E.J., 1992. Nature and origin of minerals in anthracite from Eastern Pennsylvania. Ph.D. Dissertation, Illinois University. Gunnarsson, I. and Arnórsson, S., 2000. Amorphous silica solubility and the thermodynamic properties of H4SiO4 in the range of 0° to 350°C at Psat. Geochim. Cosmochim. Acta, 64(13): 2295–2307. Hem, J.D., 1989. Study and interpretation of the chemical characteristics of natural waters. U.S.G.S. Water Supply Paper, No. 2254. Johnson, J.W., Oelkers, E. and 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 bar and 0 to 1000°C. Comput. Geosci., 18: 899–947. Kelly, W.R. and Matisoff, G., 1985. The effects of gas well blow out on groundwater chemistry. Environ. Geol. Water Sci., 7: 205–213.
ELSE_TSANG_CH023.qxd
326
8/21/2005
9:21 AM
Page 326
Chemical Interactions between Waste Fluid, Formation Water, and Host Rock
Kulik, D.A. and Aja, S.U., 1997. The hydrothermal stability of illite: implications of empirical correlations and Gibbs energy minimization. In: D. A. Palmer and D.J. Wesolowski (Eds), Proceedings of the Fifth International Symposium on Hydrothermal Reactions, Gatlinburg, Tennessee, July 20–24, 1997. Oak Ridge National Laboratory, Oak Ridge, TN, pp. 288–292. Lasaga, A.C., 1981. Rate laws in chemical reactions. In: A.C. Lasaga and R.J. Kirkpatrick (Eds), Kinetics of Geochemical Processes, Reviews in Mineralogy (8) Mineral. Soc. Am., 135–169. Palandri, J.L. and Reed, M.H., 2001. Reconstruction of sedimentary formation waters. Geochimica et Cosmochimica Acta, 65: 1741-1767. Reed, M.H., 1982. Calculation of Multicomponent chemical equilibria and reaction processes in systems involving minerals, gases, and an aqueous phase. Geochim. Cosmochim. Acta, 46: 513–528. Reed, M.H. and Palandri, J.L., 1998. SOLTHERM database: A compilation of thermodynamic data from 25 to 300°C for aqueous species, minerals and gases (version: soltherm.joh dated 4/98). University of Oregon, Eugene, OR. Reed, M.H. and Spycher, N.F., 1984. Calculation of pH and mineral equilibria in hydrothermal waters with application to geothermometry and studies of boiling and dilution. Geochim. Cosmochim. Acta, 48: 1479–1492. Spycher, N.F. and Reed, M.H., 1992. Microcomputer-based modeling of speciation and water-mineral-gas reactions using programs SOLVEQ and CHILLER, In: Y.K. Kharaka and A.S. Maest (Eds), Water-Rock Interaction Balkema, Rotterdam, pp. 1087–1090. Steefel C.I. and Lasaga, A.C., 1994. A coupled model for transport of multiple chemical species and kinetic precipitation/dissolution reactions with application to reactive flow in single phase hydrothermal systems. Am. J. Sci., 294: 529–592. Xu, T. and Pruess, K., 2001. Modeling multiphase non-isothermal fluid flow and reactive geochemical transport in variably saturated fractured rocks: 1. Methodology. Am. J. Sci., 301: 16–33. Xu, T., Sonnenthal, E., Spycher, N., Pruess, K., Brimhall, G. and Apps, J.A., 2001. Modeling multiphase non-isothermal fluid flow and reactive geochemical transport in variably saturated fractured rocks: 2. Applications to supergene copper enrichment and hydrothermal flows. Am. J. Sci., 301: 34–59. Xu, T., Sonnenthal, E., Spycher, N. and Pruess, K., 2004. TOUGHREACT User’s Guide: A Simulation Program for Nonisothermal Multiphase Reactive Geochemical Transport in Variably Saturated Geologic Media. Report LBNL-55460, Lawrence Berkeley National Laboratory, Berkeley, CA.
ELSE_TSANG_CH024.qxd
8/22/2005
3:55 PM
Page 327
Chapter 24
WATER–ROCK GEOCHEMICAL CONSIDERATIONS FOR AQUIFER STORAGE AND RECOVERY: FLORIDA CASE STUDIES J.D. Arthura, A.A. Dabousa, and J.B. Cowartb a
Florida Department of Environmental Protection—Florida Geological Survey, Tallahassee, FL, USA b Department of Geological Sciences, Florida State University, Tallahassee, FL, USA
24.1 INTRODUCTION Three aquifer storage and recovery (ASR) facilities in southwest Florida are investigated to further understand water–rock geochemical interactions during the ASR process. Facilities included in this study are the Rome Avenue ASR (Hillsborough County), Punta Gorda ASR (Charlotte County), and Peace River ASR (DeSoto County). All of these utilize the Oligocene Suwannee Limestone as the storage zone. Combined results from multiple-cycle tests suggest that As, Co, Fe, Mn, Mo, Ni, V, and U are mobilized from the aquifer system matrix into the injected waters. Of these metals, only As is of concern with respect to water quality standards. Mobilization is most apparent during the recovery phase of a cycle test at the ASR well. Arsenic and U mobilization are the most consistent and well-documented trends, with maximum concentrations exceeding 112 and 12 µg/L, respectively. Successive cycle tests indicate that maximum observed As concentrations decrease with time; however, this preliminary observation holds true only where both cycle-test injection volumes are similar and exposure of “new” aquifer matrix to the injected water is minimal. This result is not only desired, but expected, assuming that the As source is a fixed and consistently depleted concentration within the aquifer matrix and not replenished due to changes in redox or pH conditions, mixing, or changes in flow paths. In contrast, data from paired cycle tests, where the second injection input volume is greater, reveal different results. Arsenic concentrations in the second-cycle test are equal to or greater than those of the first-cycle test due to the exposure of input waters to a larger volume of previously unaffected (e.g., unleached) aquifer matrix. Mineralogical and chemical characterization of storage zones in the Floridan Aquifer System (FAS) has been determined through a variety of analytical methods. Storage zone carbonates are dominantly calcite and dolomite with minor clay minerals and organic material, and trace amounts of quartz, gypsum, and arsenian pyrite. Within the Suwannee Limestone, As is associated with trace metals such as Fe, V, Ni, Mo, Sb, Sc, and U. Many of these metals are also associated with arsenian pyrite. Although arsenian pyrite is among the sources of As in the aquifer matrix, preliminary sequential extraction studies suggest that Fe oxides and organics may also contain As and associated metals. A principal mechanism for metals mobilization during the ASR process is change in the redox environment due to injection of oxygen-rich surface waters into a native aquifer under reducing conditions. Arthur et al. (2002) suggest additional variables affecting this mobility. These include native and input water chemistry and related parameters, aquifer matrix
ELSE_TSANG_CH024.qxd
328
8/22/2005
3:55 PM
Page 328
Water–Rock Geochemical Considerations for Aquifer Storage and Recovery
chemistry/mineralogy, input water-matrix contact time and number of cycle tests, and sitespecific hydrogeology (e.g., pore/conduit geometry, dynamic pathways). Owing to concerns regarding maximum contaminant levels, the design, construction, and operation of ASR facilities, including monitor-well placement and monitoring schedules, should take into account the possibility of water–rock interaction and mobilization of metals into recovered waters. 24.2 HISTORICAL OVERVIEW As Florida addresses water-resource shortages owing to expanding population growth and periodic droughts, application of ASR technology is increasingly considered a cost-effective solution. The first ASR facility in Florida was permitted for operation in 1982. In 1993, nine ASR facilities existed, and one was proposed. Ten years later, 37 facilities are in operation and 12 are proposed. Several types of ASR facilities exist in Florida, including treated drinking water, raw groundwater, reclaimed water, and partially treated surface water. ASR not only helps meet the increasing demands for drinking water, but it has several other applications in industry, agriculture, and environmental restoration. A significant example of the latter application is the Comprehensive Everglades Restoration Plan (CERP). More than 300 ASR wells are proposed in southern Florida to capture ~1.7 billion gallons (BG) per day and store the water in the FAS until it is needed. Although early operational testing of ASR wells in Florida (early 1980s to mid-1990s) emphasized engineering aspects such as recovery efficiencies, more recent emphasis is on geochemical and microbiological aspects of ASR operations. Initial discussions on the role of ASR in CERP, for example, led to identification of several uncertainties by state and federal review committees (ASR Issue Team, 1999; National Research Council, 2001). Fies et al. (2002) provide an overview of these uncertainties, some of which specifically address implementation of ASR at a regional scale (i.e., the southern Florida peninsula). Regardless of scale, however, an important issue identified in the review process pertains to changes in water quality during movement and storage within the aquifer system. This particular issue was identified partly owing to results of ongoing research at the Florida Geological Survey (FGS). The focus of this paper is to provide a summary and update of FGS research on water–rock interactions during ASR activities in southwest Florida. Additional aspects of water-quality changes during ASR are summarized in Pyne (2002). Most ASR facilities in Florida utilize confined permeable zones of the FAS as the zone in which recharged water is stored. Reducing conditions usually characterize native groundwater in these zones. Source (i.e., surface) waters for ASR contain higher dissolved oxygen (DO) than native groundwater. Once source waters are introduced into a reduced aquifer, selective leaching, mineral dissolution, or desorption processes may release metals into the injected water. For example, Green et al. (1995) report up to 25 ppm U within FAS limestones, and that more than 30% of the U can be leached from the aquifer matrix under oxidizing conditions in the laboratory. Recognizing the water quality implications for ASR, the FGS and the Florida Department of Environmental Protection (FDEP) initiated geochemical research on ASR. Results of earlier phases of this research are presented in Cowart et al. (1998), Williams et al. (2002), and Arthur et al. (2001, 2002). 24.3
RESEARCH GOALS
Goals of this ongoing research include: (1) investigate water–rock interactions during ASR with an emphasis on trace-metal mobilization into recovered water; (2) assess these
ELSE_TSANG_CH024.qxd
8/22/2005
3:55 PM
Page 329
24.5 Water-Quality Changes
329
interactions within varying hydrogeologic settings (i.e., different aquifer systems and matrix compositions [chemical/mineralogical]); (3) evaluate the effect of repeated ASR cycle testing and other ASR practices (e.g., borehole acidization) on these interactions; (4) explore applications of U isotopes to identify source waters (injected, native, and pore water) and mixing; and (5) provide the FDEP and CERP with scientific knowledge on which to base ASR design, regulatory, and operational decisions. ASR wells located in the Florida peninsula comprise the focus of our research. Results summarized herein are based on our work at three of these facilities: Rome Avenue ASR (Hillsborough County), Punta Gorda ASR (Charlotte County), and the Peace River ASR (DeSoto County). 24.4 HYDROGEOLOGIC SETTING All of the ASR facilities in this study utilize the Oligocene Suwannee Limestone as the storage zone. The Suwannee Limestone is predominantly a permeable grainstone that comprises the upper part of the FAS in much of southwestern Florida. In general, low-permeability sediments above the Suwannee Limestone storage zones are carbonate sediments (variably siliciclastic and phosphatic) of the Oligocene to Pliocene Hawthorn Group. Confining strata beneath these storage zones are lower permeability, finer grained carbonates of either the Suwannee Limestone or the Eocene Ocala Limestone. The Suwannee Limestone locally contains dolostone, as well as minor amounts of quartz sand, and trace amounts of finely disseminated organics and pyrite (Arthur et al., 2000; Price and Pichler, 2002). Further details on the stratigraphy of the area are described in Green et al. (1995) and Arthur et al. (in review). 24.5 WATER-QUALITY CHANGES Hydrochemical data obtained from ASR cycle tests, when evaluated in time-series diagrams, allows for comparison of water-quality changes during injection (recharge), storage, and recovery. For example, in a pair of cycle tests at the Punta Gorda ASR, Ca concentrations are low in recharged water, and high in native water (Fig. 24.1). As such, during the recovery phase of the test, a mixing curve is evident as water is recovered to higher native concentrations. In this example, any subtle trends reflecting Ca mobilization by dissolution are overprinted by the mixing curve. On the other hand, concentrations of As are low in both recharged and native water, but an increase in concentration is observed during recovery. Upon ruling out anthropogenic factors, mobilization of As from the aquifer matrix is indicated. Although As concentrations appear to decline at Punta Gorda with a subsequent cycle test, this is not always the case. Successive cycle tests from ASR-7 (Fig. 24.2), which is part of an eight-well configuration at Rome Avenue, show relatively consistent U and As mobilization. This graph, however, can also be misleading because recharge volumes vary from 4 million gallons (MG) to 132 MG. Although recovery volumes for each of these cycles exceeded input volumes, cycle tests for other ASR wells are designed to leave water in storage. In cases where earlier cycle tests recharge and recover volumes that roughly equal or exceed that of subsequent cycle tests, the levels of As and U decline with time (Fig. 24.3; compare Spring 2001 with Spring 2002). This observation holds true when no recharge water remains in storage (as calculated by volume) between cycle tests. In cases where more
ELSE_TSANG_CH024.qxd
330
8/22/2005
3:55 PM
Page 330
Water–Rock Geochemical Considerations for Aquifer Storage and Recovery
Fig. 24.1. Punta Gorda cycle tests: As and Ca distributions. Arsenic concentrations in recharge and native groundwaters are less than 10 µg/L.
Fig. 24.2. Arsenic and U semi-log variations in recovered water during three cycles tests at well ASR-7, Rome Avenue.
recent cycle tests involve greater recharge volumes than earlier tests, concentrations of recovered As are usually greater (Fig. 24.3; compare Fall 2000 to Spring 2001). Arsenic and U mobilization are the most consistent and well-documented trends observed at all sites in our study, with concentrations in recovered waters up to 112 and 12 µg/L, respectively. Cycle-test results also indicate mobilization of Co, Fe, Mn, Mo, Ni, and V from the aquifer system matrix into the recharged waters (Arthur et al., 2002; present study).
ELSE_TSANG_CH024.qxd
8/22/2005
3:55 PM
Page 331
24.6 The Aquifer System Matrix
331
Fig. 24.3. Arsenic concentrations for three cycle tests from ASR-4, Rome Avenue. Legend indicates relative recovery periods; input volumes are shown in million gallons (MG). In all cases, recovery volumes exceeded recharge volumes.
In addition to individual wells showing As variability in recovered water across multiple cycle tests, a high degree of As variation is observed within a single cycle test within a well field. For example, at the Peace River ASR facility during a 5-day recovery period, water from three wells located within 0.3 km of each other exhibit more than a threefold range in As concentrations, with maximum observed values from 24 to 88 µg/L. The samples represent recovery of slightly more than half of the recharged volume, which was from the same source for all three wells. 24.6 THE AQUIFER SYSTEM MATRIX Mineralogical and chemical characterization of the Suwannee Limestone (storage zone) has been determined through a variety of methods. Mineralogy was determined through binocular description, petrography, X-ray diffraction, scanning electron microscopy, and energy-dispersive X-ray microprobe analyses. To date, we have analyzed 37 samples using a 64-element, multi-method analytical technique (XRAL and Activation Laboratories). Arthur et al. (2001) present geochemical data for 22 samples. Analyses of the remaining samples are provided in Table 24.1. A summary of selected Rome Avenue whole-rock chemistry correlations that are significant at the 99% confidence level [Pearson’s r (two-tailed), p ⫽ 0.01, n ⫽ 15, r ⬎ 0.64] includes: (1) As and Fe2O3, loss on ignition (LOI), C (total), V, Ni, Mo, Sb, Sc, and U; (2) U and As, Sr, V, Cr, Mo, Sb, Sc, Th, and several rare-earth elements; and (3) Fe2O3 and LOI, As, Mo, Ni, P2O5, Sb, Sc, Th, and V. Two examples of these
418 19 5 16 8
512 16 4 11 4
468 18 3 26 19
1.03 0.12 0.21 0.002 0.70 55.2 n.d. 0.07 0.015 0.03 41.1 98.4 0.13 11.8
25 ASR3 253.5
513 16 2 n.d. 2
0.40 0.03 0.04 0.002 0.72 54.9 n.d. n.d. 0.010 0.02 43.2 99.3 n.d. 12.0
26 ASR4 303.5
569 14 3 7 2
0.45 0.04 0.06 0.002 0.71 54.9 0.02 n.d. 0.010 0.02 43.2 99.3 0.13 12.1
27 ASR4 305.0
599 13 5 5 2
0.51 0.06 0.06 0.002 0.69 55.0 0.02 n.d. 0.010 0.03 43.3 99.6 0.15 12.0
28 ASR4 306.0
496 21 7 12 3
1.25 0.08 0.06 0.002 0.72 54.0 0.01 0.04 0.015 0.02 43.0 99.2 0.12 11.9
29 ASR4 308.0
457 18 4 19 3
0.75 0.10 0.09 0.002 0.59 54.9 0.01 n.d. 0.015 0.03 43.0 99.5 0.08 11.8
30 ASR5 251.0
457 13 4 n.d. 3
0.78 0.07 0.06 0.002 0.65 54.9 n.d. n.d. 0.010 0.02 43.5 99.9 0.06 11.9
31 ASR5 255.0
500 19 3 13 2
0.69 0.07 0.07 0.002 0.63 54.7 0.03 0.04 0.015 0.02 43.1 99.4 0.1 12.1
32 ASR5 258.0
515 21 10 5 2
2.22 0.13 0.07 0.002 0.85 53.6 0.02 0.02 0.015 0.02 42.4 99.3 0.05 11.9
33 ASR6 322.0
35 ASR6 360.0
36 ASR6 372.0
593 38 9 7 3
489 n.d. 6 n.d. 1
543 14 6 10 2
2.17 0.65 0.53 0.27 0.12 0.07 0.10 0.06 0.06 0.002 0.002 0.002 0.72 0.79 0.69 53.2 55.2 55.8 n.d. n.d. n.d. 0.06 0.02 n.d. 0.026 0.010 0.010 0.02 0.02 0.02 42.7 43.3 43.1 99.2 100.0 100.1 0.39 0.08 n.d. 12.1 12.1 12.0
34 ASR6 339.0
509.9 18.0 5.0 11.5 3.9
0.88 0.10 0.09 0.00 0.71 54.7 0.02 0.04 0.01 0.02 43.0 99.5 0.1 12.0
Avg.
3:55 PM
520 12 4 7 3
0.94 0.12 0.18 0.002 0.78 55.0 n.d. n.d. 0.016 0.03 41.9 99.0 0.17 11.8
24 ASR3 255.0
332
0.47 0.41 0.12 0.13 0.15 0.09 0.004 0.002 0.73 0.73 55.0 54.6 n.d. n.d. n.d. 0.03 0.010 0.010 0.04 0.03 43.4 44.2 99.9 100.0 0.13 0.07 11.9 12.0
23 ASR2 408.0
Table 24.1. Whole-rock geochemical analyses of representative samples from Rome Avenue ASR well field
22 ASR2 404.0
8/22/2005
Major elements (wt%) SiO2 Al2O3 Fe2O3 MnO MgO CaO Na2O K2O TiO2 P2O5 LOI TOTAL Org (C) Tot (C) Trace elements (ppm) Sr Zr Ba V Ni
Fgs no. Well no. Depth (ft. bls)
ELSE_TSANG_CH024.qxd Page 332
Water–Rock Geochemical Considerations for Aquifer Storage and Recovery
3 n.d. 8.3 0.9 1 3.1 3 0.2 0.3 0.4 2.7 0.08 0.5
3.6 4 4 0.65 0.1 0.35 0.06 6
3 5 8.8 1.1 1 2.8 5 0.4 0.2 0.5 2.9 0.19 n.d.
3.5 3 3 0.65 0.1 0.36 0.06 5
4.7 3 4 0.79 0.1 0.54 0.09 8
5 n.d. 26.8 1.3 11 0.5 25 0.8 0.5 0.8 9.5 0.14 1.1
0.8 n.d. n.d. 0.13 n.d. 0.11 0.02 n.d.
7 6 13.1 1 2 0.7 5 0.1 0.1 0.1 1.6 0.15 n.d.
0.9 n.d. n.d. 0.15 n.d. 0.12 0.02 2
5 5 16.3 0.6 2 0.5 6 0.1 0.2 n.d. 1.5 0.06 0.4
0.9 1 n.d. 0.17 n.d. 0.15 0.03 2
2 6 15 1.1 2 0.6 8 0.1 0.2 0.2 1.5 0.06 0.5
1.1 1 1 0.19 n.d. 0.15 0.02 3
2 4 13.4 0.7 3 1.1 17 0.2 0.2 0.2 1.8 0.06 n.d.
3.9 3 3 0.72 0.1 0.5 0.08 7
3 10 22.8 1 4 n.d. 15 0.3 0.3 0.5 8.6 0.07 n.d.
4.2 3 4 0.81 0.2 0.71 0.11 8
3 n.d. 24.5 0.8 2 0.7 11 0.2 0.4 0.7 7.2 0.07 n.d.
4.6 4 4 0.89 0.1 0.68 0.1 9
4 n.d. 28.2 1 2 0.7 11 0.1 0.4 0.7 6.8 0.07 n.d.
1.8 2 2 0.3 n.d. 0.23 0.03 6
17 5 9.5 1.2 0 0.7 7 0.1 0.2 0.3 1.5 0.08 0.4
2.2 3 2 0.38 n.d. 0.25 0.04 3
6 n.d. 17.1 0.7 2 2.3 8 0.1 0.3 0.5 3.1 0.06 n.d.
1.2 2 1 0.21 n.d. 0.15 0.02 4
4 n.d. 12.1 0.9 0 0.7 5 0.1 0.2 0.3 1.8 0.11 n.d.
0.8 1 1 0.12 n.d. 0.06 0.01 1
3 6 5.8 0.6 1 0.7 8 0.2 0.1 0.2 3.3 0.06 n.d.
2.6 3 3 0.46 0.1 0.33 0.05 5
4.6 5.6 16.2 0.9 2.9 1.1 10.5 0.2 0.3 0.4 4.2 0.1 0.6
3:55 PM
4.4 3 4 0.8 0.1 0.53 0.08 8
2 4 22 1.3 11 0.8 23 0.5 0.4 0.6 8.9 0.07 n.d.
8/22/2005
n.d.: not detected; FGS: Florida Geological Survey; bls: below land surface.
Cu Pb Cr Co As Br Mo Sb Sc Th U S Cd Rare-earth elements (ppm) La Ce Nd Sm Tb Yb Lu Y
ELSE_TSANG_CH024.qxd Page 333
24.6 The Aquifer System Matrix 333
ELSE_TSANG_CH024.qxd
8/22/2005
334
3:55 PM
Page 334
Water–Rock Geochemical Considerations for Aquifer Storage and Recovery
correlations (Fig. 24.4) suggest that the V and Mo are associated with As-bearing phases such as arsenian pyrite. The y-axis intercept, however, indicates that V and Mo are associated with either non-As bearing pyrites or other minerals/phases present in the rocks. Arthur et al. (2001) report an As–Fe association in the Suwannee Limestone, which is consistent with observations of arsenian pyrite (e.g., Arthur et al., 2002; Price and Pichler, 2002). In these samples, pyrite occurs in two forms (Fig. 24.5): intergranular and intra-granular euhedral to subhedral grains averaging less than 1 µm in size, and framboidal pyrite masses on the order of 10–20 µm in size. In an effort to identify the distribution of As in these carbonate rocks, modified sequential extraction procedures (Chunguo and Zihui, 1988; Thomas et al., 1994) were applied. Four extraction sequences were completed to isolate metals distributions within the rock matrix: (1) water soluble, (2) bound to carbonates, (3) bound to Fe and Mn oxides, and (4) bound to insoluble material (e.g., organics and pyrite). Figure 24.6 shows normalized results of sequential extractions for three duplicate high-As limestone samples from the ASR-3 well (Rome Avenue). These preliminary results, as well as sequential extraction analyses of carbonates from other ASR sites, substantiate the strong association of As with the insoluble fraction.
24.7 DISCUSSION Water-quality data from more than 15 cycle tests from three ASR facilities in southwestern Florida indicate significant mobilization of metals into stored and recovered water during ASR. Of perhaps greatest significance is the mobilization of As. Most of our analyses of recovered waters exceed 10 µg/L, which is the newly adopted U.S. Environmental Protection Agency maximum contaminant level (MCL) for As in drinking water. Compliance to this new standard will be required by January 2006. Arsenic mobilization during artificial recharge is not unique to Florida (e.g., Ruiter and Stuyfzand, 1998; Stuyfzand, 1998).
Fig. 24.4. Covariants of As with V and Mo in the Suwannee Limestone (Rome Avenue samples).
ELSE_TSANG_CH024.qxd
8/22/2005
24.7 Discussion
3:55 PM
Page 335
335
Fig. 24.5. A through C: Backscatter electron images of Suwannee Limestone, Floridan Aquifer System. Pyrite (white) occurring as: (A) framboidal mass along pore spaces within carbonate matrix (Rome Avenue, ASR 5); (B) finely disseminated intergranular and intragranular subhedral crystals (Rome Avenue, ASR 3); (C) element map showing pyrite grain (upper left) and distribution of Fe, As, and S (note: As contrast is poor due to Mg interference); (D) secondary electron image of framboidal mass in core chip (photo courtesy: Roy Price). Scale bar ~10 m.
Fig. 24.6. Normalized results of sequential extractions for heavy metals for three duplicate limestone samples from well ASR-3, Rome Avenue.
ELSE_TSANG_CH024.qxd
336
8/22/2005
3:55 PM
Page 336
Water–Rock Geochemical Considerations for Aquifer Storage and Recovery
Not only do recovered waters contain elevated As and other metals, but water quality changes during recharge within storage-zone monitor wells are observed (Arthur et al., 2002). During injection, mobilized chemical “fronts” (i.e., breakthrough curves) move past storage-zone monitor wells in varying degrees of intensity. Several processes likely affect this variation, including leaching/dissolution, mixing/dilution, precipitation/sorption, microbial activity, and phase transformations. Regarding the latter, for example, Nicholson et al. (1990) report that long-term pyrite oxidation in carbonate-buffered solutions results in formation and accumulation of FeOOH-coating on the pyrite surface, thereby reducing the rate of pyrite oxidation. This “armoring” effect may contribute to attenuated As with distance from the ASR well, and to declines in As with successive cycle tests. On the other hand, “flushing” of matrix As, via oxidation of arsenian pyrite from approximately the same aquifer storage-zone volume during repeated cycle testing, likely accounts for the observed reductions as well. In either case, when subsequent recharge events involve greater input volumes, recovered water typically contains higher As concentrations. From a regulatory viewpoint, this observation suggests that cycle tests utilizing recharge volumes greater than those expected during production would yield valuable results concerning the lateral transport and fate of these metals, if a sufficient monitoring program exists. Uranium concentrations somewhat mimic those of As during recovery, but to a lesser degree, and they are not always synchronous. Closer examinations of cycle-test data reveal that maximum U concentrations may either precede or occur simultaneously with maximum As concentrations suggesting different geochemical processes and kinetics (Arthur et al., 2002). For example, the U concentrations may be influenced by carbonate leaching or dissolution as well as oxidation of other constituents (e.g., organic matter) in the aquifer matrix. These differences may also be a function of complex and dynamic groundwater flow pathways that vary from site to site and perhaps from cycle to cycle. It is also noteworthy to mention that the U alpha activity ratio 234U/238U is useful for identifying mixing and evolution of waters during ASR, as well as leaching of the aquifer matrix (Cowart et al., 1998; Arthur et al., 2001; Williams et al., 2002). Geochemical and mineralogical studies have focused on concentrations and sources of As in FAS carbonates (Arthur et al., 2001, 2002; Price and Pichler, 2002; Price, 2003; Lazareva, 2004). Within the Suwannee Limestone, As correlates significantly with trace metals such as Fe, V, Ni, Mo, Sb, Sc, and U. Numerous trace metals have been associated with framboidal and euhedral diagenetic pyrite, including As, Co, Cu, Cr, Mo, Ni, V, and Zn (Raiswell and Plant, 1980). Given these associations, many of the observed geochemical correlations are consistent with pyrite control. Other minerals or phases containing As in these carbonate rocks, such as organic material, are also present (Price, 2003). Welch and Stollenwerk (2003) provide an expanse of information on the behavior of As in the hydrogeologic systems. Several lines of evidence support mobilization of As from pyrite: (1) the correlation of several transition metals common to pyrite with As and Fe in the aquifer matrix (e.g., Fig. 24.4); (2) the presence of arsenian pyrite (Fig. 24.5; Price and Pichler, 2002); (3) mobilization of As and Fe within recovered recharge water; and (4) redox conditions conducive to pyrite oxidation. In a mixed carbonate-siliciclastic aquifer in South Carolina, Mirecki et al. (1998) report reductions in DO during storage and recovery, indicating similar oxygenbuffering processes. Although a dominant agent for As mobilization is the oxidation of arsenian pyrite, other mobilization agents may exist, such as changes in pH, desorption, dissolution, and microbial processes. Moreover, other sources/sinks for mobilized metals likely exist.
ELSE_TSANG_CH024.qxd
8/22/2005
Acknowledgments
3:55 PM
Page 337
337
Preliminary sequential extraction data suggest that dominant As-bearing phases reside within the “organic plus insoluble residue” fraction of the matrix, which includes sulfide minerals. Results summarized in Figure 24.6 also demonstrate that the “organic plus insoluble residue” fraction is strongly associated with Al, As, Cd, Cr, Co, Fe, La, Ni, Pb, Se, Sr, Th, and U, whereas Zn is strongly associated with carbonate minerals. Organic material is recognized as a source/sink for U. These results also suggest that sources of arsenic may include organics, as well as pyrite, which has been confirmed by Price (2003). Future extraction studies are planned to further isolate As-bearing phases (i.e., separate sulfides from organics and other insolubles). Preliminary extraction results also provide evidence for presence of As and other mobilized metals in “non-sulfide” fractions of the aquifer matrix (e.g., the carbonate and Fe-oxide fractions). As such, additional bench-scale leaching and extraction studies, microprobe analyses, and geochemical modeling are under way.
24.8 CONCLUSIONS Mobilization of metals from the FAS matrix during ASR activities is apparent, and is likely due to a complex interaction of geochemical processes. The chemically heterogeneous aquifer contains metal-rich phases in sufficient amounts to become mobilized by oxidation or other processes to yield concentrations in recovered waters that may exceed MCLs. Evaluation of cycle test data demonstrates that mobilization varies not only between cycle tests in a given well, but also within wells in the same well field, and among ASR wells in the same region utilizing the same lithostratigraphic unit as a storage zone. Arsenian pyrite is among the sources of As and other trace metals in the aquifer; however, preliminary sequential extraction studies and work by other researchers suggest that phases such as organics contain As and other metals. Moreover, organic material may contain U, which is also thought to be associated with carbonate minerals in the Suwannee Limestone. Further work evaluating cycle tests utilizing different aquifers, sequential extraction studies, and geochemical modeling are needed to enhance our understanding of mobilized metals during ASR. With this knowledge hopefully comes an improved ability to minimize these geochemical effects on water quality, and facilitate more cost-efficient ASR operations in Florida by reducing the need for post-recovery treatment.
ACKNOWLEDGMENTS This research is funded by the Florida Department of Environmental Protection— Underground Injection Control (UIC) Program, which administers funds from the U.S. Environmental Protection Agency. We are grateful for the logistical support and expertise of Richard Deuerling, Judy Richtar, and George Heuler of the UIC program. Numerous individuals have been helpful in coordination and sample collection for this project. Among the several individuals, agencies, and companies to whom we are grateful are Ed Fox and Laura Cintron (Hillsborough County Water Department), Brian Fuller (City of Punta Gorda), Mark McNeil (CH2M Hill), Kevin Morris (Peace River Regional Water Supply Authority), Harris Smith (City of Tampa Water Department), Camp Dresser & McKee, Inc., CDM Missimer, and MWH Global, Inc.
ELSE_TSANG_CH024.qxd
8/22/2005
338
3:55 PM
Page 338
Water–Rock Geochemical Considerations for Aquifer Storage and Recovery
REFERENCES Arthur, J.D., Fischler, F., Kromhout, C., Clayton, J., DeWitt, D., Kelley, M., Lee, R.A., Li, L., O’Sullivan, M., Green, R., Thompson, R. and Werner, C., 2005. Hydrogeologic framework of the Southwest Florida Water Management District. Florida Geol. Surv. Bull. 68: 95 (in review). Arthur, J.D., Dabous, A.A. and Cowart, J.B., 2002. Mobilization of arsenic and other trace elements during aquifer storage and recovery, southwest Florida. In: U.S. Geological Survey Artificial Recharge Workshop Proceedings, Sacramento, California, April 2-4, 2002, U.S. Geological Survey Open File Report 02-89, pp. 44–47. Arthur, J.D., Cowart, J.B. and Dabous, A.A., 2001. Florida Aquifer Storage and Recovery Geochemical Study: Year Three Progress Report, Florida Geological Survey Open File Report 83, 46pp. Arthur, J.D., Cowart, J. and Dabous, A., 2000. Arsenic and uranium mobilization during aquifer storage and recovery in the Floridan aquifer system. Florida Peninsula: Geological Society of America Abstracts with Programs, 32(7): 356. ASR Issue Team, 1999. Assessment Report and Comprehensive Strategy: Aquifer Storage and Recovery Issue Team: A Report to the South Florida Ecosystem Restoration Working Group, http://www.sfrestore.org/issueteams/asr/documents/asrreport.htm. Chunguo, C. and Zihui, L., 1988. Chemical speciation and distribution of arsenic in water, suspended solids and sediments of Xiangiiang River, China. Science Total Environ. 77: 69–82. Cowart, J.B., Williams, H.K. and Arthur, J.D., 1998. Mobilization of U isotopes by the introduction of surface waters into a carbonate aquifer. Geol. Soc. Am. Abstracts with Programs, 30(7): A-86. Fies, M.A., Renken, R.A. and Komlos, S., 2002. Considerations for regional ASR in restoring the Florida Everglades, USA. In: P.J. Dillon (Ed), Management of Aquifer Recharge for Sustainability, Proceedings of the 4th International Symposium on Artificial Recharge, Adelaide, Australia, September 22–26, 2002, A.A. Balkema Publishers, The Netherlands, pp. 341–346. Green, R., Arthur, J.D., and DeWitt, D., 1995. Lithostratigraphic and Hydrostratigraphic Cross Sections through Hillsborough and Pinellas Counties, Florida. Florida Geological Survey Open File Report 61: 26pp. Lazareva, O., 2004. Detailed Geochemical and Mineralogical Analyses of Naturally Occurring Arsenic in the Hawthorn Group. M.S. Thesis, University of South Florida, Tampa, FL, 128pp. Mirecki, J.E., Campbell, B.G., Conlon, K.J. and Petkewich, M.D., 1998. Solute changes during aquifer storage and recovery testing in a limestone/clastic aquifer. Ground Water. 36(3): 394–403. National Research Council, 2001. Aquifer Storage and Recovery in the Comprehensive Everglades Restoration Plan. National Academy Press, Washington, DC, p. 58. Nicholson, R.V., Gillham, R.W. and Reardon, E.J., 1990. Pyrite oxidation in carbonatebuffered solution: 2. Rate control by oxide coatings. Geochem. et Cosmochim. Acta. 54: 395–402. Price, R.E. and Pichler, T., 2002. Oxidation of framboidal pyrite as a mobilization mechanism during aquifer storage and recovery in the Upper Floridan Aquifer, Southwest Florida. Eos Transactions, Am. Geophy. Union. 83(47): F521.
ELSE_TSANG_CH024.qxd
References
8/22/2005
3:55 PM
Page 339
339
Price, R.E., 2003. Abundance and Mineralogical Associations of Naturally Occurring Arsenic in the Upper Floridan Aquifer, Suwannee Limestone. M.S. Thesis, University of South Florida, Tampa, FL, 74pp. Pyne, R.D.G., 2002. Water quality changes during aquifer storage and recovery (ASR), 2002, Considerations for regional ASR in restoring the Florida Everglades, USA. In: P.J. Dillon (Ed), Management of Aquifer Recharge for Sustainability, Proceedings of the 4th International Symposium on Artificial Recharge, Adelaide, Australia, September 22–26, 2002, A.A. Balkema Publishers, The Netherlands, pp. 65–68. Raiswell, R. and Plant, J., 1980. The incorporation of trace elements into pyrite during diagenesis of black shales, Yorkshire, England. Econ. Geol. 75: 684–699. Ruiter, H. and Stuyfzand, P.J., 1998. An experiment on well recharge of oxic water into an anoxic aquifer. In: J.H. Peters (Ed.), Artificial Recharge of Groundwater, A.A. Balkema, Rotterdam, Netherlands, pp. 299–304. Stuyfzand, P.J., 1998. Quality changes upon injection into anoxic aquifers in the Netherlands: Evaluation of 11 experiments. In: J.H. Peters (Ed.), Artificial Recharge of Groundwater, A.A. Balkema, Rotterdam, Netherlands, pp. 283–292. Thomas, R.P., Ure, A.M., Davidson, C.M., Littlejohn, D., Rauret, G., Rubio, R. and LopezSanchez, J.F., 1994. Three-stage sequential extraction procedure for the determination of metals in river sediments. Anal. Chim. Acta. 286: 423–429. Williams, H., Cowart, J.B. and Arthur, J.D., 2002. Florida Aquifer Storage and Recovery (ASR) Geochemical Project: Year One and Year Two Progress Report. Submitted to Bureau of Water Facilities Regulation, Florida Department of Environmental Protection, January 1999, Florida Geological Survey, Report of Investigation No. 100, p. 131. Welch, A.H. and Stollenwerk, K.G. (Eds), 2003. Arsenic in Ground Water Geochemistry and Occurrence. Kluwer Academic Publishers, Boston, MA, pp. 475.
This page intentionally left blank
340
ELSE_TSANG_CH025.qxd
8/23/2005
2:11 PM
Page 341
Chapter 25
GROUTING WITH MINERAL-FORMING SOLUTIONS—A NEW TECHNIQUE FOR SEALING POROUS AND FRACTURED ROCK BY DIRECTED CRYSTALLIZATION PROCESSES G. Ziegenbalg TU Bergakademie Freiber g—Freiberg University of Mining and Technology, Institute of Technical Chemistry, Freiber g, Germany
25.1 INTRODUCTION Based on the idea of copying natural crystallization and precipitation phenomena, a new technology has been developed allowing the preparation of supersaturated, mineral-forming solutions. If these solutions are placed in soil or rock formations, sealing is achieved by the in situ growth of minerals such as gypsum, anhydrite, calcite, or barite. The process can be applied in a similar w ay for immobilization of dissolved heavy metals, as well as the overlay of reactive mineral surfaces with insoluble minerals. This chapter summarizes the fundamentals of the technology and gives an overview of the first lar ge-scale application of BaSO4-forming solutions to immobilize contaminants in the former Koenigstein uranium mine of Wismut GmbH (Germany). The sealing of porous rock or soil formations, as well as fractures or joints, is of great importance to direct groundw ater flow, to stop brine inflows into mines during shaft building, or to encapsulate contaminated areas. Enhanced oil reco very is possible if water-bearing zones can be sealed. In many cases, conventional grouting or injection processes are used. These can be applied to improve soil strength, shut-off seepage, or construct grout curtains. Typical grouts are based on suspensions (cement, clay, bentonite) or chemicals such as water glass, acrylamide, polyurethanes, or epoxides. Water glass is particularly useful in many cases. Mixing with CaCl2 solutions or organic chemicals results in the formation of hard or soft gels, depending on the concentrations of the solutions used (Karol, 1990). Systems consisting of colloidal silica can be applied in a similar way. The silica sol is converted into a gel by the addition of electrolytes such as CaCl 2 or MgCl2. For all materials, a typical first step is for the pore or fracture space to be completely f illed with grout. Setting or hardening then leads to sealing. It is important that no shrinkage occurs during or after this process. The penetration radius of the material is determined by its setting time or by viscosity development during hardening. When suspensions are used, the particle diameter determines the penetration beha vior of the material. Suspensions can penetrate a formation only if the diameter of the pores or fractures is at least three times lar ger than the a verage diameter of the suspended solids (Gaitzsch, 1992). Otherwise, penetration is only possible for a limited period of time. A filter cake is formed around the injection hole, and the grout cannot be brought into the zone that has to be treated. All these facts result in the conclusion that formations characterized by permeability lo wer than 10−5 m/s are ungroutable (Karol, 1990). The penetration radii are small in dense formations. Man y injection holes are
ELSE_TSANG_CH025.qxd
342
8/23/2005
2:11 PM
Page 342
Grouting with Mineral-Forming Solutions—A New Technique for Sealing
necessary over short distances. In some cases, sealing or permeability reduction cannot be achieved by conventional materials and techniques. F or example, homogeneous sealing of porous sandstone formations with conventional grouts is highly difficult. In nature, however, many processes are known that lead to complete closure of large as well as micro flow paths. Apart from flow-path blocking by suspended fines, crystallization or precipitation plays an important role in such processes. The formation of carbonates and sulfates—such as CaCO 3, CaSO4∗2H2O, CaSO4, or BaSO4—has led in many cases to complete closure of flow paths. These processes can cause huge problems—during the injection of solutions into deep horizons, for example. Fast blocking of a storage horizon occurs when incompatible solutions are mixed. On the other hand, crystallization processes resulting in permeability reduction can support remediation processes. Such phenomena are typically coupled-dissolution/precipitation processes. A well-known method to reduce the acidity of tailings or waste-rock dumps is to overlay them with alkalinity-generating materials such as limestone, dolomite, or fly-ash. Dissolution of these materials, by rain for example, leads to alkaline solutions that can neutralize acidic zones. As a result, secondary minerals such as iron, or aluminum hydroxide and gypsum, are formed. They can plug the flow paths and protect deeper zones against oxidation and leaching. Similar reactions are possible when acid mine drainage penetrates into soil formations containing CaCO3. Dissolution and precipitation processes tak e place. It is well documented that the formation of gypsum, and of hydroxides, can result in clogging of flow paths (Blowes et al., 1991; Chermak and Runnels, 1996; Rammlmair, 2002). Further solution transport is interrupted and in filtration of acid mine drainage is stopped. Although gypsum is characterized by a relatively high solubility, long-time stable sealing has been pro ven. However, natural mineral-forming processes are slow, and only small amounts of mass are transported. Crystallization processes can cause huge problems in technical operations, as well as during heating or boiling of water. It is widely known that growing carbonate, hydroxide, or sulfate minerals can lead to complete closure of pipes—even those with a large diameter. Based on fundamental investigations of the described crystallization phenomena, a new technology was developed to allow the sealing of flow paths by crystallization processes similar to those occurring in nature. Solutions supersaturated in slightly soluble sulf ates or carbonates are prepared by using precipitation inhibitors. If such solutions are used as grout, directed crystallization of minerals tak es place during the flow of the solutions through the formation to be sealed. Supersaturated solutions of fer sealing of flow paths that cannot be sealed by conventional materials. Additionally, crystallization or precipitation phenomena can be used as a tool to immobilize in situ contaminants in rock or soil formations. This chapter summarizes the fundamentals of the technology and provides an overview of large-scale applications that ha ve been realized.
25.2 FUNDAMENTALS Mineral precipitation tak es place when concentrations of dissolved constituents exceed the solubility limits. The kinetics of crystallization depend on the nature of the formed mineral as well as the degree of supersaturation. F or example, BaSO4 precipitation occurs very fast if the solubility product (kSP = 10−10 mol2/l2, at 25°C) is exceeded. Due to the low solubility of BaSO4, even traces of barium cause BaSO 4 precipitation in sulfate-containing solutions. The gravimetric method of sulfate determination is based on that. Sulfate is quantitati vely converted into BaSO 4 by the addition of an excess of BaCl2. In contrast, solutions supersaturated
ELSE_TSANG_CH025.qxd
8/23/2005
2:11 PM
Page 343
25.2 Fundamentals
343
with respect to gypsum are stable for longer periods, but if a critical concentration is achieved, fast crystallization takes place, too. Gypsum has a solubility of 2.5 g/L at 25°C. The course of precipitation is changed fundamentally if the so-called inhibitors are added. Such compounds offer the possibility to stabilize solutions characterized by concentrations that are far from equilibrium. Temporarily stable, supersaturated solutions are formed. Mixing BaCl2 and MgSO4 containing solutions results in fast, quantitative BaSO4 formation. In the presence of a suitable inhibitor, however, clear, supersaturated solutions are obtained. The natural solubility of BaSO4 is approximately 2 mg/L (at 25°C); in the presence of suitable precipitation inhibitor, up to 400 mg/L can be stabilized temporarily. It is possible to synthesize supersaturated solutions in the same manner, leading to gypsum or anhydrite formation, or to the crystallization of CaCO3 (Fig. 25.1). The inhibitor compositions are proprietary. In general, the inhibitors influence nucleation and nuclei growth. Inhibitors suitable to prevent spontaneous crystallization, when mixing incompatible solutions, are adsorbed on formed crystal nuclei. Growth of the nuclei is blocked, and the supersaturation remains stable for a limited period of time. Inhibitors do not increase the solubility of a mineral; they change the kinetics of the crystallization. Decomposition of the supersaturation is temporarily prevented. Apart from using inhibitors, there are many other possibilities for obtaining solutions characterized by concentrations above the solubility of a mineral—slow cooling of a saturated solution, for example. During grouting, interactions with pore solutions present in the treated formation result in deactivation of the inhibitor, as with precipitation or complex formation. In addition,
Without precipitation inhibitor 2+
−
2+
Ca + 2 Cl
+ Mg
+
Ca(OH) 2(solid) + 2 H −
Ba 2+ + 2OH Ca
2+
+ 2 Cl
+ SO42 +
+ 2 H+ +
−
+
+ 2 Na +
−
− SO42 − SO42
CO32
→ → → →
−
CaSO4*2H2O + Mg2+ + 2 Cl
// H2O
CaSO4*2H2O
// H2O
BaSO4
// H2O +
−
CaCO3 + 2 Na + 2 Cl
// H2O
spontaneous, uncontrolled precipitation In the presence of a suitable precipitation inhibitor Ca 2+ + 2 Cl
−
+ Mg2+ + SO42
Ca(OH) 2(solid) + 2 H
+
+ SO4
2−
−
Ba 2+ + 2 OH + 2 H+ + SO42 Ca 2+ + 2 Cl
−
−
+ 2 Na+ + CO32
−
→ → → →
−
Ca2+ + SO42- + Mg2+ + 2 Cl 2+
Ca
+ SO4
2−
Ba2+ + SO42
// H2O
−
−
// H2O // H2O
−
Ca2+ + CO32 + 2 Na+ + 2 Cl
// H2O
Clear, temporary stable solutions
Deactivation of the inhibitor
BaSO4; CaSO4*2H2O Formation of layers, single crystals or sludge depending on the solution composition
Fig. 25.1. Fundamentals of the preparation of CaSO4 and BaSO4 supersaturated solutions.
ELSE_TSANG_CH025.qxd
8/23/2005
344
2:11 PM
Page 344
Grouting with Mineral-Forming Solutions—A New Technique for Sealing
decomposition reactions such as hydrolysis can destroy the inhibitor. The supersaturation breaks down, and growing single crystals or layers are formed, depending on the conditions in the flow paths. During the flow of supersaturated solutions through soils or rock formations, adsorption phenomena also play an important role in the deactivation of inhibitors. The inhibitors are adsorbed on surrounding minerals and the solution becomes unstable. Crystallization takes place. During this process, the inhibitor is incorporated into the growing minerals. The effect of inhibitors on the progress of gypsum and BaSO4 precipitation is shown in Figures 25.2–25.4. Precipitation from supersaturated solutions can be adjusted by the
25
CaSO4 [g/l]
20
0.2 g/l inhibitor 0.3 g/l inhibitor 0.4 g/l inhibitor 0.5 g/l inhibitor
15
10
5
0 0
10
20
30
40
50
60
time [h]
Fig. 25.2. Progress of gypsum crystallization over time in the presence of various inhibitor concentrations.
400 350
BaSO4 [mg/l]
300 Initial BaSO4 concentration
250
340 mg/l 250 mg/l 170 mg/l
200 150 100 50 0 0
20
40
60
80
100
120
140
160
time [hours]
Fig. 25.3. Progress of BaSO4 crystallization over time in solutions containing 100 mg/L inhibitor and various initial BaSO4 concentrations.
ELSE_TSANG_CH025.qxd
8/23/2005
2:11 PM
Page 345
25.2 Fundamentals
345
140,00
BaSO4-concentration [mg/l]
120,00 100,00 inhibitor A
80,00
inhibitor B 60,00
inhibitor B
40,00
without inhibitor
20,00 0,00 0
25
50
75 100 time [h]
125
150
Fig. 25.4. Progress of BaSO4 crystallization over time in relation to inhibitor used (inhibitor concentration: 60 mg/L).
composition of the inhibitor, its concentration, and the overall solution composition. It is possible to synthesize solutions containing, for example, up to 80 g/L CaSO4 or 1200 mg/L CaCO3. Mixing of BaCl2 or CaCl2 containing solutions with soluble sulfates, such as MgSO4 or Na2SO4, results in the formation of solutions containing MgCl2 or NaCl as byproducts. A favorable way to synthesize “pure” BaSO4 or CaSO4 solutions is given if the corresponding hydroxides are mixed with diluted sulfuric acid. A suitable inhibitor is essential, otherwise spontaneous precipitation occurs. If supersaturated solutions are used as grout, crystallization takes place in the penetrated flow paths. The following effects can be observed (Fig. 25.5): ● Blocking of flow paths by single grown crystals (Figs 25.6 and 25.7). ● Formation of slowly growing layers leading to a step-by-step closure of the flow paths, and to the protection of mineral surfaces against oxidation, leaching, etc. ● Precipitation due to interactions with pore water present in the formation. Sealing is achieved by crystallization processes similar to those occurring in nature. It is important to clarify the differences of such a technology compared to conventional grouting processes: ● Pure solutions with a viscosity similar to natural pore water are used. The penetration behavior is much greater than that of suspensions or plastics, so formations with lower permeability can also be treated. ● In contrast to conventional grouts, in which sealing is achieved by one injection, mineralforming solutions lead to a step-by-step closure of flow paths. Layers growing on the flow paths are formed, as well as single crystals. Long injection times are possible, penetrating areas far away from the point of injection when necessary. ● In most cases, sealing is achieved with minerals that are naturally present in the formation. There is no environmental risk. ● The stability of the seal depends on the characteristics of the formed minerals as well as the composition of the pore solution. Due to the extreme low solubility and the natural sulfate content of groundwater, BaSO4 (barite) can be regarded as the stable phase in most
ELSE_TSANG_CH025.qxd
346
8/23/2005
2:11 PM
Page 346
Grouting with Mineral-Forming Solutions—A New Technique for Sealing formation of layers
formation of single crystals
precipitation as result of mixing with present pore water
Fig. 25.5. Schematic illustration of flow path reduction by induced precipitation in a porous medium (not to scale).
Sandstone grains
Formed gypsum crystals
Fig. 25.6. Gypsum crystals grown in porous sandstone.
cases. Dissolution does not take place. In comparison, gypsum is a mineral with relatively high solubility. In areas where groundwater or pore solutions are undersaturated in CaSO4, dissolution is possible. This process, however, depends on the flow conditions and especially the flow rate. Grouting with solutions supersaturated with gypsum means that, after crystallization, a solution remains in the treated area that is saturated in gypsum. When sealing or a high reduction in permeability is achieved, further solution transport is impossible or very slow. Diffusion processes will dominate, and these are slow. It can be assumed that the formed gypsum seal will be stable for a long time.
ELSE_TSANG_CH025.qxd
8/23/2005
2:11 PM
Page 347
25.3 Sealing of Porous or Fractured Rock Formation by Induced Crystallization
347
Sandstone grains
Formed gypsum crystals
Fig. 25.7. Gypsum crystals blocking flow paths.
25.3 SEALING OF POROUS OR FRACTURED ROCK FORMATION BY INDUCED CRYSTALLIZATION Column tests with sea sand (<200 µm) were performed to demonstrate the sealing capacity of gypsum-forming solutions,and to determine the change of permeability during continuous injection. Characteristic data from the columns are summarized in Table 25.1. The gypsum-forming solution was prepared by continuous mixing of 0.5 mol CaCl 2 and MgSO4 solutions in a volume ratio of 1:1. After that, the solution was pumped continuously with a flow rate of 100 mL/hour through the column. The permeability was determined at specific time intervals by measuring the outflow at constant injection pressure. It is clear from Figure 25.8 that gypsum precipitation has led to a continuous decrease of the permeability of the column. In the same manner, the experiment demonstrated that long injection times can be realized without blocking the entrance to the column. This was also seen after the end of the test by analyzing the CaSO4 content of the sand in relation to its distance from the input source. Most of the gypsum was formed in the middle of the column (Fig. 25.9). Based on the CaSO4 input/output concentrations and molar v olume of gypsum, it was possible to determine the amount of pore space filled. Seven liters of solution containing 35 g/L were introduced into the column, leading to a reduction of the permeability from 11 to 2 m/day. This was achieved by filling only 12% of the pore space with gypsum. Crystals blocking the flow paths in their narrowest locations have caused sealing. Figure 25.10 shows a column containing coarse sand, which was sealed by gypsum. The crystals formed were strongly connected with the surrounding rock and also ga ve the sand pack a high mechanical stability (Graupner , 2002). Although they can be stable for long periods, supersaturated solutions are metastable systems. Their stability is not only influenced by the solution composition,but also by mechanical motion such as stirring, shaking, or pumping through pipes. In general, increased mechanical movement results in accelerated crystallization. The same is true in the presence of suspended solids. For practical applications, it is important for the concrete solution composition to be adjusted for field conditions. Normally, the stability of supersaturated solutions decreases with temperature. If suitable inhibitors are a vailable, however, it is possible to prepare systems characterized by
ELSE_TSANG_CH025.qxd
8/23/2005
348
2:11 PM
Page 348
Grouting with Mineral-Forming Solutions—A New Technique for Sealing Table 25.1. Characteristics of the column experiments Column length Column diameter Amount of sand Pore volume CaSO4 concentration in solution Stability of the solution (in the absence of sand) Flow rate Inhibitor concentration
97 cm 5.3 cm 3.4 kg 810 mL 34 g/L ⬎ 24 hour 100 mL/hour 1.4 g/L
9
12
k f [m/day]
8 10
7
8
6 5
6
4
4
3 2
2
1
0
CaSO4 concentration in the output [g/l]
10
14
kf [m/d] CaSO4 concentration
0 0
2
4
6
8
volume of CaSO4 solution [litre]
Fig. 25.8. Permeability change during injection of a sand column with a gypsum-forming solution and eluate concentration.
1,4
mg CaSO4 / g sand
1,2 1 0,8 0,6 0,4 0,2 0 0
20
40
60
80
100
length of the column [cm]
Fig. 25.9. Gypsum distribution in a column treated with CaSO4 supersaturated solutions.
ELSE_TSANG_CH025.qxd
8/23/2005
2:11 PM
Page 349
25.3 Sealing of Porous or Fractured Rock Formation by Induced Crystallization
349
Fig. 25.10. Column with coarse sand that was sealed completely by gypsum.
high stability at temperatures above 100°C. Figure 25.11 shows the course of CaSO4 precipitation at 120°C in relation to the initial concentration. While at 90°C, the solutions are stable for hours, but at 120°C, the addition of 1 g/L inhibitor allows stabilization for only 2 hours before crystallization starts. If solutions with an initial concentration of 7.5 g/L are used, the CaSO4 content decreases within 4 hours to 4.8 g/L. This is still 9.6 times higher than the equilibrium concentration. Further crystallization occurs within 24 hours. One possible use for such solutions would be for sealing sandstone at high temperatures. Experiments in a specially constructed apparatus ha ve shown that penetration of sandstone at 90°C with a pressure gradient of 3.12 bar/m results in a reduction of the permeability from 0.33 to 0.088 mD. This was achieved by flowing three pore volumes through the core. As shown in Figure 25.12, CaSO4 solutions can penetrate dense materials to form precipitates. Due to the high solubility of most salt minerals, brine or water inflows into potash or rock salt mines are highly dangerous. Ev en small inflows can jeopardize the whole mine. If the inflow brine is undersaturated with respect to surrounding salt rock, dissolution processes will result in a fast increase in the diameter of the flow paths, with higher inflow rates as a consequence. This leads to high demands on possible grouting agents. Dissolution processes must not take place, and it is important that small inflows are sealed too. For the sealing of brine inflows into potash or salt mines CaSO4 supersaturated solutions based on NaCl, or NaCl/KCl saturated solutions, were specially developed (Ziegenbalg and Crosby, 1997). The solubility of gypsum in a NaCl saturated solution is 4.5 g/L at 25°C. While gypsum crystallization takes place in NaCl saturated solutions, syngenite (K2SO4∗CaSO4∗H2O) or gypsum are formed in NaCl/KCl saturated solutions. Crystallization progress depends on the composition and amount of added inhibitor , as well as the overall solution composition and temperature (Fig. 25.13). It is possible to prepare solutions containing 20 g/L CaSO4 with a stability of more than 10 days. The maximum CaSO4 supersaturation that can be stabilized in NaCl saturated solutions is around 40 g/L at present. Density and viscosity of the supersaturated solutions are similar to pure NaCl or NaCl/KCl saturated solutions. The achievable permeability reduction depends on the characteristics of the formations as well as on the composition of the solution used. Large-scale field tests have proved that sealing can be achi eved both in salt rock and in zones surrounding salt deposits that consist of carbonates or sandstone (Gaitzsch, 1992; Ziegenbalg and Crosb y, 1997). In contrast to other techniques, mixing with solutions circulating in the formation to be sealed is not necessary. For example, a widely used technique to stop brine inflows into potash mines is grouting with highly concentrated CaCl2 solutions. Sealing is or can be achieved after mixing with the brine present in the formation due to the formation of
ELSE_TSANG_CH025.qxd
8/23/2005
350
2:11 PM
Page 350
Grouting with Mineral-Forming Solutions—A New Technique for Sealing
CaSO4 [g/1000 g solution]
12 10 initial CaSO4 concentration
8
9.5 g/l 7.5 g/l 6.5 g/l 5.0 g/l
6 4 2 0 0
1
2 time [h]
3
4
Fig. 25.11. CaSO4 precipitation at 120°C in the presence of an inhibitor concentration of 1 g/L.
Fig. 25.12. Dense sandstone after treatment with a CaSO4-forming solution at 90°C.
gypsum, or the salting out of NaCl (Kuehne, 1993). Difficulties connected with such a process are: ● Salt-out and precipitation processes during mixing of two solutions may occur spontaneously. ● It is impossible to control their rate of progress. ● Directed mixing of two solutions in porous or fractured rock is highly difficult. A plug flow is often formed. Mixing and crystallization take place only partially in the boundary zone between the solutions, which means that the greater part of the injected solution does not lead to precipitation or salting out.
ELSE_TSANG_CH025.qxd
8/23/2005
2:11 PM
Page 351
25.4 In Situ Immobilization by Crystallization Processes
351
25 24 23
CaSO4 [g/l]
22 21 20 19 18 17 16 15 0
50
100
150
200
250
300
time [hours] inhibitor A
inhibitor C
inhibitor B
inhibitor D
Fig. 25.13. Progress of gypsum crystallization over time in NaCl saturated solutions containing various inhibitors in a concentration of 800 mg/L.
Gypsum- or syngenite-forming solutions, prepared with precipitation inhibitors, contain all components necessary for mineral formation. Precipitation takes place from the whole solution. For example, 1 m³ of solution containing 25 g/L of CaSO4 can form approximately 25 kg of gypsum. This results in a decrease of the pore space in the treated formation by 10 liters. At present, the large-scale application of gypsum-forming solutions for the reduction of brine inflows into potash mines is under development.
25.4 IN SITU IMMOBILIZATION BY CRYSTALLIZATION PROCESSES Mineral-forming solutions can be used not only for solving geotechnical problems by sealing, but also for immobilization of diluted or soluble contaminants. Calcium carbonateforming solutions, and grouts leading to BaSO4 precipitation, are of primary interest in this context. CaCO3-forming solutions can be used for in situ neutralization of acidic pore water as well as for the construction of zones acting as reactive walls, such as pH buffer. The use of BaSO4 solutions offers the possibility of overlaying reactive mineral surfaces with a mineral characterized by an extremely low solubility. In addition, it is possible to fix heavy metals as coprecipitates or as solid solutions based on BaSO4. Apart from the already discussed use of precipitation inhibitors for the stabilization of supersaturated solutions, it is possible to construct reactive systems, leading to directed Fe(OH)3, Al(OH)3, or Ca(OH)2 precipitation. Sodium aluminate solutions can be used as sources for Al(OH)3 precipitation. These are strongly alkaline solutions. In the case of neutralization, for example, by removing OH−, Al(OH)3 precipitation takes place. This can be
ELSE_TSANG_CH025.qxd
8/23/2005
352
2:11 PM
Page 352
Grouting with Mineral-Forming Solutions—A New Technique for Sealing
achieved by the addition of compounds, which decompose by consuming hydroxide ions. The reaction leads to a decrease in pH and results in timely, controlled Al(OH)3 formation: Proprietary compound ⫹ OH⫺compound 2 [Al(OH)4]−Al(OH)3 ⫹ OH− As for the already-described systems, it is typical for Al(OH)3-producing systems to be clear solutions with low viscosity. Favorable applications are seen in the in situ construction of adsorption barriers, for the cleaning of heavy metal-containing plumes, for example. The application of grouts supersaturated in BaSO4 was first tested and is now used in large scale in the former Koenigstein uranium mine of the Wismut GmbH company (Germany). Uranium production began in this mine in the 1960s by conventional underground mining. Due to decreasing ore contents, a step-by-step transition to specially developed underground in situ leaching technologies with diluted sulfuric acid took place. Several in situ leaching technologies were used, but only two will be discussed here: ● In Situ Leaching (ISL): hydrodynamic in situ leaching of low-permeable rock via patterns of boreholes (Fig. 25.14). ● Block-pressure leaching: This technology is based on hydrostatic or infiltration leaching of blasted low-permeability rock in hydraulically isolated blocks. These were characterized by volumes in the range of 100,000–1,000,000 m³. To construct such blocks, sections of the ore-containing horizon were blasted and then separated by dams from the other mine areas. In all cases, leaching was performed with a solution of 2–3 g/L of H2SO4. After mining operations ended, it was necessary to develop a strategy for safe closure of the mine. This is highly important because acidic pore waters containing high levels of heavy metals are present in large parts of the mine. One way to prevent groundwater pollution is through in situ immobilization of the contaminants. Such a process has to meet the following demands: ● The immobilization must lead to slightly soluble minerals that are stable under expected flooding conditions. ● An immobilization must not result in an additional input of contaminants to the treated formation.
diluted H2SO4
Output s olution
Fig. 25.14. Schematic illustration of the ISL technology used in the Koenigstein mine of the Wismut GmbH company (not to scale).
ELSE_TSANG_CH025.qxd
8/23/2005
2:11 PM
Page 353
25.4 In Situ Immobilization by Crystallization Processes
353
The process has to be applicable under typical underground mining conditions. Several field tests were performed on massive sandstone formations and blasted material before large-scale application could be realized. The immobilization of areas where uranium recovery was based on the hydrodynamic in situ-leaching technology was tested through the use of BaSO4 supersaturated solutions. These were prepared through step-by-step mixing of Ba(OH)2 solution, inhibitor solution, and diluted H2SO4. In two technical tests, approximately 5000 m3 of BaSO4 solution, containing between 100 and 280 mg/L of BaSO4, were injected into a test area of approximately 40,000 m3 of sandstone. The solution was pumped into the sandstone. After passing the formation, the output solution was collected at a lower horizon. Injection progress is summarized in Figure 25.15. Step-by-step increase of the injection pressure was necessary to force the solution into the formation, indicating that the permeability was significantly reduced. The dimensions of this effect were surprising because the absolute amounts of BaSO4 injected into the formation were low in relation to the pore volume. The test showed that it is not necessary to fill the flow paths with secondary minerals to create a seal. Even closure by single crystals leads to a drastic reduction of the permeability. After finishing BaSO4 grouting, the test area was treated for 3 weeks with freshwater. Constant volume flow of 20 m3/day was delivered at a constant pressure of approximately 4.2 bar. Signs indicating dissolution processes were not found. The produced seal remained stable; otherwise, a reduction of the injection pressure, followed by an increase of the volume flow, would have occurred. A comparison with tests in which similar sectors were flushed with water has shown that the eluates of the BaSO4 treatment are characterized by lower heavy metal content and faster increases in pH values. ●
7
250 BaSO 4 treatment
6 200
150
4 washing with water
100
3
pressure [bar]
3
volum e fl ow [m /d]
5
2 50
volume flow 1 pressure 16 .11
6. 11
11 .11
1. 11
27 .10
22 .10
17 .10
7. 10
12 .10
2. 10
27.9
22.9
17.9
7.9
12.9
2.9
28.8
0 23.8
0
date
Fig. 25.15. Progress of injection pressure and volume while grouting ISL-leached areas with BaSO4 solutions.
ELSE_TSANG_CH025.qxd
354
8/23/2005
2:11 PM
Page 354
Grouting with Mineral-Forming Solutions—A New Technique for Sealing
Solutions based on Ba(OH)2, a precipitation inhibitor, and Na2SO3 were used to treat blasted rock material encapsulated in blocks (Fig. 25.16). The blocks contained an unknown amount of acidic pore water and were in contact with moisture and air for years. To reduce further mobilization of metals due to oxidation, sodium sulfite was added to the injection solution as a reducing agent. Oxidation resulted in the formation of sulfate, but spontaneous BaSO4 precipitation was prevented due to the presence of an inhibitor. The first test was performed with a solution characterized by an average composition of 115 mg/L of SO32− and 200 mg/L of Ba2+. To increase the immobilization capacity, sodium silicate solution was added until a final SiO2 concentration of 120 mg/L was reached. Within 6 days, a total of 3540 m3 of BaSO4-forming solution was injected into the block. Subsequently, about 1200 kg of BaSO4 was formed within the block. Solution preparation occurred without any problems under typical mining operation conditions. The solution remained in the block for 10 days. After that time, the solution was drained off with a volume flow of 20 m3 hr−1 initially. In the second step, the block was filled with BaSO4forming solution and drained again. After draining, the block was flushed with freshwater in order to demonstrate the stability of the formed secondary minerals. The amounts of contaminants discharged during the different treatment periods are summarized in Figure 25.17. A stepwise reduction of discharged quantities is apparent. The first immobilization step resulted in solutions containing a total of 11.6 kg uranium, 363 kg iron, and 101 kg zinc. After the second treatment step with BaSO4-forming solutions, a total output of only 8 kg U, 184 kg Fe, and 52 kg Zn was observed. The flushing of the treated block with freshwater did not lead to an increased release of contaminants. To the contrary, a further reduction of the total load was observed. This demonstrates the stability of the formed immobilization products against leaching. Otherwise, a significant increase of released quantities would have been observed. Thermodynamic calculations and column tests in the laboratory have shown that the fixation of heavy metals is based on the following three mechanisms: ● Overall pH increase and formation of hydroxides as well as hydroxysulfates ● Overlaying of reactive mineral surfaces by insoluble minerals such as BaSO4 ● Coprecipitation as well as formation of mixed crystals based mainly on BaSO4. To evaluate effectiveness of the immobilization, it is necessary to compare the amounts of contaminants discharged after treatment with those that would be expected when washing the block only with water. To obtain the required data, a similar block was treated with water in a second test. Because it is impossible to compare the absolute eluate concentrations due to
Blasted rock
Injection hole
Collection hole
Grout plant
Fig. 25.16. Schematic illustration of the treatment of leached blocks with BaSO4 solutions (not to scale).
ELSE_TSANG_CH025.qxd
8/23/2005
2:11 PM
Page 355
25.4 In Situ Immobilization by Crystallization Processes
355
700
discharged U, Zn, Al, Fe [kg], Cr, Cd, Cu [g]
600 500 400
1. Immobilization 2. Immobilization flushing with water
300 200 100 0 Cr
Cd
Cu
Al
Fe
U
Zn
Fig. 25.17. Total amounts of contaminants discharged during BaSO4 treatment and final water flushing of leached blocks in the Koenigstein mine.
inhomogeneities in the rock material, the development of the concentrations was used to evaluate the effects of the BaSO4 treatment. The following procedure was used. First, the block was filled with water. After a retention time of 7 days, the block was emptied. A total solution volume of 1233 m3 was collected. The starting uranium concentration was 5.47 mg/L. After 762 m3, the solution was characterized by a uranium concentration of 4.25 mg/L. In other words, after collecting 61.8% of the total eluate volume, the uranium concentration decreased to 77.7%. After the outflow of 1049 m3 of solution, the uranium concentration was 10.5 mg/L. That means, after collecting 85% of the total eluate volume, the solution was characterized by uranium concentrations equivalent to 192% of the starting concentration. The final solution contained 13.3 mg/L of uranium. The same procedure was applied to determine the theoretical uranium concentration when flushing the “BaSO4 block” with water. For example, the starting concentration was 9.8 mg/L uranium. After collecting 61.8% of the solution, the uranium content should have decreased to 7.5 mg/L when flushing with water. The concentration found after flushing with BaSO4 solution was 4.7 mg/L. The complete results of the calculations are given in Figure 25.18. The course of the uranium concentrations is represented in relation to the amount of collected eluate. The points give the calculated uranium concentrations when flushing the block with water, the triangles characterize the analytically determined values for eluates obtained after the BaSO4 treatment. The main difference was that the immobilization resulted in a significant drop in the concentrations, whereas flushing with water only produced solutions with slightly decreased concentrations. The effect of the immobilization is evident not only in the change in concentrations, but also in the total amounts of discharged contaminants. In both immobilization stages, the discharged amounts of contaminants were 50–70% lower than would be expected when flushing the block only with water (Fig. 25.19). Results from the field tests were the basis for the decision to apply the developed technology in selected areas of the Koenigstein mine. Three prepared, but unleached, blocks were immobilized by applying approximately 100,000 m3 of BaSO4-forming solution. The blocks were prepared for leaching in the 1980s, but leaching was not carried out due to the decision to stop uranium production in 1990. Contact with air and moisture was possible for
ELSE_TSANG_CH025.qxd
8/23/2005
356
2:11 PM
Page 356
Grouting with Mineral-Forming Solutions—A New Technique for Sealing 30
Uranium [mg/l]
25 20 15 10 5 0 0
10
20
30
40
50
60
70
80
90 100
collected solution [% of total eluazte volume] Uranium concentration after flushing with BaSO4 solution
Uranium concentration after flushing with water
Fig. 25.18. Comparison of the change in effluent uranium concentrations after flushing with BaSO4 solution and after flushing with water only.
35
discharged uranium [kg]
30 25 20 15 10 5 0 1. Immobilization
2. Immobilization
uranium discharged after immobilization discharged uranium in the case of flushing with water
Fig. 25.19. Comparison of the amounts of uranium discharged through immobilization and through flushing only with water.
ELSE_TSANG_CH025.qxd
8/23/2005
2:11 PM
Page 357
25.5 Conclusions
357
more than 15 years. When flushed with water, the formation of highly concentrated, acidic solutions would be expected. The goal of treatment with BaSO4-forming solutions was to stop oxidation processes and reduce the discharge of contaminants during the later mineflooding process. Figure 25.20 gives an impression of the grout plant used during largescale application of supersaturated BaSO4 solutions. As in the field tests, the solution used was composed of Ba(OH)2, Na2SO3, water glass, and a precipitation inhibitor. The treatment of the blocks was undertaken between December 2001 and June 2002. Because the solutions remained in the filled sectors of the mine, only limited data were available for analysis. Samples from the mine showed concentrations that were dramatically lower than those found in areas where water flushing was applied.
25.5 CONCLUSIONS The technology of sealing and immobilization by directed crystallization processes can find many applications, both to solve geotechnical tasks and to remediate contaminated areas. The technology combines many advantages, such as: ● Treatment and sealing, or immobilization, of areas where conventional grout materials cannot be applied. ● Sealing and immobilization can be achieved by minerals naturally present in the formation. ● Immobilization processes can be coupled with permeability reduction. Different technologies—such as spraying, penetration grouting, pressure grouting, infiltration from ponds or reservoirs—can be used to transport the mineral-forming solutions into the area that has to be treated. The simple preparation of the solutions can be performed both on the surface and under conditions typical for underground mining. The total cost for chemicals is low, and only standard equipment is necessary. Only non-toxic, environmentally friendly compounds are used. There are many opportunities to direct and control the process of mineral formation. This can be achieved both by selection of the components for the grout preparation, and by their concentrations. All grout or penetration agents are pure solutions with a viscosity similar or equal to natural groundwater. Good penetration or infiltration even
Fig. 25.20. View of the grout plant used for preparation of the solutions.
ELSE_TSANG_CH025.qxd
8/23/2005
358
2:11 PM
Page 358
Grouting with Mineral-Forming Solutions—A New Technique for Sealing
occurs with soils of low permeability. The first large-scale applications have demonstrated successful implantation of the technology for immobilizing contaminants and for sealing porous formations.
ACKNOWLEDGMENTS The summarized investigations were partially carried out together with Wismut GmbH, or were funded by the European Commission in the “CRYSTECHSALIN” project.
REFERENCES Blowes, D.W., Reardon, E.J. and Cherry, J.A., 1991. The formation and potential importance of cemented layer in inactive sulfide mine tailings. Geochim. Cosmochim. Acta, 55: 965–978. Chermak, J.A. and Runnels, D.D., 1996. Self-sealing hardpan barriers to minimize infiltration of water into sulfide-bearing overburden, ore and tailing piles. In: Tailings and Mine Waste, 1996 Proceedings of International Conference on Tailings and Mine Waste, pp. 265–273. Gaitzsch, H., 1992. Thesis, TU Bergakademie Freiberg. Graupner, U., 2002. Unpublished results of the CRYSTECHSALIN project, TU Bergakademie Freiberg. Karol, R.H., 1990. Chemical Grouting, 2nd edn., Marcel Dekker, New York. Kuehne, R., 1993. Control of inflows by efficient use of oceanic solution equilibrium. Kali Steinsalz, 11(3–4): 85–89. Rammlmair, D., 2002. Hard pan formation on mining residuals. In: B.J. Merkel, B. PlanerFriedrich and C. Wolkersdorfer (Eds), Uranium in the Aquatic Environment, Proc. Mining and Hydrogeology III, 15–21 September 2002, Freiberg, Springer, Berlin, pp. 173–182. Ziegenbalg, G. and Crosby, K.S., 1997. A review of a pilot test to reduce brine inflows with controlled crystallization of gypsum at the IMC K2 brine inflow. Miner. Resour. Eng. 6(4): 173–184.
ELSE_TSANG_CH026.qxd
8/23/2005
6:42 AM
Page 359
IV. LIQUID WASTE INJECTION
359
This page intentionally left blank
360
ELSE_TSANG_CH026.qxd
8/23/2005
6:42 AM
Page 361
Chapter 26
INJECTING BRINE AND INDUCING SEISMICITY AT THE WORLD’S DEEPEST INJECTION WELL, PARADOX VALLEY, SOUTHWEST COLORADO K. Mahrera, J. Akea, L. Blocka, D. O’Connella, and J. Bundyb a
U.S. Bureau of Reclamation, Denver, CO, USA Subsurface Technology, Inc., Houston, TX, USA
b
26.1 INTRODUCTION Deep, high-pressure fluid injections induce earthquakes. Most are microearthquakes, detectable only by instrumentation (i.e., seismometers) in neighboring wells at or near the injection depth. In deep injections, we estimate that nearly 0.1% of the events can be recorded by standard earthquake seismometers at the surface, and less than 15 km from the injection well. A small percentage of the surface-recorded events can be felt by humans if the injection exceeds injection volume, injection pressure, and duration thresholds (e.g., Baisch et al., 2002). In the early 1960s, deep high-pressure injection near Denver, Colorado, induced over 1300 surface-recorded earthquakes. The largest was a magnitude M5.6 in a region where few felt earthquakes had occurred since the late nineteenth century (Healy et al., 1968). Beneath a remote section of southwestern Colorado, the Paradox Valley Unit (PVU), a U.S. Bureau of Reclamation Project injects high-pressure waste brine injectate into an ∼4.3 km deep limestone formation in order to reduce salinity in the Colorado River. Since inception in 1991, injection has induced over 4100 surface-recorded earthquakes, the largest being a magnitude M4.3; more than 99.9% of the surface-recorded events are microseismic, below human detection (M2.5). 26.2 THE PROJECT The Colorado River supplies water to ∼23 million people and about 1.6 million ha in southwestern U.S. At present, excess salinity in the Colorado River causes damage in excess of $500 million/year. Without rigorous intervention, damage is predicted to exceed $1 billion by the second decade of the twenty-first century (Barnett, 1999). Normal Lower Colorado River salinity is >700 mg/L (ppm) total dissolved solids (TDS); for reference, freshwater is defined as <450 mg/L tds. Without intervention, the Dolores River, a tributary of the Colorado, adds ∼2 × 108 kg tds/year to the Colorado salinity. The Dolores receives its TDS, mainly sodium chloride (i.e., salt), by seepage from a briny aquifer as the river traverses Paradox Valley, a salt diapir. This Paradox Valley brine (PVB) has a TDS load of about 250,000 mg/L, which is eight times the salinity of seawater. To reduce the influx, PVU extracts the brine from the aquifer before entering the Dolores, treats it, and injects it, ∼4.3 km below the surface into the Mississippian-aged Leadville Limestone. By the end of 2003, PVU will have injected ∼4 × 109 l of brine, disposing of ∼7 × 108 kg of salts.
ELSE_TSANG_CH026.qxd
8/23/2005
362
6:42 AM
Page 362
Injecting Brine and Inducing Seismicity at the World’s Deepest Injection Well
PVU includes a 4.8 km injection well, the world’s deepest disposal well; ∼90 shallow extraction, monitoring, and testing wells along the Dolores in Paradox Valley; a treatment and pumping facility; and a local 15-station seismic network. PVU injects around the clock (with the exception of two 20-day shutdowns each year) at an estimated benefit of ∼$136/metric ton of salt disposed, and an operating cost of ∼$76/metric ton. (For more details, see Bundy, 2003.) Until January 2002, PVU injected 70% PVB mixed with 30% freshwater, 0.26 metric tons of salt per m3 of injectate emplaced; from January 2002 until present, PVU injected and continues to inject, at 100% PVB, 0.37 metric tons of salt per m3 of injectate emplaced. 26.3 LOCAL GEOLOGY PVU sits ∼2.5 km south of Bedrock, CO (pop. ∼100, Fig. 26.1) on the western edge of Paradox Valley, a northwest-trending, collapsed diapiric salt anticline. The Valley, which is about 40 km long and 8 km wide, and underlain by ∼6 km of interbedded salts and shales, is one of a number of local basins that began forming about 250 million years ago when mountain uplifts created lateral stresses on the intervening sedimentary formation. Faults and fractures formed along weak axial zones. Subsequent stress relaxation, combined with the weight of overlaying strata, forced deeply buried salt to flow up into the faulted area, creating the diapiric anticline. Subsequent uplifts (present underlying stratigraphy dips ∼15° to the east), extensions, and unequal erosion created the current topography of Paradox Valley. The Dolores River flows across Paradox Valley near its midpoint. PVU’s primary injection target, the Leadville Formation, is a locally vuggy, highly fractured, tight dolomitic limestone with an effective (i.e., fracture-based) porosity less than 6%. Within the Leadville and surrounding formations is the Wray Mesa Fault system. The Wray Mesa underlies and borders the west side of the anticline. This relationship is shown in Figure 26.2, the initial cross section developed by Bremkamp and Harr (1988) prior to the drilling of the injection well
Injection Well
Fig. 26.1. Location map of Bedrock, CO; Paradox Valley; and the Paradox Valley Injection Unit. Dashed line ( ) traversing Paradox Valley denotes the surface expression of the cross section shown in Figure 26.2
----
ELSE_TSANG_CH026.qxd
8/23/2005
6:42 AM
Page 363
26.4 Injection Well and Operations
363
Fig. 26.2. Initial interpretation (Bremkamp and Harr, 1988) of the Paradox Valley cross section prior to drilling the PVU Injection Well. The cross section is normal to the northwest axis of Paradox Valley (see dotted line in Fig. 26.1). PVU-induced seismic source locations (1991–2002) are projected onto the cross-section plane. Implied normal (i.e., main through going) faults of Wray Mesa Fault System also shown.
and based on sparse surface seismic and local well log data. Based on the induced seismic data, the locations of these faults may not be correct. However, the existence of these faults is not in question. This fault system, which dips ∼80° to the northeast, has its primary faults striking subparallel to Paradox Valley’s long axis at ∼N55°W and has an extensive secondary fracture system. The injection well was sited to intersect the Wray Mesa system. A pressure analysis of existing deep wells in Paradox Basin by Bremkamp and Harr (1988) predicted fluid migration (i.e., a pressure gradient) along faults and fractures trending to the northwest. Our analysis of the induced seismic event locations and migration over time confirms this prediction.
26.4 INJECTION WELL AND OPERATIONS The study of the Paradox Valley–Dolores River brine seepage problem began in 1971, and by the late 1970s, the project operated more than 90 shallow (12–21 m) brine extraction, monitoring, and test wells bordering the Dolores River. Today, PVU extracts brine using nine of these wells, and injects using the ∼4.9 km (16,000 ft) PVU salinity control well No. 1 (Fig. 26.3). As noted in Figure 26.3, Well No. 1 is perforated between 4.3 and 4.9 km deep; the perforation rate is ∼20 perforations/m. The perforation interval includes ∼200 m of the Leadville, underlying Devonian sandstones, and ∼70 m of Precambrian aphanitic schist. Following perforation, a series of freshwater mini-hydraulic fractures indicated that fracturing occurred between 27 and 29 MPa surface pressure (Envirocorp, 1995), which corresponds to ∼69 MPa (∼10,000 psi) bottom-hole pressure. (Note: 69 MPa fracture pressure corresponds to fracture pressure prior to injection; noting the years of injection at Paradox, the formation has tightened, and fracture pressure is now above 69 MPa.) Because it is a brine disposal system, PVU had to qualify for a U.S. Environmental Protection Agency (EPA) Class V permit (Bundy, 2003). Testing the well to qualify for the permit began in 1991; the testing plan called for seven pump-and-shut-in tests that varied
ELSE_TSANG_CH026.qxd
364
8/23/2005
6:42 AM
Page 364
Injecting Brine and Inducing Seismicity at the World’s Deepest Injection Well
North, km 0.5
0
Triassic
Wellbore 1
0 -0.5
0
0.5 East, km
Depth, km
Perforations
Permian
2 Pennsylvanian Mississippian
3
Devonian
4 5
Cambrian Pre-Cambrian
6 -0.5
0
1 North, km
Plan View
Cross Section
Fig. 26.3. Profile of Paradox Valley Injection Well No. 1. Cross section is viewed from the east, looking west.
injection rate and injectate chemistry, tested the well integrity, and confirmed the formations’ proclivity to take injectate. By late 1995, the tests confirmed the system, and the EPA granted the permit. Around-the-clock injection began in July 1996. Wellhead safety rating restricts surface pressure to or below 34.5 MPa (5000 psi). During pumping, surface treatment includes filtering (0.076 mm filter), adding corrosion inhibitor (after 1997), and mixing 70% PVB with 30% freshwater before injecting. The brine was diluted to 70/30 (specific gravity = ∼1.119) because geochemical lab analysis predicted that 100% brine, when contacting the connate fluids and the Leadville Limestone at preinjection downhole pressure and temperature, would likely precipitate anhydrite (i.e., CaSO4) in the perforation zone, inhibiting injection (Kharaka et al., 1997). In 2001, after five years of (nearly) continuous injection, temperature logs indicated that the near-wellbore region had cooled sufficiently so that, in the near-wellbore region, the risk of significant anhydrite precipitation was abated. With the risk of precipitation in the near-wellbore, especially the perforation zone, considered unlikely, PVU began injecting 100% PVB plus inhibitor (specific gravity = ∼1.17) in January 2002. To date, we have seen no adverse effects from pumping 100% PVB. 26.5 PARADOX VALLEY SEISMIC NETWORK Based on concerns over injection-induced earthquakes, PVU installed a seismic network, the Paradox Valley seismic network (PVSN), in 1985. The network operates two three-component seismometers (a third is planned) and 13 vertical-only seismometers (Fig. 26.4). On the surface, the network uses water-tight vaults that hold 1 Hz seismometers (i.e., Teledyne Geotech S-13s, 1-Hz velocity transducers). The network is loosely arranged in two concentric circles surrounding the injection well; one circle is within ∼10 km of the well, the other
ELSE_TSANG_CH026.qxd
8/23/2005
6:42 AM
Page 365
26.6 PVSN Recording Sensitivity
365
Fig. 26.4. Locations of the 15 Paradox Valley seismic network sites (blue triangles). The white triangle is a planned 16th monitoring site. White squares are local municipalities. The gold star is an injection well on the southwest edge of Paradox Valley.
within ∼25 km. All sites have extremely low cultural noise, allowing microearthquake detection to M0.5, and reliable source location for M 0.0. The data indicate incomplete earthquake detection below M0.7 (Mahrer et al., 2001). Each PVSN station continuously radios analog data to Nucla, CO, ∼30 km southeast of Bedrock, for digitization and telemetry to the Denver Federal Center (DFC). At DFC, the data are event-detected, classified, and archived for further analysis, after finding and mapping preliminary seismic source locations.
26.6 PVSN RECORDING SENSITIVITY Regarding the issues of recording sensitivity, surface monitor, and number of events induced, Figure 26.5 shows a plot of the number of recorded events as a function of injected volume for seismically monitored hydraulic fracture injections and the Paradox injection. The hydraulic fracture data are from Phillips et al. (2002) and are recorded in situ within a few hundred meters of the injections at or near the depth of injection in observation wells. The minimum projected magnitude of the Phillips et al. data is between M2.0 and M3.0 (personal communication, Phillips). In Figure 26.5, the diagonal dashed line through the Phillips data shows the general linear trend of these data (i.e., on a log–log plot). We copied and displaced a line parallel to the Phillips et al. data, but through the Paradox data, which
ELSE_TSANG_CH026.qxd
8/23/2005
366
6:42 AM
Page 366
Injecting Brine and Inducing Seismicity at the World’s Deepest Injection Well 1.0E+07
Paradox Continuous Injection ('96-'02) + (Cumulative) Injection Tests Sedimentary (Phillips et al.)
1.0E+06 Crystalline (Phillips et al.)
Number of Events
1.0E+05
Paradox Injection Tests
1.0E+04
1.0E+03
1.0E+02
1.0E+01
1.0E+00 1.0E+00
1.0E+01
1.0E+02
1.0E+03 1.0E+04 Injected Volume, cu-m
1.0E+05 1.0E+06
1.0E+07
Fig. 26.5. Number of events recorded as a function of injected volume at hydraulic fracture treatments (Phillips et al., 2002) and in Paradox Valley (red). Squares and triangles show hydraulic fracture data recorded in situ at the injection depth and within a few hundred meters of the injection well. Note the type of rock given in legend. Diamonds are the Paradox data (tests and continuous pumping) recorded at surface by PVSN. The dashed line in hydraulic fracture data is fitted; the dashed line through Paradox data is a fitted line parallel to the hydraulic fracture data line. The circled value (∼2 million) is the projected number of events that would be recorded at Paradox if recording at Paradox were done in situ at depth, near the injected well.
show a similar linear trend. From the end of the Paradox continuous pumping data (i.e., the total number of events recorded by PVSN), we projected vertically (the arrowed line) to the Phillips data (see circle in Fig. 26.5). The circled value in Figure 26.5 approximates the number of events that we estimate could be recorded at Paradox in situ in wells neighboring the injection well. This value is ∼2 × 106 and shows that PVSN recording sensitivity corresponds roughly to 0.1% of the events M3.0 and greater induced at Paradox.
26.7 SEISMICITY AND INJECTION With regard to the injection at PVU, PVSN has monitored seismicity over three sequential time periods: preinjection (1985–1991), injection testing (1991–1995), and continuous injection (1996–present). 26.7.1 Preinjection Generally speaking, the Paradox Valley region is seismically inactive (Wong et al., 1996). Prior to PVSN, the nearest seismic monitoring stations were more than 160 km from the
ELSE_TSANG_CH026.qxd
8/23/2005
6:42 AM
Page 367
26.7 Seismicity and Injection
367
proposed PVU site. Between 1850 and 1983, the National Oceanic and Atmospheric Administration Website lists 24 events within 160 km of the injection well. The largest event was M5.5, about 110 km east of the site; earthquakes less than M4.0 prior to 1900, and M3.0 after 1900 probably were not detected prior to PVSN. In ∼6 years prior to injection (1985–1991), PVSN detected six minor earthquakes, within the 19,000 km2 covered by PVSN. None of these events was within 10 km of the injection well, and no earthquakes were felt (Block et al., 2001). 26.7.2 Injection Tests Between July 1991 and April 1995, PVU ran seven injection tests that pumped continuously between 25 (3600 psi) and 32 MPa (4600 psi) surface pressure. Each injection was followed by a well-head shut-in and pressure fall-off period (see Table 26.1 and Fig. 26.6). During the first 10 days of Injection Test 1, PVSN located 20 (induced) earthquakes within 0.25 km of the well. The first event was recorded 3 days after Test 1 began. Table 26.1 gives the injection volume, pumping duration, injectate (i.e., injected fluid), and number of seismic events induced by the injection tests. Note that after the acid stimulation of the formation, the number of induced events per test or per injected volume increased dramatically. Figure 26.6 shows the wellhead injection pressure, the number of events per day, and the total number of events recorded during each test (boxed number on bottom of graph). Through Test No. 6, all the seismicity was located within 2 km of the injection well. By the end of Test No. 7, the seismogenic zone had extended to the west by an additional 2 km. 26.7.3 Continuous Injection Continuous injection (i.e., continuous disposal of PVB) began in May 1996. Between May 1996 and October 2004, PVSN located 3435 induced earthquakes, giving a total of more than 4100 earthquakes induced by the PVU injection and recorded at the surface. The first
Table 26.1. Injection Tests 1991–1995* Test. no.
Injected volume, m3
Pumping duration, days
Injectate,
1 2 3 4 — — 5 6 7
11,000 16,000 54,000 42,000 38 34 54,000 89,000 354,000
14 12 54 47 — — 28 41 242
0%:100% 33%:67% 67%:33% 0%:100% 28% HCl acid stimulation Freshwater flush following acid‡ 70%:30% 70%:30% 70%:30%
20 9 16 0 — — 81 170 370
Total
620,072
438
—
666
*
No. seismic events
%PVB† : %Freshwater
See also Figure 26.6. † PVB Paradox Valley brine (260,000 mg/L). ‡ Injection well surface pressure became negative (i.e., “went on vacuum”) following water flushing of acid into formation.
ELSE_TSANG_CH026.qxd
368
8/23/2005
6:42 AM
Page 368
Injecting Brine and Inducing Seismicity at the World’s Deepest Injection Well
Fig. 26.6. Injection Tests at PVU: pressure (blue) and events per day (red) as functions of time. Blocked numbers near the top of the figure give injection test numbers. Blocked numbers at the bottom of the figure give the number of seismic events recorded during neighboring tests, and during shut-ins following the tests.
event recorded at the surface occurred 111 days following the inception of continuous injection. Since continuous pumping began, about a dozen of the events were felt by humans; the first felt event occurred in August 1997. In May 1999 an M3.5 and in June 1999, an M3.6 event, the largest induced event up to that time, occurred. In May 2000, the largest induced earthquake, a magnitude M4.3, occurred. Responding to events, PVU instituted mitigation procedures to reduce the proclivity to produce larger events. Based on the seismicity and the need to maintain economic viability, PVU has instituted three procedural changes in pumping schedules, resulting in four pumping phases, Phase I through Phase IV: ● Phase I—From May 1996 until the M3.6 event in June 1999, PVU injected about 1290 L/min (345 gpm) at ~33 MPa (4800 psi) surface pressure and ~80 MPa (11,600 psi) downhole pressure, while injecting 70% PVB mixed with 30% freshwater. ● Phase II—Following the June 1999 M3.6 event, PVU augmented injection to include the same injection pressure and rate as Phase I plus a 20-day shutdown (i.e., “shut-in”) every 6 months (i.e., in December–January and in May). The intent was for the injectate from the pressurized fractures and faults to diffuse into the formation rock matrix (i.e., in situ stress relaxation). As discussed below, this reduced the amount of seismicity, but did not reduce the proclivity to produce large seismic events. ● Phase III—Following the May 2000 M4.3 earthquake, PVU reduced the injection rate ∼33% to 870 L/min (230 gpm), leading, initially to a ∼10% reduction in surface pressure (but same bottom-hole pressure). Together, the biannual 20-day shutdowns and lower injection rate reduced earthquake production. From 1998 through the M4.3 event, PVSN recorded an average of 81 earthquakes/month; following the reduced injection in late June 2000 through the end of 2002, that average dropped to less than nine earthquakes/month. In 2002, the average was five/month, with no induced events in April 2002.
ELSE_TSANG_CH026.qxd
8/23/2005
6:42 AM
Page 369
26.8 Seismicity and Local Geology
369
Phase IV—As noted above, PVU began injecting 100% brine at the Phase III injection pressure and rate in January 2002. This is the present injection schedule: 100% PVB at ∼870 L/min (230 gpm) and a 20-day shutdown every 6 months. (Note: At the inception of Phase IV, the surface pressure was ∼30 MPa (∼4400 psi), and downhole pressure was ∼79 MPa (∼11,500 psi). By November 2003, the surface pressure had slowly increased to ∼32 MPa (∼4650 psi), with the accompanying downhole pressure also increasing by ∼2 MPa (∼290 psi). Figure 26.7 shows monthly injection volumes and monthly earthquake production. These data are for events greater than magnitude M0.0 from the beginning of continuous injection in mid-1996 through the end of 2002. Note the pumping phases and the correspondence with the two largest earthquakes. In conjunction with Figure 26.7, Figure 26.8 shows the pumping phases, the bottom-hole pressure, the cumulative number of seismic events recorded by PVSN, and the larger magnitude events (i.e., M 2.0) as a function of time. Figures 26.7 and 26.8 show that the progression of pumping strategies from Phase I through Phase IV has reduced the overall rate of event production and controlled the production of large earthquakes. In Phase II, the overall production of events was reduced, but the change in the production rate of large events was negligible. Implementing Phase III reduced the large events, but also hurt the economics of salt tonnage injected. Phase IV increased the economic benefit by injecting more salt without increasing the large earthquake production. ●
26.8 SEISMICITY AND LOCAL GEOLOGY PVSN identifies some characteristics of fluid flow by mapping earthquake locations. We analyzed nearly 650 earthquakes and developed a three-dimensional (3-D) velocity model
75
Biannual 20-day shut downs
Injected Volume (Megaliters)
50
Reduced Injection Rate
25
0 200
1996
1997
1998
1999
2000
M 3.7 150
2001
2002
No. of Earthquakes (M≥0) I
M 4.3 II
III
IV
100 50 0
Fig. 26.7. Continuous pumping at PVU: monthly injected volume and monthly number of induced earthquakes as functions of time. Note: The four shaded areas, pumping phases I–IV, are discussed in the text.
ELSE_TSANG_CH026.qxd
8/23/2005
Page 370
Injecting Brine and Inducing Seismicity at the World’s Deepest Injection Well 12000
6
10000
5
8000
4
6000
3
4000
2
1
2000 I 0 7-96
Event Magnitude
Num. Events Bottomhole Pres., psi
370
6:42 AM
II
III
IV 0
7-97
7-98
7-99
7-00 Date
7-01
7-02
7-03
Fig. 26.8. Cumulative number of events since 1996 (black line). Bottom-hole pressure and large magnitude events as functions of time. Roman numerals in shaded areas indicate the four pumping phases, discussed in text.
using a progressive 3-D velocity-hypocenter inversion (Block et al., 2001). Using this model and cross-correlations determined from shear minus compression wave travel times (Block et al., 2001), we precisely located about 3500 events. Figure 26.9 shows a plan view of these locations; in the figure, the axes are centered on the injection well. The data in Figure 26.9 include 1991–2002 earthquakes, specifically the source locations. Our analysis of these events shows focal mechanism (i.e., moment) solutions of most corresponding to shear failures that align with the strikes of the Wray Mesa secondary faults and fractures. A few of these focal mechanisms align with the main Wray Mesa faults. The dashed lines in Figure 26.9 indicate our interpretation of likely main through-going Wray Mesa faults, based on the seismic locations and strikes from Bremkamp and Harr (1988). Note how well the strikes of the main faults of the Wray Mesa align with the terminations of seismic source locations. We interpret this to mean that the normal faults (i.e., main faults) of the Wray Mesa system are generally aseismic conduits for fluid movement, either injectate or connate fluids. The aseismicity of the Wray Mesa occurs because the main system is along a principal stress direction (i.e., has no shear stress component to be liberated by the effective pressure increase from fluid pressure) (Block et al., 2001). In Figure 26.9, the source locations divide into two seismically disconnected regions: (1) a large group or cloud asymmetrically surrounding the injection well, and (2) a smaller cloud ∼8 km to the northwest, along the Wray Mesa fault trend. From these data, we infer that fluid-pressure perturbations have migrated at least 8 km from the injection well. Figure 26.9 also shows quasi-parallel lineaments of earthquakes. These lineaments and their relative locations illustrate the secondary conduits of the injectate: large faults and fractures of the Wray Mesa system that are seismically activated.
ELSE_TSANG_CH026.qxd
8/23/2005
6:42 AM
Page 371
26.8 Seismicity and Local Geology
371 North, km
6
4
2
0 -10
-8
-6
-4
-2
0
2
4
-2
-4
Fig. 26.9. Plan view of induced earthquake locations and the earthquake-implied direction of the main (throughgoing) Wray Mesa Fault system (dashed lines). Axis origin is the wellhead.
Figures 26.10 and 26.11 show cross sections of the near-wellbore seismicity viewed looking north and west, respectively. These figures also show simplified stratigraphic cross sections, which are based on logs from the injection well and the known general stratigraphic dip (∼15° to the east). These cross sections do not include the normal faults of the Wray Mesa system, as shown in Figure 26.2. Figures 26.10 and 26.11 show that the earthquakes are confined to a relatively narrow depth range that dips (∼15°) to the east following the inferred depth interval of the Leadville. The grouping of events ∼0.5 km to the southwest of the injection well agrees with the inferred location and with the trend of one of the Wray Mesa system faults. A second group of earthquakes ∼2.5 km southwest of the well is probably an unidentified fault. Note that the deepest earthquakes near the well occur below the Leadville within the Precambrian. It should be noted that in Figures 26.10 and 26.11, there are few very shallow events. These are most likely artifacts of the processing for events with small signal-to-noise ratios. One tenet of the processing is cross-correlating seismic signals (Block et al., 2001). High correlation coefficients mean improved event locations in comparison to preliminary velocity model locations. These shallow events have very low cross correlations. We do not feel these depths are accurate nor should these events be considered in the overall interpretations. Using 346 earthquakes, we calculated seismic-source characteristics, which suggest mainly strike-slip motion with the minimum principal stress pointing northeast and subhorizontally. The recorded induced seismicity is not tensile or Mode I fracturing. Pressure analysis during well completion indicated relatively high deviatoric stresses, suggesting that at the injection depth, the maximum and intermediate principal stresses are nearly equal and considerably larger than the minimum principal stress. Fractures observed in oriented cores agree with the direction of the principal stresses. The strikes of the hypocenter-defined lineaments generally agree with the focal mechanism fault planes. The lack of fault planes oriented
ELSE_TSANG_CH026.qxd
8/23/2005
372
6:42 AM
Page 372
Injecting Brine and Inducing Seismicity at the World’s Deepest Injection Well 0
Triassic Mag<1
1
1<=Mag<=2
Permian
Mag>2
Depth, km
2 Pennsylvanian
3
Mississippian Devonian Cambrian
4
5 Pre-Cambrian
6 -5
-4
-3
-2
-1 0 1 Distance, km
2
3
4
5
Fig. 26.10. Cross section looking north, showing near-wellbore induced seismicity (sorted by magnitude) and stratigraphy (implied from the well log sans Wray Mesa faults). The wellbore is represented by the central slanting line.
0
Triassic
Mag<1 1<=Mag<=2
1
Permian
Mag>2
Depth, km
2 Pennsylvanian
3
Mississippian Devonian Cambrian
4
5 Pre-Cambrian
6
-5
-4
-3
-2
-1
0 1 Distance, km
2
3
4
5
Fig. 26.11. The cross section looking west, showing near-wellbore induced seismicity (sorted by magnitude) and stratigraphy (implied from the well log). The wellbore is shown by the central nearvertical line.
parallel to the major through-going faults of the Wray Mesa system suggests these N55°W striking planes are major fluid conduits but lack sufficient shear stress to produce slip (i.e., detectable earthquakes). They are favorably oriented for dilation, normal to the minimum principal stress.
ELSE_TSANG_CH026.qxd
8/23/2005
6:42 AM
Page 373
26.10 Findings
373
26.9 POROSITY AND RESERVOIR LIFETIME In addition to the daily economics of brine disposal, a project concern is how long the formation(s) will continue to take injectate. As originally conceived, PVU is a 100-year project. As a first-cut model, we assumed no new fractures were being created and that all of the injectate was dispersed into the native porosity, uniformly within the seismogenic zone. The seismogenic zone is ∼30 km3; the injectate volume is ∼3.6 × 10−3 km3. To disperse this injectate volume requires a porosity of ∼0.01%. Based on local geology and core samples, the effective porosity of the Leadville Limestone was estimated to be 6% (Bremkamp and Harr, 1988), meaning that after more than a decade of injecting, we seem to have barely begun tapping the available porosity of the reservoir. Note that this simple model assumes (1) that the available porosity is defined by the seismogenic zone, and (2) that there is no new fracture growth. Concerning assumption (1), Baisch et al. (2002) argue that the available porosity in a deep injection can substantially exceed the seismogenic zone. Concerning assumption (2), we know that the injection pressure is in excess of that which would create new fractures. Cumulatively, these assumptions mean a likely reservoir life greater than that predicted by this simple model.
26.10 FINDINGS ● ●
●
●
●
●
●
●
●
●
● ●
●
Injection pressure at Paradox Valley exceeds fracture pressure. The induced seismicity at Paradox illuminates an extensive system. The system is a nonsymmetric maze of fractures, faults, joints, etc., and certainly does not demonstrate the traditionally hydraulic fracture picture of two vertical symmetric fractures emanating from opposite sides of the injection well. Surface-recorded seismic events are slips on preexisting faults, joints, and planes of weakness, not “new-fracture” openings. There are two seismic event clouds: one asymmetrically surrounding the well, and one displaced ∼8 km to the northwest of the injection well along the trend of the known Wray Mesa fault system. More than 99.9% of the over 4100 surface-recorded events induced at the Paradox Valley injection since 1991 have magnitudes less than M2.0 (human detection threshold M2.6). The best estimate of surface recording sensitivity by surface-based PVSN is 0.1% of the induced events with M> −3.0. The largest seismic event, an M4.3 in May 2000, occurred after ∼4 years of continuous injecting. Rates of seismicity are not uniform; there are extended (i.e., multiday) quiet periods, and multihour to multiday active periods and active swarms. The seismic swarms at Paradox are like typical earthquake swarms that usually culminate in one large event, some foreshocks, and a few aftershocks. Seismic events occur as isolated events and in swarms; swarms can occur over hours to days in a single location. Seismic events do not correlate with any detectable surface pressure changes. The seismicity is pervasive with activity still occurring within the interior of the existing seismic clouds; seismicity does not only show growth of the seismic cloud. Event depths are vertically contained.
ELSE_TSANG_CH026.qxd
8/23/2005
374 ●
●
●
● ●
●
●
6:42 AM
Page 374
Injecting Brine and Inducing Seismicity at the World’s Deepest Injection Well
Epicenter patterns align with (i.e., terminate along) major through-going faults of the Wray Mesa system and predicted hydraulic gradients of the target formation. The major fault system aligns with the principal stress direction and acts as fluid conduits showing only minor, if any, surface-recordable seismicity. The seismic locations map the secondary faults and fractures of the Wray Mesa system; the fault planes defined by focal mechanism solutions (i.e., moment tensors) align with the strikes of the faults and fractures. The overall rate of seismic event production changes with injection pressure. Economically reasonable 20-day shutdowns (i.e., “relaxing the reservoir”) somewhat reduce the proclivity for large events. The percentage of brine (i.e., %PVB) in the injectate does not affect seismicity, except through change in bottom-hole pressure due to change in specific gravity of injectate. The storage of injectate is facilitated by the injection pressure exceeding the fracture pressure; the seismically illuminated faults and fractures can only accommodate a few percent of the injectate volume.
26.11 RETROSPECTIVE In addition to the geophysically interesting results, this project shows that by coordinating seismic data with injection operations, we have adjusted brine disposal to maintain economic viability while reducing the proclivity for felt earthquakes. During the 24 months (January 2001 through December of 2002) of pumping at the lower rate, PVSN recorded six earthquakes with M2.0 or greater, and all were below M2.8; pumping at the higher injection rate during 1999–2000, PVSN recorded 36 M2.0 or greater earthquakes, including the M3.5, M3.6, and M4.3 earthquakes. However, reduction is not elimination, and felt earthquakes will occasionally occur. PVU is an economically successful project for reducing the Dolores River’s contribution to the Colorado River’s salinity. At normal groundwater flow, PVU extracts between 40 and 60% of the Paradox Valley brine seepage, depending on how the climatic conditions affect the aquifer. As a result of the success of this project, we are considering a second injection well. ACKNOWLEDGMENTS This work is possible through the continued support of Andy Nicholas, Project Manager, U.S. Bureau of Reclamation Paradox Valley Unit, Bedrock, CO. Thank you! We also recognize Paul Osborne of the EPA-Region 8 for his suggestions and discussions, and Dee Overturf and Tom Bice of the U.S. Geological Survey—Denver for their support of field equipment. REFERENCES Baisch, S., Bahnhoff, M., Ceranna, L., Tu, Y. and Harjes, H.-P., 2002. Probing the crust to 9 km depth: Fluid-injection experiments and induced seismicity at the KTB Superdeep Drilling Hole, Germany. Bull. Seis. Soc. Am., 92(6): 2369–2380. Barnett, J.A., 1999. Executive Director of the Colorado River Basin Salinity Control Forum, Testimony Before the Water and Power Subcom. of the House Resources Committee, Oct. 21.
ELSE_TSANG_CH026.qxd
References
8/23/2005
6:42 AM
Page 375
375
Block, L., Ake, J. and Mahrer, K., 2001. The association between seismicity induced by deep-well injection, injectate migration, and tectonic stresses at Paradox Valley, Colorado. Seis. Res. Let., 72(2): 286. Bremkamp, W. and Harr, C.L., 1988. Area of Least Resistance to Fluid Movement and Pressure Rise Paradox Valley Unit, Salt Brine Injection Project, Bedrock, Colorado (unpublished report contracted by the U.S. Bureau of Reclamation). Bundy, J., 2003. Update—World’s deepest Class V disposal well in its 17th year. Proceedings of the Second International Symposium on Underground Injection Science and Technology, Lawrence Berkeley National Laboratory, LBNL-53836, Symposium Abstracts (with Proceedings CD), Berkeley, CA, Oct. 22–25. Envirocorp, Report of Evaluation of Injection Testing for Paradox Valley Injection Test No., Envirocorp Project No 10Y673 (Unpublished report to U.S. Bureau of Reclamation). Healy, J.H., Ruby, W.W., Griggs, D.T. and Raleigh, C.B., 1968. The Denver Earthquakes. Science, 161: 1301–1310. Kharaka, Y.K., Ambats, G., Thordsen, J.J. and Davis, R.A., 1997. Deep well injection of brine from Paradox Valley, Colorado: Potential major precipitation problems remediated by nanofiltration. Water Resour. Res., 33(5): 1013–1020. Mahrer, K., Block, L. and Ake, J., 2001. 2000 Status Report-Paradox Valley Seismic Network Paradox Valley Project Southwestern Colorado, U.S. Dept. of Interior, Bur. of Reclam., Tech. Mem. D8330-2001-007. Phillips, W.S., Rutledge, J.T., House, L.S. and Fehler, M.C., 2002. Induced microearthquake pattern in hydrocarbon and geothermal reservoirs: Six case studies. Pure and Appl. Geophys., 159(1–3): 345–369. Wong, I.G., Olig, S.S. and Bott, J.D.J., 1996. Earthquake potential and seismic hazards in the Paradox Basin, southeastern Utah, geology and resources of the Paradox Basin. Utah Geo. Assoc. Guidebook, 25: 241–250.
This page intentionally left blank
376
ELSE_TSANG_CH027.qxd
8/23/2005
2:18 PM
Page 377
Chapter 27
EVALUATION OF RESERVOIR INFORMATION IN RELATION TO EARTHQUAKES IN ASHTABULA, OHIO H. Gerrisha and A. Nietob a
U.S. Environmental Protection Agency, Chicago, IL, USA Departments of Geology and of Civil and Environmental Engineering, University of Illinois, Urbana-Champaign, IL, USA
b
27.1 INTRODUCTION Attempts to link earthquakes that occurred near Ashtabula, Ohio, in 1987, 1989, 1990, 2000, and 2001 to industrial waste injections cite the proximity of an injection well to the earthquake hypocenters and the absence of seismic events before an injection began. To determine the reservoir characteristics of the injection zone, we reviewed core analyses, geophysical well logs, pressure transient test results, and operational records. Using an appropriate hydrologic model, we calculated the water pressure distributions within the reservoir, which are likely to be higher than those which actually existed. We did not find a persuasive correlation between calculated water pressure and earthquake timing. We have concluded that causal relations between the injection operation and the Ashtabula earthquakes are not obvious. We also assigned reasonable geomechanical characteristics to the rock masses in the region to evaluate if the earthquakes were induced by the strike-slip reactivation of a preexisting fault. We have proposed a hypothesis whereby these earthquakes are caused by extensional or Mode I failures parallel to the contemporary east–north–east major principal stress in the region. One of the main purposes of this chapter is addressing the ever-increasing acceptance of the injection-induced seismicity along the south shore of Lake Erie. Although the general mechanism for injection-induced seismicity is firmly founded in theory and field observations, we believe that the evidence for (or lack of) causal relations at these injection sites needs to be examined very critically. We hope our effort constitutes an initial step in that direction.
27.2 HISTORICAL OVERVIEW An earthquake sequence including a magnitude mb ⫽ 5.0 main shock occurred near Chardon in northeastern Ohio about 10.5 miles (17 km) south of a newly constructed, but unlicensed, nuclear power plant on January 31, 1986. The event aroused concerns about the safety of the plant. It was noted that the seismic activity had occurred 7.5 miles (12 km), at a depth of 6600 ft (2000 m), from a facility operated by Calhio Chemicals, Inc. (now Arvesta) near Perry, Ohio, where two injection wells are used for the disposal of industrial
ELSE_TSANG_CH027.qxd
378
8/23/2005
2:18 PM
Page 378
Evaluation of Reservoir Information in Relation to Earthquakes in Ashtabula, Ohio
wastes. The facility had injected 4.5 ⫻ 109 gal (1.2 × 106 m3) of nonhazardous wastes with wellhead pressures up to 1625 psi (11.2 MPa) since operations began in 1975 (Nicholson et al., 1988). Beginning on July 13, 1987, a second sequence of earthquakes occurred approximately 3000–6000 ft (1–3 km) from an injection facility operated by Reserve Environmental Services (RES), and about 28 miles (45 km) from the epicenters of the 1986 sequence. The nearness of the epicenters to the injection well suggests triggering. The RES injection well was plugged in 1994, but additional sequences of earthquakes near the original location have occurred as recently as July 2003. Seeber et al. (2003) have argued that all activity near the RES site was caused by the injection activities. Nicholson and Wesson (1990) offered the following three criteria for evaluating the likelihood that seismicity is induced: “First, there was a very close geographical association between the zone of fluid injection and the earthquakes in the resulting sequence. Second, calculations based on the measured or inferred state of stress in the earth’s crust and the measured injection pressure indicated that the theoretical threshold for frictional sliding along favorably oriented, preexisting fractures likely was exceeded. Third, a clear disparity was established between any previous natural seismicity and the subsequent earthquakes, with the induced seismicity often being characterized by large numbers of small earthquakes that persist as long as elevated pore pressures in the hypocentral region continue to exist.” We believe that no proof has been offered to definitively connect the seismic activity in the region around Ashtabula, Ohio, to injection activity. We are not convinced that there is no causal relationship, only that the preponderance of evidence suggests that the activity following closure of the wells was not induced and the occurrence of the later activity so near the location of the earlier activity casts doubt that the injection activity played any role. This chapter describes additional evidence bearing on whether the water pressure increases in the reservoir are likely to have triggered the nine sequences of seismicity between 1987 and 2003.
27.3 REGIONAL GEOLOGY AND TECTONICS The southeastern shore of Lake Erie is in the North American Craton, at the northeast limit of the Appalachian Foreland Basin; as such, it is underlain by subhorizontal Paleozoic sedimentary rocks from the late Cambrian to the late Devonian age with very gentle average dips to the southeast, toward the deeper portion of the basin. Under the sedimentary sequence lies the Grenville Complex. This complex has not been studied in detail, but the limited number of deep wells that penetrated it indicates that it is composed of gneisses and occasional schists (Lidiak, 1996). Aeromagnetic maps show that there are some major lineaments in the region (Braile, 1989). Some of these lineaments have been used to explain the possible presence of seismogenic faults. In particular, the Akron magnetic boundary is thought to be associated with the reactivation of the northeast Ohio seismic zone. On the other hand, Hinze (1996) discussed the sources of errors as well as the quality of gravity and magnetic data that lead to ambiguous interpretation. Both gravity and magnetics are subject to interpretational limitations at both anomaly and regional scales (Grauch, 1993). For instance, we interpret the Akron magnetic boundary as a contact between two lithological provinces with contrasting sets of
ELSE_TSANG_CH027.qxd
8/23/2005
2:18 PM
27.4 Chardon Earthquakes
Page 379
379
elastic properties. These changing material properties could provide the heterogeneities required for extension crack propagation and growth (Cai et al., 1999; Cundall et al., 1996) Seeber and Armbruster (1993) have also associated seismic activity with pre-Cambrian basement features for seismic zones east (Attica, New York) and west (Anna, Ohio) of the area of interest. However, no obvious faults have been found cutting the sedimentary rock sequence in the area of interest, based on the existing subsurface control (a limited number of deep wells), and on several seismic reflection sections run by RES. However, we are aware that some recent work (study for the Perry Nuclear Power Plant, proprietary seismic data, and subsurface mapping by Baranski, 2002) suggests the presence of faulting near the Chardon and Ashtabula sites. Except for the Perry Nuclear Power Plant study, these assertions about faulting are based on interpretative subsurface work and may reflect a bias for finding faulting. We believe that there indeed remain many unmapped faults in the area of interest, but we question whether these are capable faults. Thus, our preferred position is that there is no evidence for capable faults in the region and that the Akron magnetic boundary is the manifestation of a lithologic discontinuity. This allows one to consider earthquake mechanisms other than shear slip. Carter et al. (1996) show that the oil and gas pools in southern Ontario are controlled by faults that displace the pre-Cambrian basement surface and the lower Paleozoic sediments, but do not extend above the middle Devonian. Furthermore, there is no evidence that any of the very mild seismic activity of this region is connected to these faults. The establishment of planar trends in microseismicity focal studies of the region (Armbruster et al., 1987; Seeber et al., 2003) does not prove preexisting faulting; it can simply indicate the development of Mode I failures along a tension joint. Locally, the lithology at the Chardon site differs from that at the Ashtabula site. The Chardon site is underlain by diorite and schist, and the Ashtabula site by metagranite, quartzite, and gneiss (Baransky, 2004, personal communication); however, it can be shown that the two sites have very similar geomechanical properties (e.g., Hoek and Brown, 1980).
27.4 CHARDON EARTHQUAKES The January 31, 1986, earthquake was followed by a series of aftershocks that persisted into April 1986. Many of these earthquakes were less than 7.5 miles (12 km) from the Calhio facility, which uses Class I injection. The main shock’s hypocenter was at a depth in the range of 23,000 ft (6000–8000 m) (Nicholson et al., 1988), far below the depth of the injection wells. Because of the proximity of the epicenter to the Perry Nuclear Power Plant, the sequence of seismic events incited investigations that are of interest to us. The U.S. Geological Survey conducted the most comprehensive investigation (Nicholson et al., 1988). They used mathematical reservoir modeling to determine that the pressure increase due to injection at the hypocenter was likely less than about 50 psi (345 mPa), and may have been just a few psi (30 mPa). Based on the lack of large numbers of small earthquakes, the long lag between the beginning of injection and the seismic activity, and low induced pressures in the hypocentral region, they concluded that the injection wells were most likely not responsible for triggering this sequence of earthquakes. Injection has continued since this sequence of earthquakes, but until June 2003, there had been no additional sequences of earthquakes in the vicinity of the Calhio wells; this increases the weight of evidence pointing toward a natural origin for the 1986 events.
ELSE_TSANG_CH027.qxd
380
8/23/2005
2:18 PM
Page 380
Evaluation of Reservoir Information in Relation to Earthquakes in Ashtabula, Ohio
27.5 ASHTABULA EARTHQUAKES When an earthquake occurred near Ashtabula, Ohio, on July 13, 1987, researchers from Columbia University’s Lamont-Doherty Earth Observatory (LDEO) moved portable seismographs into the area. As a result, there are many well-constrained data on aftershocks (see Map 27.1). However, there was no comprehensive investigation similar to the one Nicholson et al. (1988) performed, accounting for reservoir properties and injection effects following this seismic activity. Complete information about the events is not generally available. The locations of aftershocks of 36 events in the 1987 sequence were determined by using data from a number of temporary networks (Seeber et al., 1993, 2003), which range from 0.5 to 3.5 miles (0.9–5.7 km) from a well used by RES to dispose of a calcium-chloride brine. The location of the main-shock event of July 13, 1987, is not precisely known because of the lack of nearby instruments available at the time of its occurrence; it is believed to be about 1.4 miles (2.25 km) southwest of the RES injection well among the well-located aftershock hypocenters. The hypocenters cluster in a narrow, E–W striking, vertical zone about 1 mile (1.6 km) long and extending from a depth of about 1–2.2 miles (1.7–3.5 km). The indicated fault was named the Ashtabula fault. Seeber et al. (2003) interpret the first motion stereoplots as consistent with a left lateral fault, consistent with the ENE direction of the regional major principal stress. However, these data can be also interpreted as an ENE extensional fracture, as discussed below. Additional shocks were detected in 1989, 1990, 1992, 1995, and 2000 (see Map 27.2). According to Seeber et al. (2003), the hypocenters may lie on the Ashtabula fault, but westward from the original cluster. No portable seismic arrays were deployed, and the hypocenter locations are subject to considerable error. The activity increased in 2001 when two distinct sequences were recorded. The largest (Mblg4.3) shock of the multiyear series occurred on January 26. On June 3, a second sequence began just two days after a deployment of a seismograph network by LDEO. The
Map 27.1. Map of Ashtabula seismic hypocenters and epicenters.
ELSE_TSANG_CH027.qxd
8/23/2005
2:18 PM
Page 381
27.6 Possibility that Injection Triggered Earthquakes
381
Map 27.2. A map of additional shocks that were detected in 1989, 1990, 1992, 1995, and 2000. According to Seeber et al. (2003), the hypocenters may lie on the Ashtabula fault, but westward from the original cluster.
network collected data generated by 51 separate events. Thirteen accurately determined hypocenters defined a tabular region striking N96°E, approximately parallel to and 2.5 miles (4 km) south of the Ashtabula fault. 27.6 POSSIBILITY THAT INJECTION TRIGGERED EARTHQUAKES The occurrence of earthquakes close to an injection well revived the discussion of the possibility that the Chardon earthquake was induced, and some researchers (Seeber and Armbruster, 1993), including Nicholson and Wesson (1990), have become increasingly certain that all sequences of earthquakes near the town of Ashtabula were triggered by brine injection at the RES disposal facility. Other researchers (Fischer, 1990) have pointed out that the pattern of seismic activity at the Chardon site and, by extension, the Ashtabula site, is not typical of other sequences of allegedly induced earthquakes such as the Denver and Rangely, Colorado, and Dale, New York, series. The arrays of seismographs deployed following the detection of activity in 1986, 1987, 2001, and 2003 (Nicholson et al., 1988; Seeber and Armbruster, 1993; Seeber et al., 2003) were capable of detecting very weak seismic activity so that there is a wealth of information about the location of events for several short intervals. In contrast, previous to 1986 and between each deployment, location data for seismic activity in the coverage of the area was poor despite the operation of a seismograph at John Carroll University in Cleveland from 1902 to 1992. In 1999, the Ohio Seismic Network (OhioSeis), a network of seismographs sponsored by the Ohio Department of Natural Resources, was launched. The network now consists of 23 stations, with the nearest being 12 miles (19 km) from the RES well. Only four OhioSeis sites are within 62 miles (100 km) of Ashtabula.
MO
1 7 7 7 7 7 7 7 7 7 7 7 7 7 8 8 8 1 7 9
YR
1983 1987 1987 1987 1987 1987 1987 1987 1987 1987 1987 1987 1987 1987 1989 1989 1989 1990 1990 1990
7 5 13 20 19 7 5 19 23 18 7 14 6 4 16 16 4 23 23 6
HR
46 58 5 53 39 52 49 0 49 25 47 51 2 49 12 50 7 3 4 13
MN 58 52.31 23.6 5.46 19.44 12.88 19.41 8.5 14.5 11.98 27.26 11.67 25.54 40.66 48.75 30.74 48.64 4.89 38.01 4.89
SEC 41.75 41.884 41.899 41.899 41.899 41.898 41.903 41.899 41.902 41.899 41.902 41.902 41.895 41.895 41.898 41.893 41.9 41.902 41.902 41.902
LAT N 81.02 80.7 80.768 80.768 80.768 80.757 80.758 80.768 80.75 80.768 80.75 80.75 80.751 80.751 80.758 80.752 80.761 80.799 80.879 80.749
LON W
2.6
2
DP 3.3 2.2 2.9 2.2 2.1 3 3.8 2.3 2.4 2.8 2.4 2.8 2.4 2.7 2.8 2.9 2.2 2.2 2.3 2.3
MAG 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
MT
IV
MMI
FA
LAKE ASHT ASHT ASHT ASHT ASHT ASHT ASHT ASHT ASHT ASHT ASHT ASHT ASHT ASHT ASHT ASHT ASHT ASHT ASHT
County
NCE JCU JCU JCU JCU JCU JCU JCU JCU JCU JCU JCU JCU JCU JCU JCU JCU JCU/OSN JCU/OSN JCU/OSN
Source
56140.01769 7121.731557 5912.769097 5912.769097 5912.769097 4401.94931 4169.528532 5912.769097 3044.858101 5912.769097 3044.858101 3044.858101 3993.568027 3993.568027 4541.209798 4398.80155 4805.464276 10551.24771 23068.30935 2901.707856
Dist
2:18 PM
22 13 13 13 13 13 13 13 13 13 14 14 16 16 1 1 3 1 24 26
DA
382
8/23/2005
Table 27.1. Seismological data for earthquakes of magnitude > 2.0, with epicenters within 18.6 miles of the RES well
ELSE_TSANG_CH027.qxd Page 382
Evaluation of Reservoir Information in Relation to Earthquakes in Ashtabula, Ohio
1990 1992 1992 1992 1992 1995 1995 1998 1998 2000 2000 2000 2001 2001 2001 2001 2001 2001 2001 2001 2002 2003
11 3 3 3 4 2 4 1 1 6 6 10 1 1 1 1 1 1 6 6 8 2
18 26 28 31 7 23 9 27 30 7 7 20 20 26 26 26 26 26 3 5 17 10
9 3 8 1 1 9 11 0 4 6 6 23 2 3 5 3 5 3 22 8 8 5
20 43 22 54 35 32 37 38 59 55 19 26 5 45 11 11 36 3 36 27 26 34
52.81 15.27 44.06 52.11 22.12 11.99 29.01 30.24 18 8.42 18 26.54 7 25 5 30 58 20.63 46.39 15 31.89 43.07
41.902 41.87 41.86 41.86 41.883 41.87 41.97 42.03 41.97 41.88 42.01 41.86 41.88 41.87 41.87 41.87 41.87 41.87 41.87 41.88 41.79 41.95
80.79 80.87 80.91 80.86 80.85 80.8 80.75 80.99 81.07 80.71 80.78 80.79 80.78 80.76 80.76 80.76 80.76 80.76 80.76 80.76 80.98 80.72 2.5 2.5
2.3 2.5 2.9 2.5 2.0? 2.9 2.4 3 2.4 2.4 2 2.5 2.6 2.2 2 2 3.2 4.5 3.2 2.2 2 2.4
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 86
JCU/OSN JCU/OSN JCU/OSN JCU/OSN JCU GSC/OSN GSC/OSN GSC GSC GSC/OSN GSC GSC/OSN OSN OSN OSN OSN OSN OSN OSN OSN OSN GSC/OSN
9149.662915 23049.47188 29673.29289 22448.66162 19211.6824 13292.03454 13414.06838 47945.04 54584.42677 6817.410743 22820.83866 13557.03551 9516.423463 9081.900805 9081.900805 9081.900805 9081.900805 9081.900805 9081.900805 7302.527408 46170.4816 9107.538206
8/23/2005 2:18 PM
III VI III II NF
III
ASHT ASHT ASHT ASHT ASHT ASHT ASHT LAKE LAKE ASHT ASHT ASHT ASHT ASHT ASHT ASHT ASHT ASHT ASHT ASHT ASH ASHT
ELSE_TSANG_CH027.qxd Page 383
27.6 Possibility that Injection Triggered Earthquakes 383
ELSE_TSANG_CH027.qxd
384
8/23/2005
2:18 PM
Page 384
Evaluation of Reservoir Information in Relation to Earthquakes in Ashtabula, Ohio
Armbruster et al. (1987) of LDEO indicate that no activity was known within 30 kilometers (18.6 miles) of Ashtabula prior to 1987. However, a comparison of location data tables developed by LDEO to those by the OhioSeis (Table 27.1) indicates that many of the LDEO epicenters are up to 3 miles (5 km) from the OhioSeis locations. Map 27.1 shows the distribution of seismicity in the region of the Chardon and Ashtabula earthquakes. Table 27.1 includes significant data from each of the periods of activity, but there seems to be no comprehensive listing including all events. The National Earthquake Information Center list of earthquakes within 18.6 miles (30 km) of the RES well records ten separate events ranging from 6 to 18.6 miles (10–30 km) in addition to the many shocks and aftershocks that are arguably injection-induced. However, one eventually must admit that the quality of the epicentral data is really not sufficient to be certain of the presence or absence of the 30 km (18.6 miles) “clean” zone. But even if one accepts its presence, one must ponder the significance of the temporal–spatial relation at the Ashtabula site between the start of injection (1986) and initiation of nearby seismic activity (1987), as this is a most seductive “line of evidence” if one wishes to demonstrate causality. The presence of clean zones is not uncommon in the region. In the north-central area of the current (2003) Ohio Epicentral Map of the Ohio State Network, one notices clusters of recent seismic activity centered on areas, 20 km in radius or larger, that were “clean” before a certain date (1926 for the cluster north of the boundary between Lucas and Wood Counties, 1936 for the cluster south of the boundary between Sandusky and Seneca Counties, and 1940 for the central portion of Ashland County). If injection operations had began shortly before those years, should such operations be considered responsible for those earthquakes? Because the induced Denver earthquakes have been very well documented and studied (e.g., Hseih and Bredehoeft, 1981), a tendency among investigators is to find similarities between a case in question and those events. However, there are some fundamental differences between the Denver earthquakes and the Chardon and Ashtabula earthquakes: • At the Rocky Mountain Arsenal, the fluid was injected directly into fractured, crystalline basement, whereas in northern Ohio, the injectate would have to find its way from the Conasauga/Rome/Mt. Simon, an injection interval with extensive, although low, matrix porosity and permeability, to the hypocentral depths in the crystalline basement via postulated fractures. Further, to reach the fracture on which the more recent hypocenters at Ashtabula occurred, the pressure front must pass the fracture on which the earlier earthquakes occurred and about 2.5 miles of additional sedimentary reservoir. The nearer fracture would serve to absorb and spread the injection-induced pressure along its extent, and to allow pressure to be lost through a greater extent of matrix porosity in the sedimentary rocks. Because we do not know the extent of the fracture nor any of its physical characteristics except its orientation, we cannot reasonably estimate its attenuating effect on the spread of pressure through the sedimentary rock and beyond. • In the Denver area, the crystalline rock of the reservoir is known to be fractured in core samples; furthermore, evidence exists that the fracture system is an extension of the fractures of the Rocky Mountains front range (Hseih and Bredehoeft, 1981); no evidence of faulting exists in the northern Ohio events, except for some alignment of microseismicity, which we discuss later. • There are strong well-hydraulics analytical evidence (Hseih and Bredehoeft, 1981) and seismic-activity clustering that indicate the Denver reservoir is a wide strip of fractured rock—a fault zone, in all probability. In the northeastern Ohio cases, no such evidence exists; instead, the evidence suggests a tabular, horizontal reservoir of broad extent.
ELSE_TSANG_CH027.qxd
8/23/2005
2:18 PM
Page 385
27.7 An Alternative Source Mechanism for the Ashtabula Earthquakes
385
• Even though seismicity near Denver spreads along the axis of the fracture zone, the incidence of seismicity was always highest near the injection well, where the induced pressure was highest, and the latest seismicity was near the injection well. Hsieh and Bredehoeft (1981) estimated a critical threshold pressure required for seismic activity based on the pressure in the reservoir near the injection well when seismicity ended. As will be shown, there is no such relationship between pressure and seismicity at Ashtabula. • The fault system of the Front Range has been active in the recent geological past (Zirbes, 2001); again, there is no evidence for current or recent movement on a fault system in passing through Ashtabula, Ohio, other than the earthquakes themselves. Our hypothesis for an earthquake mechanism in the area is also presented below. • Despite the lack of rock mechanics measurements at the Denver site itself, knowledge of the geology of the reservoir (nonporous, faulted rock-mass; major principal stress in the region vertical and equal to the weight of the overburden; e.g., Zoback and Zoback, 1980) constrains quite well both the mechanical properties of the reservoir and the prevailing state of stress; thus, it is possible to evaluate the effect of water pressures on the mechanical stability of the reservoir. Nicholson et al. (1988) and Nicholson and Wesson (1990) discussed the difficulties of evaluating the rock mechanical aspects for northeastern Ohio. They center on the uncertainties in estimating values of the critical mechanical parameters necessary to demonstrate shear failure of the rock mass (state of stress, shear strength of the rock mass, presence of pre-earthquake discontinuities, transmissivity, and storativity). In conclusion, we believe that attempts to compare suspect sites with the well-known Denver case require careful scrutiny.
27.7 AN ALTERNATIVE SOURCE MECHANISM FOR THE ASHTABULA EARTHQUAKES Seeber and Armbruster (1993) have interpreted the focal mechanism to be a strike slip along a preexisting near-vertical fault plane, with approximately E–W strike. This is reasonable in as much as the computed hypocentral locations of the Ashtabula events appear to align in an E–W direction along a band 1–0.6 miles (1.5 ⫻ 1.0 km) long (Fig. 27.1, Seeber and Armbruster, 1993), and such an orientation is compatible with double-couple, left-lateral slip, based on one interpretation of the composite first-motion stereoplot for those events. The strength envelopes (A through E, Fig. 27.2) were obtained from standard rock mechanics literature, as follows: The intact strength envelope for typical unweathered gneisses, metagranites, granites, diorites, quartzites, and similar massive igneous and metamorphic rocks (A) was developed using the original Hoek–Brown criterion, m ⫽ 24 and s ⫽ 1 and qu ⫽ 35,000*0.75 psi (5170 mPa), where m and s are empirical constants and qu is the uniaxial compressive strength of typical gneiss corrected for size (Hoek and Brown, 1980). The peak shear strength envelope (B) for a fracture that has not undergone shear displacement (e.g., joint) was obtained using the Barton criterion, with JRC ⫽ 20, qu ⫽ 35,000*0.75 psi, and φr ⫽ 35°, where JRC is the joint roughness (Hoek and Brown, 1980). Furthermore, the attitude of the slip is also roughly compatible with the known orientation of the principal horizontal stresses in the region (N65–70°E for σH), but if, and only if, one assumes residual shear strength for the discontinuity. Figure 27.1 shows a series of Mohr circles and Coulomb strength (failure) envelopes relevant to the effective stress and strength conditions at the Ashtabula and Chardon deep-well injection sites. It can be demonstrated that although the Chardon site is underlain by diorite
ELSE_TSANG_CH027.qxd
386
8/23/2005
2:18 PM
Page 386
Evaluation of Reservoir Information in Relation to Earthquakes in Ashtabula, Ohio
Fig. 27.1. Coulomb strength envelopes and Mohr stress circles for various assumed conditions in pre-Cambrian rocks in Ashtabula and Chardon (see text).
and schist, and the Ashtabula site by metagranite, quartzite, and gneiss, the two sites have very similar geomechanical settings (e.g., Hoek and Brown, 1980). The initial-stress circles (dashed circles) were obtained from our own data on in situ stresses in the eastern United States (Cole, 2003) and are within 10% of the stress values employed by Nicholson et al. (1988). The effective stress circles (dashed lines) corresponding to the epicentral locations at the time of the earthquakes were calculated and discussed below or were obtained from Nicholson et al. (1988), and φr is the residual or minimum friction angle (tan φr ⫽ µ or friction coefficient). Discontinuities that have experienced previous shear displacements are faults; the available strength is strictly frictional and given by tan φr. Three linear envelopes are plotted with different values of φr . Envelope C corresponds to the mean value of the Byerlee plot or tan 40.4° ⫽ 0.85; envelope D represents tan 35° ⫽ 0.70, a value commonly used in rock engineering for residual strength of unweathered gneiss. Finally, envelope E is for tan 31° ⫽ 0.60, which is the lower boundary of the Byerlee plot. In fact, for the normal stress level of interest [about 7400 psi (51 Mpa)], all the test results on the Byerlee plot fall above that last envelope. This is the value used by Nicholson and Wesson (1990) to evaluate the causal relationship between the Chardon earthquakes and the Calhio injection operations. Fischer and Greene (1989) and Fischer (1990) assume a value of φr ⫽ 30°. Figure 27.1 then indicates that the probabilities of shear failure, triggered by the injection operations, along a preexisting fault are low. Of course, probabilities of failure along preexisting joints or intact rock are even lower. However, the most important question about shear failure along a fault oriented in such a way is the likelihood that it will be found very close to the RES well. This reflects the belief that “. . . for the purposes of seismic hazard assessment, it would seem judicious to assume that a critically oriented fault is always
ELSE_TSANG_CH027.qxd
8/23/2005
2:18 PM
Page 387
27.7 An Alternative Source Mechanism for the Ashtabula Earthquakes
387
present, since to show otherwise is generally difficult” (Evans, 1987). This assumption gives the gneiss rock masses of the Grenville Complex the shear strength of sand. Although one could always make the case for small seismogenic faults that would escape detection, the assumption of omnipresent, unfavorably located, preexisting faults is not substantiated by surface or subsurface mapping of the lower Paleozoic/pre-Cambrian sequences in the region of interest. Well-controlled faults that cut the pre-Cambrian and lower Paleozoic of the oil and gas fields of southern Ontario have a north–north-east strike in the western portion and a more EW strike in the eastern part of this area, and are not associated with seismicity; in fact, tight subsurface control indicates they have been inactive since postTrenton time (Carter et al., 1996). The question of ad hoc reactivation of these faults at residual strength after being dormant, and presumably without the benefit of diagenetic processes, for hundreds of millions of years, is also troublesome. Rather, we propose that the pre-Cambrian sequences of the region have been continuously failing, irrespective of fluid injection, under the same stress field that is assumed to create shear slips along preexisting faults. We reinterpret the focal mechanism for the Ashtabula (and Chardon) earthquakes as nondouble couple events; specifically, we interpret these earthquakes as caused by tensile ruptures along essentially vertical fractures, so that the first-motion dilatational and compressional distributions fall along a tennis ball array, rather than a beach ball array as shown in Figure 27.2. The mechanics of nondouble couple earthquakes have been widely discussed (e.g., Frohlich, 1994), and evidence from mining seismology indicates that seismic events with this source mechanism can release substantial amounts of energy. Eldenger and his co-workers (e.g., Eldenger, 1982, 1985, 1993) have discussed the coincidence of the attitude of the most prominent joint system in many areas of eastern United States with the measured direction of the major principal stress, σH. Eldenger and co-workers have found that joints in the northeastern United States are essentially parallel to the orientation of σH, and that they are not shear but tensile or, in fracture mechanics
Fig. 27.2. Proposed nondouble couple (tensile focal mechanism for the Ashtabula (and Chardon) earthquakes. This focal mechanism obviates the need to invoke ubiquitous preexisting discontinuities in all directions, at residual shear strength (c ~ 0; φ ~ 30°).
ELSE_TSANG_CH027.qxd
388
8/23/2005
2:18 PM
Page 388
Evaluation of Reservoir Information in Relation to Earthquakes in Ashtabula, Ohio
terms, Mode I features. Furthermore, he believes they have to be formed in an effective tensile stress field. We propose that this set of essentially vertical fractures is the product of tensile failure, but within an all-around compressive field, and that these failures manifest themselves as seismic events in Ashtabula and surrounding areas. Cai et al. (1998) have demonstrated that a tensile model for brittle rocks under compressive loads explains the observed depletion of S-wave energy in high-quality microseismic records, and the relationship between physical fracture sizes and measured microseismic energy, whereas a shear model does not. The orientation of the contemporary major principal stress, SH, in that portion of the North American Plate corresponding to the northeastern United States is well established (Sbar and Sykes, 1973; Zobak and Zobak, 1980, 1989). The average orientation of SH is about N65°E; however, in our area of interest it appears to trend more easterly, at about N75°E (World Stress Map, Edeilberg University, Karlsruh, Germany, 2003). Hinze (1989) indicates a value of about N70°E. These orientations are confirmed by our own studies (Cole, 2003). One important aspect of the proposed focal mechanism, in the context of these earthquakes, is that it obviates the need to assume ubiquitous preexisting discontinuities in all directions, at residual shear strength (c ~ 0; φ ~ 30°). This assumption is not borne by surface or subsurface field observations and, in our opinion, would have prevented the development of a majority of demonstrably safe deep-well disposal operations in the eastern United States. But perhaps the most significant aspect of this focal mechanism is that, unlike mechanisms that revolve around Mohr–Coulomb theory, it is probably minimally affected by any changes in effective stresses, but depends instead on elastic-strain energy considerations. The effect of pore pressure increases on the propagation of Mode I fractures is not a simple matter of effective stress reduction, as it is with shear slip along preexisting faults. Walls of tensile fractures are not simply pushed apart by fluid, as they are in hydraulic fracturing. The strain energy that drives Mode I crack propagation is derived from compressional deformation of uncracked rock matter, as described in Biot’s poroelastic parameter
αb ⫽ (1 ⫺ βi/βc), where βi and βc are the compressibilities of the uncracked and cracked rock matter, respectively. This parameter controls the increase in compressive stress, and hence in the strains, in the rock framework due to increases in pore-water pressure. Obviously, values of the parameter depend on assumptions regarding the conditions of the rock mass prior to the increase in water pressure. Commonly cited values of elastic modulus (reciprocal of compressibility) for quartz, fused silica, and feldspar are generally between 8.70 and 10.88 ⫻ 107 psi (6.0 and 7.5 ⫻ 105 MPa) (e.g., NITS, 2003; Tosoh Quartz, 2003): these values can be considered as reasonable approximations for 1/βi. Laboratory values of Young’s modulus for unweathered granites and high-grade gneisses (e.g., Liu et al., 2002; Katz et al., 2000) can be considered representative of the rocks with microcracks that populate the crystalline basement; these also vary normally between 7.252 and 10.88 ⫻ 103 psi (5.0 and 7.5 ⫻ 105 MPa) (1/βc). Therefore, the value of αb above would be very small, and the changes in stress in the rock skeleton due to an increase in pore-water pressures would also be very small. In the specific case of Ashtabula, the relevant question then would be, How much strain could a very small fraction of 420 psi (2.9 MPa)—our calculated value of maximum water-pressure increase at the epicentral location of the 1987 event—induce in rock substances with a modulus on the order of 8.88 ⫻ 107 psi (6.0 ⫻ 105 MPa)?
ELSE_TSANG_CH027.qxd
8/23/2005
2:18 PM
Page 389
27.8 Description of the Injection Zone
389
One could argue that the recurrent seismicity indicates that the area may be on the verge of failure and that the injection-induced pressure increase is “the straw that broke the camel’s back.” However, this type of argument is unverifiable and does not move the discussion of the problem of causality forward. Thus, we strongly believe that before accepting any particular explanation for the 1987 or similar events, further studies of the characteristics of the ongoing seismicity and the neotectonism of the area are required.
27.8 DESCRIPTION OF THE INJECTION ZONE The RES well was drilled to a depth of 6060 ft (1847 m), with 7 in. casing set to 5506 ft (1678 m), 37 ft (11.25 m) below the top of the Conasauga Sandstone, and within a zone that has both porosity and permeability. The Conasauga, 5469–5620 ft (1662.5–1713 m), is a dolomitic sandstone interbedded with dolomite, which the porosity log shows has strata 10–20 ft thick with porosities near 7%. The Rome Formation, 5620–5880 ft (1713–1792 m), is a dolomite with low porosity. The lowest sedimentary formation is the Mt. Simon, 5880–5980 ft (1792–1823 m). According to the core description, the Mt. Simon Formation is a fine-grained quartz sandstone. The porosity log indicates that at least 100 ft (30.5 m) of the Mt. Simon Sandstone is porous, with thin beds having in excess of 15% porosity. These porosity measurements are not high; at an average of about 10%, they are typical of many older sandstones that also have adequate permeability to serve as reservoir rocks. Conventional cores were cut from 5500 to 5588 ft (1676.4–1170.0 m) and from 5902.0 to 5919.5 ft (1799.0–1804.3 m). Only 10 samples from the Mt. Simon were analyzed for porosity and permeability. The porosity measurements range up to 15.6% and tend to support the porosity log measurements. The sample from 5911 ft (1801.7 m), with a permeability of 36 mD (0.1 m/s), was the only one with a permeability greater than 3 mD (0.03 m/s). In addition, 42 sidewall cores were collected and analyzed, but only a graphical representation of the results appears to be preserved. These cores indicate higher porosity and hydraulic conductivity than do the conventional cores, but their value is suspect. For modeling, we chose a porosity of 10%. Various permeability indicators such as spinner surveys, radioactive tracers, and temperature logs indicate permeability through the porous sections. Radioactive tracer surveys were inconclusive in defining permeable zones. Temperature logs (Fig. 27.3) indicate that fluid leaves the well bore throughout the interval between 5500 and 5980 ft (1672 and 1822.7 m), but that the bulk of the fluid exits the well bore between 5870 and 5910 ft (1789 and 1802 m), still well above the pre-Cambrian unconformity. Based on all information, we set the net thickness at 80 ft (24.5 m). In 1990, 1992, and 1993, pressure falloff tests were conducted. All of the tests indicate low transmissivity, with the average being 376 mD ft (transmissibility ⫽ 1.1 m2/s). Each also indicates radial flow throughout; in the case of the 1990 test, this is a period of at least 120 hours after shut in (Fig. 27.4). The pressure data have been plotted against Horner time for each of the tests, and the straight-line portions of the curves extrapolated to infinite shutin time (Fig. 27.5). For each test, the radius of investigation extended beyond 400 ft (125 m). In addition, there is a significant negative skin factor, which leads to pressure reductions in the range of 1000 psi (6.8 MPa) at high injection pressures. The formation water is a sodium-chloride brine with 240,000 mg/L total dissolved solids, yielding a specific gravity of 1.169. The measured hydrostatic pressure prior to injection
ELSE_TSANG_CH027.qxd
390
8/23/2005
2:18 PM
Page 390
Evaluation of Reservoir Information in Relation to Earthquakes in Ashtabula, Ohio
Temperature, deg F
130
Temperature Measurements Through RES Injection Zone
125 120 1993 1994
115 110 105 100 5000
5200
5400
5600
5800
6000
Depth, ft.
Fig. 27.3. When liquid is injected through a cased well, the surrounding rock is affected by conductive heat transfer. This largely affects rock very near the well bore, and the return of the temperatures along the well bore to natural geothermal is relatively rapid. Where there is significant infiltration, the time required to return to normal geothermal temperatures is much longer. We reason that most of the injected liquid left the well bore between 5850 and 5950 ft.
Pressure Increase, psi
1000
LOG-LOG PLOT 1990 PRESSURE FALL-OFF TEST
100
10 0.01
0.1
1
10
100
1000
Time, hrs Pressure Change, psi
Derivative
Fig. 27.4. This plot of pressure and its derivative following a temporary halt of injection indicates radial flow through 120 hours of measurement of pressure fall off following a lengthy injection period.
was 2733 psi (18.84 MPa) at 5950 ft (1813 m). All later pressure measurements cited in this chapter are changes from this pressure.
27.9 INJECTION ACTIVITY IN THE ASHTABULA AREA According to information provided by the Ohio Environmental Protection Agency (EPA) and confirmed by the Ohio Department of Natural Resources, the RES well was at 41°54′28″ (41.9078) north latitude and 80°43′56″ (80.7322) west longitude. RES injected
ELSE_TSANG_CH027.qxd
8/23/2005
2:18 PM
Page 391
27.10 Propagation of Injection-Induced Pressure Effects MONTHLY INJECTION DATA
2000
1,500
1500
1,000
1000
500
0
500
Dec-1987 Volume Average IP
Dec-1989 High IP
Dec-1991 Low IP
Dec-1993
INJECTION PRESSURE, psi
INJECTED VOLUME, gals Thousands
2,000
391
0
July 13, 1987
Fig. 27.5. IP is injection pressure measured at the surface and reported to the Ohio EPA.
90 ⫻ 106 gal (3.4 ⫻ 105 m3) of calcium-chloride brine having specific gravity, which tended to range from 1.1 to 1.15, into the Conasauga, Rome, and Mt. Simon formations (Cambrian and pre-Cambrian), between the depths of 5469 and 6000 ft (1667 and 1828.8 m), between May 1986 and June 1994. Figure 27.6 shows monthly data reported to the Ohio EPA. The overall average was 21 gpm (4.77 m3/s). During the first year of injection, the average rate was 30 gpm (6.81 m3/s), and the cumulative injection volume at the time of the July 13, 1987, earthquake was 1.89 ⫻ 107 gal (7.15 ⫻ 104 m3). Based on chemical engineering correlations (Perry and Chilton, 1973), viscosity of the brine is estimated to be 1.5 cp at reservoir temperature. Injection through the well ended on June 20, 1994. The pressure increase just before shut-in was 1237 psi (8.53 MPa). Unfortunately, no measurements of pressure were recorded between the end of injection operations and the final measurement on December 5, 1994, when plugging was initiated. By that time, the pressure had declined to within 216 psi (1.49 MPa) of the pressure measured before any injection.
27.10 PROPAGATION OF INJECTION-INDUCED PRESSURE EFFECTS Figure 27.6 shows the pressure data from the three pressure falloff tests plotted against their respective Horner time functions, each data set yielding a linear trend following dissipation of the effects of pressure-sensitive permeability. Horner time at 1.0 is an approximation of a relatively infinite period following a rate change. The pressure data, and all other data reviewed, tend to confirm the prevailing view that the Mt. Simon is a blanket sandstone with consistent reservoir properties in which flow from a vertical well will be radial. This model invites extrapolation of pressure trends within the reservoir. Using measured reservoir parameters, a model of the injection reservoir has been developed. There are uncertainties, but the simulations of pressure change based on this model match observed effects reasonably well. There are uncertainties regarding elements contributing to our estimates, such as for the viscosity of the calcium-chloride brine at reservoir conditions, but these don’t affect the simulation because the values can be treated as groups. On the other hand,
ELSE_TSANG_CH027.qxd
8/23/2005
392
2:18 PM
Page 392
Evaluation of Reservoir Information in Relation to Earthquakes in Ashtabula, Ohio Trends of Pressure Decline 4500 1990 Data Pressure, psia
4000
1990 Extrapolation 2001
3500
1992 Data 1992 Extrapolation 2001
3000
1993 Data 1993 Extrapolation
2500
2001 2000 1E+0 1E+1
1E+2 1E+4 1E+6 1E+3 1E+5 1E+7 Horner Time
Fig. 27.6. Standard plots of pressure decline versus Horner time, illustrating the progress of pressure decline expected by 2001 at the RES well bore, relative to the final pressure level at a theoretical infinite shut-in time. The plots indicate that the pressures at the well head were only a few tens of psi from the final pressure, which is what existed prior to injection.
our model does not consider the effects of vertical migration out of the injection reservoir because no means to estimate values for important parameters governing vertical movement are available. Moreover, our simple, empirical calculations cannot provide effective treatment for three-dimensional analysis. Calculations performed, but not included in this chapter, suggest that an equilibrium state with no additional outward pressure development occurred before the end of the injection period. This important phenomenon cannot be ignored by those who wish to link effect to cause by extrapolating pressure effects of injection through time and space. The rates of pressure rise and fall both temporally and spatially due to changes in injection rates vary, depending upon the reservoir. For injection into a stratigraphic reservoir in which a horizontal, uniformly permeable stratum is over and underlain by much less permeable strata, the pressure declines radially away from the well in all directions, and the pressure changes logarithmically with increasing distance. If injection is into either (case 1) a single fracture or a fracture zone with finite permeability and effectively infinite length with no loss of fluid through permeability of its walls or (case 2) into a fracture with infinite permeability that loses fluid to its walls, the pressure changes in one dimension (in two opposite directions), and the pressure gradient within the case-1 fracture and perpendicular to the case2 fracture is linear. There are many combinations of flow regimes possible. A radial flow pattern can be confined laterally, perhaps by impermeable faults on two sides, so that a transition of flow patterns from radial to linear occurs if there is injection into the system through a well. Where a well penetrates a fracture with limited extent in a porous, permeable layer, injection pressures will respond as in a linear flow system shortly after a change in injection rate; however, a transition from linear to radial flow will occur as the area affected by the rate change increases beyond the ends of the fracture, and pressure responses reflect the change. Because the testing at the RES site shows that the flow system within several hundred feet of the injection well is radial and that fractures would add a possibility of linear flow at greater distances, we believe that we need to only consider radial and linear flow patterns.
ELSE_TSANG_CH027.qxd
8/23/2005
2:18 PM
Page 393
27.11 Geological Framework for Seismicity
393
Although we may be able to estimate pressure distribution in the injection reservoir itself, the seismicity occurs along faults in the pre-Cambrian basement from which there is some vertical separation from the zone that appears to be accepting most of the injected liquid. If the injection triggers the seismicity, it is obvious that the liquid gains access to the fracture at some point, and increases the hydraulic pressure within the fracture in order to induce seismic activity. For practical purposes, we have no way of knowing the dimensions and permeability of the fracture system or the degree of hydraulic connection between the fracture and the permeable matrix. These factors are important in determining the arrival time and magnitude of pressure change at a potential hypocenter. Because we cannot know these things, we might reasonably postulate that the fracture is open to the injection zone throughout its length or, at the other extreme, that there is a hydraulic connection between the injection zone and the fracture at either the location directly above the hypocenter or at the fracture’s closest approach to the injection well. For the purpose of estimating pressure at the hypocentral areas, we can assume that the fracture permeability is more than that of the matrix and no more than infinite. We will assume that the fracture permeability is infinite, at least between the point on the fracture nearest the RES well and the various hypocenters. The resulting inferences can serve as a test case or point of reference so that we can reasonably discuss the probability of causation. It is important to remember that the assumptions result in the greatest possible pressure changes and earliest arrival times at all points along the fracture. Although it is true that the pressure effects due to injection continue to move outward after injection ceases, the highest pressure must always be at the well bore itself. This allows us to know or estimate a maximum pressure in the reservoir. Figure 27.7 illustrates the radial pressure distribution at times that are important to our discussion. A number of authors (Hsieh and Bredehoeft, 1981; Nicholson et al., 1990; Seeber et al., 2003) have pointed out that the pressure front resulting from injection continues to spread outward and might induce seismicity even after injection has ended. Using our radial-flow model with low porosity and permeability, pressures can be extrapolated, based on our assumptions about the fracture, to the hypocenters at appropriate times. Figures 27.8–27.10 show the results of simulations through the year 2002. Other simulations were also carried out through periods of 50 and 1000 years. The simulations show that, through the years, the pressures at all distances converge as they decline toward the original reservoir pressure, but even after 1000 yr, there would still be a 1 psi (6.89 ⫻ 10–3 MPa) increase in pressure at distances from 1 to 20,000 ft (6.1 km) in a “perfect” reservoir.
27.11 GEOLOGICAL FRAMEWORK FOR SEISMICITY Seeber et al. (2003) have located the fracture along which the 1987 earthquake cluster occurred, although it is not known if any error analysis was performed to include anisotropy effects (for instance, seismic velocities parallel to σH and to the strike of the major joint set are probably much greater than those perpendicular to them). The epicenter of the 1987 main shock was about 7400 ft (2.25 km) from the RES well, but the nearest approach of the fracture to the RES well is about 2300 ft (700 m) from the well. The 2001 sequences are not so well constrained; the hypocenters of just 13 of the 51 events have been accurately located (Seeber et al., 2003). However, the epicenter of the main shock of the sequence that began on June 3, 2001, has been accurately located, and is 3.8 miles (6.15 km) from the RES well. The fracture along which the seismicity occurred is nearly parallel to the Ashtabula fault,
ELSE_TSANG_CH027.qxd
8/23/2005
394
2:18 PM
Page 394
Evaluation of Reservoir Information in Relation to Earthquakes in Ashtabula, Ohio Radial Pressure Profiles 1000
Pressure Change, psi
800
600
400
200
0
0
5000
10000
15000
20000
Distance from RES Well, ft Jul 1987
Dec 1994
Dec 2005
22,000 ft
Jun 1994
Jan 2001
2,300 ft
Data Dec 1994
Fig. 27.7. Profiles showing pressure distribution around the RES well at the time of the 1987 earthquake, the end of injection, closure of the injection well, and the time of the 2001 earthquake. Short vertical lines indicate the distances from the injection well along lines perpendicular to extensions of the fractures along which the hypocenters were located.
Pressure Histories at Close Approaches of Fractures Active in 1987 and 2001
Pressure Change, psi
600 500 400 300 200 100 0
Dec-86
Dec-89
Dec-92
Dec-95
Dec-98
Dec-01
Time 2300 ft
15,000 ft
Jul 13, 1987
Jan 31, 2001
Fig. 27.8. Comparison of pressure increases at the near approaches of extensions of the assumed fractures, along which the 1987 and 2001 series of seismic events occurred. Notice that the pressure increase at the time when failure occurred in the 1987 events is almost six times greater than the increase at the time of the 2001 series.
but is 2.5 miles (4000 m) farther south. An extension of the fracture just slightly beyond the easternmost epicenter must be made for the fracture to reach the point in which line perpendicular to it might include the location of the RES well.
ELSE_TSANG_CH027.qxd
8/23/2005
2:18 PM
Page 395
27.12 Sensitivity Analysis
Pressure Change, psi
600
395 Pressure Histories at Close Approach of Fracture and at 1987 Hypocenter
500 400 300 200 100 0
Dec-86 Dec-89 Dec-92 Dec-95 Dec-98 Dec-01 Time 2300 ft
7400 ft
Jul 13, 1987
Fig. 27.9. Comparison of simulated pressure histories at the nearest approach to the RES well of the extension of the assumed fracture along which the 1987 series of events appeared to occur, and at the hypocenter of the main shock (distance estimated by averaging the distance to each of the well-located hypocenters in the series), assuming radial flow to each location. The range of pressure is from 419 to 61 psi on July 13, 1987. These were the highest pressures to that date at these locations. Maximum pressure increases at those locations are 569 in early 1990 and 225 in early 1992, respectively.
Pressure Histories at Close Approach of Fracture and at 2001 Hypocenter Pressure Change, psi
600 500 400 300 200 100 0 Dec-86 Dec-89 Dec-92
Dec-95 Dec-98 Dec-01 Date
15,000 ft
22,000 ft
Jan 26, 2001
Fig. 27.10. Comparison of simulated histories at the nearest approach to the RES well of the fracture along which the two 2001 series of events appeared to occur, and at the hypocenter for the January 26 main shock. The range of pressures is from 72 to 46 psi on January 26. The highest values for pressure increase at these locations are 91 psi in late 1995, and 47 psi in early 1999.
27.12 SENSITIVITY ANALYSIS We conducted a demonstration of the validity of the model as a conservative predictor, which tends to result in of pressure increases at the more distant hypocenters. The model
ELSE_TSANG_CH027.qxd
396
8/23/2005
2:18 PM
Page 396
Evaluation of Reservoir Information in Relation to Earthquakes in Ashtabula, Ohio
Pressure Change, psi
Sensitivity to Variation of Reservoir Property Pressure History at 15,000 feet 150 100 50 0
Dec-86 Dec-89 Dec-92
Dec-95 Dec-98 Dec-01 Date
Base
Thin
Moderate
Jan 31, 2001
Thick
Fig. 27.11. The sensitivity check included four scenarios based on alternative interpretations of logging and well test results. The base case includes porosity of 11%, permeability of 4.7 mD, and accumulated thickness of 80 ft. The thick-zone scenario increases the thickness to 300, and decreases the average porosity to 7%, while maintaining a transmissivity of 376 mD ft. The thin and moderate scenarios include 14 and 10% porosity, and 30 and 100 ft of thickness, respectively. Notice that in each case, the highest pressure attained occurs prior to 2001, the latest in mid-1996.
predicts a pressure difference on December 5, 1994, of 276 psi (1.90 MPa) rather than the measured pressure difference of 216 psi (1.49 MPa) (Fig. 27.11). We found that we could match the measured pressure by increasing the transmissibility group by a factor of 0.34. The effect of this change at the distance (15,000 ft; 4500 m) from the RES well of the assumed near-approach of the fracture along which seismicity began on January 26, 2001, was to reduce the increase in water pressure on that date from 72 to 60 psi (0.50–0.41 MPa). In addition, the timing for reaching peak pressures at 15,000 ft (4500 m) changed from 91 psi (0.63 MPa) in November 1995, to 84 psi (0.58 MPa) in February 1995, both long before the advent of seismicity along this fracture. Our sensitivity test varies the numerical value of the storativity group while maintaining a constant value for the transmissibility group by simultaneously changing the permeability and thickness because that approach results in greater relative pressure variations and later peak pressures at hypocentral distances, although the maximum pressures are lower. Figure 27.11 shows that variations of significant magnitude do not result in qualitatively dissimilar results. Based on this sensitivity analysis, we conclude that the model predicts likely, but overstated, values of pressure increases resulting from injection. Therefore, the model is suitable for making the sort of projections necessary to consider the role of pore pressure in the occurrence of the Ashtabula area earthquakes. 27.12.1 Case 1—The 1987 Seismic Sequence Figures 27.8–27.10 show the results of our simulations of pressure at the assumed nearapproach and epicenter access points for the 1987 and 2001 seismicity sequences. Vertical lines at the times of the main shocks of the series indicate the pressures at the hypocenters
ELSE_TSANG_CH027.qxd
8/23/2005
2:18 PM
Page 397
27.14 Discussion
397
at the times of the events. The pore pressure increase at the 1987 main shock hypocenter could be as great as 419 psi (2.89 MPa). If the fracture does not have infinite permeability, then the pressure change would have been less, but no less than 61 psi (0.42 MPa), because that is the pressure increase that would occur at that point and time if there is hydraulic connection only at the location above the epicenter. 27.12.2 Case 2—The June 2001 Seismic Sequence It is immediately obvious that the change in pressure of 419–61 psi (2.89–0.42 MPa) at the onset of seismicity in 1987 was likely several times greater than the pressure increase of 72–46 psi (0.50–0.32 MPa) at the onset of either of the 2001 sequences. Although the 1987 and 2001 sequences were on different fractures, the fractures were nearly parallel, and the Coulomb failure criteria should be similar. In addition, the loss of pressure in transit to the more distant fracture resulting from the action of the nearer fracture as a pressure sink increases the considerable conservatism of our estimates of pressure at the hypocenter of the earthquakes lying along the more distant fracture.
27.13 DISCUSSION It is very clear that pore-pressure change at the hypocenter of the 1987 main shock might have been sufficient to induce seismicity on a susceptible fracture; however, the pattern of activity is dissimilar to the pattern at Denver, where events seem to have occurred randomly across the active faults whenever the injection pressure was elevated. There may have been several sequences of seismic activity along the Ashtabula fault, but each of those sequences consisted of a main shock and a series of aftershocks, as areas that experienced increased stress as a result of the main-shock occurrence, and other aftershocks reacted through small movements. Even successive main shock believed to have occurred on the Ashtabula fracture might have been induced more by the occurrence of the original 1987 movement than by pore-pressure increases. We note that those events followed the 1987 event by more than 1 year. If the fracture permeability was sufficiently high to transmit nearly undiminished pressure to the 1987 main-shock hypocenter, it should also have transmitted the pressure very rapidly through the additional distance to the sites of the 1989 and 1992 events, if they were, in fact, on the same fracture. If the fracture did not have a very high permeability, then the pressure could not have been close to 419 psi (2.89 MPa), which was possible at the hypocenter at the time of the July 13, 1987, earthquake because a pressure gradient existed within the fracture. In the event there is no single access point, then the fracture also serves as a pressure sink because of leak off along the entire fracture where it penetrates the reservoir zones. These effects cannot be quantified because of our lack of specific knowledge about the fractures, but their existence would certainly reduce the pore pressure available to reduce the cohesion of the fracture walls. Any “demonstration” using water pressure increases and the Mohr/Coulomb theory (reduction in effective stresses) needs to show evidence of discontinuities at residual strength, and no such evidence is offered. On the other hand, we have introduced here an alternative focal mechanism that provides at least conceptual evidence for properly oriented discontinuities. No assumption of preexisting discontinuities in all directions is needed.
ELSE_TSANG_CH027.qxd
398
8/23/2005
2:18 PM
Page 398
Evaluation of Reservoir Information in Relation to Earthquakes in Ashtabula, Ohio
27.14 CONCLUSIONS The evidence for an induced origin for the 1987 Ashtabula earthquakes is proximity of epicenters to the injection wells. Evidence against induction by injection includes the sparsity of events following the onset of seismicity, the lack of any correlation between the level of either injection rate or pressure to seismicity, and the beginnings of later sequences of earthquakes near Ashtabula years after injection-induced pressure effects began to decline at the hypocenters. We believe that we cannot conclusively demonstrate that the 1987 main shock was not induced by injection pressure, although a reasonable application of the Coulomb failure theory indicates that it was not. We believe that we can very reasonably argue that none of the later events was directly induced by injection. That being the case, we think that we can also argue that the 1987 sequence is probably not induced, because the Ashtabula area is clearly subject to natural seismic activity. The argument of noninduction is supported by the new focal mechanism, Mode I tensional failure, which at least a priori is not obviously affected by pore-water pressures, but by elastic strain energy considerations. The authors conclude that the evidence linking the various post-1987 sequences of earthquakes near Ashtabula to injection by RES is incomplete. No authors espousing causation by injection have attempted to evaluate quantitatively the physical effects of that injection. The frequency of seismic activity has clearly not been related to injection activity, so any claims of causality must be considered tenuous. A lack of seismicity, in the absences of an injection by RES, cannot be demonstrated. We show that the pressure at the hypocenters of the later main shocks is low, and is likely only a fraction of the pressure that might have existed at the hypocenter of the 1987 main shock and the occurrence of several possibly unrelated seismic events within the 30 km (18.6 mi) “clean zone.” This removes all but the circumstantial evidence that an earthquake occurred near an injection well following an initiation of injection. When one looks at the peppering of noninduced epicentral locations within and surrounding the “clean zone,” one wonders if it was not a matter of time before a seismic event had occurred near the problem area, irrespective of any human activity. Perhaps probabilistic evaluations of the likelihood of an earthquake happening within a certain unit area, irrespective of cause, for a subregion of northeastern Ohio, might shed some light on this important issue. ACKNOWLEDGMENTS The authors thank Chuck Lowe and the Ohio EPA for providing information related to the injection operations; Michael C. Hansen who operates the Ohio Seismic Network, provided information, and reviewed the chapter; John Armbruster and Leonardo Seeber of LDEO who have done much research in this area, provided the authors with their latest work prior to publication, and were willing to discuss the issues; and Mark Baransky of the Ohio Department of Natural Resources who reviewed the draft. Any shortcomings are the responsibility of the authors who heeded advice as they believed appropriate. REFERENCES Ahmad, M. and Smith, J.A., 1988. Earthquakes, injection wells, and the Perry Nuclear Power Plant, Cleveland, Ohio. Geology, 16: 739–742.
ELSE_TSANG_CH027.qxd
References
8/23/2005
2:18 PM
Page 399
399
Armbruster, J.G., Seeber, L. and Evans, K.F., 1987. The July 1987. Ashtabula earthquake (Mb ⫽ 3.6) sequence in north eastern Ohio and a deep fluid injection well, (Eastern Section), October 7–9. Baransky, M., 2004. Personal communication. Braile, L.W., 1989. Crustal structure of the continental interior. In: L.C. Pakiser and W.D. Mooney (Eds), Geophysical Framework of the Continental United States. Geol. Soc. Am. Mem., 172: 285–315. Byerlee, J.D., 1978. Friction of rock. Pure Appl. Geophys. 116: 615–626. Cai, M., Kaiser, P.K. and Martin, C.D., 1999. A tensile model for the interpretation of microseismic events near underground openings. In: S. Talebi and P.T. Volumens (Eds), Seismicity Caused by Mines, Fluid Injections, Reservoirs and Oil Extraction, Birkhauser, Basel, pp. 67–92. Cai, M., Kaiser, P.K. and Martin, C.D., 1998. Investigation of stress-path-dependency of stress-strain relations of jointed rock. Technical Research Report for Tokyo Electric Power Services Co. Ltd., Tokyo, Japan 98-I-IR, Geomechanics Research Centre, Laurentian University, Sudbury, Canada. 53p. Carter, T.R., Trevail, R.A. and Easton, R.M., 1996. Basement control on some hydrocarbon traps in southern Ontario. In: B.A. van der Pluijm and P.A. Catacosinos (Eds), Basement and Basins of Eastern North America: Boulder, CO. Geol. Soc. Am., vol. 308, pp. 95–107. Cole, T.C., 2003. The Role of Regional Horizontal Stresses in the Formation of Valley Stress Relief Features. Ph.D. Thesis, Department Civil & Environmental Engineering University of Illinois at Urbana-Champaign, to be submitted. Cundall, P.A., Ptyondy, D.O. and Lee, C.A., 1996. Micromechanics-based models for fracture and breakout around the Mine-by Tunnel. J.B. Martino and C.D. Martin (Eds). Proceedings of International Conference on Deep Geology. Disposal of Radioactive Waste. Can Nuclear Soc., Toronto, pp. 113–122. Eldenger, T., 1982. Is there a genetic relationship between selected regional joints and contemporary stress within the lithosphere of North America? Tectonics, 1: 161–177. Eldenger, T., 1985. Loading paths to joint propagation during a tectonic cycle: An example from the Appalachian Plateau, U.S.A.. J. Struct. Geol., 7: 459–476. Eldenger, T., 1993. Stress Regimes in the Lithosphere. Princeton Univ. Press, Princeton, NJ. Evans, D.M., 1968. The Denver area earthquakes and the Rocky Mountain arsenal disposal well. Mt. Geol., 3(1): 101–112. Evans, K.F., 1987. Assessing Regional Potential for Induced Seismicity from Crustal Stress Measurements: An Example from Northern Ohio. National Center for Earthquake Engineering Research Technical Report, NCEER-0025, Buffalo. Fischer, J.A., 1987. Injection wells and the January 31, 1986 Ohio earthquake. Proceedings of the Pacific Conference on Earthquake Engineering, New Zealand, pp. 173–184. Fischer, J.A., 1990. Proceedings of Fourth U.S. National Conference on Earthquake Engineering, Palm Springs, California, pp. 649–658. Fischer, J.A. and Greene, R.W., 1989. In: J.R. Watters (Ed.), Engineering Geology and Geotechnical Engineering. pp. 267–277. Frohlich, C., 1994. Earthquakes with non-double-couple mechanisms. Science, 264: 804–809. Gillespie, M., 2002. Waste well blamed for Ashtabula Quakes. Cleveland Plain Dealer, 01/21/2002, Cleveland. Grauch, V.J.S., 1993. Limitations on digital filtering of the DNAG magnetic data set for the conterminous U.S. Geophysics, 58: 1281–1296.
ELSE_TSANG_CH027.qxd
400
8/23/2005
2:18 PM
Page 400
Evaluation of Reservoir Information in Relation to Earthquakes in Ashtabula, Ohio
Hamiel, Y., Liu, Y., Lyakhovsky, V., Ben-Zion, Y. and Lockner, D., 2002. Visco-elastic damage rheology model: Theory and experimental verification. Geophys. Int. J., 159: 1155–1165. Hinze, W.J. and Hood, P.J., 1989. The magnetic anomaly map of North America; A new tool for regional geologic mapping. In: A.A. Bally and A.R. Palmeer (Eds), The Geology of North America: An Overview: Boulder, Colorado, Geological Society of America, Geology of North America, Vol. A, pp. 29–38. Hinze, W.J., 1996. The crust of the Northern U.S. Craton: A search for beginnings. In: B.A. van der Pluijm and P.A. Catacosinos (Eds), Basement and Basins of Eastern North America: Boulder, Colorado. Geological Society of America, Special Paper 308. Hoek, E. and Brown, E.T., 1980. Underground Excavations in Rock. Inst. Min. Met., London. Hsieh, P.A. and Bredehoeft, J.S., 1981. A reservoir analysis of the Denver Earthquakes—A case study of induced seismicity. J. Geophys. Res. 86: 903–920. Katz, O., Reches, Z. and Roegiers, J.-C., 2000. Evaluation of mechanical rock properties using a Schmidt hammer. Int. J. Rock Mech. Min. Sci., 37: 723–728. Lidiak, E.G., 1996. Geochemistry of subsurface proterozoic rocks in the eastern midcontinent of the United States: Further evidence for a within-plate tectonic setting. In: B.A. van der Pluijm and P.A. Catacosinos (Eds), Basement and Basins of Eastern North America. Geol. Soc. Am. Spec. Pap., Vol. 308, pp. 45–66. Liu, Y., Lyakhovsky, V., Ben-Zion, Y. and Lockner, D., 2002. Visco-elastic damage rheology model: Theory and experimental verification, Geophys. Intern. J., January 2002, Submitted. Nicholson, C., Roeloffs, E. and Wesson, R.L., 1988. The northeast Ohio earthquake sequence of 1 January 1986: Was it induced? B. Seismol. Soc. Am. 78(1): 188–217. Nicholson, C. and Wesson, R.L., 1990. Earthquake hazard associated with deep well injection—A report to the U.S. Environmental Protection Agency. U.S. Geological Survey Bulletin, 1951, 74p. NITS, National Institute of Standards and Technology, 2003. Ohio Geology, 2001(3): 1–3 (http://ois.nist.gov/srmcatalog/catalog/numrpt.cfm). Perry, R.H. and Chilton, C.H., 1973. Chemical Engineers’ Handbook, 5th edn. McGrawHill, New York. pp. 12–46. Quartz, T., 2000. TOSOH Quartz Group, The Worldwide Quartz Network, Weiss Scientific Glass Blowing Company, 14380 NW Science Park Drive, Portland, Oregon 97229. www.tosoh quartz.com/engineering3.html, Portland, OR. Reed, C., 2002. Triggering quakes with waste. Geotimes, March. Roeloffs, E., Nicholson C. and Wesson R.L., 1989. Comment on “Earthquakes, injection wells, and the Perry Nuclear Power Plant, Cleveland, Ohio.” Geology, 382–384. Sbar, M.L. and Sykes, L.R., 1973. Contemporary compressive stress and seismicity in eastern North America: An example of intra-plate tectonics. Geol. Soc. Am. Bull., 84: 1861–1882. Seeber, L. and Armbruster, J.G., 1993. Natural and induced seismicity in the Lake Erie–Lake Ontario Region: Reactivation of ancient faults with little neotectonic displacement. Geographie Physique Quaternaire, 47(3): 363–378. Seeber, L., Armbruster, J.G. and Kim, W., 2003. A fluid-injection triggered earthquake sequence in Ashtabula, OH: Implications for seismogenesis in stable continental regions, Bull. Seis. Soc. Am., vol. 94, 88–99. Seeber, L., Armbruster, J.G. and Kim, W., 2004. A fluid-injection-induced earthquake sequence in Ashtabula, Ohio: Implications for seismogenesis in stable continental regions. Bull. Seis. Soc. Am., 94(1): 76–87.
ELSE_TSANG_CH027.qxd
References
8/23/2005
2:18 PM
Page 401
401
Sinnott, R.W., 2003. Virtues of the Haversine, Sky, and Telescope, 68(2), 159, 1984, Tosoh Quartz Ltd., Durham, England. Zirbes, M., Earthquake History of Colorado, NEIC (http://wwwneic.cr.usgs.gov/neis/states/ colorado/colorado_history.html), 2001. Zoback, M.L. and Zoback, M.D., 1980. State of stress in the conterminous United States. J. Geophys. Res., 85: 6113–6156. Zoback, M.L. and Zoback, M.D., 1989. Tectonic stress field of the Continental United States. In: L.C. Pakiser and W.D. Mooney (Eds), Geophysical Framework of the Continental United States. Geol. Soc. Am. Mem., Vol. 172, pp. 523–539.
This page intentionally left blank
402
ELSE_TSANG_CH028.qxd
8/22/2005
4:16 PM
Page 403
Chapter 28
INJECTION OF BRINE FROM CAVERN LEACHING INTO DEEP SALINE AQUIFERS: LONG-TERM EXPERIENCES IN MODELING AND RESERVOIR SURVEY J. Zemkea, M. Stöwera, and M. Borgmeierb a
Untergrundspeicher-und Geotechnologie-Systeme GmbH, UGS Mittenwalde, Germany E-On Hanse AG, Hamburg, Germany
b
28.1 INTRODUCTION One of the preconditions for constructing salt caverns for storage purposes is the removal of brine from the leaching process. At locations where no options of using the brine industrially on site or to discharge the brine into rivers or the sea exist, new ways have to be found and implemented for environment-friendly brine disposal. An alternative solution is to dispose of the brine in geological formations, especially by injection into a deep aquifer. To ensure an economical and safe injection into deep aquifers under long-term conditions, geological requirements, such as the existence of extended aquifer reservoirs with suitable flow properties and tight cap rock, have to be fulfilled. The geological conditions for brine disposal have to be investigated by exploration work. The scope of work to be carried out is comparable with exploration of aquifer structures for natural gas storage. New technical solutions have to be adopted for disposal-well completion to ensure an optimal injection process and to avoid corrosion. Above-ground facilities for leaching and brine disposal are characterized by their compact construction. Elaborate in situ brine filtration and brine conditioning installations have to be erected so that the brine can be injected into the formation over a long period of time without changing the characteristics of the injection horizon. During the injection period, pressure development in the aquifer is permanently controlled by measurements in observation wells. Results are compared with analytical calculations to limit the pressure below the maximum permitted pressure in the formation. In addition, intensive simulation studies are performed to gather information about pressure distribution in the structure for the time period when the forecasted brine volume will be injected. Results and experiences concerning brine disposal in deep aquifer reservoirs are presented in this chapter, referring to geological and technological investigations carried out by Untergrundspeicher-und Geotechnologie-Systeme GmbH (UGS) for two cavern storage projects in Germany in recent years. 28.2 HISTORICAL OVERVIEW New technological innovations are necessary to build a cavern storage facility in the interior of a country without the option of using brine industrially on site, discharging the brine into rivers, or building a pipeline to the open sea. Injection into deep-aquifer reservoirs has been proposed as an alternative solution for environment-friendly discharge of the brine. In
ELSE_TSANG_CH028.qxd
404
8/22/2005
4:16 PM
Page 404
Injection of Brine from Cavern Leaching into Deep Saline Aquifers
recent years, UGS performed geological and technological investigations for brine disposal in such reservoir structures for two storage sites in Germany. The underground storage site in the Kraak salt dome, owned by our customer E-On Hanse AG, officially began operating at the end of September 2000. The supply of a working gas volume of approximately 160 ⫻ 106 m3, subdivided into three caverns, is planned by the year 2004. The leaching operation started in January 1997. The first cavern was commissioned in winter 2000/2001. Brine drainage is simultaneously continued in the other two caverns, so that a total geometric cavity of 1.4 ⫻ 106 m3 will be reached at the end of the leaching period. The exploration work carried out in 1994 and 1995 was not only focused on the Kraak salt dome, but was also extended to a nearby deep-aquifer reservoir with a forecasted injection volume on the order of 9 ⫻ 106 m3. The reservoir recently put into use has a formation water volume of approximately 1.8 ⫻ 109 m3 and is structurally limited. In 2002, new concepts for the possible extension of the storage site, and associated options for long-term brine disposal, were investigated. Based on optimized structural interpretation of the diapir and experience from the ongoing leaching process, a total of 23 caverns with a geometric cavity of 10.4 ⫻ 106 m3, and a total brine volume of 73 ⫻ 106 m3, was predicted. Therefore, extended reservoir capacity for brine disposal is needed. Extensive simulation studies were carried out for two alternative deep-aquifer reservoirs that are both structurally unlimited (open-reservoir type). For another customer, EWE, seismic and drilling exploration work was performed in recent years at the salt pillow of Rüdersdorf, near Berlin, and the injection location of Heckelberg, about 50 km from the cavern location. For brine disposal, sandstone layers in four intervals, with a summarized thickness of approximately 150 m, were evaluated in a mainly unlimited reservoir, with high porosity and permeability, at a depth of 900–1300 m, providing favorable conditions for brine injection. Locations for both projects are shown in Figure 28.1. Detailed information presented in this paper will mainly refer to the cavern storage project in Kraak in North Germany. An areal overview of the above-ground facilities at the Kraak storage site is given in Figure 28.2. The entire installation is controlled from the UGS in Reitbrook (80 km away).
28.3 GEOLOGICAL REQUIREMENTS FOR BRINE DISPOSAL IN POROUS AQUIFERS The following geological requirements have to be fulfilled to ensure an economical and safe brine injection in deep aquifers under long-term conditions: ● An aquifer reservoir of sufficient areal extension must be available to allow the placement of an appropriate amount of brine. Trough-shaped structures have the advantage of entrapping the injection fluid for reducing the risk of uncontrolled brine migration into other aquifer horizons (Fig. 28.3). ● Favorable reservoir properties, like high layer thickness and good porosity, as shown in Table 28.1, are necessary to dispose of the brine with low injection pressure and minimize the number of injection wells. ● The aquifer structure must be covered by a tight cap rock with areal integrity. This is especially required for reservoirs with closed-boundary conditions where a high-pressure gradient must be realized in order to reach sufficient brine disposal capacity (Fig. 28.4). Consequently, the exploration work was not only performed at the Kraak salt dome, but was simultaneously extended, with the same scope of investigations, to the injection formation.
ELSE_TSANG_CH028.qxd
8/22/2005
4:16 PM
Page 405
28.4 Preliminary Investigations
Gas storage locations in pore space
405
in caverns
in operation
in operation
planned
planned
Fig. 28.1. Location of Kraak and Rüdersdorf storage sites on the map of Germany.
28.4 PRELIMINARY INVESTIGATIONS
●
Preliminary investigations for brine disposal in deep-aquifer structures mainly include: Seismic and drilling exploration work to select the most promising aquifer structure.
ELSE_TSANG_CH028.qxd
406
8/22/2005
4:16 PM
Page 406
Injection of Brine from Cavern Leaching into Deep Saline Aquifers
Fig. 28.2. Areal overview—above-ground facilities—Kraak storage site.
Investigations of core samples: — from salt cavern wells to determine salt composition and the insoluble fraction; — from disposal wells to examine the rock properties of permeability and porosity, and the capillary properties of the reservoir and the cap rock. ● Formation testing to determine the reservoir properties and the capacity of the injection wells. ● Laboratory investigation for the evaluation of possible interactions between the brine to be disposed of, and the formation fluid and rocks in the hydraulically linked area. ● Comparison of different well-completion techniques for optimizing the brine injection. ● Evaluation of different technological solutions for a comprehensive mechanical and chemical method of brine conditioning. Based on these investigations (results are summarized in other research and technical reports), technical solutions for brine discharge by injection into geological formations at Kraak and Rüdersdorf were developed and established. ●
28.5 TECHNICAL SOLUTIONS 28.5.1 Disposal-Well Completion The following main aspects of well completion have to be considered: ● Use of stainless casing and liner elements in the reservoir intervals ● Use of coated casing and tubing elements for the injection string ● Intensive cleaning of the well, after first perforation, by producing formation water and adding nitrogen to avoid the corrosion process ● Installation of packers to isolate specific layers.
ELSE_TSANG_CH028.qxd
8/22/2005
4:16 PM
Page 407
28.6 Monitoring and Simulation Program for Brine Disposal
407
Schwerin
Sewage treatment plant Schwerin
Dogger reservoir 5 km
0
120
250 km2
Gt S 2/87 Gt S 5/88
0
140
0
21 km
160
1800 2000
Gt S 3/87
Gt S 1/88
2200 2300
18 km
fresh water brine Gt
Kraak
K 101/94 102/97 103/99
injection well
Kraak salt dome 7 x 4.5 km
Fig. 28.3. Kraak salt dome and areal extension of Dogger reservoir near Schwerin. The pipeline from the cavern storage site to the injection wells is about 21 km long. The fresh water pipelines are also shown.
28.5.2 Surface Disposal Facilities Brine conditioning includes (Fig. 28.5): ● Chemical conditioning to avoid plugging reactions by decreasing the pH-value and oxygen content in the solution. ● Microfiltration of the brine to a particle size of 0.2 m to avoid plugging effects in the sandstone pores of the reservoir. ● Adding fresh water to the brine to limit brine density before pumping it to the disposal site. After the leaching process, the brine is transported into a closed tank, and then in a pipeline system, avoiding any contact with atmospheric oxygen. The pipeline for brine transportation consists of plastic material.
28.6 MONITORING AND SIMULATION PROGRAM FOR BRINE DISPOSAL Suitable geological structures for brine disposal were found in an area north of the Kraak salt dome in Mesozoic sandstones of the Aalen (Middle Jurassic) and Rät (Upper Triassic) formations. The reservoir is well known from former geothermal projects. The horizons are
ELSE_TSANG_CH028.qxd
8/22/2005
408
4:16 PM
Page 408
Injection of Brine from Cavern Leaching into Deep Saline Aquifers Table 28.1. Reservoir properties — Aalen sandstone (brine-injection formation)
Reservoir parameter
Project
Reservoir area, F Effective thickness, heff Porosity, ⌽ Formation water volume, V Initial reservoir pressure, pinit Max. reservoir pressure Pressure increase, 䉭p Specific disposal capacity Disposed brine volume
2
km m % 106 m3 bar bar bar 106 m3/bar 106 m3
10/2002 265 10 . . . 60 25 1,810 130
195.0 65.0 0.226 14.5
156.63 26.63 ca. 0.22 5.9
Fig. 28.4. Cross section of brine-disposal area.
Fig. 28.5. Brine conditioning and disposal flow-path schematic.
located at a depth of about 1300–2000 m. Five geothermal wells had been drilled through the formation. Some of them were reopened and used for brine injection. The total surface area covers approximately 250 km2, and conservative calculations have shown that cumulative injection volume could be expected on the order of some 10 ⫻ 106 m3 brine. The aquifer is isolated from groundwater horizons by overburden layers, which in this case consist of argillaceous marl and Oligocene clay (Rupel).
ELSE_TSANG_CH028.qxd
8/22/2005
4:16 PM
Page 409
28.6 Monitoring and Simulation Program for Brine Disposal
409
An extensive measure and monitoring program, for pressure and volume development in the aquifer structure, was created to keep the massive underground fluid invasion under control. The following items have to be observed: ● Temperature distribution in the aquifer by discontinuous measurements in the injection wells. ● Pressure development in the reservoir with two observation wells. Permanent measurements at the wellhead with memory units in addition to discontinuous registration of well-bottom hole pressure. ● Volume development in three injection wells by online registration of fluid rates. ● Changes in ground level (fine leveling) above the structure. Periodic measurements of ground movement at the surface, depending on the injected volume. The monitoring program is completed by analytical model calculations to enable a prognosis on further pressure development, depending on injection rates and cumulative injected volume. This is necessary to keep the pressure increase in the aquifer below 65 bar, which is the maximum value allowed by mining authorities. Since injection started in 1997, there have been 7.138 ⫻ 106 m3 (as of December 2003) of brine injected in three wells with rates in the range of 100–150 m3/h (Fig. 28.6). Results from analytical model calculations show a pressure increase of approximately 5 bar per 1 ⫻ 106 m3 of brine volume injected. This corresponds very well with the observed data. The total amount of brine from the leaching process is transported via a pipeline to three injection wells (Ug S 5A/93, Gt S 2/87, Gt S 3/87). In the past, two of them were mainly used for brine injection. The proportion of brine disposed by each well is given in Figure 28.7.
Fig. 28.6. Average injection rates and cumulative injected volume of brine (1997–2003).
ELSE_TSANG_CH028.qxd
410
8/22/2005
4:16 PM
Page 410
Injection of Brine from Cavern Leaching into Deep Saline Aquifers
Fig. 28.7. Cumulative volume for different wells, and percentage of total brine injected.
An advanced monitoring program was accomplished using the Schlumberger GeoQuest reservoir-modeling software application. The 3-D reservoir simulator—originally used in the oil and gas industry—was adapted to the problem of brine disposal in a deep aquifer. The model for the structure was initialized with two layers (upper and lower Aalen horizon) and closed-boundary conditions. Reservoir rock and fluid properties were taken from the geological reservoir interpretation and laboratory measurements. The simulation grid and the initial pressure distribution are shown in Figure 28.8. The matching process of injection history was controlled by observed pressure data from the injection and monitoring wells. The geometric distribution of reservoir parameters in the model was set up using geostatistical methods. In the simulation run, the brine was marked with tracers to visualize the fluid invasion into the aquifer. A local grid refinement was applied to take into account that pressure changes are much higher near the injection wells. After adjusting the property data, and the porosity and permeability distribution during the history-matching process, the results of the simulation run corresponded extremely well with the measured pressure data. Results from simulation showed good evidence for the entrapment of brine in a closed-aquifer structure, since the measured data could only be matched with closed outer-boundary conditions. The pressure distribution and its development in time were visualized with the simulation model. Prognosis runs were also made to verify the analytical-model calculations. The resulting pressure rates from the analytical-model calculation, as well as those from the simulation model, are shown together with observed pressure data in Figure 28.9. Permanently measured pressures at the wellhead, and periodically registered bottom-hole pressures from downhole tools, were accurately matched with the calculated values. This is evidence that the assumed model parameters are valid. The simulation model is periodically updated, and can be used as a helpful planning tool for the optimization of the injection process. Injection rates at the different wells can be
ELSE_TSANG_CH028.qxd
8/22/2005
4:16 PM
Page 411
28.6 Monitoring and Simulation Program for Brine Disposal
411
Fig. 28.8. 3-D simulation grid and initial pressure distribution for the injection horizon (Aalen formation).
Fig. 28.9. Pressure increase in the aquifer structure with respect to brine injection—results from model calculation and observed data.
adjusted depending on the calculated pressure distribution in the model. Prognosis calculations for wellhead pressure could give information about necessary pumping capacities for injection, and for decisions concerning the construction of a new pumping station or some simulation of injection wells.
ELSE_TSANG_CH028.qxd
412
8/22/2005
4:16 PM
Page 412
Injection of Brine from Cavern Leaching into Deep Saline Aquifers
Prognosis calculations with the simulation model have shown that the leaching process and the brine injection in the aquifer can be extended as planned without reaching critical pressure values in the reservoir. Simulation models provide good evidence to ensure economical and safe brine injection into deep aquifers under long-term conditions.
28.7 RESULTS AND EXPERIENCES GATHERED 28.7.1 Kraak Storage Project In the period from 1997 to August 2003, nearly 6.3 ⫻ 106 m3 of brine were injected with three wells into the two horizons of the deep-aquifer structure near the Kraak salt dome. The formation water volume of the aquifer is about 1.8 ⫻ 109 m3. The injected brine was entrapped in a closed structure. The pressure in the formation increased, corresponding to the compressibility of the porous reservoir. The facilities for brine conditioning and microfiltration were operated without any serious problems over the last years. A stable brine injection regime could be reached. It was shown that brine disposal in deep-aquifer layers could be a good alternative solution for locations where there are no options for industrial brine processing or for discharging the brine into rivers or the open sea. 28.7.2 Rüdersdorf Storage Project In recent years, seismic and drilling exploration work was performed at the Rüdersdorf salt pillow, and at the Heckelberg aquifer location (about 50 km away from the cavern location). Four sandstone layers, with a total thickness of approx. 150 m, exhibiting high porosity and high permeability, are available at depths ranging from 900–1300 m, providing favorable conditions for brine injection. The aquifer structure is supposed to be unlimited. An efficient concept for use of the four partially separated sandstone layers was developed and offered to the customer. Some additional investigations for optimizing the brine-conditioning process were performed with respect to changed parameters at the new location. UGS is now involved in the construction and operation of the leaching and brine disposal facilities for its customer, EWE. The project includes the drilling of two cavern wells and two disposal wells. Leaching operations and brine disposal in the deep underground were started this year.
ELSE_TSANG_Ch029.qxd
8/22/2005
4:18 PM
Page 413
Chapter 29
USE OF DEEP GEOLOGIC HORIZONS FOR LIQUID WASTE DISPOSAL AT POWER COMPLEXES IN CENTRAL RUSSIA B.P. Gorbatenkoa, A.D. Turkovskiya, A.I. Rybalchenkob, M.K. Pimenovb, E.P. Kajminc, and E.V. Zacharovac a
Kalinin Nuclear Power Plant, Udomlia, Russia All-Russia Designing and Research Institute of Production Engineering (VNIPIPT), Moscow, Russia c Institute of Physical Chemistry of the RAS, Moscow, Russia b
29.1 INTRODUCTION Thermal and nuclear power plants for the production of electricity, and power plants for supplying heat to big cities, are large water consumers. Water used in technological processes at these plants must meet defined requirements for chemical composition, so water treatment (“softening”) is accomplished with ion-exchange equipment. During periodic regeneration of this ion-exchange equipment, wastewater containing salts of calcium, magnesium, and other components are generated. Cold, long winters in Russia require the generation of large quantities of heat, so hundreds or thousands of cubic meters of waste form every day as a result of water treatment. Traditional methods of liquid waste management include their discharge into surface lakes, ponds, and rivers, but this practice of waste disposal cannot be applied everywhere due to environmental protection policy. Well injection of such waste into deep geological formations (reservoir horizons) is one of the methods for isolating waste from fresh surface and shallow underground drinking water supplies. Deep horizons, as a rule, contain saltwater (brines), which is why exchanging one type of saltwater for another, though different in composition, will not cause any significant changes in the geologic formation. The existence of a favorable geological structure is a necessary condition for the application of deep-well injection. There are many regions in Russia where geological formations are suitable for deep-well liquid waste injection. Deep-well injections of liquid industrial waste are regulated by several acts (laws) of the Russian Federation. The main acts are: “On Mineral Resources” (2000), “On the Environment” (2002), and “On the Waste Production and Consumption” (1998). The basic requirement for deep-well injection is the localization of waste in boundaries of exclusion zones or allotment.
29.2 GEOLOGICAL CONDITIONS FOR WASTE INJECTION IN THE CENTRAL PART OF RUSSIA In the central part of European Russia, known geologically as the Russian Platform, vast territories of the big sedimentary basin are characterized by favorable conditions for deepwell injection of liquid industrial waste. Deep geological sections of those territories are composed of sedimentary rocks from the Carboniferous, Devonian, and more ancient periods.
ELSE_TSANG_Ch029.qxd
414
8/22/2005
4:18 PM
Page 414
Use of Deep Geologic Horizons for Liquid Waste Disposal
Deep sedimentary formations include porous permeable horizons, able to contain and retain limited volumes of injected waste, and layers of low-permeable rocks (mainly clays and salts) isolating permeable horizons from surface and shallow fresh water aquifers. Similar conditions are observed in the Moscow region, and in almost the whole European part of Russia, excluding the Voronezh Crystalline Massif and the Northwest Territories at the border with the Scandinavian shield. At the Voronezh Crystalline Massif, magmatic and metamorphic rocks outcrop on the surface. In the Northwest Territories, the thickness of sedimentary rocks decreases in the direction of the Scandinavian shield. From the east, the area of sedimentary rocks is limited by the Ural Mountains. Deep permeable horizons contain saltwater with mineralization from several tens to hundreds of grams per liter, and refer to zones of stagnation water exchange. Groundwater flow in such horizons is very low, and flow velocities are estimated to be of the order of fractions of a meter per year.
29.3 GEOLOGY OF THE KALININ NUCLEAR POWER PLANT The Kalinin Nuclear Power Plant is located in the northwest part of the Russian Platform, on the west wing of the Moscow Syneclise. Geological structure includes Archean, Proterozoic, Cambrian, Ordovician, Devonian, Carboniferous, and Quaternary rocks and sediments. Three structural complexes are marked out in geological sections from bottom to top: ● The first stage is represented by highly dislocated crystalline rocks of the basement. ● The second stage is composed of Riphean sediment occurring unconformably on the firststage rocks. ● Third-stage units of sedimentary cover begin with Vend sediments of the Valdai series occurring on dissected surfaces of Riphean and overlying sediments. The basement surface is very complicated, connected with the junction of two structures of the second order: Nelidovo–Torjok and Valday graben. The deep-well injection site is located 15–20 km from the eastern edge of the graben, beyond the boundaries of regional faults. The basement in the site injection area is complicated by faults of the second order. The nearest fault is located at a distance of 1 km from site injection. The center of site injection is located between faults. The basement between the faults is not disturbed and has subhorizontal surface bedding at a depth of 2750–2800 m. For waste injection according to geological exploration results, the recommended reservoir horizon is in Riphean sediments (complex of Tiscresco–Baltic sediments), bedding in the central part of the site within depth intervals of 1283–1352 m, composed of fine-grained sand and sandstone. Above the reservoir horizon (injection zone), low-permeable, practically water-confining sediments occur, which include Ordovician and Narovsky horizon sediments of the Devonian Period (carbonate and clay rock, gypsum) with total thickness of 370 m. This formation separates the injection zone from a buffer horizon consisting of Shvetoisky–Starooskolsky and Semiluksky horizons of the upper-middle Devonian. The buffer horizon is isolated from above-lying permeable horizons by clay formations of the upper Frankian sediments of the upper Devonian, with a total thickness of 250 m. Carboniferous and Quaternary sediment is bedded in the upper part of the geological section. The reservoir horizon occurs in a zone of troubled water exchange and contains water with a mineralization of 220 g/L. Mineralization of buffer horizon water is 80 g/L. Absolute heads (elevation above sea level) of static underground water levels in the reservoir horizon, and in the buffer horizon, differ significantly and are characterized accordingly as ⫺3 m and
ELSE_TSANG_Ch029.qxd
8/22/2005
4:18 PM
Page 415
29.4 Characteristics of Waste and Site Injection
415
⫹ 87 m. Depth of the static water level in the reservoir horizon is 174–176 m. The large distinction between mineralization and the absolute heads of underground water levels is evidence of isolation for reservoir and buffer horizons. The complex of sedimentary rocks beginning with the Vend sediments is not disturbed by tectonic faults, as confirmed by data gained during well drilling and the results of helium investigations in underground water. Helium concentration was determined to be n ⫻ 10⫺5 mg/L. According to the results of well investigations, and test-filtration data, the reservoir horizon has the following characteristics (average values): ● Total thickness: 65 m ● Effective thickness: 35 m ● Effective porosity: 0.14 ● Transmissivity: 138 m2/d ● Hydraulic conductivity: 2.5 m/d ● Piesoconductivity or head diffusivity coefficient (ratio of hydraulic conductivity to elastic storage): 1.4 ⫻ 10 m2/d ● Specific discharge of well: 35 m3/d MPa Available data on geologic structure, and hydrogeological conditions of the site area, testify to reliable isolation of the reservoir horizon from shallow horizons of fresh water, and to the possibility of waste localization in the deep horizon. Indicated parameters of the reservoir horizon will allow the injection of waste with acceptable pressures on the well heads.
29.4 CHARACTERISTICS OF WASTE AND SITE INJECTION Liquid waste for injection is created by the equipment for demineralization of natural water used for replenishment of technological systems in the first and second loops of the nuclear reactor. Treated water is taken from a surface water reservoir, which is why the waste contains calcium and magnesium salts. This main category of waste also includes decontamination solutions from different sources: vat residues from evaporation equipment containing boric acid, solution used for the regeneration of exchange capability in ion–exchange materials, spent washing solution from steam generators, and distillates which contain radioactive tritium. The total volume of waste for disposal is 1050–1325 m3/d. According to norms, the composition and salt concentration of these wastes classifies them as of low toxicity. Tritium-specific activity in the total volume of waste will be 200 Bq/L (average). Taking into account the tritium half-life of 12.3 years, 25–30 years after injection the tritium activity will decrease below threshold values for referring to this waste as radioactive waste (77 Bq/L). In addition to tritium, other radioactive nuclides are also contained in unbalanced solutions. They are: cobalt-60, strontium-90, cesium-137, etc. Their specific activity is ten times lower than established norms. Earlier, when only two reactors were operated at the Kalinin Nuclear Power Plant, nonradioactive waste was discharged into the Sjezha River. In accordance with the requirements of regulating state bodies, the discharge of wastes into surface water had to stop after the third reactor began operating in 2004. Discharge of salt-containing wastes into the cooling pond system will result in salinization of the water and deterioration of the circulating cooling system. After concentration, liquid radioactive waste formed at the nuclear plant is bitumenized and then stored in a surface repository.
ELSE_TSANG_Ch029.qxd
416
8/22/2005
4:18 PM
Page 416
Use of Deep Geologic Horizons for Liquid Waste Disposal
The deep-well injection site includes five injection wells and five observation wells up to 1300 m deep in the reservoir horizon. There are also observation wells for above-lying horizons. Surface construction includes the equipment for treatment of waste prior to injection, pumping equipment with a pressure of 2.5 MPa for waste injection, high-pressure pipeline for supplying treated waste to the injection wells, and a station for control and management. Surface construction of the injection site, and the injection wells, is located within the boundaries of the sanitary-protective zone for Kalinin Nuclear Power Plant. For deep-well injection of liquid waste, exclusion zones (allotment) are registered. Exclusion zone boundaries are 3 km in diameter.
29.5 WASTE TREATMENT FOR DEEP-WELL INJECTION Waste treatment is done to prevent a decrease in rock permeability in the vicinity of injection wells in the reservoir horizon. The treatment includes sedimentation, filtration, coagulation, and pH correction. The waste is accumulated in a receiving reservoir from which high-pressure pumps direct it to the injection wells. Effective pressure for waste injection will be 0.5–1.5 MPa. Taking into account the filter zone resistance of the injection well, maximum pressure is evaluated as 2.5 MPa. The permeability of injection zones is preserved.
29.6 ESTIMATION OF DEEP-WELL INJECTION CONSEQUENCES Analyses of reservoir horizons filled by injected waste, the development of pressure fields, and long-term operating experience with functioning injection sites shows that waste localized within the boundaries of the allotment does not have any impact on the environment beyond the boundaries of the sanitary-protection zone—soil cover, vegetation, surface water, shallow underground water, and population. According to forecasting calculations, as a result of waste injection, pressure in the reservoir horizon will be 0.10–0.15 MPa on the wall of the injection well, which will exceed the natural pressure by 10–15%, and by 1–2% up to several tens of meters from the well. Such pressure changes will not cause any geodynamic phenomena, or induce the seismicity that is seen in long-term seismic observations of the Mining and Chemical Combine (Krasnoyarsk-26) area. The level of underground water in the wells is established at depths of 170–180 m. When a well shaft is filled with waste, the salt content and density of which is correspondingly lower than the underground water, the water level will rise but remain below the surface. This means that if the well head were to fail suddenly, sealing discharges from the head of the well would be practically impossible. The calculations were performed on waste distribution within the reservoir horizon for a 30-year period of deep-well injection, taking into account hydrodynamic dispersion and different densities of the waste and underground water. Formation of a dispersion zone is characterized by a uniform density increase in the mix of waste and underground water, extending from waste-filled areas of the horizon to peripheral parts of the waste contour. It was determined that formation of a dispersion zone creates a decrease in the influence of density differences on the regularity of waste distribution within the reservoir horizon. After 30 years of injection, the distance of the waste marker from the center of injection will be
ELSE_TSANG_Ch029.qxd
8/22/2005
4:18 PM
Page 417
29.8 Conclusion
417
1000–1200 m. After injection is stopped, injected-waste migration in the reservoir horizon will occur with a lateral velocity of about 0.5–1.0 m/year. The upper horizons of fresh water are reliably isolated from the inner tubing of the injection well through which waste is injected into the reservoir horizon. The interval of 0–140 m is completed with three casings, the interval up to 350 m is completed with two casings. The annuli between the outside of the casings and the rock are cemented. After operations are completed, the injection site will be shut down. Technology for well abandonment was previously developed and checked while abandoning wells at other injection sites.
29.7 EQUIPMENT FOR USING A NATURAL UNDERGROUND SOLUTION IN MOSCOW At the plant for heating and electric energy generation in the south of Moscow (West Biruljevo), deep horizons within depth intervals of 135–1470 m are used for extracting saltwaters used for regeneration by the water treatment (demineralization) equipment. Liquid wastes formed with calcium and magnesium salts are injected into the same horizon, but through another well located 500 m distant from the first well used for extracting saltwater. This technology made it possible to stop buying sodium chloride, stop processing run-off, and stop discharging it into the Bitsa River, a tributary of the Moscow River.
29.8 CONCLUSION Results from studies of deep geological formations in European Russia, as well as our injection experience and expertise, show that liquid salt wastes from nuclear power plants can be successfully injected into deep geological formations. The volume of liquid wastes that is suitable for injection at atomic power plants can be increased through the use of new effective sorption materials developed by the Russian Federation. By applying these materials in the preprocessing of liquid radioactive waste, total radioactive activity can be reduced to levels below those established for radioactive wastes. As a result of this, nonradioactive liquid salt wastes can be injected into deep geological formations. This preprocessing and injection will decrease the physical volume of radioactive wastes remaining on the surface for further storage.
This page intentionally left blank
418
ELSE_TSANG_CH030.qxd
8/22/2005
4:20 PM
Page 419
Chapter 30
CASE STUDY: EVALUATION OF OIL FIELD AND WATER-WELL DISPOSAL-WELL DESIGNS FOR OIL SANDS FACILITY IN NORTHERN ALBERTA, CANADA Y. Champolliona, M.R. Gleixnera, J. Wozniewicza, W.D. MacFarlaneb, and L. Skulskib a
Golder Associates, Ltd., Calgary, Alberta, Canada Nexen Canada Ltd., Calgary, Alberta, Canada
b
30.1 INTRODUCTION The Long Lake Project, operated by Nexen Canada, Ltd., and OPTI Canada, Inc. (Nexen/OPTI), is one of many oil sands projects currently being developed in northeastern Alberta, Canada. The project includes a pilot and commercial phases and will extract bitumen using the Steam Assisted Gravity Drainage (SAGD) technology that involves pairs of horizontal wells, parallel to each other within the same vertical plane. The upper well injects steam into the reservoir to heat up the bitumen that flows downward into the lower well and to the surface. As part of the extraction process, large amounts of wastewater are generated, consisting mostly of produced and blowdown water. The target formation for disposal is fine- to medium-grained unconsolidated sands in the water-bearing lower portion of the Cretaceous McMurray Formation, located approximately 250 m below ground surface. Because of the nature and depth of the target formation, two distinct disposal-well designs have been evaluated at the pilot stage of the Long Lake Project: (1) perforated casing (the standard oil and gas approach) and (2) wire-wound telescopic screen (standard water-well approach). One disposal well of each design was completed and tested to assess well performance and construction and maintenance costs.
30.2 GENERAL SETTING 30.2.1
Geology
The Long Lake Project area is situated at the northeastern edge of the Western Canada Sedimentary Basin, located southwest of the Canadian shield exposure (Fig. 30.1). The regional stratigraphy from the youngest (shallowest) to the oldest (deepest) overlying Precambrian rock consists of Quaternary deposits, Cretaceous clastic formations, and Devonian limestones and evaporites (Fig. 30.2). The Quaternary deposits consist of discontinuous sandy and clayey till beds that were deposited in Paleo-channels and morainal plains shaped by glacial and fluvial processes. Cretaceous clastic formations are thickest to the southwest of the Lease, where the shale–sandstone–shale sequence of the LaBiche, Pelican, and Joli Fou formations are present beneath
ELSE_TSANG_CH030.qxd
420
8/22/2005
4:20 PM
Page 420
Case Study: Evaluation of Oil Field and Water-Well Disposal-Well Designs
Fig. 30.1. Location map.
Fig. 30.2. Geologic cross section.
ELSE_TSANG_CH030.qxd
8/22/2005
4:20 PM
Page 421
30.2 General Setting
421
Stoney Mountain. The Grand Rapids Formation subcrops beneath the Quaternary deposits over most of the Lease area where the LaBiche, Pelican, and Joli Fou formations are absent due to erosion. The Grand Rapids Formation consists of variably cemented sands and sandstones that are up to 160 m thick. The Clearwater Formation underlies the Grand Rapids Formation and consists of up to 115 m of marine shale and siltstone. The underlying McMurray Formation ranges in thickness from 50 to 100 m, consists mainly of sandstones with clay interbeds, and hosts a vast bitumen resource. The lower part of the McMurray Formation, also referred to in this paper as McMurray basal water sands, includes water-bearing channel sands and has been identified as the target formation for wastewater disposal at the Long Lake SAGD pilot operation. The thickness and grain size of the McMurray basal water sands is variable within the Long Lake area (Fig. 30.3). In the immediate vicinity of the Long Lake pilot, where the entire McMurray section is bitumen saturated, the McMurray basal water sands isopach is zero. Moving to the east and south of the pilot area, the basal water sands unit lies between the base of the bitumen-saturated McMurray section and the underlying Devonian Beaverhill Lake erosional surface. The thickness of the basal water sands generally increases to the east and southeast in the Long Lake area, while the thickness of the bitumen section of the McMurray
Fig. 30.3. McMurray basal water sands isopach, and locations of pilot-plant disposal wells WDW 9-28 and WDW 1-21.
ELSE_TSANG_CH030.qxd
8/22/2005
422
4:20 PM
Page 422
Case Study: Evaluation of Oil Field and Water-Well Disposal-Well Designs
Formation decreases. East of the bitumen edge, the McMurray Formation is entirely watersaturated, with the exception of thin gas zones, and consists of variable amounts of interbedded shales and water-bearing sands. At 9-28-85-6W4M, where disposal well WDW 9-28 is located (wire-wound telescopic screen design), the top and the bottom of the disposal zone are at 185.0 and 239.0 m below Kelly Bushing (m KB), respectively. This corresponds to a net thickness of the McMurray basal water sands of 41 m, using an approximate 27% density porosity cutoff. At 1-21-856W4M, where disposal well WDW 1-21 is located (perforated casing design), the top and the bottom of the disposal zone are at 206.8 and 274.0 m KB, respectively. This corresponds to a net thickness of the McMurray basal water sands of 31 m, using an approximate 27% density porosity cutoff. 30.2.2 Hydrogeology Groundwater flow in the Cretaceous formations is generally southwest to northeast, with groundwater recharge occurring in the Stoney Mountain area in the southwest, and discharge taking place along outcrops of the Christina and Clearwater river valleys to the northeast. The primary Cretaceous aquifers are the Grand Rapids and McMurray formations. These two formations are separated by the Clearwater aquitard. Aquifer confinement On a regional scale, the McMurray Formation is confined by the considerable thickness of the Clearwater Formation that separates the disposal interval from surficial groundwater resources and by the underlying clay-weathered top of the Devonian formations. The disposal interval is also interbedded with localized silty and clayey intervals within the McMurray Formation. These strata confine the disposal interval, and therefore fluids disposed in the disposal wells preferentially flow laterally within the disposal interval. Evidence of confinement is provided by significant hydraulic gradients between these zones under static conditions (Fig. 30.2) and by the absence of pressure response in the formations overlying the McMurray during hydraulic testing. Formation geochemistry The results of the analytical testing showed that the McMurray Formation basal water sands at both WDW 9-28 and WDW 1-21 is a sodium chloride type of water, with elevated total dissolved solids (TDS) concentrations in the order of 50,000 mg/L. This formation water also contains some dissolved hydrogen sulfide and traces of dissolved hydrocarbons. 30.2.3 Siting of Disposal-Well Locations The McMurray basal water sands were selected over other geological formations as the primary target formation due to its relative shallowness (depths of 250 m), poor water quality (TDS concentrations in the order of 50,000 mg/L), relative continuity, and great extensiveness. As explained above, the McMurray Formation is a regionally occurring unconsolidated sandstone unit that is mostly water-saturated east of the Long Lake Project area, and mostly bitumen saturated west of the bitumen edge (Fig. 30.3). The two principal factors considered for siting the locations of the disposal wells were (1) proximity to the plant area to minimize infrastructure costs (roads and pipelines) and (2) reservoir capacity. By locating the disposal wells along the bitumen edge that separates the bitumen-saturated area from the water-saturated
ELSE_TSANG_CH030.qxd
8/22/2005
4:20 PM
Page 423
30.3 Well Construction and Completion
423
area of the McMurray Formation, the large disposal capacity requirements could be achieved. Indeed, predictions using a regional groundwater flow model showed that the head buildup would be limited to 50 m after 40 years of disposal, and that the pressure front would develop eastward within the water-saturated part of the lower McMurray Formation. 30.3 WELL CONSTRUCTION AND COMPLETION 30.3.1 Permitting Deep-well disposal in Alberta is regulated jointly by the Alberta Energy and Utilities Board (EUB) and Alberta Environment (AENV). The concept and purpose of a deep-well disposal scheme must first be approved as part of a specific project (scheme) approval under the Oil Sands Conservation Act (OSCA) and the Environmental Protection and Enhancement Act (EPEA). To drill, complete, and test disposal wells, a specific approval is required under EUB Guide 56 Energy Development Application Guide and Schedules (EUB 2000a). Finally, to operate successfully tested disposal wells, an approval under EUB Guide 65 Resources Applications for Conventional Oil and Gas Reservoirs, Unit 4 Disposal/Storage (EUB 2000b), and signed by the Alberta Ministry of Environment is also required. The application for approval under Guide 65 addresses well completion, hydraulic isolation, mineral ownership, and conservation issues. In particular, specific logging requirements such as temperature surveys, annulus pressure surveys, and cement bond logs are required based on the class of disposal, per EUB Guide 51 Injection and Disposal Wells: Well Classifications, Completion, Logging, and Testing Requirements (EUB 1994). 30.3.2
Well Completion
Figure 30.4 illustrates the completion details for water disposal wells WDW 1-21 and WDW 9-28. These wells were constructed in accordance with EUB Guide 51, Injection and Disposal Wells, such that they may be used as Class Ib wastewater disposal wells. In particular, two strings of casing were installed and cemented, a production packer sealed the bottom of the injection string to the production casing, isolating the inhibiting-fluid filled part of the annulus. Finally, a wellhead rated for 21MPa working pressure was fitted to the top of the production casing to seal the well. Oil field design The completion of WDW 1-21 included perforations between 237.0 and 242.0, 244.0 and 249.0, and 267.0 and 274.0 m KB within the McMurray basal water sands. The production casing was cemented from ground surface to the total depth, and the well was stimulated by swabbing and coiled tubing. Water-well design The completion of WDW 9-28 did not include perforations, but it used a telescopic wirewound stainless-steel screen installed within the McMurray basal water sands. The production casing was installed and cemented to a depth of 196.1 m KB, more than 25 m below the deepest usable groundwater zone. The telescopic stainless-steel screen and blank riser were installed between 193 and 243.9 m KB. The total length of the installed screened interval was 17.5 m, consisting mostly of an eight-slot (8/1000 in., or 0.2 mm) screen. The slot size was
ELSE_TSANG_CH030.qxd
424
8/22/2005
4:20 PM
Page 424
Case Study: Evaluation of Oil Field and Water-Well Disposal-Well Designs
Fig. 30.4. Oil field and water-well completion diagram.
determined by a particle-size analysis of core samples, and blank stainless-steel riser pipes were installed through the interbedded silty and clayey zones within the gross disposal interval.
30.4 TEST AND ANALYSIS PROCEDURES 30.4.1 Test Procedures The principal test objective was to collect data that would allow a detailed diagnosis of the formation response to understand both local- and large-scale properties. This was achieved over the course of two test campaigns conducted following well completions and 1 year later.
ELSE_TSANG_CH030.qxd
8/22/2005
4:20 PM
30.4 Test and Analysis Procedures
Page 425
425
A local-scale phenomenon of special interest was clogging, a common problem in injection wells and exacerbated in finer-grained formations. A significant reduction in well capacity at WDW 9-28 had been identified during the first testing campaign and needed to be further understood in terms of occurrence and mechanism. Filtered and nonfiltered waters were used, and the sources of the injection water were also varied. In addition, select injection water quality parameters, including pH, dissolved oxygen (DO), and temperature, were measured to attempt to define the quality of the injection fluid. Finally, injection water samples were also collected for extensive laboratory analyses to characterize the chemical composition of the test injection fluids. Characterization of large-scale properties of the target formation, and the presence and type of boundaries were needed in addition to the local-scale well performance to better predict the long-term injection capacity. Observation points included offset standpipe piezometers and vibrating wire piezometers that were monitored during the injection and recovery periods. They included near- and far-field observation points located 80 and 2000 m from the test disposal wells. The monitoring points were completed within the target formation as well as within the formations overlying and underlying the target formation to detect possible vertical leakage. Finally, downhole pressures and temperatures were also monitored during testing using tandem recorders hung suspended to a wire line. Transient tests were performed after well development with real-time data analysis used for the optimization of the subsequent test. The test sequence was similar at both disposal wells, WDW 1-21 and WDW 9-28. Initially, a step-injection test was conducted to evaluate the well efficiency and local-scale properties (scale of a few meters). The injection rates covered the expected long-term operation requirements of the pilot project. Analysis of the collected data was used to determine a rate for the subsequent long-term injection tests. The selected rate was a compromise between the need for a large enough rate to induce a detectable formation response and the limited volume that could be reasonably delivered to the wellhead from nearby groundwater source wells. The main objective of the first long-term test was to determine the large-scale aquifer properties (scale of thousands of meters). The second constantrate injection test was conducted with unfiltered water to evaluate the disposal well’s sensitivity to physical clogging due to suspended solids in the injection fluid. Finally, step pumping and constant-rate pumping tests were conducted at WDW 9-28, which exhibited some clogging at the end of the first injection test, to assess the degree of “cleanup” that could be achieved by well redevelopment to assist in the long-term operation of the disposal wells. 30.4.2 Methods for Data Analysis Two analytical techniques beyond the standard hydrogeology approach were used (Theis and Theis derivations) for diagnosis of the formation response and derivation of hydraulic parameters (Kruseman and de Ridder, 1990). First, in addition to plotting the pressure change in log–log coordinates for transient-data analysis, the derivative of the semilog pressure change (Bourdet et al., 1983) was also displayed (Fig. 30.5). Derivative data analysis is more sensitive than an analysis of simple changes in water levels with time and provides higher resolution for the flow model identification and improves reliability in the estimate of hydraulic properties. Second, the formation responses observed at the disposal wells and at observation points (Enachescu and Wozniewicz, 2001), best represented by the derivative data, were overlaid on a normalized plot (Fig. 30.6). This direct comparison allows a visual assessment of the degree of clogging that occurs only near a wellbore during the duration of testing. The magnitude of
ELSE_TSANG_CH030.qxd
426
8/22/2005
4:20 PM
Page 426
Case Study: Evaluation of Oil Field and Water-Well Disposal-Well Designs
Fig. 30.5. Data analysis using derivatives.
Fig. 30.6. Normalized plot.
clogging is reflected by a shift in the values of transmissivity for disposal wells and observation points. For instance, if minimal clogging occurs, the curves for disposal wells and observation points are expected to be consistent early in time. Finally, the normalized plots were used to directly compare disposal-well responses between tests to understand the evolving skin effects due to changing test conditions (temperature or nature of injection fluid).
ELSE_TSANG_CH030.qxd
8/22/2005
30.5 Test Results
4:20 PM
Page 427
427
30.5 TEST RESULTS The injection test results are described in order of scale, from well efficiency (at the well) to near wellbore (includes clogging, submeter), to medium scale (tens to hundreds of meters), and to large scale (encompassing boundaries, thousands of meters). The well efficiency was expected to be greater for the water-well design (WDW 9-28) than for the oil field design (WDW 1-21), due to the much larger open area of the wire-wound screen compared to perforations. The superior open area of the water-well design usually results in a lower entrance velocity and a smaller probability for turbulent flow. One method used to evaluate well efficiency was to plot the matched skin value versus the injection rate, normally expected to show a linear trend. However, in both wells, the skin values did not show the expected trend, likely due to the clogging effects attributed to several influences. Furthermore, since the clogging was thought to occur within the formation and since the size of the openings for the two completion types were considerably larger than the formation pore size, the difference in the size of the openings was not considered important. As the lower part of the McMurray Formation proved to be a suitable disposal zone in terms of aquifer disposal capacity early in the program, the clogging issue became significant in that it was considered to be a limiting factor in the operational disposal capacity. A close examination of the disposal-well data versus observation point data on a normalized plot showed inconsistent responses. The observation point response generally showed radial-flow regime with infinite lateral extent (Fig. 30.5). In comparison, the disposal-well derivative data often displayed “humps” that could be attributed to degassing effects (Fig. 30.6). The matched transmissivity value for the disposal-well data was typically lower than for the observation points data, by up to an order of magnitude, implying that the clogging effects were reducing the local-scale permeability and, therefore, masking the undisturbed formation response (Fig. 30.6). These potential degassing effects could be the result of the following mechanisms that are not further analyzed in this chapter: (1) a chemical incompatibility between injection water and in situ formation water, (2) abrupt changes in downhole temperature causing degassing of the injection and/or formation water, and (3) entrapment of air bubbles created by the early-time vacuum conditions due to the cascading of the injection water. In addition, physical clogging was also diagnosed in some of the tests based on the exponential increase in pressure versus time. Figure 30.7 illustrates the pressure response due to physical clogging that developed during the constant-rate injection test at WDW 9-28. In this example, a rapid increase in head buildup occurred during the initial part of the test when injection water contained larger amounts of suspended solids, but as the injection water cleared up over the duration of the test, the head buildup reaches a plateau before decreasing toward the end of the test. Finally, the majority of the clogging did not appear to be permanent, as redevelopment through pumping significantly improved the near-wellbore transmissivity of this well. Due to the masking effect of the local-scale clogging in the disposal wells, aquifer properties of the undisturbed formation were derived from the pressure response at the observation points that were spaced 80 and 2000 m from the disposal wells. Analysis of the data showed a range in transmissivity between 3 ⫻ 10–4 and 2 ⫻ 10⫺3 m2/s, which was generally consistent with the transmissivity values used for the calibrated regional numerical model. The variability in transmissivity values is primarily attributed to the variations in lateral thickness of the McMurray basal water sands aquifer. For storativity, the values ranged
ELSE_TSANG_CH030.qxd
428
8/22/2005
4:20 PM
Page 428
Case Study: Evaluation of Oil Field and Water-Well Disposal-Well Designs
Fig. 30.7. Physical clogging observed during constant-rate injection test at WDW 9-28.
between 1 ⫻ 10⫺4 and 1 ⫻ 10⫺3 m2/s. No boundaries were detected over the radii of influence of the various tests (thousands of meters). Finally, the pressure response observed at the near-field observation points varied significantly within the McMurray basal water sands unit (Fig. 30.8). The difference in pressure response was attributed to the relatively poor vertical communication between the sand intervals within the gross target interval, due to the presence of interbedded clayey and silty intervals and/or to the vertical anisotropy of the sands’ hydraulic conductivity. 30.6 CONCLUSIONS The results of the two campaigns of injection testing at the Long Lake pilot project have shown that (1) the lower part of the McMurray Formation is a suitable disposal zone in terms of aquifer capacity and (2) significant clogging, apparently unrelated to well design, occurs in the near wellbore. The relatively high sensitivity of the formation to clogging might be due to the finer-grained nature of the formation. This clogging seems to be the result of suspended solids in the injected fluids and degassing effects that could be due to fluid incompatibility, air entrapment, or fluid temperature changes. This clogging causes a reduction in aquifer transmissivity near the wellbore that may limit the capacity of the disposal wells. Practically, the disposal wells must be operated under positive pressures for a formation that would otherwise allow the required disposal volumes to be disposed of under vacuum conditions. The near-wellbore clogging also masks the well inefficiencies inherent in the well designs; therefore, the hydraulic performance of the oil field versus water-well designs could not be conclusively assessed. However, the water-well design, despite its higher capital costs, allows
ELSE_TSANG_CH030.qxd
8/22/2005
References
4:20 PM
Page 429
429
Fig. 30.8. Vertical anisotropy within McMurray basal water sands (BWS) observed at vibrating wire piezometer VWP 9-28.
for redevelopment during operations, which may be required periodically due to clogging problems. In contrast, the oilfield design fills in with sand when it is not pressurized, which requires costly coiled-tubing work to rectify. Both types of disposal wells will be operated during the pilot phase of the project to further assess long-term performance and maintenance costs prior to finalizing the disposalwell design used for the commercial project. Well performance will be continuously monitored by recording flow rates and pressures at the wellheads. Further testing will be conducted when the well performances depart from the baseline well performances obtained from the injection test results. In addition, the wastewater will be regularly characterized in terms of chemical composition and suspended solids, particularly hydrocarbons. This additional testing and long-term monitoring will enable Nexen/OPTI to further understand the clogging mechanisms and appropriate remedies.
REFERENCES Alberta Energy and Utilities Board (EUB), 1994. Injection and Disposal Wells: Well Classifications, Completion, Logging and Testing Requirements, Guide 51. Calgary, March 1994. Alberta Energy and Utilities Board (EUB), 2000a. Resources Applications for Conventional Oil and Gas Reservoirs, Guide 65. Calgary, June 2000. Alberta Energy and Utilities Board (EUB), 2000b. Energy Development Application Guide and Schedules, Guide 56. Calgary, October 2000.
ELSE_TSANG_CH030.qxd
430
8/22/2005
4:20 PM
Page 430
Case Study: Evaluation of Oil Field and Water-Well Disposal-Well Designs
Bourdet, D., Whittle, T.M., Douglas, A.A. and Pirard, Y.M., 1983. A new set of type curves simplifies well test analysis. World Oil, 5: 95–106. Enachescu, C. and Wozniewicz, J.V., 2001. Transmissivity normalized plots. Abstract submitted to the National Ground Water Association Annual Meeting—Ground Water Data: Collection, Reliability, Access, and Manipulation of Basic Data, December 7–8, 2001. Nashville, TN. Kruseman, G.P. and de Ridder, N.A., 1990. Analysis and Evaluation of Pumping Test Data, 2nd edn. (Completely revised), ILRI Publication 47. Wageningen (Netherlands).
ELSE_TSANG_CH031.qxd
8/22/2005
4:23 PM
Page 431
Chapter 31
FLUID INJECTION NEAR THE WASTE ISOLATION PILOT PLANT S. Ghose U.S. Environmental Protection Agency, Washington, DC, USA
31.1 INTRODUCTION The Waste Isolation Pilot Plant (WIPP) is a deep, mined geologic repository owned and operated by the Department of Energy (DOE), and certified by the Environmental Protection Agency (EPA) on May 18, 1998, to receive shipments of transuranic (TRU) radioactive waste for permanent disposal. The facility is located approximately 26 miles east of Carlsbad, New Mexico, on federal government-owned land. The Land Withdrawal Boundary around the repository, which will be marked by passive institutional control, contains approximately 10,240 acres of land in 16 sections of T.22S and R.31E. The controlled area defined by the EPA (40 CFR 191) is limited to the lithosphere and the surface within 5 km of the outer boundary of the WIPP waste emplacement panels. In this chapter, the “controlled area” is the area of interest. The underground repository is in the bedded salt at about 2150 ft below the surface. WIPP is a disposal system with protective mechanisms of both natural and engineered barriers. The repository encompasses an area of about 38 acres in one stratigraphic level with a slight (less than a degree) southward dip. This acreage contains an experimental area to the north, eight waste emplacement panels to the south, and an operation area with four vertical shafts and connecting drifts in between. It is estimated that approximately 6.2 ⫻ 106 ft3, or 40% of the excavated volume, will be filled with the waste (Fig. 31.1). The natural resources that are economically viable in the vicinity of WIPP typically use fluid injection as a tool for production enhancement and/or disposal of fluid, and are limited to potash and hydrocarbons. Griswold (NMMBR, 1995) proposed that out of 11 potash ore zones in the area, only the fourth and tenth contained high enough grades for commercial mining. His estimates for the recoverable reserves in the WIPP area are 126 million tons of langbeinite in the fourth ore zone, and approximately 105 million tons of sylvite in the tenth ore zone. The reserve estimates are primarily based on current market demand and extraction technology. These conditions may change in the future, and the estimated quantity also will change accordingly. Foster (1974) proposed that in the part of the Delaware Basin where WIPP is located, there are 15 potential oil-bearing horizons within a depth range of 4000 –14,000 ft. Foster’s per section estimates were scaled by Sandia National Laboratory (SNL) (SAND 78) for control zones, and for the 16-square-mile WIPP land withdrawal area. According to this, the area will yield approximately 20.4 million barrels and 265 billion cubic feet of natural gas, and will distillate 3.1 million barrels.
ELSE_TSANG_CH031.qxd
432
8/22/2005
4:23 PM
Page 432
Fluid Injection near the Waste Isolation Pilot Plant
Fig. 31.1. WIPP mine plan.
The area around WIPP has been reasonably active for oil and gas recovery in the past. There are approximately 56 oil and gas wells within a radius of 10 miles (SAND88-1452). The nearest well is about 3 miles to the south from the facility. These wells generally extend as much as 14,000 ft down to the Pennsylvanian Formation. The closest potash mine is approximately 3 miles from the facility. EPA regulation 40 CFR Part 191 (1985) requires that the disposal site should not be located in an area with recoverable natural resources unless it is proven that the favorable characteristics of the location compensate for the possibility of being disturbed for resource exploitation in the future. In addition, the EPA’s criteria for compliance, 40 CFR Parts 194.32 and 194.33, require that the performance assessment consider drilling that might affect the disposal system during the regulatory time frame (10,000 years). Through an extensive performance assessment that included a probabilistic treatment of fluid injection among other intrusion scenarios, the WIPP site demonstrated the ability to contain waste and meet the applicable criteria. The objective of this chapter is to discuss the following issues as they pertain to the area near WIPP: (1) why consider fluid injection, (2) the mechanics and practice of fluid injection, and (3) the rock mechanical response and potential environmental impact of fluid injection. 31.2 MECHANICS AND PRACTICE OF FLUID INJECTION IN THE AREA The mechanics and practice of fluid injection have a direct bearing on the mechanical response and environmental impact on the surrounding areas. Therefore, an adequate explanation of the conditions and background information related to the operations in the oil fields around WIPP are necessary. Fluid injection is a site-specific operation. The controlled area is located in a relatively younger oil field that is not well developed. The oil and gas activities are limited, and the secondary recovery process is less prominent. Since available data is very sparse, this study also includes a larger area that surrounds the controlled area but is still within the Delaware Basin. In general, oil and gas production is influenced by product demand, economics, geology, reservoir characteristics, the availability of fluid or a substitute product, and regulatory restrictions. The size of a secondary recovery operation is primarily influenced by the profit margin and reservoir characteristics.
ELSE_TSANG_CH031.qxd
8/22/2005
4:23 PM
Page 433
31.2 Mechanics and Practice of Fluid Injection in the Area
433
The pay zones in the controlled area are relatively small and are mostly stratigraphic traps. The Livingstone Ridge area, which is located partly in and adjacent to the controlled area, shows that the average production from a well is about 89,000 barrels and 116 million cubic feet of gas. The average cutoff limit for economic production is 150 barrels per month, and the minimum pay zone thickness is 40 ft (Broadhead, 1993, 1995). The average life of the primary recovery process is less than 10 years, and through the secondary recovery process, an almost similar time period can be obtained. The water cut in New Mexico oil is typically high, varying between 10 and 90%, with the majority over 50%. The average API grade of the oil is 52. For this study, fluid-injection operations are categorized as either production- or disposal-based operations. In a production-based operation, fluid injection is used to improve the depletion rate in both primary and secondary recovery phases. Solution mining is a specialized application of fluid injection in mineral extractions. Disposal-based operations include the brine that has been generated as a by-product of oil and gas production. 31.2.1 Production-based Operations Primary recovery Primary recovery is the period of oil and gas depletion from a single well when artificial energy is not needed to maintain the economic level of production. Prior to the completion of a well, several technical procedures are used to prepare the pay zone for the easy depletion of oil and gas. Hydraulic fracturing (hydrofracturing). Hydraulic fracturing, or hydrofracturing, is a standard operating procedure used to increase the recoverable oil and gas reserve by increasing the porosity and permeability of a pay zone. In this process, a specially blended fluid is pumped at a high rate, with enough pressure to initiate a fracture and to generate a fracture propagation. The main rock types for hydrofracturing in the WIPP area are sandstone, dolomite, and limestone. The mechanical properties and textural, compositional, structural, and thickness variations of these rocks affect fracture initiation pressure, fracture propagation, and fracture treatment procedures. Based on studies conducted by SNL, it has been established that the fracture initiation pressure for the interbeds in the area is close to the in situ stress. Although the geology and depth are different compared to the commercial hydrofracturing horizons, these estimates still offer an approximate idea of the failure criteria. The application of compressional pressure over the minimum in situ principal stress will create fracturing (tensile strength of the rock will play a role in this failure) and generate propagation. The failure of strata will equilibrate pressure. The fluid injection for hydrofracturing is a controlled operation, and pressure is regulated at a rate that is optimum for the necessary fracture design to maximize depletion of the pay zone. Fractures in these conditions typically propagate from the well (from the maximum pressure point), in two opposite directions on a vertical plane. Considering the reservoir characteristics in the area (described in the following section), mini-hydrofracturing (an injection rate of tens of barrels per minute) should be appropriate for this kind of operation. After fracture initiation, slurrymixed fluid is pumped to keep the fractures open. The flow of oil in primary recovery will be largely influenced by the tightness and initial permeability of the rock. Once the natural flow of oil and gas (during primary recovery) falls to or approaches an economic cutoff level, the secondary recovery process is introduced (if economics are favorable).
ELSE_TSANG_CH031.qxd
434
8/22/2005
4:23 PM
Page 434
Fluid Injection near the Waste Isolation Pilot Plant
Secondary recovery Secondary recovery is the second phase of oil and gas depletion from a producing reservoir. In this process, artificial energy (the medium of which can be air, water, or nonreactive gas) is applied through a borehole to drive oil and gas toward surrounding receiving boreholes. The depletion schedule in producing oil fields is a continuous operation, since interruptions or breaks in production are economically disruptive. In the oil fields surrounding WIPP, water is typically used as a medium for displacement. Pressure maintenance and waterflooding are two commonly used operations for secondary recovery in the area; from an operational viewpoint, these two are essentially the same, and are treated as such in this chapter. The in situ stress, pore pressure, and tensile and compressive strengths of the materials in the WIPP area are mostly within the favorable range for fluid injection. Past injection practices have been modified due to new regulatory restrictions. An approximate estimate of waterflood oil (oil displaced by waterflooding) is expected to be about $5 more than primary recovery hydrofracturing. This includes the cost of water, the cost of injecting water into the reservoir, and necessary chemicals. It is expected that some water will be recirculated from the primary production phase. At today’s rate of $28 for a barrel of crude oil, the profit margin is adequate. However, this will decrease significantly when the price of crude drops to $11 per barrel (the price in the late 1990s) and the incentive for secondary recovery reduces accordingly. The cost of waterflooding can vary with the increasing depth (bigger size pump), porosity, permeability, and water and gas saturation. However, due to the specific characteristics of the reservoirs (discussed in the next section) a large-scale waterflood is not anticipated in this area. Mechanics of the secondary recovery operation are site specific. The key features of the process, with special reference to the reservoirs of the area, are presented here. The practical aspect of the waterflood operation requires thorough planning for the right borehole pattern, injection rate, and injection pressure, and the shortest flow path to receiving wells. A pilot operation may be used to determine the most effective conditions for the recovery process. The New Mexico Oil Conservation Division (NMOCD) monitors these operations and provides specific requirement and guidance for each individual site. The injection rate and pressure, if not properly monitored or regulated, can disrupt the natural in situ stress equilibrium, create deformation, and disrupt and intermix with the natural groundwater. Waterflooding and pressure maintenance require at least two boreholes (for input and output) for operation in a producing field. However, larger fields, multiple producing horizons, and fields with existing natural barriers require more boreholes for economic optimization of an operation. It is in the economic interest of the producer to utilize the existing boreholes and existing patterns. The existing borehole patterns in the area exhibit a 40-acre spacing, which is a regulatory requirement, with mostly incomplete grids (or undeveloped fields), and are drilled in a linear orientation. A maximum of 16 boreholes is permitted in a section. Some of the sections have four boreholes, each toward the margin in north–south and east–west directions. This arrangement appears to be adequate for a linear pattern. Patterns and input boreholes are adjusted as the field matures. An optimum rate of fluid injection is determined by using characteristics such as pay zone thickness, horizontal and vertical permeability, oil viscosity, and well radius and pattern. However, it is important to maintain a balance between the production and fluid injection rates. Injection pressure is regulated by NMOCD: the maximum wellhead pressure recorded in the area ranges between 640 and 1613 psi. However, for economic reasons, this is also closely watched by the operator, as excessive pressure might initiate fractures and propagate outside the pay zone.
ELSE_TSANG_CH031.qxd
8/22/2005
4:23 PM
Page 435
31.2 Mechanics and Practice of Fluid Injection in the Area
435
31.2.2 Disposal-based Operations Saltwater disposal A considerable amount of oil is produced in the New Mexico portion of the Delaware Basin. Approximately 4.3 billion barrels and 19.7 trillion cubic feet of gas were produced up to 1996. In 1996, New Mexico produced 174,000 barrels of crude and condensate. It is evident that as the oil field matures, a portion of the liquid waste will be used in the recovery process; however, a majority (more than 50%) will be disposed in the nonproducing horizons at shallower depths. Therefore, demand for saltwater disposal by fluid injection will increase as the production increases. Within the area of interest, Class II Underground Injection Control permits authorize the injection of the brine produced as a by-product from oil production into the Bell Canyon and or deeper formations. Approximately 5000 permits have already been issued so far in Lea County, New Mexico alone. The injection wells in the area are about 5 years old, on average, and range in depth between 3820 and 8344 ft (Bell Canyon Formation). The injection rate is determined by the operator according to the needs, operation costs, bearing capacity of the formation, and the condition of the well. The overall injection rate varies between 5 and 3000 barrels per day, with the majority between 500 and 2000 barrels per day. Injection pressure is the primary concern in any fluid injection operation. The state of New Mexico permits, monitors, and restricts the overall pressure used in an operation. The risk of fluid traveling to an unpermitted horizon is greatly enhanced if the pressure is not regulated. Typically, an operator’s economic interest would not suffer if fluid had moved to a new horizon. NMOCD recommends 0.2 psi for each linear foot of well depth to the top of the fluid injected zone. Moreover, the Division prescribes the maximum injection pressure, which is based on the geology, groundwater, and producing wells in the vicinity, for each injection site. NMOCD also requires annual testing of the tubing and packer (Bradenhead test), and a mechanical integrity test (MIT) every fifth year of disposal operation. Solution mining Solution mining is an in situ mechanical dissolution of an ore body and its recovery on the surface for commercial use. This is a specialized application of fluid injection. In this process of extraction, fluid is injected through a borehole to the subsurface mineralized bed, and the dissolved mineral in the fluid is pumped back to the surface through another, or sometimes the same, borehole. Injection of warm fluid is preferred to increase the dissolution rate. Both input and output boreholes are located on the same horizontal plane, at a predetermined distance from each other. The input and output boreholes are connected by a fracture, which is created by a hydraulic fracturing process. This is a site-specific operation and is controlled by geology, the depth to the ore body, thickness, ore grade, economics, and regulations. The rate of injection depends on the size of the operation, depth, and solubility. The maximum pressure of injection is regulated by the state and is used in every operation of fluid injection. In the area surrounding WIPP, there are several commercially extractable potash ore zones ranging between 885 and 1400 ft from the surface in the McNutt potash ore zone of the Salado Formation. At the present time, there is no active solution mining operation in the area. Solution mining creates uneven and irregular underground cavities that remain unsupported in an imbalanced stress area. This can lead to subsidence. The mode, magnitude, timing, extent, and mechanics of subsidence are mainly dependent on the geology, rate of dissolution, dimension, and geometry of the cavity.
ELSE_TSANG_CH031.qxd
436
8/22/2005
4:23 PM
Page 436
Fluid Injection near the Waste Isolation Pilot Plant
31.3 GEOLOGY AND RESERVOIR CHARACTERISTICS WIPP is located in the northern part of the Delaware Basin. The geological history of the area is typically divided into three different categories that are broadly equivalent to three geologic eras. The first category (Paleozoic era) is dominated by nonhydrocarbonbearing crystalline rocks. The second category (Mesozoic era) is a period of marine submergence, basin subsidence, structural deformation, reef development, and evaporite formation. This has the maximum potential and was favorable to hydrocarbon formation and accumulation (trap). The third phase (Cenozoic era) is relatively more stable and marked by continental environments. Dissolution, erosion, and continental types of depositions are the predominant processes. One of the objectives of this chapter is to discuss the impact of fluid injection on the environment around WIPP. The intensity and magnitude of the effects depend on the reservoir size, and the rate and pressure of fluid displacements. A description of potential oil and gas plays will help to understand the nature of future oil production activities expected to occur in the area of interest. Among the various formations (at the present time), the Delaware Mountain Group, located approximately 2000 ft below WIPP, is the primary target for depletion. A list and brief description of representative sample plays from potential oil-bearing horizons are provided below, along with a short summary of the geological deposition sequence ● Delaware Mountain Group. The sandstones of the Brushy Canyon, Cherry Canyon, and Bell Canyon Formations exhibit the strongest indications of potential plays. In general, the thickness of individual pay zones varies between 15 and 40 ft. The sandstones have 14–25% porosity; 9.6 ⫻ 10 16 to 4.8 ⫻ 10⫺14 impermeability; 35–45 API oil gravity; 7–26% recovery efficiency; and 35–65% water saturation, with an initial pressure of 9.8–24.9 MPa. ● Bone Spring Carbonates. The pay zones are 20–30 ft thick and 5480–9720 deep, and have an initial pressure of 20–28.7 MPa and a temperature of 128–180°F. ● Strawn Group. The carbonate reservoirs can be 10–50 ft thick or substantially more. Porosity varies between 2 and 9%; permeability, 2.1 ⫻ 10⫺15 to 2.0 ⫻ 10⫺13 m2; temperature, 112–163°F; and initial pressure, 11.9–28.5 MPa. ● Atoka Sandstone. The plays in this formation are deeper, and vary between 8500 and 14,000 ft. These can be overpressured reservoirs, predominantly containing gas. Porosity is between 1 and 17%; permeability, 9.9 ⫻ 10⫺15 to 2.2 ⫻ 10⫺14 m2; gravity, 59 API; and initial pressure, 21–73 MPa. ● The Wolfcampian Carbonate Play. The limestone reservoirs are noted for oil and gas productions. The plays are 8000–13,000 ft deep, with 5–10% porosity, 1⫻10⫺15 to 1.2 ⫻ 10⫺13 m2 permeability, 130–193°F temperature, and 20–77 MPa initial pressure.
31.4 MECHANICAL RESPONSE AND ENVIRONMENTAL EFFECTS OF FLUID INJECTION The site-specific prediction of the environmental effects and rock deformation due to fluid injection is difficult. This requires a large database to interpret the natural anisotropy of WIPP-area rock characteristics. The area has plastic-like halite beds, and brittle units such as anhydrite beds associated with a variety of silicate and carbonate rocks. Each unit individually and also in combination will react to applied stress differently. Since the present area of interest is very close to the TRU waste repository site, a primary concern is to maintain its integrity to isolate the waste.
ELSE_TSANG_CH031.qxd
8/22/2005
4:23 PM
Page 437
31.4 Mechanical Response and Environmental Effects of Fluid Injection
437
To create a disruption in the natural environment and affect the integrity of WIPP would require an energy source to create pathways for rapid movement of injected fluid or gas toward the target. Fluid injection provides a potential energy source. All operative procedures related to fluid injection, including drilling (if necessary), well completion, hydrofracturing, acid treatment, waterflooding, pressure maintenance, plugging activities, and disposal of liquid waste (brine), introduce fluid to the subsurface under varying degrees of pressure. Among these operations, disposal of liquid waste is the most likely mechanism capable of creating damage, because in this type of operation, some operators might try to complete the assignments in the shortest possible time with higher injection rates, and try to use the same well for longer periods of time to optimize costs, without considering possible damages. Large-scale solution-mining operations that are not properly designed and managed can also damage the environment. These two operations can initiate and propagate fractures, which can be pathways for fluid movement. Fractures are the most visible response of the rock deformations subjected to pressure by fluid injection. In order to create noticeable disruption, fractures should be able to propagate and collate outside the point of origin. Fracture conductivity, density, direction, and distance of propagation are important properties for fluid movements. Certain rocks in the area, such as limestone and other brittle carbonates, might favor the propagation of energy introduced through the injection, and other rocks such as shales and clay bands might hinder the propagation. In situ stress difference, impermeable layers, and variable material properties can also resist the propagation and can terminate it if the energy source is not adequate. It is also possible that closure of the injection operation might not terminate the fracture propagation; fluid might continue to move through the new or pre-existing fractures until it is interrupted by adverse conditions. SNL experiments (Beauheim et al., 1993) in the Salado Formation (repository) showed that significant variations exist in fracture initiation and extension pressures, in two test holes, at 25 m apart. The Linear Elastic Fracture Mechanics (LEFM) model has been used in the WIPP performance analysis and by interested stakeholders to predict the behavior of hydrofractures in salt and anhydrite. However, the conclusions are not free from controversy. In 1991, a saltwater blowout was encountered in the “Rhodes–Yates Field,” which is located approximately 40 miles southeast of WIPP. The blowout took place while drilling through the Salado Formation, at 2281 ft. The source of this brine was decided, through litigation, to be injected fluid from waterflooding in the immediate vicinity. The pathway of fluid movement was suspected to be the anhydrite beds. A similar encounter was also reported in the “Vacuum Field” 32 miles northeast of WIPP. Based on modeling work related to the WIPP performance assessment for certification and verification by EPA, geological barriers at the WIPP site, frequency and location of fluid injection, improved technology of drilling and monitoring, and the New Mexico State regulations, EPA decided that the fluid injection is of low consequence for certification; however, the monitoring of activities near to the site is ongoing and will continue. Review of all these components is beyond the scope of this study. Instead, a brief description of the mechanics of fluid movement is presented. In order to affect the integrity of WIPP, a large-scale movement of fluid under pressure would be required. For example, injected fluid in the Bell Canyon Formation (widely used and comparatively shallower) has to travel approximately 3000 ft vertically and 1.5 miles horizontally to reach the WIPP repository. In a controlled operation, a vertical extension of the pathway to outside the target zone is unlikely. At the depth of Bell Canyon, only vertical fractures are feasible. The fractures will develop in the plane normal to the plane of the least principal stress (Hubbert and Willis, 1957). The lithostatic stress is in the maximum principal
ELSE_TSANG_CH031.qxd
8/22/2005
438
4:23 PM
Page 438
Fluid Injection near the Waste Isolation Pilot Plant
stress direction, and the fractures will follow the direction of least resistance. A vertical pathway would be needed to bring the fluid up to the level of the repository and then transmit horizontally to WIPP. It has been suggested that a mechanical failure in the casing, cement, or plugging can bring the fluid outside the drill casing and into contact with the permeable horizons. Marker Bed 139 and other similar horizons can act as conduits for fluid movement. These are brittle formations that develop fractures due to differential strength with respect to the host formation. Field tests conducted by SNL (Beauheim et al., 1993) showed that fracturing took place when the injected fluid pressure reached close to the in situ stress, which is typically less than the lithostatic stress. The primary breakdown pressure was 11.6 MPa. The fracture pattern in general was close to the pre-existing fracture in Master Bed 139. The fractures did not extend for long distances; however, the storage capacity and permeability were substantially increased. The operations in the WIPP area are typically small and are not capable of introducing a substantial amount of energy to the subsurface. In order to affect the integrity of WIPP, vertical and horizontal pathways would need to remain open and connected. Fluid pressure close to in situ stress has to be generated for fluid movement in Marker Bed 139 or similar lithological units. In addition, the fracture flow path should be able to overcome the adverse effects of variations in thickness, mineral fillings, and creep closure on the surrounding halite; however, an unlikely combination of the following events and processes can increase risks from fluid-injection leakage. For example, consider the possibility of injection-well leakage going on for a very long period of time without being detected, with a significant amount and high rate of fluid flow exceeding regulatory restrictions. The fluid would move up into thief zones without losing force and connect to a marker bed with long, open, and continuous fractures, when repository pressure is less; however, the probability that all these factors would come into play would be very low. There are five possible aquifers in the area: the Capitan, the Culebra, the Magenta, the Dewey Lake, and the Santa Rosa; the groundwater production data from these geologic units are unknown, as these are primarily used to supply water for livestock. In an unlikely combination of events discussed above, the water from these aquifers could be affected by mixing with other aquifers and oilfield brine, which would reduce the utility until corrected. Water level in the wells could also rise due to fluid flow, and water temperature might increase if water is mixed with the fluid used in solution mining. Based on these factors, it can be concluded that the effect of fluid injection on the environment and WIPP is of low consequence.
REFERENCES Anderson, R.Y. and Powers, D.W., 1978. Salt Anticlines in Castile-Salado Evaporite Sequence, Northern Delaware Basin, New Mexico, New Mexico Bureau of Mines and Mineral Resources, Circular 159. Bredehoeft, J., 1997. The Hartman scenario: Implication for WIPP. Report prepared for the N.M. Attorney General, Copy on File at Sandia National Laboratory WIPP Files, WPO #45839. Beauheim, R.L., Wawersik, W.R. and Roberts, R.M., 1993. Coupled permeability and hydrofracture tests to assess the waste-containment properties of fractured anhydrite. J. Rock Mech., 30(7): 1159–1163. Broadhead, R.F. and Speer, S.W., 1995. Oil and gas in the New Mexico part of the Permian Basin. Rosewell Geological Society Symposium of Oil and Gas Fields of Southeastern New Mexico.
ELSE_TSANG_CH031.qxd
References
8/22/2005
4:23 PM
Page 439
439
Broadhead, R.F. and Speer, S.W., 1993. Oil and gas in the New Mexico part of the Permian Basin. New Mexico Geological Society Guidebook. 44th Field Conference, Carlsbad Region, New Mexico 7 West Texas, October 6–9, pp. 293–300. Brown, J., 1995. Cabin Lake field study. Rosewell Geological Society Symposium of Oil and Gas Fields of Southeastern New Mexico. Cheeseman, R.J., 1978. Geology and Oil/Potash Resources of Delaware Basin, Eddy and Lee Counties, New Mexico, New Mexico Bureau of Mines and Mineral Resources, Circular 159, Docket: A-93-02, Reference #18. David, E.K., 1977. Springs Upper Pennsylvanian Gas. Symposium of the Oil and Gas Fields of Southeastern New Mexico, Supplement, Rosewell Geological Society. Dean, W.E. and Anderson, R.Y., 1978. Salinity Cycles: Evidence of Subaqueous Deposition of Castile Formation and Lower part of Salado Formation, Delaware Basin, Texas, and New Mexico, New Mexico Bureau of Mines and Mineral Resources, Circular 159. Foster, R.H., 1974. Oil and gas potential of a proposed site for the disposal high-level, radioactive waste. Open File Report, Contract No. Af(40-1)-4423, Oak Ridge National Laboratory. Grant, P.R. and Foster, R.W., 1989. Future Petroleum Provinces in New MexicoDiscovering New Reserves, New Mexico Bureau of Mines and Mineral Resources. Griswold, G.B., 1977. Site Selection and Evaluation Studies of the WIPP, Los Medanos, Eddy County, NM, Sandia National Laboratories, Albuquerque, SAND77-0946. Griswold, G.B., 1995 (New Mexico Bureau of Mines and Mineral Resources), Method of potash reserve evaluation, In: Evaluation of Mineral Resources at the Waste Isolation Pilot Plant (WIPP) Site, Vol. 2, Chapter VII. Carlsbad, NM: Westinghouse Electric Corporation, Waste Isolation Division. Hoose, G. and Dillman, G., 1995. Sand Dunes Delaware, W; Sand Dunes Delaware S. Rosewell Geological Symposium of Oil and Gas Field of Southeastern NM. Hubbert, M.K. and Willis, D.G., 1957. Mechanics of hydraulic fracturing. Am. Inst. Mech. Eng., 210. James, A.D., 1985. Producing characteristics and depositional environments of Lower Pennsylvanian Reservoirs, Parkway—Empire South Area, Eddy County, NM. Am. Assoc. Petroleum Geologists, Bull. V, 69(7): 1043–1063. Larson, K. and Fewell, M., 1997. Development of the interbed fracturing model implemented in BRAGFLO and its parameter values used in the 1996 CCA performance assessment. Memo to Margaret Chu, March 12. May, B., 1995. Lost Tank Delaware, field summary. Rosewell Geological Society, Symposium of Oil and Gas Fields of Southeastern NM. May, B., 1995. Livingstone Ridge Delaware. Rosewell Geological Society, Symposium of Oil and Gas Fields of Southeastern NM. Mendenhall, F.T. and Gerstle, W., 1993. WIPP anhydrite fracture modeling. Unpublished Report, Sandia National Laboratory. OGCI, 1996. Short course, Smith, Tulsa, OK. Silva, M.K., 1996. Fluid injection for saltwater disposal and enhanced oil recovery as a potential problem for the WIPP. Proceedings of June 1995 Workshop and Analysis, EEG-62. Stoelzel, D.M. and O’Brien, D.G., 1997. The Effects of Saltwater Disposal and Waterflooding on WIPP, WPO #40837, Sandia National Laboratory, Docket No. A-9302, Ref. #661. Stoelzel, D.M. and Swift, P.N., 1997. Supplementary Analyses of the Effects of Saltwater Disposal and Waterflooding on WIPP, WPO#44158, Sandia National Laboratory.
ELSE_TSANG_CH031.qxd
440
8/22/2005
4:23 PM
Page 440
Fluid Injection near the Waste Isolation Pilot Plant
Thomerson, M.D. and Catalano, L.E., 1995. Reservoir characteristics and petrophysical analysis of the upper brushy canyon sandstones, East Livingstone Ridge Delaware field, Lea County, NM. Roswell Geological Society Symposium of Oil and Gas Fields of Southeastern NM. U.S. Department of Energy, 1996. WIPP Compliance Certification Report, Docket: A-9302, II-G-1. Warpinski, N.R. and Hansen, F.D., 1998. Hydraulic fracturing analyses applied to WIPP. Memo to Melvin Marietta. February 10.
ELSE_TSANG_Ch032.qxd
8/23/2005
12:29 PM
Page 441
Chapter 32
INJECTION OF ORGANIC LIQUID WASTE IN A BASALTIC CONFINED COASTAL AQUIFER, REUNION ISLAND J.-S. Martial, J.-L. Join, and J. Coudray Laboratoire des Sciences de la Terre, Université de La Réunion, Saint-Denis, Ile de La Réunion, France
32.1 INTRODUCTION Effluents from distilleries are significant and require manufacturers to own their own waste treatment facilities. Every year, the Savanna Distillery, in Bois Rouge, on the Eastern Reunion Island in the Indian Ocean, produces about 80,000 m3 of liquid wastes. The organic load of the effluent is noteworthy. One liter of alcohol generates about 15 L of liquid wastes (Poggi-Varaldo, 1992). Dissolved organic carbon (DOC) is the parameter used to quantify organic waste concentration (Leenheer et al., 1976). A statistical study carried out in India on distilleries shows (Rao and Viraraghavan, 1985) that the average DOC is around 100,000 mg/L, and that the total nitrogen is on an average 1100 mg/L. In most cases, rum producers discharge the rum effluent into seawater. Rum distilleries, mainly located in the Caribbean islands, are frequently located near the shore. Thus, ocean disposal is the most common way of dealing with liquid wastes. Studies show (EPA, 1977) that since rum effluent contains high levels of organic carbon compounds, it lowers the dissolved oxygen level in the receiving waters, resulting in anoxic conditions. Besides ocean disposal, other technologies exist, but they all have limitations. Land application can be difficult to put into practice because of the significant runoff that occurs on recently formed volcanic islands such as Reunion (Join, 1991). This is essentially caused by strong topographic slopes associated with tropical rainfalls (over 2000 mm/yr). Moreover, the cost of transportation to a designated site often winds up being a substantial indirect investment. Biological treatment of high-strength distillery slops has shown to be reliable and efficient, reaching 70% DOC removal and 8 m3/day biogas productivity. Nevertheless, the final effluent does not usually meet the criteria that would allow for direct discharge into water bodies, and thus requires further treatment. Moreover, we have to take into account the sludge production that lengthens the treatment process. The economic feasibility of such a complex process is questionable for a small industry such as a rum distillery (Poggi-Varaldo, 1992). The process of underground disposal of fluids by means of injection wells is used extensively in many of today’s industries. These industries include petroleum, chemical, food, and product manufacturing; geothermal energy development; and many small specialty plants and retail businesses. Underground disposal by injection has been advocated (Cleary and Warner, 1970) as a more convenient and economic way of dealing with the pollution issue than other alternatives. Nevertheless, injection requires specific geological and hydrological conditions for it to be done properly. Microbial degradation of organic waste occurs under anaerobic (reducing) conditions within the injection aquifer, producing a large amount of gas. Methane and sulfides resulting
ELSE_TSANG_Ch032.qxd
442
8/23/2005
12:29 PM
Page 442
Injection of Organic Liquid Waste in a Basaltic Confined Coastal Aquifer, Reunion Island
from sulfate reduction are present in very high concentrations. Note that degradation is the highest on the periphery of the plume, where the effluent is diluted with native water (Ditommaso and Elkan, 1976), and that undiluted waste may inhibit bacterial growth. The iron content of the water is also a limiting factor in microbial degradation. 32.2 THE CASE OF THE SAVANNA DISTILLERY For 10 years, liquid waste from the Savanna Distillery was injected into a deep well, beginning in November 1992, when a borehole was drilled 30 m from the shore. In the case of this distillery, our purpose is to constrain the hydrogeological setting of the aquifer after these 10 years of injection, which should then help us to determine how different factors (rate of dilution, well plugging, gas production) explain both the current and future evolution of the waste management methods used. The injection zone at the distillery is a confined, brackish, 100 m-deep aquifer that discharges into the ocean 2 km downstream. Injection flow rate is between 15 and 25 m3/hour (66–110 gal/min). Injection pressure has increased since the very beginning of production in 1993. During this first year, injection ran without requiring any overpressure. The following years showed a quite rapid increase in injection pressure, imprecisely recorded. Injection pressure reached 3 bars in less than 3 years. Some adjustments were made to limit the rise of pressure. Prior to injection, the liquid waste goes through a centrifugal machine, and afterwards stays in a settling basin for degassing. The injection pressure remains around 4 bar (60 psi) until the end of a production season. At the Savanna Distillery, the DOC of injected wastes ranges from 80,000 to 100,000 mg/L; the effluent is acidic (mean pH ⫽ 4.75), dense (1.03–1.04), and hot (30–40°C), and contains a high volume of total suspended solids ((TSS) – around 2000 mg/L) (see Fig. 32.1). Ionic analysis of the injected effluent also shows a significant amount of K⫹ to be linked with the organic matter.
32.3 THE HYDROGEOLOGICAL SETTING OF BOIS ROUGE—A RARE ASSET The Savanna Distillery is located on the eastern side of Reunion Island, in the Indian Ocean. The area of Bois Rouge corresponds to the extreme north of the Rivière du Mât alluvial plain (Fig. 32.2). There, the sedimentary contribution of Rivière du Mât is 100 m thick, with the upper 20 m of this sedimentary deposit made of very coarse alluvium (metric boulders) and the remaining 80 m consisting of sandy clay loam (Fig. 32.3). The sedimentary pile overlays the late basaltic lava flows from the Piton des Neiges shield volcano (Billard, 1977). Two aquifers can be found in the region (Fig. 32.4). The first aquifer is shallow and unconfined, and runs through the coarse alluvium. The second one runs under the low-permeability sediments and through the volcanic medium (basaltic fissured lava flows and scoria). This second, confined aquifer contained brackish water prior to the first injections.
32.4 EXPERIMENTS AND RESULTS In the 2000 production season, the Laboratoire des Sciences de la Terre de l’Université de la Réunion (LSTUR) began a hydrological survey of boreholes drilled in
ELSE_TSANG_Ch032.qxd
8/23/2005
12:29 PM
Page 443
Fig. 32.1. Characteristics of injected effluent.
Fig. 32.2. Simplified geological map and wells location.
ELSE_TSANG_Ch032.qxd
444
8/23/2005
12:29 PM
Page 444
Injection of Organic Liquid Waste in a Basaltic Confined Coastal Aquifer, Reunion Island
Fig. 32.3. Geological setting of the Bois Rouge area.
Fig. 32.4. Hydrogeological cross section at Bois Rouge: (1) very coarse alluviums; (2) sandy clay loam; (3) Piton des Neiges lava flow series.
the studied area. The superficial aquifer is monitored by means of eight observation wells, while the deep, confined aquifer receiving the injectate is monitored by means of two observation wells (see Fig. 32.4). Whereas Mauroux and Barrera (1992) have conducted a predictive study, we aim, in our survey, to appraise the suitability of this kind of waste management for volcanic coastal aquifers. Liquid waste injection is atypical in France—indeed, since 1998, French legislation concerning groundwater has prohibited any wastewater injection, except in the case of the mining industry. Experimental data
ELSE_TSANG_Ch032.qxd
8/23/2005
12:29 PM
Page 445
32.4 Experiments and Results
445
obtained through this survey will allow us to determine this method’s relevance in terms of environmental management. 32.4.1 Hydrodynamic Characterization of the Deep Confined Aquifer We conducted a 64-hour pump test, at a flow rate of 20 m3/hour, in the lateral deep observation well. Because of the borehole’s size, we could not pump with a greater discharge. The apparent transmissivity deduced from the interpretation of the drawdown, strongly affected by tidal variations, was 1 ⫻ 10⫺2 m2/s (see Table 32.1). This value is comparable to the average transmissivity (Martial et al., 2000) found in boreholes drilled in volcanic media on Reunion Island. Thus, we concluded that hydrodynamic conditions had been preserved 137 m east of the injection site, despite 10 years of injection. Because the discharge rate was not fast enough to provide significant variation at the neighboring wells, the storage capacity of the basaltic aquifer was tested by means of a tidal fluctuation analysis. The tide was monitored in the port of Sainte-Marie, located 9 km west of the injection site. The water level recorded at two observation wells showed a damping of 24.2% for the well closest to the shore and 26.7% for the upstream well. The storage capacity was calculated as 1 ⫻ 10⫺6. The distance to the discharge zone calculated from the damping analysis (2 km) is consistent with the geology, and the confinement of the lower aquifer was confirmed. Moreover, note that the water level is 4 m over sea level at 35 m from the shore. The average hydraulic gradient calculated between the observation wells was 0.17% for the deep aquifer, and was always different from the one measured for the upper aquifer. For preliminary analysis of the apparent transmissivity around the well, the pressure increase measured at the injection well during the production period was interpreted inversely, as if for a long-term pump test, Our analysis showed that the apparent transmissivity of the host formation in the vicinity of the injection well decreased from 1 ⫻ 10⫺2 to 1 ⫻ 10⫺4 m2/s. This evolution of the host formation is caused by the gradual plugging of the aquifer. The extensive amount of total suspended solids (Fig. 32.5) is chiefly responsible for the decrease in rock porosity. Moreover we have to take into account the consequences of gas production (essentially methane and sulfide) due to microbial activity. However, the different adjustments made at the Savanna Distillery (centrifugation and degassing of the effluent prior to injection) have helped to create a steady state with regard to injection pressure over the last 4 years. Assuming that the regional geological setting provides optimal initial conditions for underground disposal of organic effluents, the durability of such waste management methods is critically dependent on these adjustments (Saripalli et al., 2000). Table 32.1. Summary of hydrodynamic parameters Layer 1
Layer 2
Layer 3
Materials
Coarse alluvium
Sand; silt; clay
Thickness Aquifer type Transmissivity Downstream boundary Upstream boundary
20 m Unconfined 1.2 ⫻ 10⫺2 Ocean Very variable
80 m Unconfined ? Ocean Unknown
Fractured lava flows and porous scoria Screened on 30 m Confined (S ⫽ 1 ⫻ 10⫺6) 1 ⫻ 10⫺2 Discharge in ocean at 2 km offshore Gradient: i ⫽ 0.17%
ELSE_TSANG_Ch032.qxd
446
8/23/2005
12:29 PM
Page 446
Injection of Organic Liquid Waste in a Basaltic Confined Coastal Aquifer, Reunion Island
Fig. 32.5. Chemical evolution of the injected effluent during the nonproduction period of the year 2001.
32.4.2 Waste Migration Investigations To verify whether the effluent migrates upstream, similar to the observations by Kaufman and McKenzie (1975) concerning the Floridan aquifer, we monitored Observation Well #1, located 350 m upstream. DOC analysis, started in 1993, found ranges between 0 and 10 mg/L. The electric conductivity, recorded weekly, was 231.5 µS/cm ⫾ 5.6. Over the 2year period of this survey, the physical and chemical properties of the site displayed the characteristics of uncontaminated native groundwater. We also tested the lateral expansion of the liquid plume. Major ionic analysis of the collected samples during the long-term pump test performed at the lateral observation well (Table 32.2) indicated that the injected effluent did not significantly affect this observation well. The potassium concentration, which could be evidence of waste contamination, remained very low throughout the pump test (11 mg/L for F#2 versus 7000 mg/L for the injection well). On the other hand, sodium and chloride are the prevailing ions, and show typical brackish water content. Thus, these results would suggest that the lateral spread of the effluent plume is strongly confined. With regard to the downstream discharge of the aquifer into the ocean, Bigot (2000) recently conducted an ecological survey based on the evolution of the benthic fauna since 1994, to monitor the arrival of the effluent plume. The sediment is collected by means of a Van Veen bucket at 11 geographically referenced points, at depths ranging from 20 to 200 m. This survey showed that after 7 years, at a depth of 120 m, a significant increase in animal biomass occurred, while biodiversity was maintained. From this, we could conclude that the deep biological stocking structure has been upgraded. These ecological results were interpreted as indirect evidence of the offshore arrival of waste. This waste is to be considered as strongly diluted and degraded by anaerobic microorganisms—otherwise, the consequences to the marine biotops would have been more serious. Nevertheless, because we are
ELSE_TSANG_Ch032.qxd
8/23/2005
12:29 PM
Page 447
32.5 Discussion
447
Table 32.2. Major ionic composition (in meq.) of injected effluent and observation wells
Injection well (Inj) Upstream Obs. well (#1) Lateral Obs. well (#2)
Na⫹
K⫹
Mg2⫹
Ca2⫹
F⫺
Cl⫺
SO42⫺
NO3⫺
HCO3⫺
37.6
207
78.1
69.6
35
119
30.7
0.568
196
1.01 15.3
0.05
0.47
0.61
0.006
0.74
0.83
0.007
1.43
0.25
4.021
1.81
0.031
16.12
1.4
0.02
1.52
not able to sample the discharging effluent in the ocean, we cannot estimate how much the effluent has been diluted and/or degraded by microbiological activity. During the nonproduction period of 2001, we began to study how the major ionic chemistry of samples collected at the injection well had evolved. Samples were collected after 10 min of discharge (at a rate of about 1 l/s) from the injection well, which was under pressure. From December 6, 2000, to July 12, 2001, 40 samples were collected. The last 12 samples were collected according to the same experimental protocol, but following the collection of the sample, instead of closing the injection well valve, the manufacturer opened it slightly, to release the gas and thus reduce the pressure within the injection well before the next production season began. The samples collected during the first period maintained relative chemical stability. On the other hand, the last 12 samples that were followed by 24 hours degassing showed a very significant decrease in K⫹, SO42⫺, and F⫺, while Na⫹ increased. During these 12 degassing operations, liquid effluents were also released with the gas. Since we were able to record the degassing discharge only for the last six samples, we were not able to establish an actual dilution trend. Nevertheless, the farther the sampling from the injection well, the greater the dilution. We believe that, close to the injection well, the dilution and biodegradation of the effluent was very limited, because the organic contents remained constant. This finding may mean that, in the vicinity of the injection well, the clogging due to the suspended solids associated with the liquid waste significantly reduced the dilution of the effluent by the native water that circumvents this lower permeability zone, and that the microbial degradation starts when the effluent is diluted enough.
32.5 DISCUSSION Because of the scarcity of extensive impervious layers, confined aquifers are not common in recently formed volcanic islands. The coastal volcano-sedimentary formations induce the confinement of underlying volcanic aquifers (Join, 1991). Only the alluvial fans located at the outlets of the major streams can provide an adequate quantity of fine sediments to generate a regional impermeable layer. This is particularly true where alluvial accretion is made of very fine elements, in the very distant part of the alluvial fan, such as at Bois Rouge. The upstream observation well is, for instance, the only artesian well on the island. Because the Bois Rouge area is located on the eastern side of the island, the prevailing wet trade winds (alizées) provide 3 m of annual rainfall on the coast (and more inland), which causes a significant recharge. Martial et al. (2000) demonstrate statistically that hydraulic gradients were three times greater on the windy side of Reunion Island than on the dry western side.
ELSE_TSANG_Ch032.qxd
448
8/23/2005
12:29 PM
Page 448
Injection of Organic Liquid Waste in a Basaltic Confined Coastal Aquifer, Reunion Island
The initial brackish properties of the lower aquifer of Bois Rouge make it unsuitable for drinking purposes. The highly confined aquifer, associated with sizable recharge and high transmissivity, are among the many assets that contribute to considering the Bois Rouge area as a favorable site for liquid waste injection. In addition, the 2 km distance to the ocean discharge lengthens the travel time and thus the bacterial degradation of organic wastes. However, the transposition of such environmentally safe waste management facilities to another location on Reunion Island would be rather risky, since the confinement conditions in the Bois Rouge area do not exist elsewhere on the island. Conversely, other islands may easily meet these conditions. For instance, in Hawaii, Voss and Souza (1987) have described how low-permeability sedimentary caprock overlays pervious lava flows along some of the coastal margins. The most frequent aquifer pattern found in recently formed volcanic islands is that of an unconfined, highly transmissive, basaltic aquifer. In such a geological context, it is hard to imagine how injection of such an organic load could not be harmful to the groundwater resources and to the coastal biological environment. Nevertheless, on Mauritius Island, tracer tests were performed in 1999 to develop this kind of solution for domestic waste management of Port-Louis. The other kind of hydrologic structure that can be found in recent volcanic islands is the “volcano-detritic aquifers,” located at the outlet of a very active erosion area (called “cirques” at Reunion Island). In these cases, the aquifer has a variable, semi-confined character. According to the grain size of the volcanic sediments, the groundwater bodies are more or less confined. Thus, the relatively impervious layers present a lateral extent that in most cases may not be enough to ensure that the injected waste would not contaminate the upper aquifers. The last coastal aquifer type we encounter at Reunion Island, and in most tropical islands, is that of groundwater bodies located beneath the lagoon and its coral reefs. Join (1988) demonstrates that the St. Gilles volcanic aquifer is confined, and that consequently we could imagine that underground injection is a possible solution for this area, even with the numerous environmental issues owing to rapid growth. We have to test the ability of the groundwater to discharge in the ocean because, in contrast to the eastern side of the island, the recharge may not be sufficient. These different hydrologic patterns will be modeled in the next part of our study, to establish the conditions that ensure a suitable operation of an underground injection system, in light of the environmental issues surrounding such a system.
32.6 CONCLUSION Regarding the deep-well injection of liquid waste by the Savanna Distillery that took place from 1992 to 2001, lateral and upstream surveys show no evidence of any contamination coming from the injection well. Our recent marine ecological survey describes a slight and progressive enrichment of the sediment at a depth of 120 m. The survey data are consistent with the hypothesis of a slow migration toward the outlet of a confined aquifer, 2 km from the shore, where the effluent that reaches the ocean has been diluted by surrounding native water and degraded by means of underground microbial activity. The long-term evolution of this waste management method remains unknown. The adjustments made prior to injection seem to contribute to slowing down the decline in performance of the concerned facilities (because they tend to prevent a significant increase in the plugging of the host formation) and to limiting the injection of gases within the liquid wastes.
ELSE_TSANG_Ch032.qxd
8/23/2005
References
12:29 PM
Page 449
449
The numerical simulation of the system must deal simultaneously with the native water, injected effluent, saline intrusion, and gases produced. Moreover, the temporal decrease of hydraulic conductivity will have to be taken into account if the model is to describe the evolution of well plugging. In addition, our lack of knowledge concerning the depth of the aquifer’s lower confining bed prevents us from studying the possible downward migration of the effluent, whose initial density is comparable to seawater.
ACKNOWLEDGMENTS We thank M. Laurent BROC and the staff of Distillerie de Savanna for the technical assistance and useful data they have given us.
REFERENCES Bigot, L., 2000. Suivi environnemental des rejets de la “Distillerie de Savanna” sur le site de Bois Rouge, St André—Ile de la Réunion, Evolution des écosystèmes marins entre 1994 et 1999, Analyse de l’impact sur le milieu et les peuplements benthiques. Rapport final d’étude—Rap, ARVAM pour le compte de la Distillerie de Savanna. Billard, G., 1977. Carte géologique de l’île de la Réunion, éditions B.R.G.M. Cleary, E.J. and Warner D.L., 1970. Some considerations in underground waste water disposal. Am. Water Works Assoc. J., 62(8): 489–498. Ditommaso, A. and Elkan, G. H., 1973. Role of bacteria in decomposition of injected liquid waste at Wilmington, North Carolina. Underground Waste Manage. Artif. Recharge, 1: 585–599. Join, J.-L., Pomme, J.-B., Coudray, J. and Daesslé, M., 1988. Caractérisation des aquifères basaltiques en domaine littoral. Impact d’un récif corallien. Hydrogéologie, 2: 107–115. Join, J.-L., 1991. Caractérisation hydrogéologique du milieu volcanique insulaire; Le Piton des Neiges—Ile de la Réunion. Thèse de Doctorat, Université de Montpellier II. Kaufman, M.I. and McKenzie, D.J., 1975. Upward migration of deep-well waste injection fluids in Floridian aquifer, south Florida. J. Res. U.S. Geol. Survey, 3(3): 261–271. Leenheer, J.A., Malcom, R.L. and White, W.R., 1976. Investigation of the reactivity and fate of certain organic components of an industrial waste after deep-well injection. Environ. Sci. Tech., 10(5): 445–451. Martial, J.S., Fevre, Y. and Join, J.L., 2000. Analyse synthétique des forages du Programme Départemental de Recherche en Eau, Rapp. LSTUR, Univ. Réunion, 35 p., pour le compte de HydroExpert. Mauroux, B. and Barrera, A., 1992. Evaluation de la propagation dans l’aquifère littoral par simulation du panache de pollution lié à l’injection des rejets de la distillerie de Bois Rouge, Commune de Saint André, rapp, BRGM Réunion, 92 REU 34, R35580 REU 4S 92. Poggi-Varaldo, H., 1992. A comparison of treatments for high strength distillery slops from the sugar cane industry. 47th Purdue Industrial Waste Conference Proceedings, Lewis Publishers, Inc., Chelsea, MI, pp. 789–800. Rao, T. D. and Viraraghavan, T., 1985. Treatment of distillery wastewater (spent wash)— Indian experience. Purdue Industrial Waste Conference, 40th edn, pp. 53–58. Saripalli, K.P., Sharma, M.M. and Bryant, S.L., 2000. Modeling injection well performance during deep-well injection of liquid wastes. J. Hydrol., 227: 41–55.
ELSE_TSANG_Ch032.qxd
450
8/23/2005
12:29 PM
Page 450
Injection of Organic Liquid Waste in a Basaltic Confined Coastal Aquifer, Reunion Island
United States Environmental Protection Agency (EPA), 1977. The rum industry and rum distillery wastes in Puerto Rico and the Virgin Islands: Effects on the marine environment and treatment options. U.S. EPA Report. Voss, C.I. and Souza, W.R., 1987. Variable density flow and solute transport simulation of regional aquifers containing a narrow freshwater–saltwater transition zone. Water Resour. Res., 23(10): 1851–1866.
ELSE_TSANG_CH033.qxd
8/23/2005
7:34 AM
Page 451
Chapter 33
DEMONSTRATION OF PRESENCE AND SIZE OF A CO2-RICH FLUID PHASE AFTER HCL INJECTION IN CARBONATE ROCK J.E. Clarka, D.K. Bonurab, C. Miller c, and F.T. Fischerd a
E.I. du Pont de Nemours & Co., Inc., Beaumont, TX, USA Bonura Geological Consulting, Inc., Beaumont, TX, USA c Retired du Pont Experimental Station, E.I. du Pont de Nemours and Co., Wilmington, DE, USA d 217 Hidden Lake Rd., Hendersonville, TN, USA b
33.1 INTRODUCTION From 1973 to 1992, DuPont injected acidic fluids from manufacturing operations at their Louisville, Kentucky, facility (Fig. 33.1). Fluids were injected into the Copper Ridge Dolomite, a nearly flat-lying carbonate unit found at depths below 3000 ft (Fig. 33.2). Since that time, both injection wells have been plugged and secured. Native formation fluid from the Copper Ridge Dolomite contains in excess of 10,000 mg/L total dissolved solids and is too mineralized to qualify as a source of drinking water. HCl injectate was neutralized in the subsurface by reaction with carbonate rock. The site received an approved “chemical fate” No-Migration Demonstration from the Environmental Protection Agency in 1990. Acid concentration of the injectate ranged from 0 to 10 wt% HCl, with an average concentration of ∼6 wt%. Injection occurred at ⬍50 psi (⬍345 kPa) wellhead pressure, with flow rates averaging ⬍100 gallons per minute (gpm). Calculations, laboratory experiments, and field data showed that generation of a CO2-rich phase begins at acid concentrations approaching 6 wt% at 100 atm (10 MPa) pressure and a downhole temperature of ∼30°C. Figure 33.3 shows the predicted relationship of wt% HCl and temperature to CO2-rich phase/no CO2-rich phase at 100 atm (10 MPa) pressure. The CO2-rich phase was less dense than the neutralized waste and rose to the top of the injection interval, particularly where a cavity had been created. The CO2-rich phase formed a “cap” in the cavity and acted as an insulating blanket, protecting the roof of the cavity from contact with acidic injectate. Thus, no further degradation of the cavity roof occurred while the cap was in place.
33.2 WIRELINE LOGGING DATA Wireline logging demonstrated the existence of the CO2-rich cap. Gradiomanometer surveys conducted annually in Injection Well #1 identified two separate fluid zones characterized by different densities (Fig. 33.4). In the cavity, the upper portion contained fluid whose specific gravity ranged from 0.5 to 0.8, and the lower portion contained fluid whose specific gravity was ∼1.0 or greater. Injection Well #1 injected acidic fluid from 1973 to 1980, freshwater from 1980 through 1983, acidic fluid from 1984 through 1992, and freshwater during
ELSE_TSANG_CH033.qxd
452
8/23/2005
7:34 AM
Page 452
Demonstration of Presence and Size of a CO2-Rich Fluid Phase
Fig. 33.1. Historical well locations at Louisville, Kentucky.
Fig. 33.2. N–S Geological section through Louisville, Kentucky, site (Cecil, 1987).
ELSE_TSANG_CH033.qxd
8/23/2005
33.2 Wireline Logging Data
7:34 AM
Page 453
453
Fig. 33.3. Diagram of HCl concentration and temperature that produces a CO2-rich phase/no CO2rich phase when HCl reacts with calcite (CaCO3) at 100 atm pressure (after Scrivner and Bennett, 1990). The triangles denote model predictions, and numbers 1 or 2 denote phases observed from experiments.
1993, prior to closure. Figure 33.4 shows that the freshwater injected between 1980 and 1983 dissolved the CO2-rich fluid phase. The CO2-rich fluid phase redeveloped following resumption of acid injection in 1984 and accumulated at the top of the cavity. Advanced sonar caliper surveys measure and delineate the cavity configurations even at distances of hundreds of feet from the borehole (Sonar and Well Testing Services Inc., Houston, TX). The advanced sonar caliper tool is specifically designed to record in aqueous fluids with a specific gravity of 1.0 or more. The receiver does not pick up signal returns that are slowed through the less-dense CO2-rich fluid phase. The interface between the two fluids is recorded as a planar surface. When the CO2-rich phase is present, the actual roof surface of the cavity cannot be seen. Injection Well #2, a backup to Injection Well #1, injected HCl acid from May 1980 to the end of 1983, and freshwater was continually injected from the end of 1983 until closure in 1993. Figure 33.5 is one of many cross sections generated from the sonar caliper survey of Injection Well #2. The roof, sides, and bottom of the cavity appear rough and irregular because of the injection of acid from 1980 through 1983 that dissolved the dolostone. Because the cavity in this well was filled with freshwater at the time of the survey, the cavity—roof, sides, and bottom—could be mapped in its entirety. Note that the cavity is not equidimensional, due to rock heterogeneity and a preferred east–west permeability direction. An isometric view of the cavity in Injection Well #2 is shown in Figure 33.6. Each “slice” represents 2 ft in height. Quite a different picture was obtained from the sonar caliper survey of Injection Well #1. The lower part of the cavity, containing aqueous fluid, was readily delineated as a rough surface. However, the upper portion of the cavity could not be mapped because it was filled with the significantly less-dense CO2-rich fluid phase. The interface between the two fluids acted as a reflecting surface to the sonar signals and mapped as a horizontal, planar surface (Figs. 33.7 and 33.8). The surface between the aqueous fluid and CO2-rich phase was ∼3081 ft measured depth at the time of this survey.
ELSE_TSANG_CH033.qxd
454
8/23/2005
7:34 AM
Page 454
Demonstration of Presence and Size of a CO2-Rich Fluid Phase
Fig. 33.4. Gradiomanometer data showing CO2-rich fluid phase (values ⬍ 1) at various measured depths in the well through time in Injection Well #1. Numbers are specific gravities of fluids.
Fig. 33.5. East–west cross section of cavity in Injection Well #2 in March 1990.
ELSE_TSANG_CH033.qxd
8/23/2005
7:34 AM
Page 455
33.2 Wireline Logging Data
455
Fig. 33.6. Southwest–northeast isometric view of cavity in Injection Well #2 in March 1990.
Fig. 33.7. South–southwest cross section of cavity in Injection Well #1 in March 1990.
Fig. 33.8. Southwest–northeast isometric view of cavity in Injection Well #1 in March 1990.
A television camera survey was run in both injection wells to determine the mechanical integrity of fiberglass reinforced pipe (FRP) injection tubing. During this survey, freshwater was injected at a constant rate to flush the borehole and avoid the possibility of damage to the camera tool. The camera confirmed the existence of the CO2-rich phase in Injection Well #1. As the camera reached the bottom of the packer and exited the injection tubing, the camera light illuminated the zone in which mixing of the injected freshwater with the CO2rich phase was taking place at the top of the cavity. This zone of mixing showed continual movement and reflections/refractions of bright light. This refraction phenomenon continued until the tool reached the approximate depth that the gradiomanometer survey had identified as the interface (3081 ft) between the different density fluids. At the interface, the camera
ELSE_TSANG_CH033.qxd
456
8/23/2005
7:34 AM
Page 456
Demonstration of Presence and Size of a CO2-Rich Fluid Phase
image became dark, because significant refraction did not result from the mixing of the freshwater and the aqueous/brine phase in the main cavity. In the absence of reflective/refractive media, the camera light was dissipated into the fluid of the cavity and simply was not bright enough to illuminate the distant walls. The televiewer screen remained dark until the camera neared the floor of the cavity. A thermal decay time (TDT) log is similar to a resistivity log, which measures chlorine present as sodium chloride in the formation water. The TDT tool generates and captures neutrons; however, the neutrons can pass through materials of well construction such as casing and cement. Thus, the TDT tool records measurements behind pipe. The tool is unaffected by borehole and casing sizes. The TDT log determines the top of the cavity adjacent to the wellbore, and thus is an important tool in determining mechanical integrity of the well. Figure 33.9 shows the cavity tops (roofs) in both injection wells over many years using the TDT tool. After initial acid injection, the cavities showed upward dissolution of several feet. With the formation of the CO2-rich fluid phase during acid injection, the cavity tops stabilized.
33.3 INTERFERENCE TESTING AND PRESSURE CALCULATIONS Interference testing of the two injection wells was conducted during April 1990. Testing involved placement of pressure transducers at depths of 3000 ft in each well to observe the response to freshwater injection into the wells. Figure 33.10 shows the pressure buildup observed in Injection Well #1 at rates of 60 and 120 gpm into Injection Well #1, and 60 gpm into Injection Well #2. Less than 0.5 psi (3.4 kPa) response was observed, ignoring the temperature-induced anomaly. Figure 33.11 shows the pressure buildup in Injection Well #2 for the same time period. A quick response was shown by Injection Well #2 to injection in Well #1. Although these wells
Fig. 33.9. Top of cavity depth in feet for Wells 1 and 2, versus time in years after initial acid injection.
ELSE_TSANG_CH033.qxd
8/23/2005
7:34 AM
Page 457
33.3 Interference Testing and Pressure Calculations
457
Fig. 33.10. Pressure buildup curve for Injection Well #1 from interference testing in April 1990.
Fig. 33.11. Pressure buildup curve for Injection Well #2 from interference testing in April 1990.
are 1500 ft apart, the pressure response was observed less than 2 min into the test. Calculated permeability of the injection interval between the two wells ranged from 40 to more than 300 D (for heights of 50 and 10 ft, respectively), depending upon thickness of the interval. An increase in injection rate from 60 to 120 gpm in Injection Well #1 did not measurably change the slope of the pressure response curve in Injection Well #2. Total increase in pressure over a period of more than 1000 min was ⬍0.9 psi (6.2 kPa), ignoring the temperature-induced anomaly. Data acquired from interference testing enabled calculation of the magnitude of downhole pressure increase resulting from the injection of a known volume of fluid. This
ELSE_TSANG_CH033.qxd
8/23/2005
458
7:34 AM
Page 458
Demonstration of Presence and Size of a CO2-Rich Fluid Phase
calculation showed that 1 gpm of injection resulted in an increase in downhole pressure of ⬍0.01 psi (⬍69 Pa).
33.4 SUMMARY AND RECOMMENDATIONS A CO2-rich fluid phase, developed as a result of HCl and carbonate rock reaction, protects the cavity roof from further dissolution. To create and maintain the protective cap, the HCl acid concentration averaged ⬎6 wt% at downhole conditions. Because the CO2-rich fluid phase was less dense than the acid injectate and native formation fluid, this phase rose to the top of the cavity. Wireline television camera surveys and sonar caliper logs showed that the base of the CO2-rich fluid phase was a distinct, horizontal plane, and the phase blanketed the aqueous liquid. Geophysical logs demonstrated the presence of this CO2-rich phase and showed that its thickness varied based on the amount of acid injected. Reservoir pressure testing, including interference testing, yielded an overall pressure/volume response of a 0.002 psi (13.8 Pa) increase for each gallon of freshwater injected. In addition to the “normal” mechanical integrity tests for Class I hazardous acidic injection wells, we recommend that the following three tests be conducted annually to determine the size and integrity of a cavity: 1. A gradiomanometer test to determine fluid-density variations and size (height) of the protective CO2-rich fluid phase. 2. A sonar caliper survey to determine cavern dimensions and, hence, permeability fracture delineation. Sonar caliper surveys can measure the parts of cavities away from the borehole where the fluids have specific gravity ⬎1. 3. Measurement of the cavity immediately adjacent to the wellbore with a TDT tool. This tool may need to be run more frequently at the start-up of acid injection as the protective cap forms.
REFERENCES Cecil, T., 1987. Geologic Evaluation of Deep Well Disposal Program. DuPont Louisville Works, Jefferson County, Kentucky (by Conoco Inc.). Scrivner, N. and Bennett, K., 1990. DuPont Louisville Works No-Migration Petition, Section 2.2: Chemical Fate of Injected Waste. Figure 3.
ELSE_TSANG_CH034.qxd
8/27/2005
7:46 AM
Page 459
Chapter 34
STABILITY ANALYSIS OF A SOLUTION CAVITY RESULTING FROM UNDERGROUND INJECTION R.W. Nopper, Jr.,a C. Millera*, and J.E. Clarkb a
E.I. du Pont de Nemours & Co., Experimental Station, Wilmington, DE, USA E.I. du Pont de Nemours & Co., Beaumont, TX, USA
b
34.1 INTRODUCTION Underground disposal of aqueous HCl in carbonate rocks at a depth of 3000 ft can result in the formation of cavities. A combined stress/failure analysis leads us to predict that the cavities will be stable against collapse, and that elastic deformation of the rock mass will be quite small at the injection site. With measurements of elastic rock properties, cavity geometry, and preexisting stresses, we employed a numerical code to calculate stresses due to the presence of the cavities. Measurements of rock cores defined the failure envelope of the formation. The net stresses predicted in the stress analysis fall well within the stable region of the envelope, demonstrating that cavities would be stable during a collapse. From 1973 until 1992, Du Pont injected acid fluids from manufacturing operations at its Louisville Works into wells 3000 ft deep. Well 1 injected freshwater during 1993 to render the waste nonhazardous. Well 2 injected acidic waste from May 1980 to the end of 1983, and injected freshwater until 1993. The waste, primarily hydrochloric acid, was injected into the Copper Ridge dolomite, a carbonate formation. This chemistry resulted in the formation of cavities. Below the packer of injection Well 1 during operation, a CO2-rich phase lay above unreacted aqueous waste (Fig. 34.1). Because Well 2 injected freshwater since 1983, it did not have a CO2-rich phase. The formation of a cavity perturbs the stresses in the surrounding rocks. The rock mass can respond in two different modes: deformation or, if the magnitude of the stress perturbation is too large, failure. Deformation can be elastic (the original form is restored upon removal of the applied stresses) or inelastic (permanent deformation occurs). In failure, a sudden and permanent redistribution of material takes place. The analysis described here demonstrates that the rock mass near the dissolution cavities at Louisville was stable against collapse and remained so for the duration of injection operations. The analysis centers around a numerical calculation of the stresses near the cavity. Much effort went into obtaining the experimental information needed as input to the numerical analysis. 34.2 METHODOLOGY OVERVIEW Our methodology in the structural evaluation was based on well-established fundamentals commonly used in rock mechanics to model excavations, and followed a sequence from standard engineering practice (e.g., Franklin and Dusseault, 1989). Essential data for the *
Retired.
ELSE_TSANG_CH034.qxd
460
8/27/2005
7:46 AM
Page 460
Stability Analysis of a Solution Cavity Resulting from Underground Injection
Fig. 34.1. Louisville Injection Well 1.
analysis include elastic rock properties, cavity geometry, stress perturbations caused by the presence of the cavity, preexisting stresses near the cavity, and criteria for rock failure. The evaluation of the structural integrity of the solution cavity at Louisville had two main parts, as outlined in Figure 34.2: (1) a stress analysis followed by a (2) failure analysis. Figures 34.3 and 34.4 show, respectively, the paths to the perturbation stresses and the preexisting stresses in more detail. To begin the evaluation, we needed to know the shape and size of the cavity. These were measured by means of a sonar logging tool suspended in the cavity, below the bottom of the casing. Next, laboratory measurements of core samples provided the elastic properties of the rock mass. Then, using a numerical modeling code, we calculated the stresses caused by formation of the cavity. This determines how the formation of the cavity perturbed the original regional stresses created by large-scale geological forces. Next, we combined the perturbation stresses with preexisting stresses to provide a description of the state of total stress near the cavity. Finally, we compared these net stresses to the stresses that the rocks can withstand without failing, determined from lab tests of core samples. If the net stresses exceed the failure criterion, then we would predict failure, and otherwise, stability. In the latter case, the numerical analysis would also give the earth deformation (e.g., surface subsidence) due to the formation of the cavity. 34.3 STRESS ANALYSIS 34.3.1
Cavity Shape and Size
To define cavity shape and size, sonar-scanning borehole-logging surveys were run in both waste disposal wells in March 1990, by Sonar and Well Testing Services, Inc. (Houston,
ELSE_TSANG_CH034.qxd
8/27/2005
7:46 AM
Page 461
34.3 Stress Analysis
461
Fig. 34.2. A diagram of the stress/failure analysis.
Fig. 34.3. The path to an estimate of the perturbation stresses.
TX). A sonar caliper is placed at successive depths below the casing and scans azimuthally, sensing echoes returning across the fluid-filled void from the rock mass. The tool also scans up and down at a fixed azimuth. These scans combine to give the three-dimensional cavity geometry.
ELSE_TSANG_CH034.qxd
462
8/27/2005
7:46 AM
Page 462
Stability Analysis of a Solution Cavity Resulting from Underground Injection
Fig. 34.4. The path to an estimate of the three principal components of stress existing prior to the formation of the cavity.
The results from the sonar surveys of Wells 1 and 2, shown in Figure 34.5 in isometric views, provide the sizes and shapes of the cavities. In Well 1, the tool indicates a flat top surface. This is not the cavity top, however, but rather the interface between the aqueous waste phase below and the CO2-rich phase above; the latter appears opaque to the sonar tool employed. Note that the true rough upper surface of the cavity at Well 2 is visible; this cavity has no CO2-rich phase and is filled with freshwater. These interpretations of the sonar data were independently corroborated by means of a borehole televiewer tool, run in March 1990. 34.3.2 Rock Elastic Properties The rock elastic properties required in the numerical stress calculation, considering the rocks to be isotropic media, are Young’s modulus and Poisson’s ratio. These were measured to an accuracy of 5% by Western Atlas Core Laboratories (WACL, 1990) using standard ultrasonic techniques applied to core samples of the dolomite just overlying the injection interval. For two core samples, the measured Young’s moduli and (dimensionless) Poisson’s ratios were 9.90 ⫻ 106 psi and 0.307 (2513 ft depth), and 1.31 ⫻ 107 psi and 0.280 (2811 ft). For purposes of the numerical models, we adopted characteristic values of 1.0 ⫻ 107 psi and 0.3 for Young’s modulus and Poisson’s ratio, respectively. 34.3.3 Perturbation Stresses We calculated the perturbation stresses using a state-of-the-art numerical modeling code, ABAQUS. This finite-element analysis package has been used for many years to solve
ELSE_TSANG_CH034.qxd
8/27/2005
7:46 AM
Page 463
34.3 Stress Analysis
463
Fig. 34.5. Results from the sonar caliper survey of March 1990.
problems in mechanical engineering and heat transfer and is very well documented and validated; it is licensed to DuPont by Hibbett, Karlson, and Sorenson Inc. (HKS) of Providence, Rhode Island. Major strengths of ABAQUS include its capability to handle problems that are nonlinear in material properties (constitutive laws), boundary conditions, and geometry. The cavity at Well 1, as shown by the sonar tool, was elongated. To reduce the amount of computation, we assumed the cavity had axial symmetry, and set its size to encompass the largest measured dimension of the cavity. This shape is conservative (for both engineering and regulatory purposes) in that it produces the highest stresses, since a laterally narrow cavity would better support its own roof. In other words, the stresses near the actual cavities should be lower than those we calculated. Figure 34.6a shows a schematic of an axisymmetric cavity. We further reduced the model size, and hence the amount of computational effort, as follows. Visualize a thin, penny-shaped volume within a mass of rock. The forces acting inward on the selected volume match those exerted outward by the rock within. When the forces balance, as they do in intact rock, no movement occurs. If the rock within the volume were now removed to form a cavity, then the forces would be thrown out of balance, and deformation or failure would occur. As Figure 34.6b shows, it is only necessary to include the part of the Earth above a thin cavity in a model if we load the top surface of the cavity with the equivalent tractions (force times the surface area) and fix the rest of the horizontal plane of the cavity in the vertical direction, so that only horizontal movement can occur in that plane. This is an application of the method of inverse tractions (Sneddon and Lowengrub, 1969; Franklin and Dusseault, 1989). We can neglect the region below the cavity in the numerical model since we do not require the stresses and deformations there. The cavity model now appears as in Figure 34.7 in schematic form. A symmetry axis lies along one vertical edge of the model; the use of axisymmetric finite elements restricts material movement along this axis to vertical. Along the other (exterior) vertical edge, material is fixed and prevented from moving. This boundary is placed sufficiently far from the cavity so
ELSE_TSANG_CH034.qxd
464
8/27/2005
7:46 AM
Page 464
Stability Analysis of a Solution Cavity Resulting from Underground Injection
Fig. 34.6. Two techniques that reduce computational effort: (a) axisymmetry; (b) inverse trachons.
Fig. 34.7. Schematic of the finite-element model, showing the boundary conditions.
that its effect on the area of interest near the cavity is negligible. The top edge of the model, representing the Earth’s surface, is a free surface and suffers deformation as calculated by the code. Along the bottom edge of the model, we applied a stress equal to the lithostatic (or effective) pressure along the segment representing the roof of the cavity. The lithostatic pressure is the weight of the overlying rock mass, exclusive of the pore fluids, and is the quantity that controls the failure of rocks (e.g., Hubbert and Willis, 1957). Over the remainder of the lower edge, material can only move horizontally. This treatment of the lower model boundary amounts to the method of inverse tractions. We performed a number of model test runs to develop a feeling for the numerical requirements of the problem. The final model of the 1990 cavity at Well 1 (the larger of the two
ELSE_TSANG_CH034.qxd
8/27/2005
7:46 AM
Page 465
34.3 Stress Analysis
465
cavities) encompassed ∼2500 ft horizontally and 3050 ft vertically, with the model cavity, 250 ft in radius, placed at the bottom. We also ran a second model to account for cavity growth (500 ft radius) through the duration of the injection operations. In these models, the earth was divided into second-order axisymmetric elements. Such elements allow for the bi-quadratic variations in the radial and vertical displacements within each element, and give good performance. We used a graded mesh, with the greatest density of elements in the vicinity of the cavity. In all, 812 finite elements were used. To gain confidence in the results, we also developed a finite-difference code and applied it to several of the scenarios described earlier. The two independent modeling approaches gave virtually identical stress distributions. For this analysis, the output included printed tables of the stresses σrr, σzz, σθθ, and σrz in the elements; displacements ur and uz at the nodes; and contour plots of these quantities. Thus, numerical values for the amount of surface subsidence were given as well as the perturbation stresses due to the cavity. Figure 34.8 gives a sample of the calculated stress distribution. As
Fig. 34.8. A sample calculation of the perturbation stresses near the 500 ft radius cavity.
ELSE_TSANG_CH034.qxd
466
8/27/2005
7:46 AM
Page 466
Stability Analysis of a Solution Cavity Resulting from Underground Injection
intuitively expected, for all components of the perturbation stresses, the large values are confined to the immediate vicinity of the cavity. If failure could occur, it would have to occur very near the cavity. 34.3.4 Preexisting Stresses We combined the perturbation stresses with the ambient, preexisting stress field to give the total stresses acting on the rock mass. Well tests and the geological literature provided the means to estimate the stresses in the vicinity of the cavities at Louisville. Figure 34.4 gives an overview of this approach. The preexisting stresses are caused by the weight of the overlying rocks (overburden), topography, and large-scale geological processes; among them, erosion, glaciation, and stresses are associated with the driving forces for plate tectonics (McGarr and Gay, 1978; Turcotte and Schubert, 1982; Franklin and Dusseault, 1989). Stress fields associated with smaller scale geological structures, such as local rift zones or thermal hot spots, might be superimposed upon these large-scale stress fields. Quantitatively accounting for the causes of stresses at specific sites is the subject of ongoing research, but some general patterns are known. Zoback and Zoback (1980, 1990) have published maps showing states of stresses in the conterminous United States. These authors compiled data generated over a wide span of depths and using a variety of techniques: earthquake focal mechanisms, wellbore breakouts, hydraulic fractures, overcoring, fault slips, and petal centerline fractures. Their compilation suggests that Louisville lies well within the midcontinental stress province, which is a large area of uniform, ENE principal stress orientation. Also, the relative stress magnitudes are essentially uniform in this region. To obtain numerical values for the preexisting stresses, first note that a vertical stress is exerted by the overburden. Since the lithostatic stress controls the fracture of rocks (Hubbert and Willis, 1957), we must subtract the part of the stress due to pore pressure. A stress gradient of 0.44 psi/ft is caused by pore fluids of density 1.01 g/cm3, an average value. An average density of 2.7 g/cm3 for the geologic section, minus the pore fluid density, gives an effective overburden stress gradient of 0.73 psi/ft. This yields the vertical stress σV at a given depth. For example, the overburden stress at the top of the cavity at 3050 ft depth is about 2200 psi. The horizontal stresses can be estimated using results from a well stimulation performed on June 21, 1971, in the Mt. Simon Formation at about 5500 ft depth. Analysis of the “frac” records showed that the instantaneous shut in pressure (ISIP) was 3300 psi. The ISIP is frequently taken as an estimate of the least horizontal principal stress σHmin (Stock et al., 1985). This yields a gradient in this quantity of 0.60 psi/ft. While the stress gradient need not be uniform throughout the section, this numerical value should provide a reasonable estimate. Removing the hydrostatic contribution gives an effective gradient in the least horizontal principal stress of 0.16 psi/ft. We used this gradient to estimate σHmin at the cavity’s depth. Finally, the maximum horizontal principal stress σHmax is expected to be somewhat greater than the vertical stress (Franklin and Dusseault, 1989). Adopting a ratio of 1.3 and subtracting the hydrostatic gradient gives an effective gradient in the maximum horizontal stress of 1.0 psi/ft. This gradient gives σHmax at the cavity’s depth. The next step is to form the total effective stresses by adding the regional stresses to the perturbation stress components. First, however, we must resolve the horizontal components of the perturbation stresses, calculated in cylindrical coordinates, into their rectangular components. The vertical stresses remain unchanged. Using tensor analysis, and denoting by θ
ELSE_TSANG_CH034.qxd
8/27/2005
7:46 AM
Page 467
34.4 Failure Analysis
467
the angle in the horizontal plane with respect to the maximum horizontal principal stress, this gives (Turcotte and Schubert, 1982) 1 1 Sxx ⫽ ᎏ (σrr ⫹ σθθ ) ⫹ ᎏ (σrr ⫺ σθθ ) cos 2θ ⫹ σHmax, 2 2 1 1 Syy ⫽ ᎏ (σrr ⫹ σθθ ) ⫺ ᎏ (σrr ⫺ σθθ ) cos 2θ ⫹ σHmin, 2 2 Szz ⫽ σzz ⫹ σV, 1 Sxy ⫽ ᎏ (σrr ⫺ σθθ ) sin 2θ. 2 34.4 FAILURE ANALYSIS 34.4.1 Failure Criteria Different microscopic mechanisms cause failure in various materials, and workers have deduced empirical laws to describe the conditions under which particular materials fail. There are standard tests that can be performed to characterize the failure criteria of a given material. Some of these are uniaxial, biaxial, and triaxial compression tests. Since it is impractical to secure enough measurements to describe in detail a large rock mass, we adopted conservative values from among the measurements. This provides another margin of safety in the stress/failure analysis. A commonly used method for predicting the onset of failure in geological materials is the Mohr–Coulomb failure criterion (Hubbert and Willis, 1957; Franklin and Dusseault, 1989). This method begins with the fact that the state of stress in a two-dimensional geometry can be represented by a circle plotted on a graph with axes of normal stress σ (horizontal axis) versus shear stress τ (vertical axis), as in Figure 34.9. The Mohr circle crosses the normal stress axis at the values of the major and minor principal stresses, σ1 and σ3. The circle gives the locus of normal and shear stress values across and along a test plane oriented at angle α with respect to the plane corresponding to the major principal stress σ1. Thus, following
Fig. 34.9. The Mohr circle, representing a state of stress, and an idealized failure envelope.
ELSE_TSANG_CH034.qxd
468
8/27/2005
7:46 AM
Page 468
Stability Analysis of a Solution Cavity Resulting from Underground Injection
Figure 34.9, as angle 2α goes from 0 to 180°, test plane angle α goes from 0 to 90°, and the normal stress goes from σ1 to σ3. The Mohr–Coulomb failure criterion is obtained by subjecting a rock sample to compression tests and by noting the stress values at which failures occur. The Mohr circle representing this state of stress is then plotted. The procedure is repeated for a number of samples under different normal stresses, giving circles of different sizes centered at different points along the σ axis. This family of Mohr circles defines the failure envelope as a tangential curve. In many cases, the outer boundary of the family of Mohr circles is approximately a straight line, as shown in Figure 34.9. For these, the failure criterion can be specified in terms of two parameters: (1) the cohesiveness c and (2) the angle of internal friction φ. The failure envelope typically meets the normal stress axis at a small negative value, since most rocks are weak under tension. For many rocks, this small “tensile cutoff” causes the failure envelope to deviate from the linear shape it possesses at larger values of normal stress, as in Figure 34.10. For the Louisville analysis, the same cores as those used in the measurement of rock elastic properties provided samples for triaxial compression tests, which gave the rock’s cohesive strength and angle of internal friction (WACL, 1990). The tests gave values of cohesiveness and angles of internal friction of 4918 psi and 48° (2513 ft depth), and 8244 psi and 47° (2811 ft), respectively. In addition, tensile cutoff of ⫺2484 and ⫺2162 psi, respectively, were measured for the samples using the splitting tensile test method. For purposes of modeling the failure envelope, we adopted conservative values of 4800 psi and 48°, with a tensile cutoff of ⫺2000 psi. 34.4.2 Test for Failure Finally, to see whether the total effective stresses could instigate failure, we calculated the values of the normal and shear stresses, which would exist for various orientations of a hypothetical failure plane at a representative sampling of locations near the cavity. This procedure
Fig. 34.10. A schematic of a realistic failure envelope with a nonlinear region expressing the weakness of rocks under tension.
ELSE_TSANG_CH034.qxd
8/27/2005
7:46 AM
Page 469
34.5 Model Results and Conclusions
469
generates a cloud of points, each corresponding to a different orientation of the test plane at the various locations, in normal stress–shear stress space. The rock mass is predicted to remain intact as long as none of these points lies above the failure envelope.
34.5 MODEL RESULTS AND CONCLUSIONS The test comparing stresses for various orientations of a test plane to the strength of rock samples near a 250 ft radius cavity appears in Figure 34.11. The solid straight line emerging from the origin is the locus of the normal stress–shear stress points that would exist in the absence of the cavity. The stress points are confined to a finite region of failure envelope space. In particular, all values of normal stress are ⬍ 3500 psi, and all values of shear stress are less than 2000 psi. The failure envelope, based on the measurements quoted above, intersects the shear stress axis at about 4800 psi, and slopes upward to the right. The failure envelope is well above the spread of calculated stress points in this case. Therefore, the prediction is that the cavity will remain intact. Mass movement will be due only to elastic deformation. Anticipated injection operations could lead to a cavity of roughly twice the size of the present-day cavity. Calculations for a cavity with a 500 ft radius produce stresses equal to those for the 250 ft cavity. Thus, the prediction is that the cavity caused by injection will not fail in the future. Given that the cavity does not fail, we can predict earth displacement, based on elastic deformation. The analysis predicts that the roof of a 250 ft radius cavity will sag by 0.76 in., and that the ground surface 3050 ft above the cavity center will subside by 0.0075 in. For a 500 ft radius cavity, the respective displacements are 1.5 and 0.059 in. These surface subsidences are negligible compared to other effects, such as Earth tides (Sheriff, 1984). In summary, we have predicted localized stresses in the vicinity of the solution cavities at Louisville. This perturbation of the preexisting stress field is insufficient to cause failure of the rock mass. Thus, the rock mass undergoes elastic deformation, and the calculated values thereof are much less than ambient deformations due to other causes.
Fig. 34.11. The failure test of a 250 ft radius cavity.
ELSE_TSANG_CH034.qxd
8/27/2005
470
7:46 AM
Page 470
Stability Analysis of a Solution Cavity Resulting from Underground Injection
ACKNOWLEDGMENTS C. H. Hales performed the corroborating finite-difference calculations.
REFERENCES Franklin, J.A. and Dusseault, M.B., 1989. Rock Engineering. McGraw-Hill, New York, NY. Hubbert, M.K. and Willis, D.G., 1957. Mechanics of hydraulic fracturing. Trans. Soc. Petroleum Engr., 210: 153–168. McGarr, A. and Gay, N.C., 1978. State of stress in the Earth’s crust. Ann. Rev. Earth Planet. Sci., 6: 405–436. Sheriff, R.E., 1984. Encyclopedic Dictionary of Exploration Geophysics. 2nd ed. Society of Exploration Geophysicists Press, Tulsa, OK. Sneddon, I.N. and Lowengrub, M., 1969. Crack Problems in the Classical Theory of Elasticity. Wiley, New York, NY. Stock, J.M., Healy, J.H., Hickman, S.H. and Zoback, M.D., 1985. Hydraulic fracturing stress measurements at Yucca Mountain, Nevada, and relationship to the regional stress field. J. Geophys. Res., 90(B10): 8691–8706. Turcotte, D.L. and Schubert, G., 1982. Geodynamics: Applications of Continuum Physics to Geological Problems. Wiley, New York, NY. Western Atlas Core Laboratories (WACL), 1990. Ultrasonic Velocity and Dynamic Moduli, Tensile Strength, High Pressure Mercury Injection, Louisville, Kentucky, Disposal Well. Report SCAL-90007, Western Atlas Core Laboratories, Irving, TX. Zoback, M.L. and Zoback, M., 1980. State of stress in the conterminous United States. J. Geophys. Res., 85(B11): 6113–6156. Zoback, M.L. and Zoback, M., 1990. Tectonic stress field of the continental United States. In: L.C. Pakiser and W.D. Mooney (Eds), Geophysical Framework of the Continental United States. Geological Society of America Memoir 172.
ELSE_TSANG_CH035.qxd
8/22/2005
5:46 PM
Page 471
V. LIQUID RADIOACTIVE WASTE INJECTION
471
This page intentionally left blank
472
ELSE_TSANG_CH035.qxd
8/22/2005
5:46 PM
Page 473
Chapter 35
LIQUID RADIOACTIVE WASTE DISPOSAL INTO DEEP GEOLOGIC FORMATIONS BY THE RESEARCH INSTITUTE OF ATOMIC REACTORS (RUSSIA) V.V. Mironov, A.M. Ulyshkin, A.S. Ladzin, and V.I. Kuprienko State Scientific Center of Russian Federation, Research Institute of Atomic Reactors, Dimitrovgrad, Russia
35.1 INTRODUCTION The State Scientific Center of the Russian Federation Research Institute of Atomic Reactors (RIAR) conducts engineering and scientific investigations into atomic power, energetic radiation and reactor materials science, chemistry and technology of transplutonium elements, technology for separate stages of the nuclear reactor fuel cycle, and investigations into using atomic science and technology achievements in other fields of the national economy. A large portion of the work conducted at RIAR focuses on liquid wastes with complex chemical and radionuclide compositions. Due to the large volumes of such wastes, and the necessity to minimize their influence on the biosphere by confining the majority of the volume of radioactive wastes in deep geological formations, a site for deepwell injection of liquid radioactive waste (LRW) of RIAR was created. This site was named the Experimental-Industrial Test Site (EITS). LRW deep-well injection (underground disposal) consists of the monitored transport of waste to reservoir horizons in geological formations, where they are neutralized due to the natural decay of radionuclides, and by physicochemical reactions of the waste interacting with holding rocks and water in the horizon. In 1958, the decision was made to undertake research and a geological survey of the LRW underground disposal problem. This meant creation of a disposal system to inject drainage water, wash water, and decontamination water into geological formations, leaving the minimum of radioactive substances on the surface in sludge storage. During this period, the LRW research at the RIAR was unprecedented. There were no historical data or experience relating to the work performance or LRW disposal, on an industrial scale, into the deep horizons of geological formations containing highly mineralized waters. For this reason, the need arose to develop new methods of physicochemical and radiochemical investigations, and deep-well sampling methods that considered the specific characteristics and problems connected with underground disposal. Likewise, new ways to model and evaluate the specific hydrodynamic conditions had to be developed to predict waste migration in the geological formations, sanitary disposal estimation methods, etc. The predictions were to be verified. Due to the necessity of solving a large number of problems, several research, project, and production organizations were invited to provide additional help and support.
ELSE_TSANG_CH035.qxd
474
8/22/2005
5:46 PM
Page 474
Liquid Radioactive Waste Disposal into Deep Geologic Formations
35.2 GEOLOGIC-HYDROGEOLOGICAL SUBSTANTIATION OF THE METHOD Geologic-hydrogeological investigations related to deep-well injection in the RIAR district were carried out in several stages. During the first stage of the work (1962–1965), the main problem involved the study of the geological structure and hydrogeological conditions to determine the geological formation available for use as a reservoir horizon for waste injection. The second stage of the study looked at the collection of the horizons from stage one to determine their possible employment for injection of LRW. The subsequent stages of the study were carried out in parallel to the experimental waste injection, and to later operation of the EITS. These stages were aimed at further detailing the geological structure of the Paleozoic deposits, and hydrogeological conditions of the area adjoining the injection site, to determine the waste migration conditions in the geological formation. In the first stage, two wells were drilled, with one open throughout the full section of the Paleozoic formations to a depth of 2264 m. The results are divided into seven water-bearing complexes. Figure 35.1 shows a geological section of the area investigated. All water-bearing complexes, except the coal-bearing complex III, are composed mainly of fractured carbonate rocks—limestones and dolomites. The coal-bearing complex comprises interbedded sandstone, aleurolites, and argillites. According to the sampling results, water-bearing complexes I and II were considered unavailable for waste disposal due to their low filtration-property indices. Water-bearing complex III demonstrated sufficiently high filtration parameters during its sampling and was recognized as a possible reservoir horizon for disposal. Complex III is located at a depth of 1440 m with a thickness of 110 m, and it is composed of sandstone with clayish shale interbeds. In the experiments conducted in the second stage, the results obtained were characteristic of water-bearing complexes, and water-bearing complex IV was recommended for waste disposal. Complex IV is located at a depth of 1130 m, with a thickness of 280 m, and is composed of limestones and dolomites. While testing the water-bearing complex III by fresh water injection, it was established that the capacity indices of this complex are considerably lower than that of productivity indices, i.e., the output during withdrawal. The capacity coefficient proved to be 20–30 times lower than the productivity coefficient and made up 2.4 –3.3 m3/(hour MPa). This phenomenon is probably related to a number of factors, including the swelling of clay minerals in the coal-bearing suite rocks due to their contact with fresh water, and the compressibility of cracks in wells near the zone—under the influence of the pressure difference that occurs during injection. The indicated circumstance requires special measures to increase a well’s capacity. At the same time, testing by injecting water-bearing complex IV showed that the acceptance coefficient is equal to, or somewhat higher, than the productivity coefficient. Sanitary reliability of LRW disposal into water-bearing complexes III and IV was determined by the availability of the regional impermeable-clay horizon (Vereyskiy) separating these complexes from upper deposits. The water-bearing complexes V and VI lie above the “Vereyskiy” horizon as buffer layers for reservoir horizons of water-bearing complexes III and IV. These complexes are separated from the surface, and Quaternary water-bearing horizons that contain fresh water, by 300 m thick clay deposits of UpperPermian Age. Water-bearing complexes III, IV, V, and VI contain waters with mineralization of 230–285 g/L of sodium chloride composition, which is not appropriate for a water supply.
ELSE_TSANG_CH035.qxd
8/22/2005
5:46 PM
Page 475
35.3 Physicochemical Investigations of Disposal Processes
PUMPING STATION WASTES PREPARATION FACILITY
475
OBSERVATION WELL
INJECTION WELL
0 60
WASTES SOURCE 300 WATER-BEARING COMPLEX VII REGIONAL WATER-RESISTANT LAYER
400
WATER-BEARING COMPLEX VI LOCAL WATER-RESISTANT LAYER
OBSERVATION WELL WATER-BEARING COMPLEX V
1050 1100
REGIONAL WATER-RESISTANT LAYER 1400 PUMPED WASTES WATER-BEARING COMPLEX IV
1500
LOCAL WATER-RESISTANT LAYER WATER-BEARING COMPLEX III
Fig. 35.1. Geological section of the injection site.
35.3 PHYSICOCHEMICAL INVESTIGATIONS OF DISPOSAL PROCESSES To successfully use deep-well injection, a preliminary study was conducted to determine the physicochemical processes that would occur when complex wastes came into contact with a real reservoir horizon. The main purpose of studying the optimal physicochemical conditions of disposal was to develop a flow chart, a means of waste preparation for disposal, and a preliminary estimate of radioactive nuclide migration under natural conditions. To solve these complex problems, an investigation was conducted that included the study of: ● The composition of underground water and rocks ● The chemical composition and physical properties of wastes ● The behavior of wastes interacting with underground waters and rocks ● Possible changes of the chemical and phasic composition of water in a reservoir horizon
ELSE_TSANG_CH035.qxd
476
8/22/2005
5:46 PM
Page 476
Liquid Radioactive Waste Disposal into Deep Geologic Formations
Changes occurring in the properties and composition of rocks during long periods of contacts with the waste ● The sorption properties of the rock and development of processes connected with radioactivity accumulation in the reservoir horizon (radiolysis, heat removal, etc.). As a result of this investigation, it was established that the average decontaminated waste is not corrosive, and its interaction with the rocks does not lead to a significant change in their composition. Tables 35.1 and 35.2 show chemical and radionuclide compositions of LRW placed in underground disposal. ●
35.4 LRW DISPOSAL AND CONTROL PROCESSES To inject waste into the reservoir horizons at the EITS, the use of four injection wells— two for each reservoir horizon (water-bearing complexes III and IV)—are needed. The LRW neutralization technologies incorporated: ● The LRW reception from the RIAR subdivisions ● The preparation for disposal ● The injection into the reservoir horizon via injection wells. Injection is conducted with a flow rate of 22.5 m3/hour and the maximum permissible pressure at the well orifice of 5.9 MPa. The average activity of the LRW is 1.8 × 106 Bq/kg. There are 37 observation wells on the EITS for observing waste distribution in the geologic medium. The observation wells are placed in three zones of the sanitary-protective area that are positioned 0.6, 3.0, and 12–13 km away from the center of injection wells, respectively. Figure 35.2 presents the scheme of observation well allocation. Constant control of waste injection is carried out in the process of the EITS operation. The flow rate and pressure at the injection-well heads are constantly monitored and recorded on tape. Systematic controls on the physicochemical properties are terminated when waste Table 35.1. Chemical composition of LRW disposed of by deep-well injection Indices pH value Alkalinity gen. (mg/L eqv.) Hardness gen. (mg/L eqv.) Sodium (mg/L) Ferrum gen. (mg/L) Phosphates (in terms of phosphorus) (mg/L) Chlorides (mg/L) Nitrates (mg/L) Sulfates (mg/L) Silica (mg/L) Oxalates (mg/L) Anion-active SAS (mg/L) Nonionogenic SAS (mg/L) Fatty acids (mg/L) Oils (mg/L) Oxidability (mgO/L) Dry residue (mg/L) Content of weighed substances (mg/L)
Mean value
Range
7.1 11.5 5.6 690 2.0 35 50 850 112 7 12 23 2.6 23 90 390 2400 23
6.6–8.5 5.3–24.2 3.1–11.7 600–3000 0.2–12.0 12–108 30–137 200–1100 29–263 4–40 3–60 10–58 1.9–6.0 10–36 80–320 220–680 1000–3400 10–60
ELSE_TSANG_CH035.qxd
8/22/2005
5:46 PM
Page 477
35.4 LRW Disposal and Control Processes
477
injection is complete. All of the EITS operational indices are noted in logbooks, from which data are regularly processed and analyzed for record documentation. Control of the radioactive and chemical waste component distribution in the geological medium is conducted by geophysical investigations and by chemical-radiochemical sampling of the underground water in the observation wells. The geophysical investigation includes the use of resistivity measurements, thermometry, acoustic logging, and gamma logging. Resistivity measurements, thermometry, and acoustic logging are performed to ensure well integrity i.e., tightness of casing string, absence of circulation behind the casing, etc. Gamma logging of the wells is the main geophysical method used to observe the removal of waste distribution. The great advantage of this method is the fact that it records gamma Table 35.2. Radionuclide composition of LRW disposed of by deep-well injection Specific activity (µCi/l) Radionuclides
Mean value
Maximum value
Cesium-137 Cesium-134 Strontium-90–yttrium-90 Europium-152 Europium-154 Cerium-144 Ruthenium-106 Cobalt-60 Volume activity of beta nuclides Volume activity of alpha nuclides
14.0 2.9 8.7 5.5 3.4 4.8 8.0 1.8 50.0 < 0.01
120.6 24.6 55.7 35.2 21.8 30.7 4.2 11.5 320.0
P-13 P-36 ZONE 2
P-12 P-14
P-35 ZONE 1 P-11 P-4 H-2 P-8 H-1 P-9 P-6 P-20 H-4 H-3
P-7
P-15 P-21 P-19
P-21
P-28
P-22
Symbols
P-2
P-20
P-16
P-17
-2 P-23
ND
P-24
Y BA
O EP
AG
OR ST
P-29
P-25 P-27
P-30 P-31
Fig. 35.2. Scheme of EITS well locations.
waste propagation in water-bearing complex IV waste propagation in water-bearing complex III injection wells in water-bearing complex III observation wells in water-bearing complex III injection wells in water-bearing complex IV observation wells in water-bearing complex V observation wells in water-bearing complex V
ELSE_TSANG_CH035.qxd
8/22/2005
478
5:46 PM
Page 478
Liquid Radioactive Waste Disposal into Deep Geologic Formations
Fig. 35.3. Average annual value of gamma radiation in the observation well. Depths in this well: 1: 1423–1424 m; 2: 1426–1428 m; 3: 1444–1445 m.
radiation, not only in the tube perforation interval, but also along the whole length of the well. Therefore, monitoring of the opened absorbing horizon, and all overlying horizons, can be carried out using one observation well. Water sampling for chemical and radiochemical analysis is conducted using an internal bathometer, allowing for sampling of underground liquid at the specified depth. The content of dry residue, mineralization value, pH value, calcium, magnesium, total alkalinity, carbonates, bicarbonates, nitrates, chlorides, sulfates, silicic acid, iron, oxidation state, and specific beta and alpha activity–as well as the content of cesium-137, strontium-90, and barium140–were determined in selected samples from the observation wells. At the same time, the spectrometric analyses for other radionuclides were conducted. More than 2.4 million m3 of wastes, with a total activity of 5 × 1015 Bq, were placed in the reservoir horizons during the EITS operation. Gamma logging and radiochemical investigations detected waste components in the reservoir horizon of water-bearing complex III in the observation wells located within the second zone of the sanitary-protective area. After the operation of water-bearing complex III was completed in 1973, the activity began to decrease at the expense of dispersion and natural decay of radionuclides, and the gammalogging currently detects 0.1 mR/hour (22 pCi/kg cm s). The decrease in activity is shown in Figure 35.3. In the reservoir horizon of water-bearing complex IV, the gamma radiation beyond the first zone of the sanitary-protective area was detected for the first time in 1977. It was detected in well P-9, located 450 m from the center of waste injection, with depths ranging from 1193 to 1195 m behind the “blind” casing string. The gamma radiation was 0.02 mR/hour. While water-bearing complex IV was filled with waste, the gamma background in this well continually increased, and is now 1.5 mR/hour. Currently, there is no increase in gamma radiation in other wells located near the center of waste injection. Signs of radioactive contamination have not been detected in the water-bearing complexes that lie above those used for injection (reservoir horizon). 35.5 CONCLUSION The results of the multiyear investigation allow us to draw conclusions about the efficiency and safety of deep-well injection (underground disposal) for LRW. We can also
ELSE_TSANG_CH035.qxd
8/22/2005
35.5 Conclusion
5:46 PM
Page 479
479
use this information to determine typical waste behavior in the geological medium, which is rather important for practical studies of problems related to the usage of different kinds of geological formations. We determined the capacity of reservoir horizons, the reliability of their isolation from the earth’s surface and shallow horizons of fresh underground waters, and the character and texture of geological structures. Sufficient mastery of waste behavior has given us positive prospects for disposal of toxic waste from other enterprises at the RIAR EITS. Technologies were developed and safely used for the experimental injection of electrolytic production wastes, pesticides, and chemical process wastes.
This page intentionally left blank
480
ELSE_TSANG_CH036.qxd
8/23/2005
12:33 PM
Page 481
Chapter 36
SAFETY ASSESSMENT OF DEEP LIQUID-ORGANIC RADIOACTIVE WASTE DISPOSAL B.G. Balakhonova, A.A. Zubkova, V.A. Matyukhaa, M.D. Noskovb, A.D. Istominb, A.N. Zhiganovb, and G.F. Egorovc a
Siberian Chemical Combine, Seversk, Russia Seversk State Technological Institute, Seversk, Russia c Institute of Electrochemistry RAS, Moscow, Russia b
36.1 INTRODUCTION Production and reprocessing of nuclear materials yields liquid-organic radioactive waste (LORW), specifically the extractants damaged by radiation-chemical impact, in addition to various oils, lubricants, and solvents containing fission products and microquantities of plutonium and uranium. Established, standard approaches to LORW treatment and disposal include either incineration in various furnaces and facilities, or incorporation of the waste in bitumen or polymer matrices. Both of the above techniques share the disadvantage of secondary radioactive waste generation. In response to the limitations of standard disposal methods, scientists have recently explored the deep injection of waste into porous geological formations (Rybalchenko et al., 1998) as a technique for isolating (from the human biological environment) the aqueous liquid radioactive waste (LRW) generated by nuclear power. Injection of LORW into deep geological formations, for disposal of alkaline LRW effluents, could increasingly be considered as an alternative LORW disposal technique. Joint injection of LORW and aqueous LRW creates a nonequilibrium multiple-phase system in the host injection zone, consisting of the host rock, aqueous-phase, and organic nonaqueous-phase solutions. To perform a safety assessment of this injection technique, we studied the joint two-phase flow of the aqueous and organic fluids through the rock in the laboratory, determined the physicochemical processes defining the retention of the LORW macro- and microcomponents and radionuclides, studied the radiation-chemical processes of the LORW component decomposition under injection zone conditions, and finally, tested, at a semi-full scale, the injection of the spent extractant into the injection zone of the LRW site injection (Balakhonov et al., 2001). Laboratory studies and monitoring of the aqueous LRW behavior in an injection zone allowed us to determine the pumping conditions that ensure the safety of this technique—and proved that the LORW components were rather effectively decomposed under injection zone conditions. Nevertheless, the ever-increasing requirements imposed on underground-injection safety, and the necessary optimization of injection conditions, gave rise to the need to quantitatively describe LORW behavior within the injection zone—to predict long-term changes in underground layer conditions. Quantitative description of the LORW behavior within the injection zone of the underground injection test site is possible with an adequate physicochemical model. In this paper, a mathematical model is presented in the injection zone conditions that developed during the joint injection of LORW and aqueous alkaline
ELSE_TSANG_CH036.qxd
8/23/2005
482
12:33 PM
Page 482
Safety Assessment of Deep Liquid-Organic Radioactive Waste Disposal
LRW. The model is based on laboratory results and on injection-zone-condition monitoring data. Simulated forecast calculations were used to assess deep LORW injection safety. 36.2 MATHEMATICAL MODEL The model we developed describes the nonequilibrium, two-phase multicomponent, nonisothermal flow of incompressible fluids, using a dual porosity approximation. Porous space is divided into two parts, mobile and immobile. The mobile part contains the mobile aqueous and organic phases and corresponds to pass-through pores, whereas the immobile part contains motionless fluids and corresponds to stagnant pores. Absorbed components of both aqueous and organic phases form the independent system parts located on the phase boundary (PB). Thus, the system consists of eight parts: (1) a mobile organic phase, (2) a mobile aqueous phase, (3) an immobile organic phase, (4) an immobile aqueous phase, (5) PB between the mobile organic phase and the rock, (6) PB between the mobile aqueous phase and the rock, (7) PB between the immobile organic phase and the rock, and (8) PB between the immobile aqueous phase and the rock. Changes in system composition occur as a result of liquid-phase convective movement, capillary impregnation, convective flow interchanges between different parts of the same phase, mass transfer between the system parts, absorption processes in the rock (e.g., sorption, ion exchange), radioactive decay, and radiation-chemical decomposition of organic substances. In terms of temperature dynamics, the system is assumed to be in local thermodynamic equilibrium. Temperature changes in the liquid waste (stratal water) host rock system occur as a result of the conductive and convective heat transfer, as well as the heat generated from radioactive decay. Basic equations for the two-phase, multicomponent, nonisothermal filtration model are as follows: i ρ Φi .UΦ ρ Φ* ρ Φi ∂ρ Φi ⫹ QΦ Θ(QΦ) ᎏ ⫹ Θ(⫺QΦ) ᎏ ᎏ ⫽⫺div ᎏ mSΦ mSΦ* mSΦ ∂t
冢
冣
冢
i ρ Φ* ρ Φi ⫹ QΦ*Φ ⫻ Θ(QΦ*Φ) ᎏ ⫹ Θ(⫺QΦ*Φ) ᎏ mSΦ* mSΦ
冢
冣
冣 (36.1)
4
i i ⫺ J Φ⫹4Φ ⫹ J Φi (Φ ⫽ 1 ⫺ 4), ⫹ 冱 J ΨΦ Ψ⫽1 Ψ⫽Φ
∂ρ Φi ~ i ⫹ JΦi (Φ ⫽ 5 ⫺ 8), ᎏ ⫽ J Φi~ Φ ⫹ JΦΦ⫺4 ∂t
(36.2)
∂(cT) ᎏ ⫽⫺div ∂t
(36.3)
U U ⫹ ᎏ 冣T⫺β grad T冣⫺q⫹W 冢冢 ᎏ S S c1
1
1
∇(k A(S1, S2)∇P) ⫽ 0,
c2
2
2
A(S1, S2) ⫽ f1(S1)兾µ1 ⫹ f2(S2)兾µ 2,
(36.4)
ELSE_TSANG_CH036.qxd
8/23/2005
12:33 PM
36.3 Results and Discussion
Page 483
483
where UΦ denotes the filtration rate for the phase Φ defined by the general Darcy’s law; ρΦi the reduced density of the ith component in the system part Φ; m the porosity of the i medium; t the time; SΦ the saturation of the system part Φ; J ΨΦ the flux density of the ith component from phase Ψ into phase Φ; QΦ the capillary impregnation flux density from the system part Φ* into part Φ; QΦ*Φ the convective flux density of the system part Φ* into the part Φ, on account of the diffusion mass transfer of macro components between all of i the system parts; Θ(x) the stepped function (Θ(x) ⫽ 1 at x ⬎ 0 and Θ(x) ⫽ 0 at x ⭐ 0); J Φ⫹4Φ i the flux density of the ith component from the liquid phase Φ onto the rock surface; JΦ the source density for the ith component in phase Φ resulting from the physicochemical reac~ ~ tions; J Φi~ Φ the effective flux density of the ith component from phase Φ into phase Φ resulting from the change in saturation of the liquid system parts; T the temperature; c the volume heat capacity of the medium defined by the heat capacities and the volume fractions of the system parts; cΦ the volume heat capacity of phase Φ defined by the heat capacities and the reduced densities of its macro components; β the heat conductivity of the medium; q the thermal flux density through the layer roof and the layer foot; W the specific energy absorption power for the energy evolved as a result of the physicochemical reactions occurred; k the permeability coefficient; µΦ the viscosity of phase Φ; A(S1, S2) the flux mobility; and fΦ(S) the relative phase permeability of phase Φ. The * sign near the phase index designates the substitution of the mobile part of the phase for the immobile one and vice versa (1↔3, 2↔4); and the ∼ sign designates the substitution of the organic phase for the aqueous one of the same mobility and vice versa (1↔2, 3↔4, 5↔6, 7↔8).
36.3 RESULTS AND DISCUSSION On the basis of the mathematical model described above, we developed software that allowed us to conduct a numerical simulation of the thermal field dynamics and organic and aqueous LRW component behavior in the injection zone. In this chapter, the results of the computer simulation are presented for the six-year joint injection of the spent organic extractant and the alkaline LRW effluent into the same injection well. We traced the behavior of the following macrocomponents: water, tributylphosphate (TBP), n-paraffin, watersoluble TBP decomposition products, and water-soluble n-paraffin products. The microcomponents under study included the long-lived 137Cs and 90Sr radionuclides and the short-lived 95Zr, 95Nb,144Ce, and 106Ru radionuclides. The organic phase displaced by the aqueous effluent is located in the stagnant pores of the injection zone within a 15 m distance of the injection well (see Fig. 36.1). Alternate aqueous- and organic-phase pumping results in the redistribution of the radionuclides among the organic phase, the aqueous solution, and the rock surface. Radioactive component behavior in the aqueous LRW effluent—the spent organic extractant—host rock system is defined by redistribution among the system parts and radioactive decay. The short-lived radionuclides get into the organic phase and are adsorbed; the long-lived radionuclides define the power flux value in the injection zone. During the injection of LRW, the gross activity of the short-lived radionuclides exceeds that of the long-lived ones by more than an order of magnitude, thereby observing the power flux maximum in the vicinity of the injection well (see Fig. 36.2). After the injection well operation terminates, the gross activity of the short-lived radionuclides decreases, and the long-lived radionuclide contribution to the power flux predominates.
ELSE_TSANG_CH036.qxd
484
8/23/2005
12:33 PM
Page 484
Safety Assessment of Deep Liquid-Organic Radioactive Waste Disposal
Fig. 36.1. Profiles of macrocomponents and the organic-phase saturation in the injection zone: (1) the organic phase saturation; (2) TBP; (3) the hydrocarbon solvent; (4) the TBP decomposition product; and (5) the decomposition product of the solvent.
Fig. 36.2. Profiles of the temperature, activity, and power flux in the injection zone: (1) total long-lived radionuclide activity; (2) total short-lived radionuclide activity; (3) temperature.
When the organic phase is completely displaced and immobilized, the quantity of organic macrocomponents is reduced, owing to radiolysis and hydrolysis. The region with the highest organic macrocomponent decomposition rate corresponds to the region of maximum activity. As a result, the distribution of the organic macrocomponent concentration in the injection zone assumes a wave shape (see Fig. 36.1). According to the simulation, in the sixyear operation of the injection well, the decomposed TBP fraction amounted to approximately 47% of the total mass injected as spent extractant and aqueous LRW effluent; whereas in the hydrocarbon solvent, the decomposed fraction reached 59% of that same mass (see Fig. 36.3). This finding indicates that the injection zone can be considered a manmade reactor for toxic organic compound decomposition. After pumping stops, the decomposition products accumulate in the aqueous phase, with their concentration proportional to the decomposition rate of TBP and n-paraffin (see Fig. 36.1). When the aqueous effluent is injected, the decomposition products form a wave. The height of this wave diminishes as the distance from the injection well increases. Cyclic LRW disposal is accompanied by temperature-field fluctuations within the injection zone. The fluctuation magnitude for the maximum temperature value is about 45°C. When pumping is stopped, the injection zone temperature rises according to the distribution of total specific activity (see Fig. 36.2). During effluent injection, a temperature wave is
ELSE_TSANG_CH036.qxd
8/23/2005
References
12:33 PM
Page 485
485
Fig. 36.3. Organic component mass: (1) injected TBP; (2) TBP mass in the formation; (3) injected solvent; and (4) the solvent mass in the formation.
formed (as a result of convective heat transfer), starting from the injection well. During the fourth year of operation, the maximum power flux growth stops because of the equilibrium attained between the short-lived radionuclide influx into the formation and radionuclide decay, and the injection zone temperature reaches its maximum value of 70°C. 36.4 CONCLUSIONS Simulation results from our mathematical model show that the organic substance distribution area is not large (not exceeding tens of meters from the injection well) and is associated with the transfer of the organic phase into an immobile condition. The presence of the extractant within the zone of the maximum radionuclide accumulation on the rock causes efficient radiationchemical decomposition of the extractant. Even the hydrocarbon extractant impervious to hydrolysis is decomposed rather efficiently. Decomposition products of TBP and the hydrocarbon solvent possess significantly higher solubility, and when the aqueous LRW is injected, they are removed from that part of the injection zone saturated by the extractant. Hence, repeated LORW injection does not increase the region of organic-phase distribution within the injection zone. The presence of the extractant injected into the injection zone does not markedly affect temperature characteristics. Under the conditions in question, the maximum temperature in the injection zone is determined by the total volume and gross activity of the aqueous and organic LRW injected into the injection zone. Under the existing injection conditions, the maximum temperature is significantly less than the LRW boiling point under formation conditions. Thus, the injection zone used for deep aqueous alkaline LRW injection could be considered a man-made radiation-chemical reactor for treating liquid-organic radioactive waste. The efficiency and safety of the LORW decomposition processes are defined by the radioactivity accumulated in the rock and the waste injection cycle length. REFERENCES Balakhonov, B.G., Zubkov, A.A., Matyukha, V.A., 2001. Mathematical modeling of the radiation-chemical decomposition of the organic contaminants in the liquid radioactive waste during the deep injection thereof. Radiokhimiya. 43(1): 82–86. (in Russian). Rybalchenko, A.I., Pimenov, M.K., Kostin, P.P., 1998. Deep Injection Disposal of Liquid Radioactive Waste in Russia. Battel Press, Columbus, OH, USA.
This page intentionally left blank
486
ELSE_TSANG_CH037.qxd
8/22/2005
5:49 PM
Page 487
Chapter 37
RESULTS OF LONG-TERM DEEP LIQUID RADIOACTIVE WASTE INJECTION SITE OPERATION AT THE SIBERIAN CHEMICAL COMBINE A.A. Zubkov a, A.S. Ryabov a, V.A. Sukhorukov a, V.V. Danilov a, and A.I. Rybalchenkob a b
Siberian Chemical Combine, Seversk, Russia All-Russia Designing and Research Institute of Production Engineering (VNIPIPT), Moscow, Russia
37.1 INTRODUCTION Liquid radioactive waste (LRW) from the Siberian Chemical Combine is disposed of at the Underground Waste Injection Site. The medium- and high-level radioactive waste injection site (Site 18a) has been in operation since 1963, and the low-level radioactive waste injection site (Site 18) has been in operation since 1967. The injection sites are located 10 km from the Tomj River. The underground repositories use the sand injection zones in the sedimentary cover of the West Siberian platform, which is sealed with aquiclude layers of argillaceous rocks. Injection is made through specially modified injection wells into the Cretaceous sediments at a depth of 270–390 m at Site 18, and at a depth of 315–340 m at Site 18a. During the entire operational period, over 40 million m3 of LRW were injected into the underground repository. The Underground Waste Injection Site was created according to the design developed by the All-Russia Scientific Research and Design Institute of Industrial Technology (VNIPI Promtekhnologii). The Federal “HydroSpecialGeologiya” Enterprise performed the geological survey of the site and construction of the wells. 37.2 CHARACTERISTICS OF THE INJECTION SITES Overall, there were 34 injection wells and 124 monitoring wells drilled at Site 18, with 12 injection wells still in operation at the site. LRW is injected continuously, but the operating conditions vary in time and repository area. Low-alkalinity LRW injected into the repository has a pH of 8.0–10.5, and total salt load of up to 30 g/L. Sodium, ammonium, and alkaline earth metals define its cation composition; the anion composition includes nitrates, sulfates, chlorides, and hydrocarbonates. Neutral LRW injected into the repository has a total salt load below 1 g/L consisting of sodium nitrate, sulfate, chloride, and hydrocarbonate. Thirty injection wells were drilled at Site 18a, of which three are currently operational, and there are 56 monitoring wells. Several injection wells are used for the injection of acidic LRW with a pH of 2–3. LRW contains up to 10–20 g/L of acetic acid, up to 90–140 g/L of sodium nitrate, up to 2 g/L of dissolved corrosion products from Cr, Fe, Ni, Al, and negligible quantities of organophosphorus compounds and silica in the form of ortho silicic acid. Fissionproduct radionuclides are also present, such as 90Sr, 134⫹137Cs, 106Ru, 95Zr, 95Nb, 144Ce, as well as uranium and trace amounts of TRU elements, such as 239Pu, 241Am, and 237Np. Several
ELSE_TSANG_CH037.qxd
488
8/22/2005
5:49 PM
Page 488
Results of Long-Term Deep Liquid Radioactive Waste Injection Site Operation
injection wells are used for the injection of alkaline LRW containing 5–15 g/L of NaOH, which represents the sodium nitrate effluents with an average salt load of 150 g/L. Sodium carbonates, sulfates, and aluminates are also present; the fission products 90Sr, 134⫹137Cs, 106Ru, 95 Zr, 95Nb, and trace amounts of plutonium, represent the radionuclide composition. 37.3 RESULTS AND DISCUSSION The following results of the multiyear operation were observed: When the waste is injected into the injection zones, there is no excessive pressure buildup sufficient for a hydraulic rupture of the formation, or for a break in the integrity of the overlying aquicludes. During the waste injection, the cone of elevated pressure in the injection zone attained an equivalent hydraulic head of 20 m above the natural level. The margins of the cone were detected by direct monitoring in wells within 5 km of the site. Even a short idle period of 3 months was sufficient for the cone of elevated pressure to retreat back within the boundaries of the site. No cone of elevated pressure developed in the overlying formations during site operation (see Fig. 37.1). 2. The contours of the radioactive waste and anthropogenically modified water distribution remain within the site boundaries. The distributions appear to be less than those predicted earlier. As the distance from an injection well increases, man-caused geochemical zoning is formed in the underground water. The zoning is associated with the dynamics of well 1.
Fig. 37.1. Dynamics of the cone of elevated pressure as a result of waste injection: (A) during the waste injection; (B) after the 3-month idle period.
ELSE_TSANG_CH037.qxd
8/22/2005
5:49 PM
Page 489
37.3 Results and Discussion
489
Fig. 37.2. Zoning in the underground water composition.
3.
4. 5.
operation, the interaction of effluents with the injection zone minerals and waters, and the differential mobility of effluent components. The distribution of radionuclides also reveals the zoning that corresponds to the experimental data (see Fig. 37.2). The composition and volume of the waste injected is restricted so that the maximum temperature of the injection zone, in the vicinity of the shut down injection well, is no higher than two-thirds of the boiling point of the fluid in the injection zone. After well operation ceases, a slow decline in temperature is observed (see Fig. 37.3). A temperature field is formed around an injection well. As a rule, by the time the well is shut down, the field is detected in an area no more than 80–100 m from the injection well, and does not cross the site boundaries. The temperature field of the entire injection area is the combination of the separate fields formed during the complete operational period of the injection wells (see Fig. 37.4). Currently, the maximum temperatures of the repository formation are associated with the injection wells shut down 3–5 years ago. Gas yield in the injection zone, which occurs because of chemical reactions, thermolysis reactions, and radiolysis of the injected effluents, does not cause the formation of either
ELSE_TSANG_CH037.qxd
490
8/22/2005
5:49 PM
Page 490
Results of Long-Term Deep Liquid Radioactive Waste Injection Site Operation
Fig. 37.3. Changes in the repository formation temperature during injection well operation.
Fig. 37.4. Temperature field map of the waste injection area.
6.
abnormal high-pressure zones or significant gas blockage of the pore space to hinder waste injection. Closure of abandoned injection wells reliably separates the injection zone from the overlying formations, and does not cause the waste to flow back through the closed well bores into the overlying horizons.
37.4 CONCLUSION The results obtained in the long-term operation of the LRW Underground Injection Site demonstrate the possibility of reliable disposal of liquid radioactive and toxic waste into sandstone injection zones.
ELSE_TSANG_CH038.qxd
8/22/2005
5:52 PM
Page 491
Chapter 38
RADIONUCLIDE DISTRIBUTION IN A SANDSTONE INJECTION ZONE IN THE COURSE OF ACIDIC LIQUID RADIOACTIVE WASTE DISPOSAL A.A. Zubkova, B.G. Balakhonova, V.A. Sukhorukova, M.D. Noskovb, A.D. Istomin, A.G. Kesslerb, A.N. Zhiganovb, E.V. Zakharovac, E.N. Darskayac, and G.F. Egorovd a
Siberian Chemical Combine, Seversk, Russia Seversk State Technological Institute, Seversk, Russia c Institute of Physical Chemistry RAS, Moscow, Russia d Institute of Electrochemistry RAS, Moscow, Russia b
38.1 INTRODUCTION Large quantities of liquid radioactive waste (LRW) are generated in the course of a nuclear fuel cycle plant’s operation, and deep injection (Rybalchenko, 1998) is one of the LRW handling techniques. In Russia, the deep injection of acidic LRW into the sandstone injection zones of deep repositories is performed at MinAtom Enterprises. To ensure the safety of deep-LRW injection, the distribution of radionuclides within the injection zone is necessarily controlled and predicted. To accomplish this, a monitoring system for the injection zone was created; laboratory studies of LRW interactions with the host rock were conducted; and mathematical models and software were designed for simulating the LRW behavior and changes in the condition of the injection zone during deep radioactive-waste injection.
38.2 PHENOMENOLOGICAL MODEL Waste is pumped into the reservoir horizon through the injection well to a depth of 280–350 m. The disposal is performed in batches. The LRW is injected into the injection zone in the following sequence: Phase 1—disposal of the nitric acid-based effluents; Phase 2—injection of the acetic acid-based solutions; and finally, repeated disposal of the nitric acid-based effluents. The radionuclide composition of nitric acid based waste (6–10 g/dm3 of HNO3) is primarily the fission 90Sr, 134,137Cs, and 106Ru products. The waste acetic acid based waste (10–20 g/dm3) contains 90–140 g/dm3 of NaNO3, dissolved Cr, Fe, Ni, Al from the corrosion products (gross content is 2 g/dm3), negligible amounts of silica; pH value of the effluents is within the range of 2–3; and the fission products of 90Sr, 134+137Cs, 106Ru, 95Zr, 95Nb, 144Ce, as well as microconcentrations of uranium and TRU elements, are also present. During deep-LRW injection, the nonequilibrium thermodynamic system of the LRWunderground water–host rock interacts with the interrelated hydrodynamic, thermodynamic, and physicochemical processes under way. The system attempts to achieve an equilibrium
ELSE_TSANG_CH038.qxd
492
8/22/2005
5:52 PM
Page 492
Radionuclide Distribution in a Sandstone Injection Zone in the Course of Acidic Liquid
condition, causing the removal of radionuclides from solution and their accumulation on the solid natural and newly formed phases (Zakharova, 2001; Zubkov, 2002). Due to radionuclide accumulation on the solid phases, the gamma-emitting power in the vicinity of the injection well within the injection zone reaches 12,000–15,000 µ R/s, and the maximum temperature reaches 170°C from an initial injection zone temperature of 10°C. Using the simulated waste effluents prepared on the nitric and acetic acid basis, we obtained the distribution-coefficient value data for the radionuclides. For the initial solid phase, we used sand from the injection zone. Sorption experiments were run in batches at the solid/liquid ratio of 1/10, for a period of 3 months, under conditions as close to the underground reservoir horizon as we could get—specifically, a pressure of 3 MPa, temperature within the range of 20–170°C, and a pH of 2–7. Experimental results are given in Table 38.1. The acidity of the injected waste decreases in time, because of interaction with the host rock minerals and radiation-induced chemical and thermochemical decomposition. Results of laboratory studies regarding the change in nitric and acetic acid concentration are shown in Figure 38.1. The experiment was conducted in an aqueous-nitrate solution in the presence of injection zone sand pretreated with acid at a temperature of 20°C.
38.3 MATHEMATICAL MODEL In our model, we used an approach based on the selection of a limited number of minerals, components, and processes, the description of which is sufficient to determine the character of the radionuclide distribution, and the thermal field within the injection zone, at any given moment of time. We considered the behavior of nine radionuclides of 90Sr, ,137Cs, 144Ce, 95Zr, 95 Nb, 106Ru, U, 239Pu, Np, and eight nonradioactive components of H, NO3, Na, Ac, HAc, and others. Interaction of the host rock with the acidic influent was described by introducing three pseudominerals combining the rock-forming minerals into groups by the dissolution rate as follows: carbonates (1); clay minerals, micas, chlorites (2); quartz and feldspars (3). Equilibrium and kinetic parameters of the radiolysis and thermal hydrolysis for nitric and acetic acids were determined from the experimental data (Egorov, 2002). To allow for
Fig. 38.1. Change in nitric (1) and acetic (2) acid concentrations in the aqueous solution of sodium nitrate in the presence of the injection zone sand at a temperature of 170°C and the dose power of 0.2 J/kg.s.
pH NaNO3 (g/dm3) 90 Sr 137 Cs 144 Ce 106 Ru 95 Zr 95 Nb 239 Pu 112–140 237 Np U 1.4 1.4
0.7 1.4 0.7 1.4 0.7 0.7 2.1
0.7 1.4 0.7 0.7 0.7 0.7 1.4 0.7 0.7
1.0
80–150
1.0
40–80
2.1–2.5 2.1–2.8
2.5–3.0 140 1.7–1.9 2.8–3.9 1.4–2.2 1.8–2.2 1.4–2.1 1.4–2.1 2.0–8.4
20–60
7–14 9.8–21
14–49 14–28 28–56 9.8–21 14–21 14–21 14–70
3–4
60–120
14–20 21–23
21–56 14–21 42–70 28–56 21–35 21–35 42–70
4–5
120–170
3–7 3–10
5–6 100 7–14 10–21 14–21 7–14 7–14 7–14 14–28
20–40
14–17 21–28
14–17 21–28 21–28 14–21 14–21 14–21 28–56
5–6
40–120
21–28 32–42
21–28 28–42 28–35 14–21 21–28 21–28 63–70
6–7
120–80
8–14 10–14
7 10 38–42 63–91 49–119 7–14 7–14 7–14
20
5:52 PM
0.7 0.7
1–2 Not present 0.5 0.7 0.6 0.7 0.7 0.7 1.4
20–40
8/22/2005
Temperature (°C)
Table 38.1. Distribution coefficient values Kd (cm3/g) for radionuclides between the liquid and solid phases, depending upon temperature and the effluent pH value
ELSE_TSANG_CH038.qxd Page 493
38.3 Mathematical Model 493
ELSE_TSANG_CH038.qxd
494
8/22/2005
5:52 PM
Page 494
Radionuclide Distribution in a Sandstone Injection Zone in the Course of Acidic Liquid
hydrodynamic dispersion, we used the dual-porosity assumption that divides porosity into the pass-through m1 and the stagnant m2. Fluid in the pass-through pores is capable of moving under the gradient of pressure and gravitational forces (mobile fluid). A part of the fluid contained within the stagnant pores does not participate in the convective movement (immobile fluid). Thus, the system under consideration consisted of the following parts: the mobile part of the liquid phase (Φ 1), the immobile part of the liquid phase (Φ 2), the host rock in contact with the mobile part of the liquid phase (Φ 3), and the host rock in contact with the immobile part of the liquid phase (Φ 4). The fluid was assumed to be incompressible; the porous medium was assumed to be incompressible and rigid. One further assumption was that the conversion of pass-through pores into stagnant ones, and vice versa, did not occur. Fluid filtration rate U was determined by Darcy’s law: k U1 grad P,
(38.1)
where k denotes the rock permeability, µ the fluid viscosity, and P the pressure. The pressure distribution was derived from the flux-conservation law assuming the stringent filtration mode approximation:
冢冣
¯ 0. div U
(38.2)
The distribution of activity within the injection zone depends on the convective mass transfer with the flux of the liquid phase, on the radioactive decay, and on radionuclide redistribution between the liquid and solid phases. In the model, the flux density Ji2 for the ith radionuclide between the liquid and solid phases was determined as follows:
冢
i
冣
i
冢 冣 i
i
Ji2 i F F i F F ( 1, 2)
(38.3) i
where i (i ) denotes the rate constant for the direct and reverse process, F is the driving force of the mass flux, and (x) the stepped function [ (x) 0 if x0 and (x)1 if i x 0] Driving force F for the mass flux of the ith radionuclide between the corresponding parts of the liquid and solid phases was assumed to be proportional to the deviation from the equilibrium concentration: i
i
i
F C, PC,
(38.3a)
i
where C, P denotes the equilibrium concentration of the ith radionuclide. The equilibrium i concentration of the ith radionuclide C, P in the liquid phase, if linearly connected with its concentration on the rock, was described as follows: i
H
i
C2 KP(T, C ) C, P (=1, 2) i
(38.4)
where C2 denotes the concentration of the radionuclide absorbed by the respective part of H the solid phase; KP(T,C ) is the distribution coefficient, being a function of temperature and pH of the solution.
ELSE_TSANG_CH038.qxd
8/22/2005
5:52 PM
Page 495
38.3 Mathematical Model
495 Hj
Rate of the hydrogen-ion concentration change J , due to interaction with the jth mineral, was described as follows: j j
J Hj (SyД) C (1, 2)
(38.5)
where denotes the parameter of the jth mineral dissolution rate, the hydrogen ion activity coefficient, and Sj the specific surface of the jth mineral at the phase boundary between the corresponding parts of the system. Reduction in the jth mineral content on account of the acidic influent neutralization was equal to the flux J Hj . The rate of the acid concentration change on account of the radiation–chemical interaci tion (J)p∂ was assumed to be proportional to the acid concentration and to the specific power flux. The rate of the acid concentration change on account of the thermochemical i decomposition (J)T was assumed to be proportional to the acid concentration. Thus, the dynamics of the component concentration change in different parts of the system were described by the following system of equations:
j
3 C1i i i i i ij m1
div(C .U)J J J J13 , 冱 1 1 12 13 t
(38.6)
j1
3 C2i i i ij m2 J 12 J 2i J 24 冱J24 , t
(38.7)
j1
m1 C3i i ,
(1m) J3i J13 m t
m2 Ci i ,
(1m) 4 J4i J24 m t
(38.8)
C3j m1 ij ,
(1m) J13 t m
m2 C j ij ,
(1m) 4 J24 m t
(38.9)
where Ci denotes the ith component concentration in the solid phase (Φ = 3, 4), Cj the jth mineral concentration (Φ 3, 4), and JΦi the rate of change for the ith component content in the Φ part of the system. The latter is defined for radionuclides by the radioactive decay i i law and for acids by the radiation–chemical and thermochemical decompositions; J13 (J 24 ) is the mass flux of the ith component between the corresponding parts of the liquid and ij ij ) is the rate of change for the ith component and jth mineral con(J24 solid phases; and J13 tent due to the neutralization process. The rate is defined for the hydrogen ion in Equation i (38.5) and equals zero for all the other components; J12 is the mass flux of the ith component from the mobile part of the liquid phase into the immobile one, assumed before to be proportional to the difference of the component concentration between corresponding parts of the system. For calculations of thermal field dynamics, we used the local heat-equilibrium approximation, assuming all the system parts in any given point of the medium have the same temperature of T. For this case, the thermal field dynamics were determined by:
(cvT)
div Tcᐍ.U div( gradT)W, t
冢 冣
(38.10)
ELSE_TSANG_CH038.qxd
496
8/22/2005
5:52 PM
Page 496
Radionuclide Distribution in a Sandstone Injection Zone in the Course of Acidic Liquid
where cᐍ denotes the volume heat capacity for the liquid phase, cv the effective volume heat capacity for the porous medium, β the effective heat conductivity of the medium, and W the specific power flux due to radioactive decay. Based on the above mathematical model, the numeric algorithm was created, followed by the software application program, UDRW. UDRW was used for the simulation of dynamics of the radionuclide and thermal-field distribution in the vicinity of the injection well. A simulation was run for the cylinder-shaped injection zone area of 20 m × 120 m. The area was divided into five highly permeable interlayers of H = 0.5 m depth, each separated from one another by the low-permeability interlayers. The permeability (k) of the highly permeable interlayers amounted to 0.1, 2, 16, 8, and 1 µm2. Highly permeable interlayers are distinctly visible in the gamma-logging results (see Fig. 38.2). It was assumed that, along the horizontal plane, the injection zone parameter distribution was uniform. LRW density was equal to 1100 kg/m3, rock density was assumed to be equal to 1400 kg/m3, gross porosity was m = 0.35, pass-through porosity was m1 = 0.15. The averaged mineral composition of the sandstone determined from the chemical and mineralogical analyses data, was as follows: quartz 65%, potassium and sodium feldspars 12%, illite 15%, kaolinite 6%, montmorillonite
Fig. 38.2. Results of gamma-logging along the injection well.
ELSE_TSANG_CH038.qxd
8/22/2005
5:52 PM
Page 497
38.4 Results and Discussion of the Simulation
497
2%, micas and chlorites 1.5%, and carbonates 1.5%. The temperature of the injected feed was 20°C; heat capacity of the solid and liquid phase was cᐍ = 4180 J/(kg K) and cs=500 J/(kg K), respectively; heat conductivity of the liquid phase was equal to βᐍ = 0.8 Wt(m K); heat conductivity of the host rock was βs = 8 Wt/(m K), which corresponds to the heat conductivity of the solid phase with a prevailing quartz component [for quartz the βs = 8.84 Wt/(m K)].
38.4 RESULTS AND DISCUSSION OF THE SIMULATION Injection of LRW with different compositions causes the formation of zones around the injection well. The zones differ in physicochemical properties (see Fig. 38.3). A zone in the immediate vicinity of the well has an elevated nitric acid concentration and low pH value— this is the nitrate zone. Adjacent to it is an area of elevated acetic acid concentration and higher pH value—this is the acetate zone. The widths of the zones are defined by the latest injected nitrate or acetate feed volume, and by the depth of the permeable interlayer. The groundwater mixing zone is located on the outer edge of the waste distribution. Small sections of acid distribution are associated with the rapid reduction of the concentrations thereof (see Fig. 38.4). During the idle period between the two injections, the
Fig. 38.3. Concentrations of nitric (1) and acetic (2) acids, and pH (3) as a function of the distance from the injection well location. Interlayer permeability of 8 µm2.
Fig. 38.4. Mass of the nitric (1) and acetic (2) acids within the injection zone as a function of time.
ELSE_TSANG_CH038.qxd
498
8/22/2005
5:52 PM
Page 498
Radionuclide Distribution in a Sandstone Injection Zone in the Course of Acidic Liquid
nitric acid is almost completely decomposed, which explains the absence of HNO3 in the mixing zone. Less intense decomposition of acetic acid accounts for the fact that acetic acid remains in the zone close to the injection well for a period of 3 years after the injection-well shutdown. The distribution of radionuclides in the injection zone close to the injection well is defined by the injection-well operating conditions (volume and radionuclide composition of the injected feed), by radioactive decay, by the depth of the permeable interlayer, and by the radionuclide redistribution between the liquid and solid phases. Absorption of the radionuclides is defined by the distribution coefficients that depend on temperature and the solution pH. The maximum activity occurs in the acetate zone (see Fig. 38.5). Heating of the medium due to radioactive decay causes maximum temperatures in the zone of elevated activity. The latter effect is, in turn, accompanied by growth of the distribution coefficients and yet more intense radionuclide transfer into the solid phase. Thus, the existing acidic-LRW disposal scheme causes concentration of the major activity in the acetate zone close to the injection well, leading to the transfer of up to 90% of the radioactivity into the solid phase during the injection-well operation period. The dynamics of the thermal field are defined by the power flux, by the heat transfer with the underground water flux, by cooling of the injection zone during injection of cold effluents, by the heat capacity of the medium, and also by the heat exchange. When injection is not under way, the thermal field is defined by the distribution of activity. The liquid phase moves primarily through the highly permeable interlayers. Each of the highly permeable interlayers has its own nitrate, acetate, and mixing zone; however, they are shifted relative to one another depending on the permeability ratio. Several power-flux maximums are formed in the injection zone, located at different depths and distances from the injection well. For the highly permeable interlayers, the maximum of activity is located farther from the injection well as compared to the low-permeability interlayers. The effect yields higher temperature gradients close to the power-flux maximums of the highly permeable interlayers and, therefore, more intense thermal dissipation. As a result, the total temperature maximum shifts in the direction of the power-flux maximum of the low-permeability interlayer, and hence to the injection well itself. Vertical heterogeneity of injection zone permeability could give rise to the formation of several local maximum temperatures associated with power-flux maximums of various
Fig. 38.5. Radionuclide activity in the fluid (1), solid phase (2), temperature (3), and acetic acid concentration (4) as a function of the distance to the injection well. Interlayer permeability of 8 µm2.
ELSE_TSANG_CH038.qxd
8/22/2005
References
5:53 PM
Page 499
499
Fig. 38.6. Temperature dependence in the observation well over time: (1) thermal logging data, (2) simulation data.
highly permeable interlayers. The location of the temperature maximum is defined by injection zone structure, and depends on the volume and composition of the injected waste. In practice, the temperature in the vicinity of the injection well is monitored by means of the geophysical well located nearby. Figure 38.6 shows the dependence of the experimental temperature readings in the observation well, and the results of computer simulation of temperature change in the geophysical well. Good agreement between the simulated and experimental curves proves the conceptual model to be adequate.
38.5 CONCLUSION The existing technological scheme of acidic-LRW injection causes the major activity to be concentrated in the acetate zone close to the injection well. This is associated with the increase in distribution coefficients of the radionuclides between the liquid phase and the rock going from the nitrate to the acetate zone as a result of the increase in pH from 2 to 4. The accumulation of the radionuclides in the area gives rise to the increase in the radiation heat emission, and hence to a temperature rise. The temperature rise leads to an increase in the distribution coefficients and to the further transfer of the radionuclides into the solid phase. Thus, a self-sustaining barrier is formed in the vicinity of the injection well, which prevents the dispersion of the radionuclides in the subsurface water.
REFERENCES Egorov, G.F., Danilov, D.I., Zakharova, E.V., Darskaya, E.N. and Zubkov, A.A., 2002. Radiation-thermal decomposition of nitric and acetic acids in the aqueous nitrate solution. Atom. Energiya, 93(1): 54–59 (in Russian). Rybalchenko, A.I., Pimenov, M.K., Kostin, P.P., Balukova, V.D., Nosukhin, A.V., Mikerin, E.I., Egorov, N.N., Kaimin, E.P., Kosareva, I.M. and Kurochkin, V.M., 1998. In: M.G. Foley and L.M.G. Ballou (Eds), Deep Injection Disposal of Liquid Radioactive Waste in Russia. Battelle Press, Columbus, OH, US.
ELSE_TSANG_CH038.qxd
500
8/22/2005
5:53 PM
Page 500
Radionuclide Distribution in a Sandstone Injection Zone in the Course of Acidic Liquid
Zakharova, E.V., Kaimin, E.P., Darskaya, E.N., Menyailo, K.A. and Zubkov, A.A., 2001. Part played by the physicochemical processes during the long-term storage of liquid radioactive wastes in the deep collector layers. Radiokhimiya, 43(4): 378–380 (in Russian). Zubkov, A.A., Makarova, O.V., Danilov, V.V., Zakharova, E.V., Kaimin, E.P., Menyailo K.A. and Rybalchenko, A.I., 2002. Man-caused geochemical processes in the sand collector layers during the liquid radioactive waste injection. Geoekologiya, 2: 143–154 (in Russian).
ELSE_TSANG_CH039.qxd
8/22/2005
5:56 PM
Page 501
Chapter 39
DEEP-WELL INJECTION MODELING OF RADIOACTIVE AND NONRADIOACTIVE WASTES FROM RUSSIAN NUCLEAR INDUSTRY ACTIVITIES, WITH EXAMPLES FROM THE INJECTION DISPOSAL SITES OF SSC RF–NIIAR AND CHEPETSK MECHANICAL PLANTS E.A. Baydarikoa, A.I. Rybalchenkoa, A.I. Zininb, G.A. Zininab, A.M. Ulyushkinc, and A.L. Zagvozkind a
All-Russia Designing and Research Institute of Production Engineering (VNIPIPT), Moscow, Russia b State Scientific Center of Russian Federation, Institute of Physics and Power Engineering, Obninsk, Russia c State Scientific Center of Russian Federation, “Research Institute of Atomic Reactors”, Dimitrovgrad, Russia. d Open Stock Company, Chepetsk Mechanical Plant, Glazov, Udmurtia
39.1
INTRODUCTION
Reservoir formations at depths ranging from 1100 to 1450 m have been used for deep-well injection disposal of radioactive liquid wastes from the SCC RF Research Institute of Atomic Reactors (Dimitrovgrad) and nonradioactive industrial wastes from the Chepetsk Mechanical Plant (Glasov) (Fig. 39.1 and Table 39.1). The reservoir formations are made of carbonate rock characterized by a complex structure of fracturing and pore space, and contain groundwater of a chloride and calcium type with a salt concentration of 230–260 g/kg. The possibility of continued waste injection into deep horizons is determined by the safety of waste disposal and localization within geological environments specified by state authorities (subsurface exclusion zones) (Rybal’chenko, 1996). Predictive calculations and modeling of waste migration on the basis of controlled observations play an important role in substantiating the safety of further injection. The subsurface waste-component distribution is known to have complex spatial and temporal features, owing to a number of factors. Waste migration, specifically in Glazov (for chemical liquid wastes) and Dimitrovgrad (for liquid radioactive wastes), is determined by several major factors: ● The considerable spatial heterogeneity of reservoir formations, represented mostly by irregularly recrystallized, fractured, cavernous limestones, and dolomites with interlayers and inclusions of various lithologies. ● The considerable thickness of the reservoir horizons (200–300 m) and injection intervals (120–230 m). ● Differences in waste and groundwater densities within the reservoir formations: density of wastes is 1.01 g/cm3 (mineralization below 20 g/L); density of stratal water is ~1.16 g/cm3 (mineralization of 200–260 g/L). ● Injection rate and pumping pressure, which vary considerably over time.
ELSE_TSANG_CH039.qxd
502
8/22/2005
5:56 PM
Page 502
Deep-Well Injection Modeling of Radioactive and NonRadioactive Wastes
Fig. 39.1. Location of the disposal sites at the Scientific Research Institute of Nuclear Reactors: NIIAR, in Dimitrovgrad, and the Chepetsk Mechanical Plant in Glazov.
To estimate the waste-component distribution within the geologic environment of the facilities under analysis for the current period and the future—accounting for the factors listed above—we have carried out a two-year predictive, numerical, epignostic study, using the GEON-3DM code (Drozhko, 1995). This code enables modeling in 3-D, taking into account the nonstationary (epignostic) conditions of groundwater flow/migration and the differences in waste and groundwater densities. This work resulted in the creation of continuous-operation mathematical models of liquid-radioactive waste injection disposal at Dimitrovgrad and Glazov. Geologic exploration data and materials for the subsurface-disposal site operation were made on the basis of those models, considering the current theoretical concepts of migration and flow processes in carbonate reservoir horizons. This numerical study of groundwater flow/migration will help in evaluating the space distribution of wastes for possible finalization of site operation (2007–2010), and for the subsequent 300-year period of containment (see Table 39.1).
39.2 MODEL FOR DEEP INJECTION DISPOSAL OF INDUSTRIAL WASTE AT GLAZOV 39.2.1 Geological-Hydrological Schematic Presentation of Natural Conditions The model for deep injection disposal at Glazov was created by taking into account the analysis of hydrological conditions of the region studied, which includes only one reservoir horizon (the Okaian-Bashkirian water-bearing complex). The area modeled by calculation is ~33 km2 (Table 39.1 and Fig. 39.2). The spatial structure of groundwater flow is characterized by sufficiently nonstationary conditions, due to the cyclic operation and nonconstant injection flow
ELSE_TSANG_CH039.qxd
8/22/2005
5:56 PM
Page 503
39.2 Model for Deep Injection Disposal of Industrial Waste at Glazov
503
Table 39.1. Site comparison Indices and characteristics
Glazov (ChMZ)
Beginning of disposal site operation Area of subsurface exclusion zone Maximum allowed injection rate Volume of disposal waste to 2000 Injection pressure Reservoir formation
1992 1973 12.56 km2 83.85 km2 2500 m3/day 550 m3/day 4,000,000 m3 2,500,000 m3 (∼120,000 Ci) ⬎25 atm (2.5 MPa) ⬎60 atm (6 MPa) Bashkirsko-oksky water- Bashkirsko-oksky waterbearing complex bearing complex (permeable zone VI) (permeable zone IV) Limestones and dolomites with impurities of gypsum, anhydrites, calcite, silicon 1420–1540 m 1110–1340 m 120 m 230 m 90–120 m 26–96 m 0.01–0.80 0.02–0.90 4–235 m2/day 1–12 m2/day 1–3 ⫻ 105 m2/day (4 ⫻ 104) to (8 ⫻ 105) m2/day
Composition of rocks (lithology) Depth of injection interval Thickness of injection interval Water level (under the surface) Effective porosity Transmissivity Pressure conductivity factor (piezoconductivity) Salinity and density of water Salinity and density of waste Major components of waste
Observed increase in ground water head due to injection at the center of disposal site and near boundaries of subsurface exclusion zone Area of model Time interval of epignostic simulation Time interval of prediction simulation Observed and calculated area and radius of waste migration at the end of the epignostic period Calculation area and radius of waste migration toward the end of the injection time Calculation area and radius of waste migration 300 years after the end of the injection
230–280 g/L 1.5–1.7 g/cm3 ⬍22 g/L, 1.01 g/cm3 K⫹, Ca2⫹, Na⫹, NH4⫹, NO3–, Cl–, SO42⫺, F–
Dimitrovgrad (NIIAR)
100–160 m and 8–10 m
200–250 g/L 1.12–1.16 g/cm3 ⬍4 g/L, 1.01 g/cm3 Sr90, Cs137, specific activity 5 ⫻ 10–5 Ci/L; Na⫹, NO3⫺, SO42⫺, PO43⫺ 160–260 m and 3 m
33 km2 1992–2000 2000–2007 (end of injection) ⫹ 300 Sfact ≈ 2–4 km2 rfact ≈ 800–1200 m Scalc ≈ 2.2 km2 rcalc ≈ 840 m Scalc ≈ 4.2 km2 rcalc ≈ 1200 m
40 km2 1973–2000 2000–2010 (end of injection) ⫹ 300 Sfact ≈ 8 km2 rfact ≈ 1600 m Scalc ≈ 9 km2 rcalc ≈ 1600–1800 m Scalc ≈ 19.6 km2 rcalc ≈ 2500 m
Scalc ≈ 7 km2 rcalc ≈ 1400–1700 m
Scalc ≈ 55.4 km2 rcalc ≈ 4200 m
rate over time. The top and bottom boundaries of the model are assumed to be impermeable: one type of condition is specified for the side boundaries—the condition of hydrostatic distribution of pressure, accounting for variation in solution density over time. Initially, the flow velocity and density of groundwater correspond to natural conditions, with zero concentration of wastes in the aquifer. Flow and migration parameters are chosen from the range of values obtained in geological exploration and operation of the storage site (Table 39.1).
ELSE_TSANG_CH039.qxd
504
8/22/2005
5:56 PM
Page 504
Deep-Well Injection Modeling of Radioactive and NonRadioactive Wastes
Fig. 39.2. Nonuniform structure of reservoir formation in Glazov, and waste contours in the plan (Layer 10).
We simulated the reservoir formation’s nonordered and nonhomogenous (nonuniform) structure by means of linear extended zones with essentially different hydraulic conductivity and effective porosity. We also simulated the random distribution of low-permeability rock inclusions (of isometric form) in permeable rock thickness, the proportion of such inclusions varying in a cross section. This complex space distribution of materials with different hydraulic and capacity properties was intended to be used to create an adequate model for the heterogeneity of deep, bedded carbonate rocks, in order to be in an agreement with the real data. The model also takes into account the cross-sectional anisotropy of carbonaterock properties—hydraulic conductivity in the vertical direction is 10 times less than that in the horizontal direction. Injection wells H-1, H-5, H-6, and H-10 are the source of waste ingression into the horizon, with the wells having different screening intervals, operating conditions, and liquid waste volumes. The calculations take into account the variation in groundwater viscosity: salt solutions within the reservoir horizon are dependent on the content of the salts therein, and a gradual increase in the salt solution density from the top to the bottom was specified for natural conditions (from 1.15 g/cm3 to 1.17 g/cm3). The density of injected solutions is 1.01 g/L, and the tracer component concentration is constant, equal to 1. Our system of mathematical model equations includes the continuity equation, a generalization of Darcy’s law for the case of variable density of flow, an equation for sodium chloride mass conservation written in terms of solution density, and an equation for waste-neutral component transport. It is assumed that the pore solution density is linearly dependent on the concentration of sodium chloride (groundwater mineralization).
ELSE_TSANG_CH039.qxd
8/22/2005
5:56 PM
Page 505
39.2 Model for Deep Injection Disposal of Industrial Waste at Glazov
505
39.2.2 Results of Numerical Studies The model adjustment (calibration) included a reiterated epignostic (1992–2000) simulation of liquid-waste injection for different variants of the horizon structure and collector properties of rock. In our calibration, we made comparisons of observed (real) and calculated plots of water table elevation over time for ten boreholes, as well as the real contamination level (from control observation data) and calculated level. As a result of the model calibration, we developed an acceptable variant of the reservoir horizon structure, with a more precisely defined hydraulic conductivity and rock porosity. In a layer of permeable rocks with hydraulic conductivity Kf ⫽ 0.15 m/day and effective porosity neff ⫽ 0.01 (Figs 39.2 and 39.3), two parallel extended zones of higher permeability are observed, with planar presentation in the northeast direction, and perpendicular presentation thereto, and with Kf ⫽ 1.5 m/day and effective porosity n ⫽ 0.09, extending over the entire area of the model and horizon thickness. At those points where the zones studied are crossing, rock permeability is higher: Kf ⫽ 2 m/day, neff ⫽ 0.09. The central section (rectangular prism shapes), where the injection wells are situated, is characterized by Kf ⫽ 1.4 m/day and neff ⫽ 0.07. The low-permeability material inclusions (Kf ⫽ 0.0005 m/day and neff ⫽ 0.1) are nonuniformly distributed in the cross section of the reservoir horizon: the intervals of horizontal arrangement (occurrence), containing 70% of such inclusions, are interlaid by those containing 10% of such inclusions. The convergence of natural observation and model data obtained as a result of calibration indicates that our groundwater flow and migration model is an adequate representation of natural conditions at the Glazov deep injection site, and that the model is suitable for predictive calculations. Predictive calculations include both: (1) modeling of the assumed period of
Fig. 39.3. Nonuniform structure of reservoir formation and contour of waste in Sections 39.1 and 39.2.
ELSE_TSANG_CH039.qxd
506
8/22/2005
5:56 PM
Page 506
Deep-Well Injection Modeling of Radioactive and NonRadioactive Wastes
continued injection of wastes to the end of operations (until 2007), with a constant flow rate of 1300 m3/day, and (2) modeling of the period for storage site conservation and plume migration (the next 300 years). Modeling results from the first prediction stage (the period of continuous injection) showed that by the end of injection-site operation (2007), the mass of disposed wastes would be entirely within the central zone volume (Kf ⫽ 1.4 m/day and neff ⫽ 0.07; Figs. 39.3 and 39.4), where the injection wells are located. The elevation of the groundwater level by 2007 will amount to 160–180 m at the center of the site, and 9 m at the boundary of the exclusion zone. In this continuous-injection operation mode, the “buoyancy” effect for wastes (gravity differentiation) is very weak. Components of wastes modeled in cross section appear to be distributed homogenously, with the contaminant migration proceeding a little faster over the layers of enhanced penetrability, and with a smaller fraction of lowpermeability inclusions. We found that the nonuniformity and anisotropy of rock hydraulic properties have a considerable effect on the distribution of waste in the reservoir horizon. The alteration (change) in groundwater head is determined mostly by the variations in hydraulic conductivity, and the velocity of waste spread by effective porosity. Density effects are noted only after the injection is terminated; the density convection under conditions of injection site conservation causes a waste-dispersing plume in the reservoir formation (three-dimensional flow) to take the shape of an inverted cone. Waste components move toward the formation top and spread rapidly over the injection zone, with enhanced penetrability. In this case, the relative concentration of the migrating substance is below 0.3. Waste migration is also notably influenced by hydrodispersion and natural flow of groundwater, resulting in a planar distribution of contamination and a gradual decrease in the concentration of the migrating substance with time.
Fig. 39.4. Location of wells and waste plume contours in reservoir formation (Layer 2) at Dimitrovgrad.
ELSE_TSANG_CH039.qxd
8/22/2005
5:56 PM
Page 507
39.3 Model of Deep Injection Disposal at Dimitrovgrad
507
By 300 years, there is a maximum spread of wastes in the northwesterly direction to 1700 m away from the site center, and the contaminant plume covers an area of ∼8 km2. However, the plume does not exceed the area of the exclusion zone, equal to 12.56 km2 (Table 39.1 and Figs. 39.2 and 39.3). In addition to the decrease in concentration of the migrating substance, observations show a decrease in reservoir pressure and a lowering of the hydraulic head. Thus, in 300 years, the hydraulic head will exceed the natural level by only 5–10 m. Our model calculations have in fact confirmed the localization of liquid wastes within the exclusion zone after injection into the reservoir horizon is complete. These calculations indicate that further operation of the deep injection site for industrial waste at the Chepetsk Mechanical Plant is possible and safe.
39.3 MODEL OF DEEP INJECTION DISPOSAL AT DIMITROVGRAD Numerical modeling of liquid radioactive waste migration at Dimitrovgrad was carried out with the objective of studying predicted subsurface localization in connection with the need to extend the designed operation time of the injection site to 2010 (Rybal’chenko, 2000). The approaches to model development and schematic representation of the site at Dimitrovgrad are similar to those described above for the Glasov site. The model includes one water-bearing complex (C3-2b-ok) used for the injection disposal of liquid radioactive wastes since 1973. The area modeled is situated within the exclusion zone at the NIIAR site, covering an area of ∼40 km2 (Figs. 39.4 and 39.5). Vertical-fluid-flow (filtration) heterogeneity in layered water-bearing carbonate rocks is
Fig. 39.5. Contour of waste in Sections 39.1 and 39.2.
ELSE_TSANG_CH039.qxd
8/22/2005
508
5:56 PM
Page 508
Deep-Well Injection Modeling of Radioactive and NonRadioactive Wastes
represented by the interbedding of high- and low-permeability layers that extend over the entire area of the model. Layers with higher permeability (zones with increased fracturing of carbonate rocks), wherein most of the industrial waste components are distributed, are determined from gamma-ray logging data. In the cross section, four high-permeability layers are assigned by the model, having 10 m, 3 m, 20 m, and 15 m thicknesses, respectively, with a characteristic hydraulic conductivity of 0.25 m/day and effective porosity of 0.02. The hydraulic conductivity of the low-permeability layers separating them is approximately two orders of magnitude smaller—0.0015 m/day; their porosity is 0.01. The model also takes into account the cross-sectional anisotropy of carbonate rocks: hydraulic conductivity in the vertical direction is one order of magnitude below that in the horizontal direction (Kfz ⫽ 0.1 Kfxy). Modeling of radioactive-component migration was carried out for Sr90 and Cs137, the most mobile radioactive waste components in terms of physical and chemical form (cations). The end time of the calculation (300 years) corresponds to the period when the content (activity) of these radionuclides would decrease to a safe level, below the admissible level for fresh water. Calculations have shown that during the injection period, the waste components for the most part fill layers with high penetrability. By 2000, the maximum migration of the radioactive components is observed southwest of the area of study, 1800 m away from the site center (the isoline of the reservoir formation solution activity is 10 ⫺9 Ci/L) (Fig. 39.4). The results obtained are in good agreement with the actual radius of waste distribution by 2000, assessed from control-observations data, equal to 1700 m. The waste-spread contour in planar presentation during the period of injection is close to circular; over time, the contamination plume extends in the direction of natural flow movement, i.e., to the southwest. By the time of the final injection (2010), the contaminant plume would occupy an area with radius less than 2.5 km and would be contained by the exclusion zone. After injection is completed, waste distribution between the layers of different permeability takes place. There are only insignificant manifestations of density convection: the “buoyancy” effect of “lighter” components is notable only 200–300 years after the beginning of waste injection. As is shown by numerical studies, the principal parameter determining the extent of waste “buoyancy” is the anisotropy of hydraulic conductivity in the vertical and horizontal directions. After 300 years, the maximum migration of waste components, registered by an isoline of activity of 10⫺9 Ci/L, does not go beyond 4200 m from the site center (Fig. 39.4), in a southwest direction. Thus, the model calculations predict containment of liquid radioactive waste within the exclusion zone over the entire 300-year period.
39.4 CONCLUSION Results from mathematical modeling of the underground disposal of liquid waste, carried out at two different sites, have confirmed the feasibility and safety of deep injection disposal of toxic and radioactive wastes into permeable aquifers. These results have been used to justify continuing such disposal of wastes and the development of measures for monitoring such a method. In the future, we intend to develop improved models for deep injection disposal and to study the influence of various anticipated adverse factors (such as faults and lithologic windows) on waste migration.
ELSE_TSANG_CH039.qxd
8/22/2005
References
5:56 PM
Page 509
509
REFERENCES Drozhko, E.G., Samsonova, L.M., Vasilkova, N.A., Petrov, A.V., Zinin, A.I., Zinina, G.A. and Ginkin, V.P., 1995. Computer model of non-stationary migration of solutions in underground waters. Proceedings of Obninsk Symposium of the 15th Mendeleev Congress on General and Applied Chemistry. Radioecological Problems in Nuclear Power and in the Conversion of Nuclear Industry. Obninsk, pp. 33–43. Rybal’chenko, A.I., Kurochkin, V.M., Baydariko, E.A., et al., 2000. Modeling of deep injection disposal of liquid radioactive wastes in Russia. J. Hydrolog. Sci. Technol., 16(14): 101–122. Rybal’chenko, A.I., Pimenov, M.K., Kostin, P.P., et al., 1996. Deep Injection Disposal of Liquid Radioactive Waste. In: J.A. Apps and C.-F. Tsang (Eds), Russia Deep Injection Disposal of Hazardous and Industrial Wastes, Scientific and Engineering Aspects. Academic Press, New York.
This page intentionally left blank
510
ELSE_TSANG_CH040.qxd
8/23/2005
6:40 AM
Page 511
Chapter 40
EFFECT OF ANTHROPOGENIC TRANSFORMATIONS OF DEEP LIQUID RADIOACTIVE WASTE REPOSITORY-CONTAINING ROCKS ON RADIONUCLIDE MIGRATION E.V. Zakharovaa, E.P. Kaimina, A.A. Zubkovb, O.V. Makarovab, and V.V. Danilovb a
Institute of Physical Chemistry of the RAS, Moscow, Russia Siberian Chemical Combine, Seversk, Russia
b
40.1 INTRODUCTION In the process of deep-injection disposal of liquid radioactive waste (LRW) into sandstone injection zones, the geological medium serves as the main protective barrier in the path of waste component migration and, above all, the migration of radionuclides. Alkaline and acidic waste effluents sent to disposal differ greatly in chemical composition from natural water (Apps and Tsang, 1996; Kaimin et al., 1996). The presence of heat-emitting radionuclides in the waste leads to a rise in temperature and, hence, activates the interaction processes in the waste-rock-natural water system. In the long term, the development of these processes during contact of the waste with the rock could either decelerate or amplify the processes of migration (Zakharova et al., 2001; Zubkov et al., 2002). In this work, we investigate the effect of the human-caused impact on changes in the absorbing capacity of rocks in deep, acidic LRW repositories. The results were obtained by means of chemical and thermodynamic modeling. 40.2 EXPERIMENTS Experiments were conducted with simulated solutions of the following composition (in g/L): NaNO3, 94.7; Fe, 0.22; Cr, 0.57; Mn, 0.20; Ni, 0.22; CH3COOH, 13.33, and HNO3, 2.16. Solutions of the given composition are injected in the pore space of the underground repository in the course of acidic-waste disposal. Before the experiments, simulated solutions were doped with the following radionuclides (in Bq/L): 137Cs, 7.0 105; 90Sr, 11.3 105; 239Pu, 10.7 105; 237 Np, 2.0·105; 241Am, 0.7 10 5, and 238U, 0.05 105. In all experiments, we used a nonradioactive rock sample taken during drilling of the injection-well. Subsurface repository rock formations are represented by sandstone beds containing quartz (55–65%), feldspar-albite, plagioclase, microcline (10–20%), micas (2–10%), chlorite (up to 2%), clay minerals-montmorillonite, kaolinite (up to 15%), calcium and magnesium carbonate (0.5–3.0%), and other minerals. Optical and X-ray methods were used to determine mineralogical content of the rock samples. Before the experiments, the rock samples were not fractionated. The rock and the solution were brought into contact in Teflon® beakers placed in an autoclave. The solid-to-liquid phase ratio was equal to 1 : 5 in all the experiments. The experiments were conducted at temperatures of 90°C and 150°C under a pressure of 3 MPa, which
ELSE_TSANG_CH040.qxd
512
8/23/2005
6:40 AM
Page 512
Effect of Anthropogenic Transformations of Deep Liquid Radioactive Waste
matches the underground repository conditions. At the end of the experiment, the solid and liquid phases were separated by centrifugation at n 800 rpm, τ 30 min, and the liquid phase was then analyzed. Thermodynamic calculations in the multicomponent and multiphase system are based upon the solution of the chemical equilibria by means of the thermodynamic potential minimization method. The simulation was run using the “Selector-C” application program package comprising the minimization algorithm implemented for the multireservoir models simulating the effluent transfer process (Karpov, 1995).
40.3 RESULTS AND DISCUSSION 40.3.1 Behavior of Radionuclides in the Vicinity of the Injection Zone Acidic radioactive waste is injected into the underground repository in batches. The process of acidic-waste injection is conducted in three phases. First, the nitrate effluents are injected (6–7 g/L), followed by the acetate effluents, and concluding with the injection of more nitrate effluents. In some years, the duration of the acidic-waste injection process itself amounted to about 3–7% of the entire injection well operational period. Such waste injection dynamics predetermine the formation of regions in the injection zone, saturated either by the individual nitrate or acetate waste, or by mixtures thereof. Accordingly, geochemical processes of different types may occur in the system. In some areas adjacent to the injection well, the temperature rose as high as 130–170°C, because of the short-lived energy-yielding radionuclides partitioned in the solid phase. The most intense interaction of the acidic HNO3CH3COOH solution of the pore space with the solid phase occurs in this very zone. Decomposition of nitric and acetic acids due to thermolysis, radiolysis and interaction with rocks, leads to a decrease of the liquid-phase acidity in the zone and a rise in the pH to 3–4 from the initial pH value of 1.3. Results of the kinetic experiments characterizing the change in simulant composition are given in Figures 40.1a and b. As a result of thermal hydrolysis, cations are isolated in the solid phase. The isolation rate is expressed by the following series: Fe Cr Ni Mn. Iron moves into the solid phase at 150°C after several hours of interaction with the rock. The accumulation of the other waste components in the solid phase depends on the contact time and temperature. It is worth noting that after 1250 hours of interaction, the solution pH increases to 3.4. The liquid-phase content of components—potassium, silicon, calcium, and magnesium (Figs. 40.1a and b)— leached by the acidic phase from the sand bed increases simultaneously. From the results of the experimental modeling performed for the sandstone bed reaction in contact with acidic HNO3CH3COOH simulant at 150°C and 3 MPa, it follows that the marked changes in mineral composition occur in just 500 hours. Chlorites grow in micas, isolated regions of hydrated micas and montmorillonite are formed in feldspars, and the iron content of the newly formed chlorites increases. Thus, the secondary mineral content grows, and the number of the cation functional exchange groups also increases, enhancing the rock absorption capability. A significant part of the accumulated waste components in the solid phase occurs as new films form on mineral surfaces, consisting primarily of the hydrous-oxide compounds of iron and chromium (Fig. 40.2). The films adhere to the grain surfaces of various minerals including quartz, which is the basic rock-forming mineral for the sandstone beds. Processes occuring in the zone in question, both in the liquid and solid phases, affect radionuclide behavior (Fig. 40.3). It should be noted that the sandstone bed sorption
ELSE_TSANG_CH040.qxd
8/23/2005
6:40 AM
Page 513
Cation content in liquid phase, mg/dm3
40.3 Results and Discussion
513
350 300 250 200 150 100 50 0 0
200
400
600 800 Time, h
(a)
Cation content in liquid phase, mg/dm3
Cr
Ni
1000
1200
1400
1200
1400
Mn
400 350 300 250 200 150 100 50 0 0
200
400
600 800 Time, h
(b) Si
K
Ca
1000
Mg
Fig. 40.1. Change in the liquid waste composition as a function of the waste-sand bed interaction time (t 150°C, p 3 MPa).
capacity is not high for the acidic salt-bearing solutions. A significant amount of sodium is present in the solutions, and this is capable of saturating the cation functional groups of the minerals. A change in the plutonium concentration in the liquid phase is associated with the formation of ferric films, as over 50% of the plutonium present in the liquid phase, like ferric ion, partitions into the solid phase in the first hours of waste interaction with the rock under the hydrothermal conditions. As the interaction time increases, other waste components accumulate in the films; primarily chromium, which participates in the formation of the hydrous-oxide precipitates possessing absorption capabilities. Simultaneously, the system pH rises, causing an increase of americium in the solid phase, basically due to the fixation of the hydrolyzed Am species on the continuously generated sorbent hydroxides. The absorption affinity of the iron–chromium films for strontium and cesium is lower than that for plutonium and americium. Absorption of uranium and neptunium in the experiments simulating the high-temperature, engineered-geochemical zone conditions does not exceed 5–10%.
ELSE_TSANG_CH040.qxd
514
(a)
8/23/2005
6:40 AM
Page 514
Effect of Anthropogenic Transformations of Deep Liquid Radioactive Waste
(b)
Fig. 40.2. (a) Formation of iron–chromium films on quartz grains 1500 magnification (5 cycles, 100 hours each, 150°C, 3 MPa), results obtained by scanning-electron Hitachi HCM-2A microscope; (b) Spectrum of the newly formed element composition on a quartz-grain surface (5 cycles, 100 hours each, 150°C, 3 MPa). Results obtained by scanning-electron Hitachi HCM-2A microscope with the microanalyzer LINK.
Fig. 40.3. Accumulation of radionuclides on the sand bed, during interaction with waste, as a function of time (t 150°C, p 3 MPa).
ELSE_TSANG_CH040.qxd
8/23/2005
6:40 AM
Page 515
40.3 Results and Discussion
515
40.3.2 Radionuclide Behavior in the Zone Farther from the Injection Borehole During the next scheduled waste injection, the reacted solutions are translocated into the adjacent zone where the temperature does not exceed 80–100°C. The reacted solutions do not contain nitric acid, and the acetate-ion content is reduced because of the various processes that occurred in the high-temperature zone. To model chemical processes occurring in the low-temperature zone, we used a solution obtained after 1250 hours of interaction with the sandstone bed under the high-temperature zone conditions (Fig. 40.1). The solution in question contains sodium nitrate and acetate, sandstone-leaching products, and waste components (macro components and radionuclides) that have not yet partitioned into the solid phase (Fig. 40.3). The pH of the solution is 3.4. This solution contacts the sandstone bed at 80°C and 3 MPa. Changes in the solution composition, as a function of the interaction time with the sand bed, are given in Table 40.1 and Figure 40.4. Low-temperature zone experiments were conducted for 1440 hours. As a result, the pH of the liquid phase increased up to 5.0. At that stage, the major chromium Table 40.1. Changes in liquid-phase composition as a function of the interaction time of reacted solutions with the sand beds (t 90°C, p 3 MPa) Time (days)
Concentrations in the solution (mg/L) Cr Ni Mn
Si
K
Ca
Mg
4 20 40 60
98 10 0.1 0.1
251 345 93 33
290 310 152 50
380 455 380 218
230 282 256 156
151 130 75 32
140 140 143 110
Fig. 40.4. Accumulation of radionuclides on the sand bed, during interaction with the transformed solutions, as a function of time (t 90°C, p 3 MPa).
ELSE_TSANG_CH040.qxd
516
8/23/2005
6:40 AM
Page 516
Effect of Anthropogenic Transformations of Deep Liquid Radioactive Waste
and nickel portion, and the minor quantity of manganese, partitioned into the solid phase. Simultaneously, after 500 hours of interaction, the liquid-phase of components leached out of the sandstone bed earlier were reduced. At first, radionuclide absorption by the sandstone bed increased gradually, then the solid-phase content of americium, cesium, uranium, and neptunium was observed to increase abruptly. Strontium was preferentially taken up by clay minerals and micas at a pH of 4–5. As a result, in the low-temperature zone, the solid phase takes up the major quantities of uranium, neptunium, cesium, strontium, and plutonium that were originally present in the simulated feed. A rather sharp fall in liquid-phase radionuclide content occurs simultaneously with the precipitation of silica. The concentration of silica, taken on an SiO2 basis in the solutions studied, exceeds the solubility of quartz under the conditions given by more than an order of magnitude (Kennedy, 1950). Naturally, the processes that occur in those solutions will be directed towards re-establishment of the silica equilibrium concentration in the system. Oversaturation of the acidic solutions with silica is associated with the formation of amorphous silica during the acidic decomposition of feldspar, the former being much more soluble than quartz (Zaraisky, 1999). Formation of the amorphous silica was proved by the IR-spectroscopy data in our experiments with the acidic simulant and feldspar under hydrothermal conditions. Studies of the transformed acidic solution after contact with the rock, solid-phase centrifugation and ultrafiltration showed that as the contact time is increased from 100 to 1000 hours, the colloidal amorphous-silica particles polymerize and grow in size from 100 to 400 nm and larger. Colloidal silica polymerization and growth is fast, and is accompanied by a sharp increase in the radionuclide concentration in the solid phase (Table 40.1). The silica in the solid phase forms species capable of participating in the sorption processes in the low-temperature zone. As the wastes contain sufficient micro quantities of transuranic elements to be an environmental hazard for hundreds of thousand of years, assessment of the long-term behavior is necessary. Thermodynamic simulation of the processes that occur in the injection zone, over different periods of injection-well operation, was performed to determine basic trends in compositional change of the rocks and solutions in the deep repository. In the thermodynamic modeling, we took into account the possibility of anthropogenic geochemical zoning during acidic-waste distribution from the injection well through the injection zone. In the process of thermodynamic simulation, the real data obtained for the entire injection period were used. Thermodynamic calculations show that during the waste interaction process with the sandstone, the carbonates are dissolved (increasing the pH of the pore fluid), the feldspar content is reduced, and the clay mineral content is increased. Initially, the carbonates in the host rock are dissolved under the impact of acidic effluents (see Fig. 40.5). In the intermediate phase of injection-well operation, the dissolution of carbonates located in the zones of the long-term nitric and acetic acid existence is practically complete. The above process causes the alkaline earth components from the nitrate and acetate zones to be removed and redeposited at some distance from the injection well. Simulated changes in the calcium content of solutions transferred from the reaction zones are shown in Figure 40.6. The feldspars are converted much more slowly. Their content in the rocks decreases gradually from one injection phase to another and attains a minimum value in the acetate zone (see Fig. 40.7). Decomposition of the natural alumosilicates leads to the formation of new clay minerals: kaolinite, montmorillonites, and illites. The most intense formation of the clay minerals takes place in the high-temperature zone, i.e., in the acetate solution zone. In general, as a result of anthropogenic rock transformations in the deep repository, the clay mineral content increases
ELSE_TSANG_CH040.qxd
8/23/2005
6:40 AM
Page 517
40.3 Results and Discussion
517
Fig. 40.5. Changes in the content of carbonates in injection zone sands during the injection-well operational period.
Fig. 40.6. Changes in calcium content of solutions transferred from the reaction zones.
(see Fig. 40.8). Among other minerals, the behavior of iron oxides should be discussed separately. As the medium in the repository is oxidizing, the iron-containing minerals are represented by goethite and hematite. Accumulation of these two minerals is slightly minimal in
ELSE_TSANG_CH040.qxd
518
8/23/2005
6:40 AM
Page 518
Effect of Anthropogenic Transformations of Deep Liquid Radioactive Waste
Fig. 40.7. Changes in the content of Na–K feldspars in the injection zone sandstone.
Fig. 40.8. Changes in the content of clay minerals in the injection zone sandstone.
the nitrate zone but is very distinctive in the acetate zone. In the acetate zone, one would expect the formation of the iron-hydroxide-oxide sorption barrier that promotes the removal of radionuclides from the liquid phase. Accumulated quantities of goethite and hematite increase with time (see Fig. 40.9). In the zone of the leaching product removal and mixing
ELSE_TSANG_CH040.qxd
8/23/2005
6:40 AM
Page 519
40.3 Results and Discussion
Fig. 40.9. Changes in the content of iron oxides in the injection zone sandstone.
Fig. 40.10. Changes in the content of amorphous silica in the injection zone sandstone.
519
ELSE_TSANG_CH040.qxd
8/23/2005
520
6:40 AM
Page 520
Effect of Anthropogenic Transformations of Deep Liquid Radioactive Waste
with the formation water, no marked increase in the iron oxide content is observed when compared to the natural rock. As a result of feldspar decomposition in the high-temperature zone (acetate zone), the silica content in the pore solutions increases. These processes ensure silica-content growth in the solutions from phase to phase (see Fig. 40.10). In the zone of removal and mixing with the formation water, silica species precipitate as the amorphous silica, thus forming the sorption barrier that restricts the radionuclide migration.
40.4 CONCLUSIONS Disposal of acidic LRW into deep underground repositories leads to the formation of a nonequilibrium, geochemical system. Complex physicochemical interactions directed towards attaining a state of equilibrium developed within the system. In general, the impact of acidic waste disposed of in the underground repository is to contribute to an increase in the rock absorption capability after the active repository operation has terminated. The said effect could explain the continuous increase in radionuclide sorption with the increased interaction time of waste with the rock under hydrothermal conditions (Figs. 40.3 and 40.4). As a result of the human-induced processes, geochemical barriers are formed. The phenomenon was proven by data obtained in the chemical and thermodynamic simulations. Experiments demonstrated the possibility of creating two barrier types. The first is associated with the formation of solid phase iron and chromium hydrous-oxide compounds; amorphous silica and the transformation products thereof form the second. Formation of both ensures the reliable localization of various radionuclides, including the most environmentally hazardous actinides.
REFERENCES Apps, J.A. and Tsang, C.-F. (Eds), 1996. Deep Injection Disposal of Hazardous and Industrial Waste: Scientific and Engineering Aspects. Academic Press, San Diego, CA. Kaimin, E.P., Zakharova, E.V., Mikerin, E.I., Kudryavtsev, E.G. and Rybal’chenko, A.I., 1996. Behavior of radionuclides in geologic formations used for underground disposal of liquid nuclear wastes. In: J.A. Apps and C.-F. Tsang (Eds), Deep Injection Disposal of Hazardous and Industrial Waste: Scientific and Engineering Aspects. Academic Press, San Diego, CA, pp. 663–668. Karpov, I., 1995. Manual of Program Complex “Selector-C,” Institute of Geochemistry SO RAS, Irkutsk. Kennedy, G.C., 1950. A portion of the system silica-water. Econ. Geol., 45(7): 629–653. Zaraisky, G.P., 1999. Conditions of non-equilibrium silicification of rocks and quartz layers formation during acidic metasomatism. Geol. Ore Deposi., 41(4): 294–307 (in Russian). Zakharova, E.V., Kaimin, E.P., Darskaya, E.N., Menyailo, K.A. and Zubkov, A.A., 2001. Part played by the physicochemical processes during the long-term storage of liquid radioactive wastes in the deep collector layers. Radiokhimiya, 43(4): 378–380 (in Russian). Zubkov, A.A., Makarova, O.V., Danilov, V.V., Zakharova, E.V., Kaimin, E.P., Menyailo, K.A. and Rybalchenko, A.I., 2002. Man-caused geochemical processes in the sand collector layers during the liquid radioactive waste injection. Geoekologiya, 2: 143–154 (in Russian).
ELSE_TSANG_CH041.qxd
8/23/2005
7:59 AM
Page 521
Chapter 41
MOLECULAR BACTERIAL DIVERSITY IN WATER AT THE DEEP-WELL MONITORING SITE AT TOMSK-7 M. Nedelkova, G. Radeva, and S. Selenska-Pobell Institute of Radiochemistry, Dresden, Germany
41.1 INTRODUCTION Investigation of bacterial communities in extreme terrestrial environments such as uranium mining wastes or other heavy metal and radionuclide contaminated sites is important for understanding the role of these communities in the biogeochemical processes of those environments. It has been reported that many bacteria can interact with toxic metals and radionuclides via oxidation (Di Spirito and Tuovinen, 1982; Lack et al., 2002), reduction (Lovley, 1993, 2002), bioaccumulation (Macaskie et al., 1992; McLean and Beveridge, 2001; Selenska-Pobell et al., 1999), and biomineralization (Douglas and Beveridge, 1998). These biotransformations lead to changes in the metals’ mobility and can strongly influence their fate in the environment (Francis, 1999; Merroun and Selenska-Pobell, 2001; Merroun et al., 2003; Selenska-Pobell, 2002). However, because of our limited knowledge of the nutrient requirements and other life necessities of most bacterial species in nature, presently only a few percent can be cultured under laboratory conditions. Fortunately, problems with culturing natural bacterial communities have been largely overcome during the last decade by the application of direct molecular approaches, such as 16S rDNA retrieval. The use of such culture-independent techniques in microbial ecology has substantially expanded the current knowledge surrounding microbial diversity and activity (Dojka et al., 2000; Hugenholtz et al., 1998a, 1998b; Ka et al., 2001; Pace, 1997; Pedersen et al., 1996a). Moreover, the cultivation of novel bacterial isolates with previously undescribed metabolic characteristics from a large variety of environments is becoming routine and is no longer particularly surprising, although it is still very exciting (Bruns et al., 2003; Cho and Giovannoni, 2004; Janssen et al., 2002; Kato et al., 1998; Liu et al., 1997; Michaud et al., 2004; Straub et al., 2001; Straus et al., 1999; Takai et al., 2001). Here, bacterial diversity was studied in water collected from the S15 monitoring well, located near to the Borehole Radioactive Waste Injection Site Tomsk-7 in Siberia, Russia. 41.2 MATERIALS AND METHODS 41.2.1 Water Sampling Monitoring well S15 corresponds to Aquifer II, which is located 290–324 m below the land surface (EC Project: FIKW-CT-2000-00105; http://www.galson-sciences.co.uk/BORIS). First, underground water from the well was pumped out by an electric pump. Then, three water samples, each with a volume of 1 L were poured into sterile glass vessels filled with nitrogen gas to keep the samples isolated from contact with the atmosphere. The biomass from the
ELSE_TSANG_CH041.qxd
522
8/23/2005
7:59 AM
Page 522
Molecular Bacterial Diversity in Water at the Deep-Well Monitoring Site at Tomsk-7
water samples was concentrated via subsequent filtration on three types of filters: one glassfiber filter with a pore size of 1.2 µm, and two nitrocellulose filters with pore sizes of 0.45 and 0.22 µm. The filters were kept frozen at –20°C for further analyses. 41.2.2 DNA Extraction Total DNA was extracted from the water samples using the method described by SelenskaPobell et al. (2001). This procedure includes direct bacterial lysis followed by purification and concentration using AXG-100 Nucleobond-type anion exchange cartridges (MacheryNagel, Düren, Germany). The DNA investigated in this study, called S15A, was recovered from the total biomass of one of the water samples, which was concentrated on the three different types of filters, mentioned above. The filters were shaken together for 1 hour at 37°C in 20 mL of buffer G3 with a pH of 8 (50 mM EDTA, 50 mM Tris-HCl, 0.5% Tween 20, 0.5% Triton X100, lysozyme 4 mg/mL) and then centrifuged. The supernatant was purified following the protocol of Machery-Nagel for extraction of bacterial genomic DNA. The obtained DNA pellet was washed with 70% ethanol, dried at room temperature, and then dissolved in 50 µL of TE buffer (10 mM Tris, 1 mM EDTA, pH 8). Using one of the other water samples from well S15, DNA was recovered separately from each of the two biomasses collected on 0.45 µm (S15B) and on 0.22 µm (S15D) filters. 41.2.3 Polyamerase Chain Reaction (PCR) Amplification The PCR amplifications of the 16S rDNA fragments from the DNA samples were carried out in a Biometra thermal cycler (Göttingen, Germany). Reaction mixtures had a final volume of 20 µL and contained 200 µm deoxynucleotide triphosphates, 2.5 mM MgCl2, 10 pmol of each primer, 1–5 ng of template DNA, and 1 U AmpliTaq gold polymerase with the corresponding 10× buffer (Perkin Elmer). The PCR primers used to amplify the 16S rRNA gene sequences were as follows: bacterial 16S7-21:F (5′-AAGAGTTTGATYMTGGCTCAG3′) and universal 16S1492-1513:R (5′-TACGGYTACCTTGTTACGACTT-3′), E. coli numbering. Prior to the amplification, the DNA was denaturated at 95°C for 3 min. This step was followed by 25 polymerization cycles consisting of three steps: 90 s at 94°C, 40 s at 55°C, and 90 s at 72°C. At the end of the reaction an extension for 20 min at 72°C was performed. 41.2.4 Construction of the 16S rDNA Libraries The amplified 16S rDNA fragments were directly cloned in E. coli using a TOPOTM–TA cloning vector (Invitrogen, Groningen, The Netherlands) following the instructions of the manufacturer. A total of 135 white colonies were randomly picked and cultured overnight at 37°C in 2 mL Luria broth (LB). The size of the 16S rDNA inserts was checked by in situ PCR using the forward M13 (-40) (5′-GTTTTCCCAGTCACGA-3′) and the reverse M13 (5′-CAGGAAACAGCTATGAC-3′) primers, followed by agarose gel electrophoresis with subsequent ethidium bromide staining. One hundred thirty clones possessing correct 16S rDNA inserts were stored as glycerol cultures at –80°C for further analysis. 41.2.5 Restriction Fragment Length Polymorphism (RFLP) Typing For screening the 16S rDNA diversity of the clone library, the PCR products were digested in parallel with three frequently cutting endonucleases: MspI, HaeIII, and RsaI (Gibco BRL;
ELSE_TSANG_CH041.qxd
8/23/2005
7:59 AM
Page 523
41.3 Results and Discussion
523
Life Technologies, Inc.; Gaithersburg, MD, USA). The digests were analyzed electrophoretically in 3.5% Small DNA Low Melt agarose gels (Biozyme, Hessisch Oldenburg, Germany). The RFLP patterns obtained were compared and grouped in 16S rDNA RFLP types. 41.2.6 Sequence Analysis One representative of each RFLP group as well as all the individual clones were sequenced. Direct sequencing of the PCR products was performed using standard bacterial 16S rDNA primers on an ABI PRISM 310 Genetic Analyser (Perkin Elmer, Foster City, CA, USA). The following program was applied: denaturation at 96°C for 2 min followed by 25 cycles of 45 s at 96°C, 15 s at 55°C, and 4 min at 60°C. The 16S rDNA sequences were compared with those available in the GenBanks by BLAST analysis. The CLUSTAL W alignment program was used for sequence alignment. Phylogenetic trees were generated based on the results of the neighbor-joining algorithm with distance analysis with Jukes-Cantor corrections according to the PHYLIP v.3.5 package (Felsenstein, 1993). The sequences were checked for the presence of chimeras by using the RDP CHECK_CHIMERA program. 41.2.7 16S rDNA Sequence Accession Numbers The 16S rDNA sequences retrieved in the S15A sample were deposited in the European Molecular Biology Library (EMBL) database under accession numbers AJ534658–AJ534692. 41.3 RESULTS AND DISCUSSION 41.3.1 Estimation of Diversity of the Cloned 16S rDNA Fragments by RFLP Typing The full-length 16S rDNA inserts of 130 clones were categorized by RFLP analysis using MspI and HaeIII enzymes. As a result, 11 groups were found that possessed identical or almost identical 16S rDNA RFLP patterns. Nineteen clones possessed inserts with individual patterns not closely related to any other in the library. One of the RFLP groups was extremely large and consisted of 85 clones. They were additionally screened with the RsaI enzyme, which also revealed identical RFLP profiles, confirming that all the clones possess the same sequence. The other RFLP groups consisted of two to five clones (see Table 41.1). 41.3.2 Sequence Analysis of the Cloned Environmental 16S rDNA Fragments Figure 41.1 and Table 41.1 represent analysis of the sequences representing the above-mentioned RFLP groups and the individual RFLP types from the 16S rDNA library of the S15A sample. Proteobacteria As evident from the results shown in Figure 41.1, the two representative sequences (S15AMN2 and S15A-MN6) from the most abundant RFLP group shared more than 99% similarity with the 16S rRNA genes of the Dechlorosoma sp. PCC. Dechlorosoma genus belongs to the Rhodocyclus group of the β-subclass of Proteobacteria and was described as consisting of (per)chlorate-reducing bacteria (ClRB) (Coates et al., 1999b). These bacteria seem to be very significant to remediation of (per)chlorate-contaminated environments because the end product of their dissimilatory (per)chlorate reduction is the harmless chloride (O’Connor and Coates, 2002). It was also reported that at the intermediate step of this
ELSE_TSANG_CH041.qxd
524
8/23/2005
7:59 AM
Page 524
Molecular Bacterial Diversity in Water at the Deep-Well Monitoring Site at Tomsk-7
Table 41.1. Affiliation of the 16S rDNA sequences of the S15A sample (1.2, 0.45, and 0.22 µm) Clone name
Accession number
No. of Closest phylogenetic relative clones (EMBL no.)
S15A-MN24
AJ534667
2
S15A-MN96 S15A-MN37 S15A-MN75
AJ534670 AJ534668 AJ534669
1 1 2
S15A-MN2 S15A-MN6 S15A-MN107 S15A-MN36 S15A-MN11 S15A-MN33
AJ534664 AJ534663 AJ534666 AJ534658 AJ534662 AJ534660
46 39 1 1 1 2
S15A-MN7 S15A-MN19 S15A-MN135 S15A-MN113 S15A-MN1
AJ534672 AJ534673 AJ534675 AJ534674 AJ534671
1 1 5 1 1
S15A-MN13
AJ534676
1
S15A-MN74 S15A-MN91 S15A-MN90 S15A-MN128 S15A-MN27
AJ534683 AJ534685 AJ534684 AJ534686 AJ534682
2 2 1 1 1
S15A-MN100 S15A-MN25 S15A-MN29 S15A-MN4
AJ534681 AJ534678 AJ534679 AJ534677
2 1 1 1
S15A-MN99
AJ534680
1
S15A-MN66
AJ534689
2
S15A-MN55
AJ534690
2
S15A-MN30 AJ534687 S15A-MN131 AJ534688
1 1
S15A-MN16
3
S15A-MN46 S15A-MN56
AJ534691
1 1
α-Proteobacteria Sinorhizobium sp. 9702-M4 (AF357225) Arsenite-oxidizing bact. BEN-5 (AY027505) Afipia broomeae F186 (U87759) Brevundimonas sp. FWC30 (AJ227796) Sphingomonas sp. BN6 (X94098) β-Proteobacteria Dechlorosoma sp. Iso1 (AF170350) Dechlorosoma sp. PCC (AY126453) Herbaspirillum sp. G8A1 (AJ012069) Delftia sp. EK3 (AJ237966) Acidovorax sp. G8B1 (AJ012071) Neisseria meningitides M7931 (AF398311) γ-Proteobacteria Pseudomonas sp. ADP (AF326383) Haemophilus segnis MCCM 00337 (AF224299) Uncultured bacterium SM2E06 (AF445725) Acinetobacter sp. V4.MO.29 (AJ244764) Acinetobacter calcoaceticus DSM30006 (X81661) δ-Proteobacteria Geobacter sp. JW-3 (AF019932) Cytophaga/Flavobacterium/Bacteroides Uncultured bacterium SHA-7 (AJ249109) Uncultured bacterium SHA-5 (AJ306736) Uncultured eubacterium WCHB1-69 (AF050545) Uncultured CFB group bacterium 8-1 (AF351234) Flavobacterium aquatile (M62797) Gram+, High G+C; Actinobacteria Uncultured bacterium KB20 (AB074931) Uncultured bacterium KB20 (AB074931) Uncultured bacterium ARFS-5 (AJ277689) Microbacterium sp. C24KA (AF287752) Microbacterium kitamiense kitami A1 (AB013920) Propionibacterium acnes #1447 (AB041617) Gram+, Low G+C Gemella haemolysans (L14326) Holophaga/Acidobacterium/Geothrix Geothrix fermentans H5 (U41563) Nitrospira group Uncultured Nitrospira sp. Clone 4-1 (AF351225) Uncultured bacterium FW118 (AF524003) Uncultured Nitrospira sp. Clone 4-1 (AF351225) Planctomycetales Uncultured bacterium MB-C2-147 (AY093482) Novel or chimeras Uncultured bacterium ZZ14AC20 (AY214201) Natronoanaerobium G-M16NWC (AJ271452)
BLAST % similarity 97 97 98 99 97 99.2 99.8 97 99 99 97 98 98 95 98 99 97 92 93 94 94 94 95 93 97 99.7 98 99 98 96 91 92.7 92.6 90 89 89
ELSE_TSANG_CH041.qxd
8/23/2005
41.3 Results and Discussion
7:59 AM
Page 525
525
Fig. 41.1. Phylogenetic tree of the 16S rDNA sequences retrieved from the S15A sample. The number of clones is shown in parentheses.
process, (per)chlorate-reducing bacteria produce extracellular O2, which can be used by hydrocarbon-oxidizing bacteria in anaerobic environments (Coates et al., 1998). The true environmental role of Dechlorosoma, however, has yet to be determined because they are ubiquitous in a broad range of environments, including pristine ones. This suggests that their
ELSE_TSANG_CH041.qxd
526
8/23/2005
7:59 AM
Page 526
Molecular Bacterial Diversity in Water at the Deep-Well Monitoring Site at Tomsk-7
distribution is not restricted by (per) chlorate availability. Recently, it was demonstrated that several (per)chlorate-reducing organisms (including Dechlorosoma species) are capable of anaerobic Fe(II) oxidation. The end product of this metabolism is amorphous Fe(III) oxide, which can immobilize heavy metals and radionuclides (Lack et al., 2002). In addition to the large group of Dechlorosoma sp. 16S rDNA specific sequences, several additional clones of the S15A library contained inserts that shared a high degree of identity with other β-Proteobacterial 16S rRNA genes. One of the sequences, S15A-MN36, showed a high degree of similarity with 16S rDNA of Delftia sp. strain EK3. The latter was isolated from a biofilm reactor capable of degrading 1,3-dichloropropene (Katsileva et al., 1999). The sequences S15A-MN107 and S15A-MN11 were related to the 16S rRNA genes of two other β-Proteobacterial strains, namely, Herbaspirillum sp. G8A1 and Acidovorax sp. G8B1. These strains were isolated from freshwater ditches and are capable of anaerobic mineralization of quaternary carbon atoms (Kniemeyer et al., 1999). Interestingly, the predominance of β-Proteobacterial sequences also was demonstrated in the samples collected from the boreholes of natural fission reactors in the Oklo region of Gabon, Africa, by Pedersen et al. (1996b). Snaider et al. (1997) reported the predominance of representatives of the Rhodocyclus group in a bacterial community found within an activated sludge of a municipal wastewater treatment plant. According to Snaider and his colleagues, members of this group are often underestimated when cultivation techniques are applied. The sequence S15A-MN33 shares an identity correlation of 97% with the 16S rRNA gene of the human pathogen Neisseria meningitides M7931 and represents a RFLP group consisting of two members. Interestingly, in recent times it was published that large nonpathogenic bacterial populations occupying various soil and water environments are phylogenetically related to different animal and human pathogens. The isolated representatives of such populations possess a high level of ecological and metabolic diversity, and their role in the natural environments has yet to be clarified (Hauben et al., 1999; Coenye et al., 2003). The second numerically predominant cluster of the library consisted of a group of five 16S rDNA cloned fragments represented by the sequence S15A-MN135 and four individual sequences, all related to γ-Proteobacterial 16S rRNA genes. The closest matching sequence (SM2E06) in the Gene Bank to the S15A-MN135 was retrieved from a travertine depositional area at Mammoth Hot Springs, Yellowstone Natural Park. The individual sequence S15A-MN7 shares a 98% identity with the 16S rRNA gene of the Pseudomonas sp. strain ADP. The latter uses atrazine as a sole source of nitrogen for growth, and, furthermore, can metabolize this compound under nongrowth conditions, which is a significant ability in the context of environmental bioremediation (Mandelbaum et al., 1995). Another individual sequence, S15A-MN19, was related to the 16S rRNA gene of Haemophilus segnis MCCM 00337 (Olsen et al., 2001). The sequences S15A-MN113 and S15A-MN1 shared an identity on the species level (i.e., more than 98%) with the Acinetobacter spp. 16S rRNA genes. Sequences closely related to the latter also were found in other subterranean environments, including different boreholes, indicating that Acinetobacter-related bacteria might be common in the groundwater (Pedersen et al., 1996a, 1996b). The closest matching sequence to S15A-MN1 was the 16S rRNA gene of A. calcoaceticus DSM30006, a strain capable of degrading polynuclear aromatic hydrocarbons (PAHs) (Mueller et al., 1997). Several sequences retrieved in sample S15 were closely related to α-Proteobacterial 16S rRNA genes of strains that are also involved in interactions with PAHs (see Table 41.1—the sequences S15A-MN75 and S15A-MN24 and their matches in the Gene Bank). The strain Sphingomons sp. BN6 was described by Nohynek et al. (1996) as PAH-degrading, whereas
ELSE_TSANG_CH041.qxd
8/23/2005
41.3 Results and Discussion
7:59 AM
Page 527
527
the strain Sinorhizobium sp 9702-M4 was characterized as transporting PAHs and toxic metals in the environment via extracellular polymers (Janecka et al., 2002). Interestingly, the next related sequence to S15A-MN24, BEN-5 represents the chemolithoautotrophic arsenite-oxidizing bacterium, recovered from a gold mine (Santini et al., 2000). The individual sequence S15A-MN37 reveals a very high identity of 99% with the 16S rRNA gene of Brevundimonas sp FWC30, which was isolated from untreated influent sewage at the Gold Bar facility in Canada (Abraham et al., 1999). The sequence S15A-MN96 shares a 98% similarity with the 16S rDNA of another member of α-Proteobacteria, Afipia broomeae F186. Only one individual sequence was found in the S15A library that matched with 97% the 16S rRNA gene of Geobacter sp. JW-3. Members of the δ-Proteobacterial family Geobacteriaceae are very effective metal reducers and were found in a large variety of environments where dissimilatory metal reduction is an important process (Anderson et al., 2003; Lovley, 2002; Nevin et al., 2003). Looking at the results presented in Figure 41.1, it is noticeable that the main part of the sequences of the S15A 16S rDNA clone library is closely related to the 16S rRNA genes of Proteobacteria, especially those of the β-Proteobacterial genus Dechlorosoma. Here an important question arises: In the retrieved populations of sample S15A, were Dechlorosomaspecies-specific 16S rDNA sequences really predominant or was their predominance in the constructed library a result of biases (described by Derakshani et al., 2001; Hansen et al., 1998) of the approach used that were a result of preferential amplification or cloning of these particular rDNA fragments? In order to answer the important question of bias surrounding the eventual “masking effect” of the “Dechlorosoma”-specific β-Proteobacterial sequences, which might have concealed the rest of the sequences of the natural bacterial populations in sample S15A, we recently analyzed a second water sample collected in parallel with the sample S15A from the same site. In contrast to sample S15A, which is the subject of this work and from which DNA was recovered from the total biomass collected on all three filters (with pore sizes of 1.2, 0.45, and 0.22 µm) (see Section 41.2), in the case of the second sample, three DNA fractions were recovered from each of the filters separately. Two 16S rDNA libraries for the DNA recovered from the filters with 0.45 and 0.22 µm pore sizes were constructed using the same primers as those used for the S15A sample. The library constructed for the DNA recovered from the biomass concentrated on the 0.45 µm filter, called S15B, was dominated by the same Dechlorosoma sp.-related sequences as those found in the S15A library but to a lesser extent (see Table 41.2). Of special interest in regard to the above question is the library constructed for the DNA recovered from the biomass collected on the 0.22 µm filter. In the 16S rDNA library of this sample, called S15D, the β-Proteobacterial sequences were not any more predominant, although they were present in the sample. Moreover, the 16S rDNA inserts of one of the RFLP groups found in the S15D library were almost identical to the 16S rRNA gene of E. coli K12. This group represented, however, only 9.8% of the total number of cloned sequences (Nedelkova and Selenska-Pobell, 2004, and to be published). About 55% of the clones of the S15D library possessed sequences that shared more than 98% identity with 16S rDNA genes of members of the Cytophaga/Flavobacterium/Bacteroides (CFB) group (see Table 41.2). Almost 14% of the S15D clones contained sequences that were very distantly related to 16S rRNA genes of an uncultured cyanobacterial group. The latter sequences were found in about 20% of the clones of the S15B library as well (detailed analyses of the S15B and S15D libraries will be soon published). From the results presented in Table 41.2, one can conclude that on one hand, the Dechlorosoma sp.-related sequences had really masked the CFB and many other 16S rDNA
ELSE_TSANG_CH041.qxd
528
8/23/2005
7:59 AM
Page 528
Molecular Bacterial Diversity in Water at the Deep-Well Monitoring Site at Tomsk-7
Table 41.2. Size of the 16S rDNA groups of sequences (given in %) found in the libraries S15A, S15B, and S15D Bacterial groups
Proteobacteria α β/Dechlorosoma sp γ δ CFB Low G+C High G+C Holophaga/Acidobacteria Nitrospira Planctomycetales Cyanobacteria Deinococcus/Thermus Termite group I OP8 OP11 TM7 Novel or chimeras
S15 clone libraries S15A: 1.2, 0.45, 0.22 µm
S15B: 0.45 µm
S15D: 0.22 µm
4.6 69/65 6.9 0.77 5.4 1.5 4.6 1.5 1.5 2.3 — — — — — — 1.5
5.1 39/28 5.8 2.2 4.4 9.5 5.1 1.4 — — 19.7 0.7 0.7 1.4 — 0.7 3.6
7.1 8.1 13.6 — 55 — 1.1 — — — 13.6 — 0.5 — 0.5 — 0.5
sequences that were not amplified or were not successfully cloned from the DNA recovered from the total biomass collected on all three filters from the S15A sample. On the other hand, it seems that the problem is not just a preferential amplification of the Proteobacteria, because in the S15D library even γ-Proteobacterial sequences were not amplified with a strong preference. We feel that the results presented in Table 41.2 indicate that the Dechlorosoma sp.-related sequences in the S15A library most probably represent a dense -Proteobacterial population. The cells of this population are either larger than 0.45 µm or they are associated with colloid particles that are larger than 0.45 µm. The main part of these bacterial cells seems, however, to be associated with colloid particles bigger than 1.2 µm. The exact size of this Dechlorosoma sp.-related population should be estimated more precisely by using real-time PCR (work in preparation in our laboratory). In addition, efforts to culture and study representatives of this species are in progress in our laboratory. Cytophaga/Flavobacterium/Bacterioides (CFB) group Seven of the clones of the S15A library possessed inserts that shared a relatively high identity (over 92%) with 16S rRNA genes of CFB representatives (see Fig. 41.1). The S15AMN91 and S15A-MN74 sequences represent two RFLP groups, each consisting of two clones. They were found to be related to the rDNA sequences SHA-5 and SHA-7 retrieved from an anaerobic 1,2-dichloropropane-dechlorinating bioreactor (Schlötelburg et al., 2000). The individual sequence S15A-MN90 of the library reveals an identity of 94% with the 16S rDNA WCHB1-69, which was found in the methanogenic zone of a hydrocarbonand chlorinated-solvent-contaminated aquifer (Dojka et al., 1998). The sequence S15AMN128 was related to CFB 16S rDNA sequence 8-1, found in coal tar waste-contaminated groundwater (Bakermans and Madsen, 2002). S15A-MN27 is the only 16S rDNA sequence
ELSE_TSANG_CH041.qxd
8/23/2005
7:59 AM
Page 529
41.3 Results and Discussion
529
in the library related to a 16S rRNA gene of a cultured CFB bacterium—Flavobacterium aquatile. As mentioned above, in contrast to the S15A library, the S15D library was dominated by clones containing CFB inserts (see Table 41.2). These inserts shared a very high identity on the level of species with several cultured Flavobacteria-related strains (Nedelkova and Selenska-Pobell, 2004, and to be published). The members of the CFB group possess a fascinating ability to change their size according to the nutrient conditions of their environment. In oligotrophic environments they were found in the form of smallsized ultra-microbacteria (Giuliano et al., 1999; Lebaron et al., 2001). Interestingly, the representatives of CFB as well as some α-Proteobacteria are able to rapidly increase their cell size and their growth rates in relation to the enrichment of the environment with nutrient sources (Lebaron et al., 2001). The results presented in Table 41.2 indicate that at the S15 site, populations of small-sized bacteria occur and that they are most probably dominated by the members of the CFB group. Actinobacteria class The number of sequences of the S15A clone library related to 16S rRNA genes of Grampositive bacteria with high G+C content was relatively high, especially Actinobacteria. Most of the retrieved rDNA sequences matched 16S rDNA sequences in the Gene Bank of notyet-cultured bacteria (see Fig. 41.1). The S15A-MN25 and S15A-MN100 sequences shared about a 94% identity with the rDNA sequence KB20, which is found in oil-contaminated groundwater (Watanabe et al., 2002). The S15A-MN99 sequence was almost identical to the 16S rRNA genes of Propionibacterium acnes, whereas S15A-MN29 was related to the ARFS-5 sequence retrieved from rice paddy soils (Lüdemann and Conrad, 2000). S15AMN4 was related to the 16S rRNA genes of an oral plaque bacterial isolate Microbacterium sp. C24KA, but shared high identity also with those of nonpathogenic actinobacterial strains found in wastewaters of a sugar-beet factory (Matsuyama et al., 1999) and in soil environmental samples (Valinsky et al., 2002). Interestingly, we were able to culture from the biomass collected on one of the 0.22 µm filters three microdiverse isolates—S15D-Iso2, S15D-Iso4, and S15D-Iso5—which were affiliated by more than a 99.5% identity with the species Microbacterium oxydans (see Fig. 41.2). As shown in the figure, all three isolates are extremely small but have slight morphologic differences. Their 16S rRNA genes possess only a few mismatches, in particular in the regions that may be connected to the observed higher growth rates of the isolates S15D-Iso2 and S15D-Iso5, when compared with those of the S15D-Iso4 isolate (to be published). Such examples of microdiversity among strains of the same species were demonstrated for many environmental bacterial populations (Moore
Fig. 41.2. Bacterial isolates S15D-Iso2, S15D-Iso4, and S15D-Iso5.
ELSE_TSANG_CH041.qxd
530
8/23/2005
7:59 AM
Page 530
Molecular Bacterial Diversity in Water at the Deep-Well Monitoring Site at Tomsk-7
et al., 1998; Prüß et al., 1999; Ruis et al., 2001; Selenska-Pobell et al., 2001). The 16S rDNA genes of such strains possess specific short sequence stretches that are characteristic for genetically distinct populations, which were adapted for optimal growth under different environmental conditions. It seems that in environments such as those studied here, a Siberian radioactive waste depository, which has been subjected to extreme environmental changes, the phenomenon of microdiversity is widespread, indicating that the bacterial populations there possess extremely high plasticity and adaptability. One sequence from the S15A library, S15A-MN66, shared an identity of over 98% with the 16S rRNA genes of the species Gemella haemolysans, which is a Gram-positive bacterium with low G+C content. Two individual sequences, S15A-MN131 and S15A-MN30, were distantly related to the 16S rRNA genes of the chemolithoautotrophs of the Nitrospira group. The closest matching sequence in the Gene Bank, called Nitrospira sp 4-1, was retrieved from the uncontaminated area of the already mentioned coal tar waste-contaminated groundwater (Bakermans and Madsen, 2002). The insert of the clone S15A-MN55 was related with 96% identity to the 16S rDNA of Geothrix fermentans, from the Holophaga/Acidobacterium phylum. This bacterium was isolated from petroleum-contaminated aquifers and was characterized as an Fe(III) reducing organism (Coates et al., 1999a). The low number of Holophaga/Acidobacterium specific sequences was somewhat surprising. We suppose, however, that this finding was not a result of biases of our method because, using the same approach, we were able to find highly abundant representatives of this phylum in soil and sediment samples from other extreme environments, such as some uranium mining waste piles (Selenska-Pobell et al., 2001, 2002). Three of the sequences in the S15A library share a low identity of 89% with the 16S rDNA of the uncultured Planctomycete bacterium MB-C2-147, retrieved at a depth of about 300 m in the deep-marine sediments in a Forearc basin (Reed et al., 2002). Two sequences, S15A-MN-46 and S15A-MN56, seem to represent novel 16S rRNA genes or chimera formations (see Table 41.1). At the moment, we are not able to decide their status, and for this reason they have not yet been submitted to the Gene Bank. Bearing in mind that only a few percent of the terrestrial bacteria have been cultured and characterized under laboratory conditions, and also the fact that horizontal transfer of 16S rDNA-specific sequences has been demonstrated for several groups of bacteria (Eardly et al., 1995; Martinez-Murcia et al., 1992; Sneath, 1993), it is sometimes difficult to discriminate between artifacts and “genetically rearranged” 16S rRNA genes that possibly represent novel bacterial lineages. A nice example is the 16S rRNA gene structure of the ferro-oxidizing chemolitoautotroph Leprospirillum ferrooxidans (Lane et al., 1992). These genes, containing stretches related to 16S rDNA of δ-Proteobacteria and of Gram-positive bacteria, possess a chimera-like structure. If they were only found in an environmental 16S rDNA clone library, they would probably be considered as chimerically up to date.
41.4 CONCLUSIONS Results presented in this study indicate that the structure of the bacterial communities at the S15 monitoring site is rather complex and diverse. Interestingly, about 75% of the sequences of the S15A 16S rDNA library shared extremely high identity at the level of species with the 16S rRNA genes of already cultured bacterial strains. About 8% of the
ELSE_TSANG_CH041.qxd
8/23/2005
References
7:59 AM
Page 531
531
sequences were not so closely related but shared more than 92% of identity with genes of cultured bacteria. About 17% of the retrieved 16S rDNA sequences in the S15A library were related to sequences of not-yet cultured bacteria. The presence of extremely large numbers of β-Proteobacterial 16S rDNA sequences in this library that were highly related to the 16S rDNA genes of Dechlorosoma sp is an indication that a Dechlorosoma-like population was predominating the water of the S15 monitoring well. The latter, however, masked the presence of a significant part of the members of the natural bacterial community when total DNA was analyzed, as was the case with the S15A library. This problem can be overcome by parallel analyses of DNA samples recovered separately from different particles fractionated from the water sample via consequent filtration through filters with different pore sizes. Many of the retrieved 16S rDNA sequences were related to the 16S rRNA genes of bacterial strains that are involved in different interactions with metals, such as Geothrix fermentans H5, Sinorhizobium sp 9702-M4, Geobacter sp JW-3, the isolate BEN-5, etc. We suppose that these sequences represent strains that might also be involved in metal/radionuclide migration processes. However, one should not overestimate the results obtained via direct molecular analysis, as the 16S rRNA genes in some bacterial groups might not always reflect the evolution of their genomes (Fox et al., 1992; Stackebrandt and Goebel, 1994). On the other hand, even if some bacterial isolates are successfully cultured and studied under laboratory conditions, one should be careful when making conclusions about their behavior in the much more complex natural environments (Minz et al., 1999). Culturing bacteria from heavy-metal-polluted and other extreme environments is, however, very important because it gives one the opportunity to study novel, not-yet-described metabolic pathways and properties of life that can help in understanding the biogeochemical processes on the Earth (Lovley, 2002; Straus, 1999). In addition, some of the isolates can serve as templates for developing in situ bioremediation procedures, as in the case of the uranium mining waste isolate Bacillus sphaericus JG-A12, which was used for construction of biological ceramics for cleaning the drain waters of uranium wastes (Raff et al., 2003).
ACKNOWLEDGMENT This work was partly supported by grant FIKW-CT-2000-00105 (BORIS) from the European Community. We thank A. Zubkov, E. Zaharova, E. Kamenev, and A. Rybalchenko for their cooperation and for supplying the water samples from the S15 monitoring site.
REFERENCES Abraham, W., Strömpl, C., Meyer, H., Lindholst, S., Moore, E., Christ, R., Vancanneyt, M., Tindall, B., Bennasar, A., Smith, J. and Tesar, M., 1999. Phylogeny and polyphasic taxonomy of Caulobacter species. Proposal of Maricaulis gen. nov. with Maricaulis maris (Poindexter) comb. nov. as the type species, and emended description of the genera Brevundimonas and Caulobacter. Int. J. Syst. Bacteriol., 49: 1053–1073. Anderson, R.T., Vrionis, H.A., Ortiz-Bernad, I., Resch, C.T., Lonh, P.E., Dayvault, R., Karp, K., Marutzky, S., Metzler, D.R., Peacock, A., White, D.C., Lowe, M. and Lovley, D.R., 2003. Stimulating the in situ activity of Geobacter species to remove uranium from the
ELSE_TSANG_CH041.qxd
532
8/23/2005
7:59 AM
Page 532
Molecular Bacterial Diversity in Water at the Deep-Well Monitoring Site at Tomsk-7
groundwater of a uranium contaminated aquifer. Appl. Environ. Microbiol., 69: 5884–5891. Bakermans, C. and Madsen, E., 2002. Diversity of 16S rDNA and naphthalene dioxygenase genes from coal-tar-waste-contaminated aquifer waters. Microbial Ecol., 44: 95–106. Bruns, A., Hoffelner, H. and Overmann, J., 2003. A novel approach for high throughput cultivation assays and the isolation of planctonic bacteria. FEMS Microbiol. Ecol., 45: 161–171. Cho, J.C. and Giovannoni, S.J., 2004. Cultivation and growth characteristics of a diverse group of oligotrophic marine Gammaproteobacteria. Appl. Environ. Microbiol., 70: 432–440. Coates, J., Bruce, R. and Haddock, J., 1998. Anoxic bioremediation of hydrocarbons. Nature, 396: 730. Coates, J., Ellis, D., Gaw, C. and Lovely, D., 1999a. Geothrix fermentans gen. nov., sp. nov., a novel Fe(III)-reducing bacterium from a hydrocarbon-contaminated aquifer. Int. J. Syst. Bacteriol., 49: 1615–1622. Coates, J., Michaelidou, U., Bruce, R., O’Connor, S., Crespi, J. and Achenbach, L., 1999b. Ubiquity and diversity of dissimilatory (per)chlorate-reducing bacteria. Appl. Environ. Microbiol., 65: 5234–5241. Coenye, T., Goris, J., De Vos, P., Vandamme, P. and LiPuma, J., 2003. Classification of Ralstonia pickettii-like isolates from environment and clinical samples as Ralstonia insidiosa sp. nov. Int. J. Syst. Evol. Microbiol., 53: 1075–1080. Derakshani, M., Lukow, T. and Liesack, W., 2001. Novel bacterial lineages at the (sub) division level as detected by signature nucleotide-targeted recovery of 16S rRNA genes from bulk soil and rice roots of flooded rice microcosms. Appl. Environ. Microbiol., 67: 623–631. Di Spirito, A.A. and Tuovinen, O.H., 1982. Kinetics of uranous ion and ferrous iron oxidation by Thiobacillus ferrooxidans. Arch. Microbiol., 133: 33–37. Dojka, M., Hugenholtz, P., Haack, S. and Pace, N., 1998. Microbial diversity in a hydrocarbon- and chlorinated-solvent-contaminated aquifer undergoing intrinsic bioremediation. Appl. Environ. Microbiol., 64: 3869–3877. Dojka, M.A., Harris, J.K. and Pace, N.R., 2000. Expanding the known diversity and environmental distribution of an uncultured phylogenetic division of bacteria. Appl. Environ. Microbiol., 66: 1617–1621. Douglas, S. and Beveridge, T.J., 1998. Mineral formation by bacteria in natural microbial communities. FEMS Microbiol. Ecol., 26: 79–88. Drancourt, M., Bollet, C., Carlioz, A., Martelin, R., Gayral, J.P. and Raoult, D., 2000. 16S ribosomal DNA sequence analysis of a large collection of environmental and clinical unidentifiable bacterial isolates. J. Clin. Microbiol., 38(10): 3623–3630. Eardly, B.D., Wang, F.S., Whittam, T.S. and Selander, R.K., 1995. Species limits of Rhizobium populations that nodulate the common bean (Phaseolus vulgaris). Appl. Environ. Microbiol., 61: 507–512. Felsenstein, J., 1993. PHYLIP, (Phylogenetic Inference Package) Version 3.5c. Department of Genetics, University of Washington, Seattle. Fox, G., Wisotzkey, J. and Jurtshuk, P., 1992. How close is close: 16S rRNA sequence identity may not be sufficient to guarantee species identity. Int. J. Syst. Bacteriol., 42: 166–170. Francis, A.J., 1999. Bioremediation of radionuclides and toxic metal contaminated soils and wastes, In: Bioremediation of Contaminated Soils. Agronomy Monograph no. 37, American Society of Agronomy, Crop Science Society of America, Soil Science Society of America, Madison, WI.
ELSE_TSANG_CH041.qxd
References
8/23/2005
7:59 AM
Page 533
533
Giuliano L., De Domenico, M., De Domenico, E., Höfle, M.G. and Yakimov, M.M., 1999. Identification of culturable oligotrophic bacteria within naturally occurring bacterioplankton communities of the Ligurian Sea by 16S rRNA sequencing and probing. Microbial Ecol., 37: 77–85. Hansen, M.C., Tolker-Nielsen, T., Giskov, M. and Molin, S., 1998. Biased 16S rDNA PCR amplification caused by interference from DNA flanking the template region. FEMS Microbiol. Ecol., 26: 141–149. Hauben, L., Vauterin, L., Moore, E.R.B., Hoste, B. and Swings, J., 1999. Genomic diversity of the genus Stenotrophomonas. Int. J. Syst. Bacteriol., 49: 1749–1760. Hugenholtz, P., Goebel, B. and Pace, N., 1998a. Impact of cultured-independent studies on the emerging phylogenetic view of bacterial diversity. J. Bacteriol., 180(18): 4765–4774. Hugenholtz, P., Pitulle, C., Hershberger, K.L. and Pace, N.R., 1998b. Novel division level bacterial diversity in a Yellowstone hot spring. J. Bacteriol., 180: 366–376. Janecka, J., Jenkins, M., Brackett, N., Lion, L. and Ghiorse, W., 2002. Characterization of a Sinorhizobium isolate and its extracellular polymer implicated in pollutant transport in soil. Appl. Environ. Microbiol., 68: 423–426. Janssen, P.H., Yates, P.S., Grinton, B.E., Taylor, P.M. and Sait, M., 2002. Improved culturability of soil bacteria and isolates in pure culture of novel members of the divisions Acidobacteria, Actinobacteria, Proteobacteria, and Verrucomicrobia. Appl. Environ. Microbiol., 68: 2391–2396. Ka, O.J., Yu, Z. and Mohn, W.W., 2001. Monitoring the size and metabolic activity of the bacterial community during biostimulation of fuel-contaminated soil using competitive PCR and RT-PCR. Microbial Ecol., 42: 267–273. Kato, C., Li, L., Nogi, Y., Nakamura, Y., Tamaoka, J. and Horikoshi, K., 1998. Extremely barophilic bacteria isolated from the Mariana Trench, Challenger Deep at a depth of 11,000 meters. Appl. Environ. Microbiol., 64: 1510–1513. Katsileva, E., Bonse, D., Krüger, A., Strömpl, C., Livingston, A. and Wittich, R., 1999. An extractive membrane biofilm reactor for degradation of 1,3-dichloropropene in industrial waste water. Appl. Microb. Biotechnol., 52: 853–862. Kniemeyer, O., Probian, C., Rossello-Mora, R. and Harder, J., 1999. Anaerobic mineralization of quaternary carbon atoms: Isolation of denitrifying bacteria on dimethylmalonate. Appl. Environ. Microbiol., 65: 3319–3324. Lack, J., Chaudhuri, S., Kelly, S., Kemner, K., O’Connor, S. and Coates, J., 2002. Immobilization of radionuclides and heavy metals through anaerobic bio-oxidation of Fe(II). Appl. Environ. Microbiol., 68: 2704–2710. Lane, D.J., Harrison, A.P., Stahl, D., Pace, B., Giovannoni, S.J., Olsen, G.J. and Pace, N.R., 1992. Evolutionary relationships among sulfur- and iron-oxidizing Eubacteria. J. Bacteriol., 174: 269–278. Lebaron, P., Servais P., Troussellier, M., Courties, C., Muyzer, G., Bernard, L., Schäfer, H., Pukall, R., Stackebrandt, E., Guindulain, T. and Vives-Rego, J., 2001. Microbial community dynamics in Mediterranean nutrient-enriched seawater mesocosms: Changes in the genetic diversity of bacterial populations. FEMS Microbiol. Ecol., 34: 255–266. Liu, S.V., Zhou, J., Zhang, C., Cole, D.R., Gajdarziska-Josifovska, V. and Phelps, T.J., 1997. Thermophilic Fe(III)-reducing bacteria from the deep subsurface: The evolutionary implications. Science, 277: 1106–1109. Lovley, D.R., 1993. Dissimilatory metal reduction. Annu. Rev. Microbiol., 47: 263–290. Lovley, D.R., 2002. Dissimilatory metal reduction: From early life to bioremediation. ASM News, 68: 231–237.
ELSE_TSANG_CH041.qxd
534
8/23/2005
7:59 AM
Page 534
Molecular Bacterial Diversity in Water at the Deep-Well Monitoring Site at Tomsk-7
Lüdemann, H. and Conrad, R., 2000. Molecular retrieval of large 16S rRNA gene fragments from an Italian rice paddy soil affiliated with the class Actinobacteria. Syst. Appl. Microbiol., 23: 582–584. Macaskie, L.E., Empson, R.M., Cheetham, A.K., Grey, C.P. and Scarnulis, A.J., 1992. Uranium bioaccumulation by Citrobacter sp. as a result of enzymatically mediated growth of polycrystalline HUO2PO4. Science, 257: 782–784. Mandelbaum, R., Allan, D. and Wackett, L., 1995. Isolation and characterization of a Pseudomonas sp. that mineralizes the s-Triazine herbicide atrazine. Appl. Environ. Microbiol., 61: 1451–1457. Martinez-Murcia, A., Benlloch, S. and Collins, M., 1992. Phylogenetic interrelationships of members of the genera Aeromonas and Plesiomonas as determined by 16S ribosomal DNA sequencing: Lack of congruence with results of DNA–DNA hybridisations. Int. J. Syst. Bacteriol., 42: 412–421. Matsuyama, H., Kawasaki, K., Yumoto, I. and Shida, O., 1999. Microbacterium kitamiense sp. nov., a new polysaccharide—producing bacterium isolated from waste water of a sugar-beet factory. Int. J. Syst. Bacteriol., 49: 1353–1357. McLean, J. and Beveridge, T.J., 2001. Chromate reduction by a pseudomonad isolated from a site contaminated with chromated copper arsenite. Appl. Environ. Microbiol., 67: 1076–1084. Merroun, M., Geipel, G., Nicolai, R., Heise, K.-H. and Selenska-Pobell, S., 2003. Complexation of uranium(VI) by three eco-types of Acidithiobacillus ferrooxidans studied using time-resolved laser-induced fluorescence spectroscopy and infrared spectroscopy. BioMetals, 16: 331–339. Merroun, M. and Selenska-Pobell, S., 2001. Interactions of three eco-types of Acidithiobacillus ferrooxidans with U(VI). BioMetals, 14: 171–179. Michaud, L., Di Cello, F., Brilli, M., Fani, R., Guidice, A.L. and Bruni, V., 2004. Biodiversity of culturable psychrophilic marine bacteria isolated from Terra Nova Bay (Ross Sea, Antarctica), FEMS Microbiol. Lett., 230: 63–71. Minz, D., Fishbain, S., Green, S.J., Muyzer, G., Cohen, Y., Rittmann, B.E. and Stahl, D.A., 1999. Unexpected population distribution in a microbial mat community: Sulfate reducing bacteria localized to the highly oxic chemocline in contrast to a eukarotic preference for anoxia. Appl. Environ. Microbiol., 65: 4659–4665. Moore, L.R., Rocap, G. and Chisholm, S.W., 1998. Physiology and molecular phylogeny of coexisting Prochlorococcus ecotypes. Nature, 393: 464–467. Mueller, J., Devereux, R., Santavy, D., Lantz, S., Willis, S. and Pritchhard, P., 1997. Phylogenetic and physiological comparisons of PAH-degradating bacteria from geographically diverse soils. Antonie van Leeuwenhock, 71: 329–343. Nedelkova, M. and Selenska-Pobell, S., 2004. Bacterial and archaeal diversity in waters at the Siberian deep-well monitoring sits., Tomsk-7. FZR-400: 22. Nevin, K.P., Finneran, K.T. and Lovley, D., 2003. Microorganisms associated with uranium bioremediation in a high-salinity subsurface sediment. Appl. Environ. Microbiol., 69: 3672–3675. Nohynek, L., Nurmiaho-Lassila, E., Suhonen, E., Busse, H., Mohammadi, M., Hantula, J., Rainey, F. and Salkinoja-Salonen, M., 1996. Description of chlorophenol-degradating Pseudomonas sp. strains KF1T , KF3, and NKF1 as a new species of the genus Sphingomonas, Sphingomonas subarctica sp. nov. Int. J. Syst. Bacteriol., 46: 1042–1055. O’Connor, S. and Coates, J., 2002. Universal immunoprobe for (per)chlorate-reducing bacteria. Appl. Environ. Microbiol., 68: 3108–3113.
ELSE_TSANG_CH041.qxd
References
8/23/2005
7:59 AM
Page 535
535
Olsen, I., Dewhirst, F.E., Paster, B.J. and Busse, H.-J., 2004. Family Pasteurellaceae. In: G.M. Garrity (Ed), Bergey’s Manual of Systematic Bacteriology, 2nd edn. Springer, New York. Pace, N., 1997. A molecular view of microbial diversity and the biosphere. Science, 276: 734–740. Pedersen, K., Arlinger, J., Ekendahl, S. and Hallbech, L., 1996a. 16S rRNA gene diversity of attached and unattached bacteria in boreholes along the access tunnel to the Äspö hard rock laboratory, Sweden. FEMS Microbiol. Ecol., 19: 249–262. Pedersen, K., Arlinger, J., Hallbeck, L. and Pettersson, C., 1996b. Diversity and distribution of subterranean bacteria in groundwater at Oklo I Gabon, Africa, as determined by 16S rRNA gene sequencing. Mol. Ecol., 5: 427–436. Prüß, B.M., Francis, K.P., von Stetten, F. and Scherer, S., 1999. Correlation of 16S ribosomal DNA signature sequence with temperature-dependent growth rates of mesophilic psychrotolerant strains of the Bacillus cereus group. J. Bacteriol., 181: 2624–2630. Raff, J., Soltmann, U., Matys, S., Selenska-Pobell, S., Böttcher, H. and Pompe, W., 2003. Biosorption of uranium and copper by biocers. Chem. Mater., 15(1): 240–244. Reed, D., Fujita, Y., Delwiche, M., Blackwelder, D., Sheridan, P., Uchida, T. and Colwell, F., 2002. Microbial communities from methane hydrate-bearing deep marine sediments in a Forearc Basin. Appl. Environ. Microbiol., 68: 3759–3770. Ruis, N., Fuste, M.C., Guasp, C., Lalucat, J. and Loren, J.G., 2001. Clonal population structure of Pseudomonas stutzeri, a species with exceptional genetic diversity. J. Bacteriol., 183: 736–744. Santini, J., Sly, L., Schnagl, R. and Macy, J., 2000. A new chemolitoautotrophic arseniteoxidizing bacterium isolate from a gold mine: Phylogenetic, physiological, and preliminary biochemical studies. Appl. Environ. Microbiol., 66: 92–97. Schlötelburg, C., Wintzingerode, F., Hauck, R., Hegemann, W. and Göbel, U., 2000. Bacteria of an anaerobic 1,2-dichloropropane-dechlorinating mixed culture are phylogenetically related to those of other anaerobic dechlorinating consortia. Int. J. Syst. Evol. Microbiol., 50: 1505–1511. Selenska-Pobell, S., 2002. Diversity and activity of bacteria in uranium waste piles. In: M. Keith-Roach and F. Livens (Eds), Interactions of Microorganisms with Radionuclides. Elsevier Sciences, Oxford, UK, pp. 225–253. Selenska-Pobell, S., Flemming, K., Kampf, G., Radeva, G. and Satchanska, G., 2001. Bacterial diversity in soil samples from two uranium waste piles as determined by repARD, RISA and the 16S rDNA retrieval. Antonie van Leewenhuek, 79: 149–161. Selenska-Pobell, S., Flemming, K., Tzvetkova, T., Raff, J., Schnorpfeil, M. and Geißler, A., 2002. Bacterial communities in uranium mining waste piles and their interaction with heavy metals. In: B.J. Merkel, B. Planer-Friedrich and C. Wolkersdorfer (Eds), Uranium in the Aquatic Environment. Springer, Berlin, pp. 455–464. Selenska-Pobell, S., Miteva, V., Boudakov, I., Panak, P., Bernhardt, G. and Nitsche, H., 1999. Selective accumulation of heavy metals by three indigenous Bacillus isolates, B. cereus, B. megaterium and B. sphaericuis, in drain waters from a uranium waste pile. FEMS Microbiol. Ecol., 29: 59–67. Snaider, J., Amann, R., Huber, I., Ludwig, W. and Schleifer, K., 1997. Phylogenetic analysis and in situ identification of bacteria in activated sludge. Appl. Environ. Microbiol., 63: 2884–2896. Sneath, P.M.A., 1993. Evidence from Aeromonas for genetic crossing over in ribosomal sequences. Int. J. Syst. Bacteriol., 43: 626–629.
ELSE_TSANG_CH041.qxd
536
8/23/2005
7:59 AM
Page 536
Molecular Bacterial Diversity in Water at the Deep-Well Monitoring Site at Tomsk-7
Stackebrandt, E. and Goebel, B., 1994. Taxonomic note: A place for DNA–DNA reassociation and 16S rRNA sequence analysis in the present species definition in bacteriology. Int. J. Syst. Bacteriol., 44: 846–849. Straub, K.L., Benz, M. and Schink, B., 2001. Iron metabolism in anoxic environments at near neutral pH. FEMS Microbiol. Ecol., 34: 181–186. Straus, M., Fuest, J.A., Kramer, E.H.M., Logemann, S., Muyzer, G., van de Pas-Schoonen, K., Webb, R., Kuenen, J.G. and Jetten, K.S.M., 1999. Missing lithotroph identified as new planctomycete. Nature, 400: 446–449. Takai, K., Moser, D.P., Onstott, T.C., Spoelstra, N., Pfiffner, S.M., Dohnalkova, A. and Fredrickson, J.K., 2001. Alkaliphilus transvaalensis gen. nov., sp. nov., an extremely alkaliphilic bacterium isolated from a deep South African gold mine. Int. J. Syst. Evol. Microbiol., 51: 1245–1256. Valinsky, L., Della Vedora, G., Scupham, A.J., Alvey, S., Figueroa, A., Yin, B., Hartin, R.J., Chrobak, M., Crowley, D.E., Jiang, T. and Borneman, J., 2002. Analysis of bacterial community composition by oligonucleotide fingerprinting of rRNA genes. Appl. Environ. Microbiol., 68: 3243–3250. Watanabe, K., Kodama, Y. and Kaku, N., 2002. Diversity and abundance of bacteria in an underground oil-storage cavity. BMC Microbiol., 2(23). (http://www.biomedcentral.com/ 1471-2180/2)
ELSE_TSANG_CH042.qxd
8/23/2005
8:06 AM
Page 537
VI. INJECTION OF SOLIDS
537
This page intentionally left blank
538
ELSE_TSANG_CH042.qxd
8/23/2005
8:06 AM
Page 539
Chapter 42
INTERNATIONAL DATABASE FOR SLURRY INJECTION OF DRILLING WASTES J.A. Veil and M.B. Dusseault Argonne National Laboratory, Washington, DC, USA
42.1 INTRODUCTION Each year, thousands of new oil and gas wells are drilled in the United States and around the world. The drilling process generates millions of barrels of waste each year (primarily drill cuttings and used drilling fluids known as muds). The drilling wastes from most onshore U.S. wells are disposed of by removing the liquids from the drilling or reserve pits and then burying the remaining solids in place (called pit burial). This practice is low in cost and has the approval of most regulatory agencies. However, there are some environmental settings in which pit burial is not allowed, such as areas with high water tables. In the U.S. offshore environment, many water- and synthetic-based muds and cuttings can be discharged to the ocean if discharge permit requirements are met, but oil-based muds cannot be discharged at all. At some offshore facilities, drilling wastes must be either hauled back to shore for disposal or disposed of onsite through an injection process. Argonne National Laboratory has recently completed a review of several methods by which drilling wastes have been injected into underground formations for permanent disposal. The results of that review are documented in a report by Veil and Dusseault (2003). Examples of the methods reviewed include injection into salt caverns, injection of wastes during plugging and abandonment of wells, injection to formations at pressures lower than the formation’s fracture pressure (subfracture injection), and injection at pressures exceeding the fracture pressure (referred to as slurry injection). The report focuses on slurry injection technology, how it is conducted and monitored, the geological conditions that favor slurry injection, and its costs. One unique feature of that report is a database describing more than 330 actual slurry injection jobs from around the world. This chapter describes and summarizes the slurry injection database. Argonne also evaluated the legal and administrative requirements used by U.S. states and federal agencies to regulate slurry injection and the other types of drilling waste injection in a separate regulatory compendium (Puder et al., 2003). In an effort to promote technology transfer and public outreach, Argonne also prepared a brief but informative brochure that describes slurry injection technology in nontechnical terms for readers not interested in the other more detailed publications (Argonne, 2003). 42.2 SLURRY INJECTION TECHNOLOGY Slurry injection involves processing solid materials to make particles of suitable size and blending them with a fluid (often seawater, collected stormwater, other fresh water, used
ELSE_TSANG_CH042.qxd
540
8/23/2005
8:06 AM
Page 540
International Database for Slurry Injection of Drilling Wastes
drilling muds, or produced water, as approved by the regulatory agency) to create a slurry. The slurry is injected into a suitable confined formation at a pressure high enough to continuously fracture the receiving formation. When injection ceases, the pressure declines as the fluid bleeds off into the formation, and the solids are trapped in place of the induced fractures. The most common forms of slurry injection involve (1) annular injection, in which the waste slurry is injected through the annular space between two casing strings into the receiving strata, and (2) injection through a dedicated disposal well completed with tubing and packer, giving access to either an open hole or a perforated casing interval at the depth of an injection formation. If the well is cased, the casing must be cemented below, through, and above the proposed injection zone to ensure the waste is confined to the intended receiving zone. Slurry injection can be conducted as a single, continuous process or as a series of smaller-volume intermittent cycles. On some offshore platforms, where drilling occurs continuously and storage space is inadequate to operate in a daily batch manner, injection must occur continuously as new wells are drilled. In these cases, injection pressures are carefully monitored so operators can be aware of changes in formation injectivity and identify incipient problems. Most other injection jobs are designed to inject intermittently (i.e., inject for several hours each day, allow the injected fluids to dissipate into the formation overnight, and then repeat the cycle on the following day or a few days later). The frequency of intermittent injection cycles depends on the rate of drilling waste generation. The intermittent approach can help induce new fractures each day rather than lengthening the original fracture. This approach minimizes the likelihood that fractures will extend outside of the targeted formation and may allow for fracture storage of a larger volume of solid material.
42.3 DEVELOPMENT OF THE DATABASE One of the goals of Argonne’s evaluation of slurry injection technology was to identify as many examples of slurry injection jobs as possible and to compile information about them into a database where they could be readily compared and analyzed. The data were collected through extensive literature review and through correspondence with oil and gas producers, service companies, and the Alaska Oil and Gas Conservation Commission. The full database, which is too lengthy to include in this paper, is included as an appendix to Veil and Dusseault (2003). It includes the following information: ● Site name and location ● Name of operator and service company performing the injection ● Geology of the injection formation and confining layers ● Type of injection process ● Depth of injection formation and injection perforations/annular injection depth ● Dates and duration of injection ● Injection rate and pressure ● Types and volumes of materials injected ● Slurry properties ● Pre-injection treatment or processing ● Problems experienced ● Costs ● Other comments ● Sources of information
ELSE_TSANG_CH042.qxd
8/23/2005
8:06 AM
Page 541
42.5 Location of Slurry Injection Jobs
541
The remaining sections of this chapter summarize the data contained in the full database. The full database and the numerous references listed in Veil and Dusseault (2003) can be examined to obtain more information on specific injection jobs. Information is not available for each data element for all injection jobs, because most of the sources included only some of the desired information. Nevertheless, injection jobs with incomplete information and potentially some subfracture injection jobs were included, because the greatest value of the database is the accumulated total weight of evidence contained therein. Even though the database is incomplete, it is believed to represent the most comprehensive publicly available source of information on drilling-waste slurry injection jobs that have been conducted around the world.
42.4 NUMBER OF INJECTION JOBS Ideally, this database would include information on all injection jobs that have ever been undertaken. However, that is not practical because much of the information is either no longer available in company records, or the companies holding the data are unwilling to share it publicly. Obtaining much of this information was a significant challenge, involving several years of urging companies to share their data. Data were welcomed in almost any form as long as they contributed to the knowledge pool in the database. The full database has 334 records. Not every record carries equal weight. While this property of the database is undesirable, it is an unavoidable artifact of the original data sources. For example, there are 37 separate records for annular injection jobs in the Alpine field and 78 more annular injection jobs in the Kuparuk field on the North Slope of Alaska, with each record reflecting a single well used to inject cuttings for a period of several weeks to several months. This is an ongoing process, with more such wells anticipated as long as drilling continues in those areas. By comparison, the single record for Statoil’s injection at the Asgard platform in the North Sea represented a composite of information from nine injection wells. This information was obtained from a published paper that did not distinguish the individual results of each well. Numerous other records for offshore platforms also summarize injection that has taken place sequentially into a series of wells. Most of the injection jobs included in the database used annular injection (296 jobs— more than 88%), while 36 (11%) of the jobs used dedicated injection wells with tubing and packer. These figures reflect the large number of annular injection jobs reported for Alaska (121 jobs, or more than one-third of all the reported jobs).
42.5 LOCATION OF SLURRY INJECTION JOBS The database lists the injection jobs alphabetically by location. Although the database includes just a portion of all the injection jobs ever conducted, it offers sufficient data to indicate the trends in slurry injection practices and demonstrates that slurry injection is a worldwide waste management practice. Locations are summarized in Table 42.1. The areas with the most representation in the database are the North Slope of Alaska (129 records), the Gulf of Mexico (66 records), and the North Sea (35 records). This distribution gives a good indication of where slurry injection is being practiced. However, only published data or data supplied by companies are included in the database. The observed distribution does not necessarily indicate that areas not listed or areas showing weak
ELSE_TSANG_CH042.qxd
8/23/2005
542
8:06 AM
Page 542
International Database for Slurry Injection of Drilling Wastes Table 42.1. Locations of slurry injection jobs based on records in the database
Location
Number of records in database
Alaska Gulf of Mexico California Other U.S. onshore North Sea Canada Latin America Asia Africa U.K. onshore
136 (North Slope, 129; other 7) 66 18 28 (Louisiana, 20; Texas, 6; Oklahoma, 1; North Carolina, 1) 35 9 (Alberta, 4; Saskatchewan, 3; Nova Scotia offshore, 2) 4 (Argentina, 1, Mexico, 2, Venezuela, 1) 20 (India, 17; Indonesia, 1; Russia/Sakhalin, 1; Thailand, 1) 17 (Tunisia, 14; Egypt, 2; Chad, 1) 1
Total
334
representation do not use slurry injection. All that can be concluded is that no information was available on injection in those areas. For example, some of the major oil-producing regions or countries (e.g., Middle East, Nigeria, Far East, Australia) are not represented at all or are poorly represented.
42.6 WHO IS DOING THE INJECTION? Most slurry injection jobs have been conducted by large oil and gas companies. The operators of the wells listed in the database include many of the major international companies, or for injection jobs conducted prior to corporate mergers, their premerger component companies. Ten large multinational companies (Arco, BP, Chevron, ExxonMobil, Conoco, Marathon, Mobil, Phillips, Shell, and Unocal or their pre- or postmerger versions) account for 236 (70%) of the entries. Slurry injection has also been practiced by smaller companies in selected parts of the world. Many of the injection jobs were actually conducted by drilling or service contractors, but frequently their identity was not listed in the sources of our data.
42.7 GEOLOGICAL INFORMATION Many of the records were weak on geological information. Generally, most annular disposal wells inject into low-permeability shale or mudstone formations, and most dedicated completed injection wells inject into high-permeability sands or sandstones. However, some cases do not follow that general trend. For example, the injection at two North Sea platforms (Eldfisk and Ekofisk) used a tubing and packer configuration, yet were injected into a Hordaland Group sequence of shales and claystones. Conversely, operators at another North Sea platform, Brent, used annular injection to inject into the Hutton Sand Layer. Annular injection of drill cuttings in Cook Inlet, Alaska, is ideal in that it is necessarily done into sands or silts because shales are absent at the depths of the surface casing shoe (2000–4000 ft). Many of the records that included geological information indicated a mixed geological profile of alternating sand and shale layers. This situation allows for better vertical containment of fracture growth because the sand layers tend to bleed off the high injection pressures.
ELSE_TSANG_CH042.qxd
8/23/2005
8:06 AM
Page 543
42.10 Injection Rate
543
42.8 INJECTION DEPTH Only about 70 of the records indicated the depth at which the slurry was injected. To the extent possible, the true vertical depth was expressed. Table 42.2 shows the number of injection jobs done in different depth ranges. Most injection jobs were done at depths shallower than 5000 ft, with many falling between 2501 and 5000 ft. The shallowest injection depth reported was 1246–1276 ft at Duri, Sumatra in Indonesia, and the deepest was 15,300 ft at an onshore well at Duson, Louisiana.
42.9 DURATION OF INJECTION It is difficult to generalize about this type of information, partly because there is a wide range in injection-campaign durations, but also because of the manner in which the data were reported. Some records clearly describe short-term injection into a single well, while others offer a composite view of multiple injection wells at a single platform or location. A few of the records list only a single month in the duration column. Presumably, this means that the total injection campaign described in that record was completed within that month. Many other records list a duration of several months to several years. The longest duration reported in the database for the same well is at the THUMS facility off the coast of Long Beach, California (on a series of artificial islands). The database reports that the injection began in 1994 and is ongoing. Annular disposal through a series of wells began as early as 1984 in Alaska’s Kuparuk Field and continues today. Dedicated grind and inject operations have been conducted since at least 1990 for real-time drilling waste (Well CC-02A in western Prudhoe Bay).
42.10 INJECTION RATE The injection rate was reported in only about 90 of the records. Many of these reported a range of injection rates. Nearly all of these records indicated that the lower end of the injection range was 5 bbl/min or less, and more than half were 2 bbl/min or less. The lowest reported lower end of a range was 0.3 bbl/min at the North Sea Asgard platform, while the highest reported lower end of a range was 16.7 bbl/min at the North Slope Grind and Inject Plant at Prudhoe Bay. Most of the upper ends of the range of injection rates were less than 4 bbl/min. The lowest reported upper end of a range was 1.9 bbl/min at the Cook Inlet NCIU facility in Alaska, while the highest reported upper end of a range was 44 bbl/min at the North Slope Grind and Inject Plant at Prudhoe Bay, Alaska. Table 42.2. Distribution of injection depths in the database Depth range (ft)
Number of records in database
<2500 2501–5000 5001–7500 7501–10,000 >10,000
14 36 8 2 3
ELSE_TSANG_CH042.qxd
8/23/2005
8:06 AM
544
Page 544
International Database for Slurry Injection of Drilling Wastes
42.11 INJECTION PRESSURE Nearly 100 of the records include data on injection pressure. About half of all the records showing injection pressure were provided by Apollo Services. All of those pressure data represent surface pressures. The other pressure values reported in the database are not necessarily consistent. In some cases the reported values are surface pressures, while in other cases they are bottomhole pressures. These values will be different for the same well. Unfortunately, much of the rest of the data entered into the database did not identify which pressure was being reported, although in general it can be surmised from the injection depth hydrostatic head correction. Like the information on injection rate, many of the pressure records are expressed as ranges. Nearly all records indicated that the lower end of the injection range was 2000 pounds per square inch (psi) or less, and more than half were 1200 psi or less. The lowest reported lower end of a range was 50 psi at an onshore well at Duson, Louisiana, while the highest reported lower end of a range was about 3000 psi at the North Sea Valhall platform. Most of the upper ends of the range of injection pressures were less than 2500 psi. The lowest reported upper end of a range was 650 psi at the North Sea Block 22/25 facility, while the highest reported upper end of a range was 5431 psi at the North Slope Grind and Inject Plant at Prudhoe Bay. This value is a bottomhole pressure, and thus it includes the full hydrostatic pressure of the column of waste during injection.
42.12 TYPE AND VOLUME OF MATERIAL INJECTED Most of the wells in the database injected drill cuttings. Many also injected other types of oil field wastes—e.g., produced sands, tank bottoms, oily wastewater, pit contents, scale and sludge containing naturally occurring radioactive material (NORM). Injected volume was one of the best-reported fields in the database. All but two of the records indicated the volume of material injected, although volume was reported in different, and not always consistent ways. Some records indicated the daily or monthly volume, while others reported the total volume going into that well. Many of the injection jobs described in the database were still ongoing at the time the information was either published or sent to Argonne; therefore, the volume reported is an underestimate of the final total volume. Another way in which the reported data vary is in whether the value is total slurry or just the solids used to make a slurry—often the reported data did not clearly indicate which. Table 42.3 shows the number of records that reported volumes within specified ranges. For the purpose of this table, data are assumed to represent the total slurry volume. The data show that more than 83% of the injection jobs in the database involved less than 50,000 bbl of slurry. The largest job reported in the database is more than 43 million bbl of slurry injected in several North Slope Grind and Inject projects at Prudhoe Bay.
42.13 SLURRY PROPERTIES The percentage of solids in the injected slurry ranged from 5 to 70%, with most values lying between 10 and 26%. The specific gravity ranged from 1.03 to 1.80 g/cm3, but the majority of values were in the range of 1.15–1.50 g/cm3. The density ranged from 8.3 to 13.3 lb/gal, although most values fell within the range of 8.6–11.5 lb/gal. Viscosity ranged from 42 to 110 s/qt (Marsh Funnel viscosity), with most values falling in the range of 50–90 s/qt.
ELSE_TSANG_CH042.qxd
8/23/2005
8:06 AM
Page 545
42.15 Problems Experienced
545
Table 42.3. Distribution of total slurry volume in the database Total reported slurry volume (bbl)
Number of records in database
⬍10,000 10,000–50,000 50,001–100,000 100,001–500,000 500,001–1,000,000 ⬎1,000,000
87 206 9 13 5 12
Total
332
42.14 PRE-INJECTION PROCESSING OR TREATMENT Muds and cuttings are separated, and the cuttings are transported to the slurrying system. The cuttings are mixed with a liquid to make a slurry. At offshore locations, seawater is typically used for this purpose, and at onshore locations, a water supply must be available. The slurry is processed through various particle-size reduction devices, and viscosifiers are added as needed to generate a slurry with the desired characteristics. For those records that describe the solids processing, centrifugal pumps with hardened edges on the impeller were the most common means of reducing the particle size and blending the slurry. In some cold weather operations, like the North Slope Grind and Inject Plant at Prudhoe Bay, solids have to be thawed before they can be ground and slurried.
42.15 PROBLEMS EXPERIENCED Problems were reported in only about 30 of the records. Some of the problems were operational ones that caused the injection process to slow or stop, while others were environmental problems that led to leakage of fluids to the ground surface or the sea floor (also referred to in some records as broaching or breaching). The most common operational problem was plugging of the casing or tubing because solids had settled out. The causes of this problem included: ● Using slurries with inappropriate viscosity ● Operating at too slow an injection rate ● Failing to clear the well bore with a clean water flush at the end of an injection cycle ● Experiencing power failures that interrupted injection cycles ● Allowing pressure to drop at the end of an injection cycle, so that solids could flow back into the well bore from the formation Another important operational problem was excessive erosion of casing, tubing, and other system components. It is not surprising that metal components show wear at higher than normal rates when solid slurries are pumped through them at high pressure. Operators learned from this experience and were able to substitute tubulars with high burst strength and to use specially hardened alloys for critical parts in the pumping system. Two other operational problems involved surface handling issues. In some cases, the injection was unable to keep up with the drilling rate, and cuttings had to be stockpiled. This situation is merely inconvenient at onshore locations, but can cause drilling to stop at offshore locations when insufficient storage capacity is available. Finally, in some cases, onsite personnel added inappropriate materials to the waste stockpile. These materials either damaged solids processing equipment or created conditions not conducive to smooth operations.
ELSE_TSANG_CH042.qxd
8/23/2005
546
8:06 AM
Page 546
International Database for Slurry Injection of Drilling Wastes
For example, shredded solid wastes (presumably trash or debris) clogged screens at a Gulf of Mexico rig (East Cameron 56 JB-3). Although the operational problems are inconvenient and costly to operators who have to stop their normal activities, environmental problems are of much greater concern. Unanticipated leakage to the environment not only creates a liability to the operator, but it also generally results in short-term to permanent stoppage of injection at that site. Furthermore, whenever injection jobs result in leakage, the confidence of regulators who must approve the practice will be diminished. Several of the largest injection jobs reported have resulted in leakage. During the pilot phase, the large Grind and Inject Plant at Prudhoe Bay operated continuously for portions of 3 years. In 1997, fluids were observed broaching to the surface at multiple locations near the injection well. Injection was stopped and leaked fluids were collected for disposal. After 4 days of broaching and a total volume of 18,000 bbl, flow to the surface stopped. Several days later, low-rate injection was started to clean up the Grind and Inject Plant. No additional broaching was observed at this point. The cause of the broaching was believed to have been intersection of the injection plume with other nearby uncemented wellbores that lead to the surface. The project demonstrated that slurry injection is effective in disposing of large volumes of drilling waste, but also highlighted the need for absolute wellbore integrity. The incident occurred during the demonstration phase of the technology. The operators of the Grind and Inject Plant drilled three new dedicated injection wells designed and constructed to minimize the potential for communication of fluids. No other wells are located within 1 mile of the injection wells. Similar problems have not been experienced with the new wells. Leakage can occur at offshore sites, too. At the North Sea Asgard platform, several wells experienced leakage at the sea floor. This leakage was presumed to be caused by poor cementing jobs. In some wells, the leakage stopped after fractures were allowed to heal, but in other wells, injection was discontinued. Although this information is not included in the database because no specific data are available, the Louisiana Department of Natural Resources has reported that since 1997 they have received 13 reports of small onshore annular injection jobs that have leaked to the surface (Wascom, 2002). Although most of the reported breaches occurred near the wellhead, one case involved leakage of 500 bbl of material to the surface about 4000 ft from the wellhead. The most likely cause of these leakage events and those described in the previous paragraph is that the fracture reaches a wellbore that has not been properly cemented or plugged. Under the high downhole pressure, the fluids will seek out the pathway of least resistance. If cracks in the cement job or geological faults are present, the fluids can preferentially migrate upward and may reach the land surface or the sea floor. The long-term injection program at the THUMS facility in California has experienced pressure increases in the injection zone. Fortuitously, the injection well was planned for injection into a series of sand layers. When the pressure rose to unacceptable levels, the operators closed in the lowermost perforations, moved up the well to the next-lowest sand formation, and recompleted the well.
42.16 ECONOMICS Veil and Dusseault (2003) include an extensive discussion on the absolute costs of slurry injection and the comparative costs of slurry injection versus other drilling-waste management methods. The full database provides additional economic information.
ELSE_TSANG_CH042.qxd
8/23/2005
References
8:06 AM
Page 547
547
ACKNOWLEDGMENTS The funding to support this project was provided by the U.S. Department of Energy, Office of Fossil Energy, National Petroleum Technology Office (NPTO), under Contract W-31-109-Eng-38. John Ford is the NPTO project manager for this work. We acknowledge his support of our efforts. We further acknowledge the efforts of the many persons and companies who contributed information to the database. We appreciate the extensive comments on the draft report and additional entries for the database provided by Jim Regg and Tom Maunder of the Alaska Oil and Gas Conservation Commission.
REFERENCES Argonne, 2003. An Introduction to Slurry Injection Technology for Disposal of Drilling Wastes. (brochure) Prepared by Argonne National Laboratory for the U.S. Department of Energy, Office of Fossil Energy, National Petroleum Technology Office, 20 pp (available for downloading at: http://www.ead.anl.gov/pub/dsp_detail.cfm?PubID=1628), September. Puder, M.G., Bryson, B. and Veil, J.A., 2003. Compendium of Regulatory Requirements Governing Underground Injection of Drilling Wastes. Prepared by Argonne National Laboratory for the U.S. Department of Energy, Office of Fossil Energy, National Petroleum Technology Office, (available for downloading at: http://www.ead.anl.gov/pub/ dsp_detail.cfm?PubID=1575), February. Veil, J.A. and Dusseault, M.B., 2003. Evaluation of Slurry Injection Technology for Management of Drilling Wastes. Prepared by Argonne National Laboratory for the U.S. Department of Energy, Office of Fossil Energy, National Petroleum Technology Office (available for downloading at: http://www.ead.anl.gov/pub/dsp_detail.cfm?PubID=1584) Wascom, C.D., 2002. Annular disposal of exploration and production wastes. Presented at the Ground Water Protection Council Annual Forum, San Francisco, CA, September 22–25.
This page intentionally left blank
548
ELSE_TSANG_CH043.qxd
8/23/2005
8:09 AM
Page 549
Chapter 43
REGULATORY REQUIREMENTS AND PRACTICES GOVERNING SLURRY INJECTION OF DRILLING WASTES* M.G. Puder, J.A. Veil, and W. Bryson Argonne National Laboratory, Washington, DC, USA
43.1 INTRODUCTION When oil and gas wells are drilled, large quantities of waste are produced. The two principal types of drilling wastes include used drilling fluids (commonly referred to as muds) and drill cuttings (rock particles ground up by the drill bit). Numerous methods are employed to manage drilling wastes, including burial of drilling pit contents, land spreading, thermal processes, bioremediation, treatment and reuse, and several types of injection processes. Some oil- and synthetic-based muds are recycled. Others, however, are disposed of, as are nearly all water-based muds, generally through pit burial (onshore) and discharge (offshore). Argonne National Laboratory (Argonne) has conducted a comprehensive feasibility evaluation of using slurry injection technology to dispose of drilling wastes for the U.S. Department of Energy. The study resulted in three documents. The regulatory portion of Argonne’s feasibility evaluation is presented in the final report Compendium of Regulatory Requirements Governing Underground Injection of Drilling Wastes (Puder et al., 2003) (the Regulatory Compendium). The companion report Evaluation of Slurry Injection Technology for Management of Drilling Wastes offers a more detailed technical discussion of the mechanisms used for slurry injection (Veil and Dusseault, 2003). Finally, the brochure An Introduction to Slurry Injection Technology for the Disposal of Drilling Wastes provides a reader-friendly overview of slurry injection (Veil et al., 2003). This chapter describes the Regulatory Compendium, including the collection and organization of the pertinent information, and the findings and trends relative to the regulations and practices governing slurry injection in the United States. Slurry injection1 involves grinding solid or semisolid drilling wastes or otherwise reducing the particle size, mixing solid particles of suitable size with a fluid (often seawater or produced water) to create a slurry, and then injecting the slurry underground at a pressure high enough to fracture the receiving formation. When the pressure is reduced, the fluid bleeds off into the formation and the solids are trapped in place in the fractures. The use of the different forms of slurry injection has increased over the past decade, primarily in oil-producing * The chapter has been created by the University of Chicago as Operator of Argonne National Laboratory (“Argonne”) under Contract No. W-31-109-ENG-38 with the U.S. Department of Energy. The United States Government retains for itself, and others acting on its behalf, a paid-up, nonexclusive, irrevocable worldwide license in this contribution to reproduce, prepare derivative works, distribute copies to the public, and perform publicly and display publicly, by or on behalf of the United States Government. 1 Others refer to this process as slurry fracture injection (trademark), fracture slurry injection, drill cuttings injection, cuttings reinjection, and grind and inject, among other names.
ELSE_TSANG_CH043.qxd
550
8/23/2005
8:09 AM
Page 550
Regulatory Requirements and Practices Governing Slurry Injection of Drilling Wastes
areas such as the North Slope of Alaska and the Gulf of Mexico. In addition to slurry injection, the Regulatory Compendium surveys other injection technologies. Subfracture injection is conducted below fracture pressure. Annular injection places the drilling wastes into the space between two casing strings, known as the annulus. Other technologies involve injection of the drilling wastes back into the well bore of origin prior to plugging, disposal in abandoned salt caverns, and disposal in coal mines.
43.2 DESCRIPTION OF THE REGULATORY COMPENDIUM The Regulatory Compendium provides a comprehensive compilation of the regulatory approaches to injection processes used to dispose of drilling wastes, particularly slurry injection. The Regulatory Compendium consists of a narrative discussion of the regulatory requirements and practices for each of the oil- and gas-producing states, an overview of requirements for operations on federal lands imposed by the Bureau of Land Management and the Minerals Management Service, a table summarizing the types of injection processes authorized in each state, and an appendix with the texts of the relevant state regulations and policies. The material included in the Regulatory Compendium was primarily derived from a review of state regulations, as well as interviews with state oil and gas regulatory officials. Source texts were obtained through web-based searches and facsimile transmittals. Most state oil and gas regulatory agencies offer access to their regulations through their home pages.
43.3 FINDINGS PRESENTED IN THE REGULATORY COMPENDIUM The Regulatory Compendium presents findings that fall into several topical themes, including the federal framework for injection operations, the regulatory and economic challenges confronting slurry injection, and the results of the survey of state regulatory requirements and practices governing slurry injection and other injection technologies. In this chapter, the state survey results compiled in the Regulatory Compendium are not replicated by state, but rather summarily described by technology. 43.3.1 Federal Framework for Injection Operations With the passage of the Safe Drinking Water Act (SDWA) in 1974, the subsurface injection of fluids came under federal regulation. In 1980, the U.S. Environmental Protection Agency (EPA) promulgated the underground injection control (UIC) regulations, which govern the different types or classes of injection wells, including those associated with the production of oil and gas under Section 1422 of the SDWA. In the wake of concerns expressed by industry as well as state oil and gas regulatory agencies, Congress amended the SDWA. Section 1425 of the SDWA gave state oil and gas regulatory agencies an opportunity to demonstrate to the EPA that their long-standing oil field injection control programs were of “equivalent effectiveness” to the federal UIC regulations. Most oil and gas production states took advantage of the option provided by Section 1425 of the SDWA. The oil field injection wells have become known as UIC Class II wells. In most producing states, drill cuttings, unrecoverable drilling muds, and well completion fluids are allowed to collect in pits, lined or unlined, and are subsequently buried in place. In 1988, the EPA made a regulatory determination clarifying that oil and gas exploration and
ELSE_TSANG_CH043.qxd
8/23/2005
8:09 AM
Page 551
43.3 Findings Presented in the Regulatory Compendium
551
production (E&P) wastes were exempt from the hazardous waste management requirements under Subtitle C of the Resource Conservation and Recovery Act (RCRA). The 1988 Regulatory Determination did not explicitly articulate an administrative linkage between the RCRA and UIC programs. Yet it was acknowledged that fluids injected back into the producing reservoir for purposes of enhanced recovery would be considered a recycling activity. UIC Class II disposal wells include those wells injecting fluids that are brought to the surface in connection with natural gas storage operations or conventional oil or natural gas production. Exempt E&P wastes were deemed acceptable for injection (disposal) into a Class II well. Consistent with RCRA, the 1988 Regulatory Determination and the recently updated publication entitled Exemption of Oil and Gas Exploration and Production Wastes from Federal Hazardous Waste Regulations include lengthy lists of exempt waste materials, including “drilling fluids, produced waters, and other wastes associated with the exploration, development, or production of crude oil or natural gas or geothermal energy.” Although the term “slurry” is not specifically listed, one can reasonably assume that the materials enumerated would be present in slurries. In particular, drilling fluids and drill cuttings are most often the primary constituents of slurries. As further evidence, EPA defines “fluid” as “any material or substance which flows or moves whether in a semisolid, liquid, sludge, gas, or any other form or state.” Thus, slurried E&P wastes are both RCRA exempt and Class II in nature. 43.3.2 Regulatory and Economic Challenges Confronting Slurry Injection Over the past decade, slurry injection has proved a viable technology in the North Slope. However, in light of regulatory and economic challenges, slurry injection has not been a top option for disposal of drill cuttings and other E&P wastes in the producing areas of the Continental United States. Regulatory Challenges The EPA’s Class II UIC regulations do not prohibit slurry injection. However, the Class II regulations specify that the initiation of new fractures and the propagation of existing fractures must occur within the injection formation. Moreover, the fracturing must not extend to the confining zone or cause migration of fluids into an underground source of drinking water (USDW). The primary regulatory barrier to the use of the slurry injection option stems from the limitation on wellhead injection pressures imposed by the Class II UIC primacy agreements between the EPA and the authorized states. The Class II UIC regulations, administered under Section 1422 or Section 1425 of the SDWA, pose more of a challenge because of interpretation rather than wording of regulations or guidance. Most State-EPA program-primacy agreements were concluded prior to the development of slurry injection technology. One of the cornerstones of the UIC program is to discourage artificially applied wellhead pressure that could cause a breach in the confining bed above the injection zone or create conduits for injected fluids to migrate into USDWs. In those states that have not received primacy over the UIC program, the EPA has allowed operators to dispose of wastes at drilling site locations into the well upon completion or abandonment, as long as the injection zone was able to accept the fluid or slurry under the column of hydrostatic pressure. The EPA has indicated that it would consider applied pressure requests and slurry injection if operators furnished proof, through geologic evidence or reservoir studies, that the confining zone above the injection formation would not be breached. State-run UIC programs are at least as restrictive as direct implementation
ELSE_TSANG_CH043.qxd
552
8/23/2005
8:09 AM
Page 552
Regulatory Requirements and Practices Governing Slurry Injection of Drilling Wastes
programs. For example, in Alaska, reviews encompass detailed confinement analyses to ensure that no breach of the confining layer occurs. Moreover, Alaskan regulations require protection of all fresh water, while EPA criteria focus on the protection of USDWs. Relative to pressures exceeding the fracture pressure, both the EPA and many states retain a cautious posture. In many older producing areas for which little documentation on the locations of pre1940 oil and gas tests, core holes, and stratigraphic tests was available, produced water injection caused formation fluids to migrate up the unplugged and poorly plugged holes. An occasional phenomenon was the stair-step effect, which occurred when produced water was injected into a deeper zone under such pressure that the fluid in a nearby, unplugged hole rose and triggered overpressurization of an artesian aquifer closer to the surface. Occasionally, several confined sand intervals at different depths above the injection zone were involved in the upward migration. As a result, unwanted flows of mineralized water into surface streams or unconfined aquifers along the subcrop have occurred. In Louisiana, several cases involving breaches of injected slurries to the surface have been observed. The breaches occurred during reserve pit annular injection operations. Louisiana responded to this problem by developing new, more comprehensive regulations for annular injection. In light of such events, many regulatory agencies reduce allowable wellhead injection pressures to between 0.5 and 0.9 pounds per square inch per foot (psi/ft) of depth. The pressures necessary for the injection of slurry regularly exceed these standards. Economic Challenges At present, operational costs appear to pose a major barrier to the extended use of slurry injection technology at most of the drilling locations in mature onshore producing areas, such as Kansas, Oklahoma, and the Appalachian states. Interviews revealed that several state regulators were surprised that independent operators would want to use slurry injection. Burial of E&P wastes and proper closure of drilling and reserve pits are cheaper than slurry injection, because facilities to reconstitute the drill cuttings into a true slurry texture are not needed. In certain offshore areas where discharges are prohibited or drilling wastes will not pass the required standards of a National Pollutant Discharge Elimination System (NPDES) permit for discharge to the marine environment, operators of offshore production sites are required to transport the materials to a shore location for treatment and disposal. Thus, in the Gulf of Mexico, several commercial waste disposal companies have established networks of facilities to receive wastes from offshore and transport them in bulk to centralized disposal sites. The costs of this service are moderate enough that most operators in the Gulf of Mexico have not chosen to employ on-site slurry injection. In Cook Inlet, the NPDES permit for produced water and water-based muds and cuttings allows discharge to the marine environment. Oil-based drilling waste is disposed of by slurry injection at platforms where a Class II well or annulus is available, or it is hauled to a platform with a Class II well. 43.3.3 Results of the Survey on State Regulatory Requirements and Practices Governing Slurry Injection and Other Injection Technologies State regulation of slurry injection and other injection technologies differs widely. Regulatory vehicles include general UIC Class II rules, specific rules tailored to a type of injection or disposal process, or policy. Some states use experimental practice permits. When regulations feature well-of-origin limitations, only the waste generated from drilling
ELSE_TSANG_CH043.qxd
8/23/2005
8:09 AM
Page 553
43.3 Findings Presented in the Regulatory Compendium
553
the original well may be injected. In certain states, requests for the application of slurry injection and other injection technologies have never been made. Slurry injection The disposal well alternative of the process involves injection to either a section of the drilled hole that is below all casing strings, or to a section of the casing that has been perforated with a series of holes at the depth of an injection formation (see Fig. 43.1, disposal well, top). Slurry injection is currently permitted on a regular basis in Alaska, Texas, and California. California often uses the experimental project approach. Wyoming and Oklahoma allow injection pressures to exceed the fracture pressure; however, the amount of fluid or slurry injectate is restricted to a single-well situation. In those cases, only the waste from the well of origin may be reinjected. Louisiana has recently finalized specific rules for Disposal of E&P Wastes by Slurry Fracture Injection. In the past, Louisiana allowed slurry injection as an experimental technology on a case-by-case basis. In general, regulation is based on the supposition that an applicant can obtain permission to use slurry injection technology if geologic conditions and other factors confirm that the confining beds will not be breached and that USDWs will not be threatened. The regulatory availability of slurry injection appears to coincide with the situation in U.S. areas that exhibit environmental, geological, or hydrogeological conditions precluding in situ burial of drilling wastes in pits. Examples include tundra or shallow water tables. Oil and gas regulatory officials from the mid-continent states signaled minimal interest in slurry injection. A
Fig. 43.1. Slurry injection through disposal well and annular injection.
ELSE_TSANG_CH043.qxd
554
8/23/2005
8:09 AM
Page 554
Regulatory Requirements and Practices Governing Slurry Injection of Drilling Wastes
few state regulators expressed concerns that formation fractures could develop in the confining beds without it being possible to ascertain the route of fluid migration. Subfracture injection This technology injects below fracture pressure. Most oil- and gas-producing states exercise their regulatory authority to allow disposal of drilling fluids and wastes by injection back into their wells of origin. In the context of wells approved for annular injection, most state agencies limit the volume of injectate by restricting the activity to a one-time disposal process. Only Indiana, by policy, prohibits any injection of drilling wastes except for those that can be used for spacers in well plugging operations. Several states indicated that state water protection agencies and water planners would be skeptical about applying the injection pressure necessary to fracture the target formation—even without fracturing any confining beds. None of the surveyed states, except Texas, indicated that dedicated wells were being used for the injection of drilling wastes at subfracture injection pressures. One commercial disposal company located in eastern Texas received permission to inject tens of millions of barrels of offshore waste into naturally fractured cap rock on the flanks of a salt dome. The injection pressures are low, and, on some occasions, the waste is drawn into the formation under a vacuum. Annular injection This technology represents the alternative to the disposal well method and involves the placement of the slurry into the annulus between two casing strings at pressures above or below fracture pressure. At the lower end of the outermost casing string, the slurry enters the formation (as shown in Fig. 43.1, annular injection, bottom). Annular injection of drilling wastes is allowed by a significant percentage of state oil and gas regulatory agencies. Annular injection is governed by regulation and policy. Several states, including Kansas, Mississippi, Alabama, Illinois, Louisiana, Texas, and Alaska, permit annular injection of drilling waste. Some states, like Alaska, which uses the term “annular disposal,” regulate annular injection of drilling waste distinctly and separately from underground injection. Annular injection is addressed independently from the UIC program as an operation incidental to the drilling of a well; it is therefore not an operation subject to UIC. Other states, like Alabama, use UIC Class II regulations. Except for Alaska, the states and the EPA limit the volume of injected waste to include only the drilling wastes generated by the well of origin. The State of Alaska allows wastes from other well locations to be brought in but requires documentation on these wastes, including their point of origin, description, and date(s) of generation. Annular injection is permitted only in wells that have surface casing and cementing to a specified depth below the base of the lowermost USDW. That depth ranges from 200 to 500 ft below the USDW (and varies by state). The extent of protected interval depends more upon the state’s disposition. Some states may favor limited annular injection. Most states and the EPA view annular injection as a short-term event, with a permitted disposal window ranging from 30 to 120 days. This time frame depends upon the state regulations governing closure of the reserve pit or upon the official completion date of the well. By comparison, Alaska limits the injectate volume to 35,000 barrels and limits the actual injection to 90 days total to be completed within 1 year from initiating the operation. States have different standards for establishing the approved injection pressure magnitude. The location of the pressure measurement is not always clearly defined. The pressure
ELSE_TSANG_CH043.qxd
8/23/2005
43.4 Conclusions
8:09 AM
Page 555
555
may be measured at the wellhead or at the bottom of the well. The bottomhole pressure will always be larger, since it includes the pressure caused by the weight of the fluid column and other pressures associated with frictional losses. Alabama’s approved maximum injection pressures are based on 90% of the test pressure applied to demonstrate mechanical integrity. In Alaska, the operator is required to determine the strength of the formations exposed below the surface-casing shoe. The allowed pressure for “annual disposal” must not exceed the downhole pressure obtained during the formation integrity test conducted below the outer casing shoe, unless a higher pressure is specifically approved. In North Dakota, the guidelines for “true annular disposal” provide that the disposal pressure cannot exceed 75% of the pressure test on surface casing, which must be pressure tested to 80% of the burst pressure rating after drilling is completed and before the production casing is run. If the surface casing pressure test fails, no injection is allowed. Plugging and abandonment Several states authorize the injection of drilling pit wastes and reserve pit wastes back into the well of origin prior to abandonment. Some states, including Texas, Oklahoma, and Wyoming, allow pressures exceeding the fracture pressure, while several others use the upper pressure limit allowed for injection of produced water under the Class II UIC program. No across-the-board standard exists for the maximum allowable pressure; individual state limitations range from 0.5 to 0.9 psi/ft of depth. Most states do not consider this category separate from the subfracture or fracture injection processes described above. They view this process as a part of plugging and abandonment rather than injection. Salt cavern disposal In the United States, disposal of drilling waste into salt caverns is currently permitted only in Texas, although Louisiana has recently finalized its cavern disposal regulations in the document Disposal of Exploration and Production Waste in Solution-Mined Salt Caverns. Cavern disposal operations in Texas appear to be successful. All of the currently permitted caverns may receive E&P wastes, including drilling wastes. Other states, including Alabama, Kansas, Michigan, Mississippi, New York, and Oklahoma, have considerable salt deposits, which have been used for solution mining of salt and for cavern storage of hydrocarbons. None of these states, however, has seriously considered the use of salt caverns for waste disposal. Disposal in coal mines Old coal mines have been used in some instances for disposing of solid wastes, such as fly ash. Several states have coal bed methane (CBM) or oil production in the areas with coal deposits. Because of the recent surge in CBM drilling and production, state officials were surveyed to learn if any drilling wastes are being disposed of in coal mines. According to oil and gas regulators, however, coal mines have never been used for the disposal of oil field or E&P wastes except in Virginia. This practice has been directly approved on a case-by-case basis by the EPA under Section 1422 of the SDWA.
43.4 CONCLUSIONS The regulation of slurry injection and other injection technologies varies greatly by state. Independent of regulatory considerations, slurry injection may not be the preferred option
ELSE_TSANG_CH043.qxd
556
8/23/2005
8:09 AM
Page 556
Regulatory Requirements and Practices Governing Slurry Injection of Drilling Wastes
of most practicing independent oil and gas operators, because of the costs involved in reconstituting the drill cuttings into a true slurry texture. Other, less expensive injection technologies, including one-time-event subfracture injection and annular injection of drilling pit fluids, seem more favored. Moreover, these practices appear more appealing to the regulatory agencies. Yet states may impose regulatory limitations on annular injection operations. The restrictions generally pertain to the volume of the injectate and the timeframe of the injection activity. Proper design and management of annular injection are critical to ensure that the fluids are contained in the receiving zone.
ACKNOWLEDGMENTS The funding to support this project was provided by the U.S. Department of Energy, Office of Fossil Energy, National Petroleum Technology Office (NPTO), under Contract W-31-109-Eng-38. John Ford is the NPTO project manager for this work. We acknowledge his support for our efforts. We appreciate the time, information, comments, and other courtesies offered by federal and state oil and gas regulators.
REFERENCES Puder, M.G., Bryson, B. and Veil, J.A., 2003. Compendium of Regulatory Requirements Governing Underground Injection of Drilling Wastes. Prepared by Argonne National Laboratory for the U.S. Department of Energy, Office of Fossil Energy, National Petroleum Technology Office, February 2003. Veil, J.A. and Dusseault, M.B., 2003. Evaluation of Slurry Injection Technology for Management of Drilling Wastes. Prepared by Argonne National Laboratory for the U.S. Department of Energy, Office of Fossil Energy, National Petroleum Technology Office, May 2003. Veil, J.A, Puder, M.G., Dusseault, M.B. and Emge, F., 2003. An Introduction to Slurry Injection Technology for Disposal of Drilling Wastes. Prepared by Argonne National Laboratory for the U.S. Department of Energy, Office of Fossil Energy, National Petroleum Technology Office, September 2003. Public Health Service Act, Title XIV (also known as the Safe Drinking Water Act), as amended, 42 U.S.C. §§300f to 300j-26 (2005); for the underground injection portions of the law, see 42 U.S.C. §§300h to 300h-8 (2005); the EPA’s implementing regulations are codified at 40 C.F.R. Parts 144 to 148.
ELSE_TSANG_CH044.qxd
8/23/2005
12:34 PM
Page 557
Chapter 44
ALASKAN UNDERGROUND INJECTION CONTROL OF SOLID WASTE DISPOSAL* T. Cutler U.S. Environmental Protection Agency, Seattle, WA, USA
44.1 INTRODUCTION Underground injection control (UIC) is a crucial part of the environmentally successful management of oil fields in the sensitive Arctic. The oil/gas/brine/mud mixture extracted from the subsurface yields large quantities of gas, along with solid and liquid wastes, that must be injected deep below the permafrost. For every barrel of crude oil transported south in the Trans-Alaska Pipeline (TAPS), over three barrels of brine and over 3 million ft3 of gas are injected. Since 1996, solid waste that requires no grinding at the Prudhoe Bay Pad 3 Class I well facility has been injected into formations below permafrost at an average rate of nearly 1 million barrels per year. During that same time period, solid waste that requires grinding at the Prudhoe Bay Surfcote Pad Class V Rule authorized well facility (formerly permitted as a Class II well) has been injected into formations below permafrost at an average rate of 400,000 yd3 per year (Collver, 2003). Waste transport is expensive and introduces an added potential for surface spills onto Arctic tundra. Year-round connecting roads are limited to the Kuparuk and Prudhoe Bay fields, and ice roads are available for only several months each spring. Surface transportation across the sparsely populated Arctic in winter is less damaging to the sensitive Arctic environment when the tundra, lakes, and streams are frozen. However, the sensitive surface tundra wetlands, with thaw penetrations of about 1 ft during summer months (July and August), are underlain and supported by over 1500 ft of permafrost, whose thickness reaches a maximum of 1800 ft at Prudhoe Bay. The “North Slope” of the Brooks Range in Alaska covers approximately 88,000 square miles, about the size of Utah, and includes eight population centers plus the petroleum development complexes. The population centers include Barrow (pop. 4581), Wainwright (pop. 546), Nuiqsut (pop. 433), Kaktovik (pop. 293), Anaktuvuk Pass (pop. 282), Point Lay (pop. 247), Atqasuk (pop. 228), and Arctic Village (pop. 152). Approximately 5000 people are engaged in the North Slope in works affiliated with the oil and gas development (AOGCC Alaska UIC Program Review, 2002). Drinking water for both indigenous communities and the geographically separate petroleum development complexes in the North Slope comes from surface accumulations, primarily lakes or desalinized water.
* Views expressed in this chapter do not necessarily represent the views of the United States Government or any of its agencies.
ELSE_TSANG_CH044.qxd
558
8/23/2005
12:34 PM
Page 558
Alaskan Underground Injection Control of Solid Waste Disposal
44.2 REGULATORY FRAMEWORK The application and regulation of Class I, Class V, and Class II underground injection in Alaska are the responsibility of the United States Environmental Protection Agency (EPA) and Alaska Oil and Gas Conservation Commission (AOGCC) (Fig. 44.1). The UIC program’s purpose, authorized by the Safe Drinking Water Act (SDWA) in 1974, is to protect groundwater quality and to ensure that fluids are injected safely, and remain where they are injected. EPA sets minimum federal requirements and manages Class I and Class V wells. The AOGCC has been delegated the state program to manage Class II enhanced oil recovery (EOR), Class II storage, and Class II disposal injection wells. The federal and state partnership is supported with a grant from EPA to AOGCC. At the same time, the regulated community in the North Slope is dominated by large integrated oil and gas companies, and is further simplified by the fact that there are only two major operators in the North Slope, ConocoPhillips, Inc. (CP) and BP Exploration Alaska, Inc. (BPXA). Both operators follow the Alaska Waste Disposal and Reuse Guide, a document supported and revised by the operators on a regular basis (http://contentstore.bpweb.bp.com/hsems/default.htm).
44.3 CLASS I AND CLASS V WELLS The North Slope Class I disposal wells managed by EPA are permitted, designed, and constructed for the permafrost subsurface. Class I wells may accept any nonhazardous fluid, municipal waste, and exploration and production waste exempted from the Resource Conservation Recovery Act (RCRA) regardless of the fluid origin. The wells are continuously cemented from the surface through the long string casing to the injection intervals, a task commonly accomplished by a three-stage cement job. In permafrost regions, Class I wells are permitted by variance to operate above the formation fracture pressure. Permit requirements include annual internal and external mechanical integrity, continuous monitoring, quarterly reporting, and a waste analysis plan. Currently, there are six Class I injection wells in the North Slope in operation. Three wells are located at Prudhoe Bay Pad 3 and are accessible by year-round road systems. Other Class I wells are located at remote facilities 25 miles east of Prudhoe Bay at Badami Field, 80 miles west of Prudhoe Bay at Colville River Delta Alpine Field near Nuiquit, and 10 miles north of Prudhoe Bay, offshore at Northstar Field. Another Class I well is permitted and has been constructed to manage fluids for the Milne Point Field, located 40 miles west of Prudhoe Bay. An additional Class I well is in the planning stage for the remote Point Thomson Field, located 50 miles east of Prudhoe Bay. EPA has rule-authorized as Class V the three injection wells at Surfcote Pad that were formerly permitted as Class II wells. The rule authorization allows for injection if nonhazardous fluids and removes the limitation that the fluids must come up from downhole. North Slope Class II wells managed by AOGCC are permitted individually. Class II wells may be designed and constructed for disposal below the permafrost or may be a modified well of opportunity. Class II wells may accept any nonhazardous fluids, as well as exploration and production waste exempted from the RCRA. However, Class II fluids are limited to fluids produced up from downhole and uniquely related to oil and gas exploration and production. Class II wells may operate above the formation fracture pressure and are tested at a minimum of every 4 years for internal mechanical integrity only. Of the 1062 Class II wells in Alaska, 1,004 (95%) are EOR (Class II-R) wells and 58 are disposal wells (Class II-D). There are no active storage (Class II-H) wells on the North Slope; however, several
8/23/2005 12:34 PM
Fig. 44.1. Location of major oil fields in the North Slope of Alaska (ADNRDOG, 2001).
ELSE_TSANG_CH044.qxd Page 559
44.3 Class I and Class V Wells 559
ELSE_TSANG_CH044.qxd
560
8/23/2005
12:34 PM
Page 560
Alaskan Underground Injection Control of Solid Waste Disposal
active storage wells are located in the Kenai Peninsula, adjacent to Cook Inlet. Class II wells may be used to inject nonhazardous waste brought up from downhole.
44.4 UNDERGROUND INJECTION REDUCES SURFACE ENVIRONMENTAL IMPACTS The traditional practice of drilling mud pits and reserve pits for storage and disposal of drilling wastes has been virtually eliminated in the North Slope. An integrated approach to managing wastes from drilling, production, maintenance operations, and camp sewage systems has been adopted there. Utilizing ball mill technologies developed in the mining sector, mechanical grinding of solids and deep-well injection employing both Class I and Class II wells are used to dispose of the waste streams from oil and gas drilling and production activities. On the North Slope, because permafrost is continuous, injection pressures are allowed to exceed the formation fracture pressures to facilitate solids injection disposal. In an effort to reduce the environmental impact of surface impoundments, the EPA has rule-authorized as Class V wells a total of three disposal wells located at Surfcote Pad. Under the former AOGCC Class II state permit, these wells were formerly restricted to receive only fluids eligible for Class II well disposal activities. These injection disposal wells accept solid waste processed by a ball mill grinding facility, located on Drillsite-4. The waste includes drilling mud, cuttings, and other materials from former reserve pits, as well as oily waste pits in the Prudhoe and Kuparuk fields that are excavated, transported, pulverized, and injected during winter months when environmental risks are minimal. Waste totals reported to EPA during the 2004 annual inspections for the period of January 1995 through March 2004 show 1.917 million yd3 of reserve pit solids, 0.420 million yd3 of ongoing solids plus 1.998 million barrels of ongoing liquids have been ground and injected. (Personal communication, Bryan Collver, BPXA, 2004) The environmental impact on surface Arctic tundra has been reduced because of the reduction in gravel-pad sizes and the fewer miles of gravel roads. The footprint of gravel pads needed for oil field development has decreased substantially as drilling technology has improved. Operators have strived toward zero-discharge status, and gravel pad space formerly needed for reserve pits to store mud and cuttings has been replaced with more compact small or mobile grind and inject systems, with injection wells designed to handle solids. Original well pads on the North Slope in Prudhoe Bay were drilled with well offsets that were commonly 100–150 ft. With slim-hole operations and other developments, well offsets have been reduced to 10 ft. Departures of downhole locations are increasing with the increased use of deviated wells, where over 10,000 acres of oil reservoir at depth are being developed from one surface pad. Alpine was developed on less than 100 acres, Badami was completed on a 26-acre pad, and recently, Northstar was operating on a pad less than 5 acres in total surface area. By eliminating the need to transport large volumes of waste offsite, remote oil fields are being developed without roads, further reducing the surface impact. However, when fixed wing aircraft access is warranted, the gravel runway impact requires an additional 100-acre footprint. Disposal needs change in remote oil fields as they mature through the construction, drilling, development, and production phases. Using the Northstar waste plan as an example, during the early stages of field development, disposal needs began with a high volume of solids generated from drilling activities, and municipal and industrial wastes from large crews during construction. Disposal needs change to predominantly liquid and municipal
ELSE_TSANG_CH044.qxd
8/23/2005
12:34 PM
44.5 Aquifer Trends
Page 561
561
wastes after drilling is completed and reservoir fluids increase in water content. On the basis of the projected 20-year life of the Northstar waste injection well plan, a total estimated 120 million barrels will be injected in injection intervals located over 5000 ft below the surface. A majority of the grinding will be performed during the initial drilling and development phase of the project. During the development phase, grinding will be needed for approximately 140,000 barrels of rock cuttings, fracture sand, vessel sludge, and sand. Prior to injection, little grinding will be needed for the 600,000 barrels of camp sewage and gray water, the 400,000 barrels of well workover fluids, the 360,000 barrels of drilling mud and fluids, and the 40,000 barrels of industrial nonhazardous waste fluids estimated to be injected over the 20-year life of the field. No grinding will be needed to dispose of 98% of all injectate for the life of the well, injectate consisting of produced water oil reservoir brine (estimated 118,500,000 barrels) and on-site storm water from rain and snow melt (estimated to be 182,500 barrels). Most of the fracture slurry injection is expected to occur during the first 2 years, when average fracture-slurry-injection rates are about 28,000 barrels per month—with maximum fracture slurry rates of 65,000 barrels per month, permitted for handling backlogs generated from plant upsets, scheduled shutdowns, well treatments, and redrilling workovers. Produced water injection rates are projected to reach 16,000 barrels per day after the first few years and remain at that level for the remaining 18 years of the 20-year life of Northstar (Northstar Permit, BPXA Waste Analysis Plan, 2001).
44.5 AQUIFER TRENDS The SDWA requires the protection of underground sources of drinking water (USDW), which may include drinkable quality water (less than 3000 total dissolved solids [TDS]) and useable quality water (3000–10,000 TDS). Alaska regulations simply identify 10,000 TDS as an upper salinity limit of an aquifer. The permafrost thickness appears to be related to localized heat flow and lithology trends in the upper part of the geologic column. Waterfilled sand-rich intervals appear to be present at Prudhoe Bay near Pad 3, where wells were drilled through 1900 ft of permafrost. On the basis of a combination of water samples and well log analyses submitted by operators, EPA and AOGCC have observed that the salinities of groundwaters below permafrost are generally greater than 10,000 TDS. EPA has determined that there appears to be no USDWs in the Colville Delta/Alpine Unit and Northstar Unit. AOGCC has determined that there appears to be no USDWs in the Duck Island Unit, Meltwater Pool, or the Prudhoe Bay Field Eastern Operating Area (EOA). EPA gave aquifer exemptions for the Kuparuk River Field in 1985 and Point Thomson Field in 2002. AOGCC gave aquifer exemptions with EPA concurrence for Prudhoe Bay Field Western Operating Area (WOA) and Milne Point Unit (Figs. 44.2, 44.3, and 44.4). On the eastern North Slope, the Tertiary nonmarine sedimentary wedge generally thickens from the west to the east, toward Point Thomson from Duck Island. Salinities in the WOA range from 9900 to 11,100 mg/L TDS, whereas Cretaceous water samples range from 36,800 to 44,100 mg/L TDS, based on Class II disposal wells with perforations at 4000–6000 ft below ground surface (BGS). On the basis of well samples from and log analyses of six wells in the WOA to the Duck Island Unit, average interval salinities of confining intervals are 18,000 mg/L of sodium chloride for the Tertiary sections from the base of the ~1500-ft-thick permafrost to depths of 3500 BGS. Average interval salinities of the same WOA lower Tertiary and Cretaceous sediments are approximately 24,000 mg/L of sodium chloride for depths of 3500–6000 ft BGS (AOGCC-Alaska UIC Program Review, 2002).
ELSE_TSANG_CH044.qxd
562
8/23/2005
12:34 PM
Page 562
Alaskan Underground Injection Control of Solid Waste Disposal
Fig. 44.2. Aquifer Exemptions and “No USDW” Determinations. North Slop of Alaska— ASP Zone 4 (AOGCC, 2002).
Fig. 44.3. Subsurface salinity. WOA to Duck Island (AOGCC, 2002).
ELSE_TSANG_CH044.qxd
8/23/2005
12:34 PM
Page 563
44.6 Geological Limits to Injection
563
Fig. 44.4. Aquifer Exemptions and “No USDW” Determinations. North Slope of Alaska—ASP Zone 3 (AOGCC, 2002).
44.6 GEOLOGICAL LIMITS TO INJECTION The physical environment and geology limit disposal, based on mechanical properties and solutions encountered in the subsurface. Wastes incidental to the drilling of wells, predominantly drilling mud and cuttings, are challenging to inject because they tend to plug pore spaces and reduce near-well permeabilities. Formations best suited to accept mud and cuttings are friable, highly porous, and permeable. Where disposal options are not available such as in some exploratory wells, mud and cuttings may be disposed of (on a single well basis, with prior approval from AOGCC) by injecting down the surface casing annulus. The operator must show that the well integrity and stratigraphic conditions will contain the waste below the casing shoe, and will not endanger USDWs. Most successful injection zones commonly show porosities over 25% with permeability of 250–4500 millidarcies (mD); however, injection pressures and rates vary. At the Class I wells at Pad 3, the Sagavanirktok Formation at the base of the permafrost receives volumes not to exceed 285,720 barrels/month and pressures not to exceed 1400 pounds per square inch (psi), which is above the fracture pressure. At the Badami Class I well, the Ugnu Formation receives volumes not to exceed 65,000 barrels per month and pressures not to exceed 3000 psi, which is above the fracture pressure. At the Surfcote wells, the Ugnu Formation receives about 30,000 barrels/day at a rate of 20–25 barrels/min at 1190 psi, which is above the fracture pressure. The upper injection zone at Alpine Class I well, in the Triassic Sag River Formation, exhibits good reservoir properties, and an average permeability of 120 mD and a lower injection zone in the Permo-Triassic Ivishak Formation, which exhibits moderate reservoir quality. The Alpine maximum permitted injection pressure is 3200 psi, with monthly injections rates of about 15,000 barrels/month at 1450–1800 psi. At the Northstar Class I well, the injection zones include the Schrader Bluff Formation, with porosities of 26–32% and permeability values of 150–2500 mD, permitted for pressures up to 3000 psi (Syed and Cutler, 2001; Fig. 44.5; Cutler, 2004 and Talib Syed Associates, 2002).
ELSE_TSANG_CH044.qxd
564
8/23/2005
12:34 PM
Page 564
Alaskan Underground Injection Control of Solid Waste Disposal
Fig. 44.5. Subsurface salinity control. Badami to Alaska State A-1 (AOGCC, 2002).
44.7 CONSTRUCTION, OPERATION, AND MANAGEMENT Experience gained over the years has been successfully applied to new well construction, to assure that operations are safe for the workers and protective of the environment. Constraints are set on design parameters requiring specialized well construction, operation, and management to assure that fluids are contained in the target injection horizons below the permafrost. Wells of opportunity have been replaced with specially constructed and continuously cemented wells. More attention has been given to increasing surface and bottomhole separations to assure thaw bulbs being separate. Attention to rock mechanics within
ELSE_TSANG_CH044.qxd
8/23/2005
12:34 PM
Page 565
44.7 Construction, Operation, and Management
565
injection zones has highlighted the importance of shut-in periods, monitoring, and careful operations. North Slope operators have overcome the unforgiving elements to make Alaska a safer oil and gas-producing region. However, injection activities have not been without incident. In 2003, the National Academy of Sciences (NAS) draft report, Cumulative Environmental Effects of Oil and Gas Activities on Alaska’s North Slope (see p. 114) in 2003, reported that at least 20 wells in Prudhoe Bay Field have experienced fluid escape to the surface around surface casing and conductor, as a result of annular injection of drilling mud. The NAS draft report stated the worst case was a pilot injection project that was a well of opportunity permitted as a Class II disposal well. On March 17, 1997, the Class II disposal well surface broached at Drillsite 4-19 when an estimated 18,000 barrels of fresh and diluted seawater surfaced around a number of adjacent wells (NAS Draft Report, 2003, p. 114). Fortunately, the volume of the spill represented a small portion of waste in comparison to the volumes routinely handled on a daily basis, and the fluid was contained at the surface and cleaned up, which reduced the impact of the release to the sensitive Arctic tundra environment. A subsequent careful review of a cement bond log (USIT) taken in May 1994 indicated a poor cement bond and a Borax pulse neutron log run in November 1994 indicated channeling. A video showed a pipe split, and records indicate a cement squeeze job was performed on the well in February 1995 (TSA, 2002). An evaluation of the pilot test well-broach incident at Drillsite 4-19 found that the release resulted primarily from a loss of well construction integrity and a close well spacing of 120 ft. Experience gained from the past has been successfully applied to the well design, construction, and operations of subsequently constructed Class I and Class II wells, as illustrated in the three rule-authorized Surfcote Pad Class V wells for solids disposal in the Prudhoe Bay area. The surface spacing of the three Surfcote wells is 300 ft and they deviate 40–50° maximum angle from vertical to provide a bottomhole separation of 3000 ft. Such spacing and deviation assure that thaw bulbs are separated at depth, and no wells are within 2000 ft of the injection zone. The wells are made up of 4300 ft of 13 5/8 in. cemented casing, plus an additional 3000 ft of cemented 9 5/8 in. casing that extends to the base of the Ugnu Formation. At the SV3 sand, an ES cementer sleeve opening to the outer annulus allows for monitoring the outer annulus with a gauge to observe for major fluid entry into the SV3 sand. Such an occurrence would indicate a major channel from the injection zone. A dipole sonic log was run on the open-hole well prior to casing the well. Annual tests performed on the well to observe ongoing conditions include shut-in temperature surveys, step-rate tests, and pressure falloff tests. The AOGCC requires mechanical integrity tests be performed every 2 years. The 7 in. tubing is tested with a caliper every summer, and injection is rotated every 10 days between three Surfcote wells that continuously inject fluid averaging 9.4 lb/gal. The three Surfcote wells are operated on a 10-day rotation, allowing fractures to close during shut-in periods to maximize well capacity over time, and are flushed for 8 hours with seawater following an injection cycle. (Personal communication, Michael Bill, ASRC, 2004) On an average, approximately 28,000 barrels of seawater are injected each day at the Surfcote Pad wells. By 2002, over 1.3 million barrels of drilling fluid from ongoing development, plus 31 million barrels of slurry (over 2.2 million yd3) consisting of 2 million tons of excavated reserve pit material and drilling solids, were injected with over 12 million barrels of treated salt water. The ball mill and injection system long-term maintenance is performed during the warmer summer months (Guo et al., 2003). Proper management of both Class I and Class II injection wells calls for proper construction to prevent leakage, use of tubing and packer, and mechanical integrity testing.
ELSE_TSANG_CH044.qxd
8/23/2005
566
12:34 PM
Page 566
Alaskan Underground Injection Control of Solid Waste Disposal
Because of the permafrost conditions in the first 1500–1800 ft, “Arctic type lite” cements have been developed for cold surface conditions that may only have compressive strengths of 500 psi or less. More conventional Class G cements are used below the permafrost, cements having compressive strengths of up to 3000 psi. Because of the special conditions particularly near the surface, cement bond logs are of particular importance in evaluating the well cement jobs. The ultrasonic imaging tool (USIT) developed by Schlumberger in 1991, which provides a continuously pulsating echo type tool, offers better coverage of the casing wall in all directions than conventional cement evaluation tools that use a fixed transducer. Class I well requirements are set in the permit and are commonly constructed with continuous cement from the surface through the long string casing to the injection intervals, which may require up to three phases to complete the cement job. AOGCC permits for Class II wells require that structural and conductor casing be cemented from the shoe to the surface, and that surface casing must be set at the base of all strata known or reasonably expected to serve as a source of drinking water, plus cementing from the shoe to the surface. In addition, intermediate and production casing is cemented from the shoe to a minimum of 500 ft above hydrocarbon and abnormal geopressured zones (AOGCC Alaska UIC Program Review, 2002). Internal mechanical integrity for tubing and packer includes standard annulus pressure tests (SAPT). Given the thermal cooling effects of the permafrost, if internal mechanical integrity tests are attempted too soon (within 4 h) after the last injection, the well may appear unstable, owing to fluid contraction from cooling of the tubing adjacent to the permafrost. To account for this cooling effect, the Class I permit generally reads as follows: In order to demonstrate that there is no significant leak to the casing, tubing, or packer, the tubing/casing annulus must be pressure tested to at least the maximum permitted injection pressure for not less than 30 min. Pressure shall show a stabilizing tendency. That is, the pressure may not decline more than 10% during the test period and shall experience less than one-third of its total loss in the last half of the test period. If the total loss exceeds 5% or if the loss during the second 15 min period is equal to or greater than one-half the loss during the first 15 min, the permittee may extend the test period for an additional 30 min to demonstrate stability.
Arctic conditions may also cause chart recorders to fail (because the pen may freeze during the test), so direct meter readings are frequently necessary. Step-rate tests are particularly important for those wells that operate above the fracture pressure. External mechanical integrity tests are commonly accomplished using oxygen-activation methods with the water flow log and temperature logs, or borax-pulsed neutron log with temperature logs (UIC permit, 2001).
44.8 CONCLUSION The preferred alternative to managing solid, liquid, and gas wastes on the North Slope is underground injection through Class I and Class II well disposal systems. Relic reserve pits are being excavated, pulverized, and injected. New fields are being planned and developed with ball mill grind and inject systems, using Class I and II wells to address nonhazardous wastes on site. This reduces the oilfield gravel pad footprint and need for gravel roads to transport waste across tundra. UIC is critical to future development of oil fields in the sensitive Arctic and cannot be accomplished without paying close attention to injection well construction, operations, and monitoring.
ELSE_TSANG_CH044.qxd
8/23/2005
References
12:34 PM
Page 567
567
ACKNOWLEDGMENTS We thank the State of Alaska Division of Oil and Gas and AOGCC for the use of their diagrams and maps, as well as Talib Syed and Dennis Thurston for their contributions, support, and review.
REFERENCES Alaska Department of Natural Resources Division of Oil and Gas (ADNRDOG), 2001. http://www.dog.dnr.state.ak.us. Alaska Oil and Gas Conservation Commission (AOGCC), 2002. Alaska UIC Program Review. Bill, M., 2004. Personal communication, Arctic Slope Regional Corporation, March 29. Collver, B., 2003. Personal communication, BP Exploration Alaska, Inc. (BPXA). Collver, B., 2004. Personal communication, BP Exploration Alaska, Inc. (BPXA), http:// contentstore.bpweb.bp.com/hsems/default.htm. Cutler, T., 2004. Annual Inspections, Surfcote Pad. Guo, X., Abou-Sayed, A. and Engel, H., 2003. Feeling the pulse of drill cuttings injection wells—A case study of simulation, monitoring and verification. Alaska, Society of Petroleum Engineers Inc. SPE 84156. National Academy of Sciences (NAS), 2003. Cumulative Environmental Effects of Oil and Gas Activities on Alaska’s North Slope. Draft Report. p. 114. Syed, T. and Cutler, T., 2001. Review of Class I Wells—North Slope of Alaska. Ground Water Protection Council Annual Forum, Reno, Nevada, September. Talib Syed Associates (TSA), 2002. Alaska UIC Program Class I Report. EPA Region 10, Seattle, WA. UIC Permit Application for BPXA’s Northstar Class I Facility and UIC Permit No. AK1I002-A Waste Analysis Plan. 2001. EPA Region 10, Seattle, WA.
This page intentionally left blank
568
ELSE_TSANG_CH045.qxd
8/23/2005
12:31 PM
Page 569
Chapter 45
DISPOSAL OF MEAT, BONEMEAL, AND RESIDUAL ASH BY INJECTION INTO DEEP GEOLOGICAL FORMATIONS V. Brkic a, I. Omrcena, S. Bukvic a, H. Gotovac b, B. Omrcenc, and M. Zelicc a
INA Oil Industry Plc., Zagreb, Croatia Faculty of Civil Engineering, University of Split, Croatia c Association of Petroleum Engineers and Geologists, Zagreb, Croatia b
45.1 INTRODUCTION For many years, waste generated in the Republic of Croatia from oil and gas exploration and production has been disposed of by injection into deep geological formations. From our vast experience in oil-industry waste disposal, we can project some possibilities for disposing of waste in other industries, specifically for the disposal of meat and bonemeal (MBM) and residual ash. As we know, since mad cow disease (Bovine Spongiform Encephalopathy, (BSE) first appeared, animal-origin proteins have been forbidden in animal feeds. With this in mind, it is essential that we dispose of existing MBM reserves as well as recently produced MBM in a safe and ecologically appropriate manner. For this purpose, laboratory research has been conducted as preparation for injection of MBM into deep wells. Furthermore, when combusting various types of industrial and municipal waste, the problem of residual ash disposal emerges. Such residual ash contains heavy metals, which (according to newly acquired experience and knowledge from the field) may also be permanently disposed of by deep injection disposal techniques. The chapter presents research into such techniques, as well as the preparation and technological procedures for disposing of the aforementioned waste. Also in this chapter, we will present a model of a plume transport generated by deep aquifer waste injection. In this model, a solute flux, defined as the mass of a solute per time and area unit from the source to the control plane, represents transport within aquifers. Because of the natural heterogeneity of geological formations, groundwater flow and transport are tortuous and unpredictable, and statistically can only be described as a random field. The solute flux is presented in terms of the first two statistical moments (sufficient for most practical purposes) as a space-time process, with time referring to the solute flux breakthrough and space referring to the transverse displacement distribution at the control plane. Velocity fluctuations lead to two basic transport mechanisms: (1) fluctuations on a scale smaller than or equal to the size of a distorted or diffused plume, and (2) fluctuations on a scale larger than a plume size, which cause the plume to “meander” relative to the mean flow direction. Two general transport modeling approaches are used: absolute, which includes both mechanisms, and relative, in which the meandering effects are removed. This transport concept is also useful for both reactive and nonreactive transport. We also present an environmental risk formulation generated from these simulated transport results. This formulation includes the entire process of selecting the risk agent source and its transport through porous media. The calculated risk level incorporates the possibility that deep-well injection could cause undesirable effects on health and environmental resources.
ELSE_TSANG_CH045.qxd
8/23/2005
12:31 PM
570
Page 570
Disposal of Meat, Bonemeal, and Residual Ash by Injection
45.2 HISTORICAL OVERVIEW The exploration of hydrocarbons in the Republic of Croatia started 100 years ago, and significant oil and gas reserves have been discovered. More than 4300 exploration and production wells were drilled into onshore and offshore sedimentary deposits. In the 1990s, Croatia initiated a program for disposing of the waste generated in the process of exploring, drilling, producing, refining, and distributing of hydrocarbons—by deep-well injection into dry exploration or depleted production wells. Deep injection technology enables not only the disposal of waste generated by the oil industry, but also the waste generated in other industries—such as the food, chemical, leather, and pharmaceutical industries. Through deep injection disposal, waste may be permanently and safely isolated into geologically appropriate wells. This technology can also work for disposal of heavy metals (contained in virtually all kinds of waste), and thus can actively support the principles of environmental protection. The overall waste disposal cost and the possibility of heavy metals polluting potable water are thereby reduced. Currently, industrial waste is permanently disposed of by applying a specific procedure developed in Croatia. Regulations on the application of deep-well injection are under preparation. Many dry and depleted wells may be appropriate for permanent disposal of industrial waste generated in various industries, and thus there is little need to pay the cost of creating new wells. However, the existing wells must be upgraded before being used for deep-well injection, and monitoring equipment must be installed. Waste disposal by injection into deep wells is a unique method, one that may be used for permanent disposal of hazardous waste without having an impact on the environment, since waste is disposed out of the biosphere. This chapter also deals with the possibility of waste disposal by incineration, as well as residual ash, meat, and bonemeal disposal (Brkic et al., 2001). 45.3 GEOLOGICAL AND PHYSICAL PROPERTIES OF THE BENICANCI OIL FIELD The Benicanci Oil Field has been designated for the injection of waste fluid. It contains mostly breccia with dolomite detritus (rarely carbonate), and in some upper parts of the formation contains conglomerate breccia of equal composition (see Fig. 45.1). For volume calculations, the value of water saturation Sw 100% has been chosen, since only the wells in the upper parts of the formation or from the less exploited part of the field are considered to be in production (with water saturation above 80%). The dolomite breccia in the Benicanci field is over 300 m thick. Field properties are as follows: ● Mean porosity, Φ 9.4%. ● Effective formation thickness, hef135 m. ● Water saturation, S 100%. w ● Mean permeability, ρ 7.4 × 10−3 µm2. Using the planimeter and the thickness chart, we calculated the volume of the permeable part of Benicanci Field over a 14.5 km2 area, which has well logging data, to be V 1960 × 106 m3. The volume of breccia over the permeable part of the field, for the corresponding drainage radius re 0.564 km, was V 134.8 × 106 m3. The value of expected gradients, using the Eaton formula with variable Poisson coefficient, was a minimal ν 0.22 and maximal ν 0.3. For waste disposal by deep well injection at the Benicanci field, well Be-28 was chosen. Field data are shown in Table 45.1. At this well, an interval from 2141 to 2042 m in depth
ELSE_TSANG_CH045.qxd
8/23/2005
12:31 PM
Page 571
45.3 Geological and Physical Properties of the Benicanci Oil Field
571
Fig. 45.1. Lithological column of Benicanci formation.
was opened, with the remaining part left as an open hole section (see Fig. 45.2). A chemical wash of this open interval was conducted with 7.5% HCl acid, after which an injectivity test was performed with formation water (Table 45.2). A total of 1,530,863 m3 of waste fluid was injected into the well. After the injection of this waste fluid, the hydrostatic pressure was measured at a depth of 2186 m. The pressure was 186.4 bar.
ELSE_TSANG_CH045.qxd
8/23/2005
572
12:31 PM
Page 572
Disposal of Meat, Bonemeal, and Residual Ash by Injection Table 45.1. Be-28 well data Field data
Be-28 well
Formation pressure gradient Injection depth Hydrostatic pressure Formation fracture gradient in depth for minimal ν Pressure at injection depth for minimal ν Formation fracture gradient in depth for maximal ν Pressure at injection depth for maximal ν Maximal injection pressure for minimal ν Maximal injection pressure for maximal ν
0.81 bar/10 m 2042 m 200.32 bar 1.17 bar/10 m 238.91 bar 1.36 bar/10 m 277.71 bar 38.59 bar 77.39 bar
45.4 DISPOSAL OF MEAT AND BONEMEAL (MBM) Feed meat and bonemeal (MBM), which contains animal protein and fat (see Table 45.3), has been banned to cattle, sheep, and goats within the European Union (EU) since 1 July 1994. The complete feedstuff ban assumes that contamination cannot be ruled out during the production of feedstuffs, and that current knowledge about transmission paths and quantities of pathogens is incomplete. Our knowledge about the conditions required to adequately destroy the BSE pathogen (infectious prions) must be considered equally incomplete. Infectious prions can currently be detected only when they appear at concentrations of at least a thousandth of the concentrations found in the brain and spinal cord of clinically infected cattle. The most recent research further suggests that the possibility of BSE being transmitted to humans is likely, and much evidence indicates a link to a new variant, called CreutzfeldJacob Disease (nvCJD). Even if findings do not yet constitute proof-positive that BSE can be transmitted to humans, the circumstantial evidence has become compelling. It should be noted that samples of brain from infected hamsters revealed traces of infectiousness even after thermal treatment at 600°C. A possible explanation under discussion is an inorganic “molecular template” capable of triggering a biological replication of the pathogen. Consequently, disposal to landfill is not an option, because simply burying the material cannot destroy all potential BSE pathogens. In addition, this form of disposal is banned for highly organic materials by EU legislation (Landfills Ordinance). Croatia produces yearly about 15,000 tonnes of MBM by processing waste of animal origin. There was no evidence of BSE-pathogen in the Croatian MBM stockpile, according to BSE-analysis conducted by the Croatian Institute of Veterinary Medicine. So far, the problem of managing the MBM has not been approached in an ecologically acceptable way: since kilns lack some basic technical requirements, they are not in a position to take over the old stock for incineration. By a resolution of the Government of the Republic of Croatia in August 2002, it has been ordered that an appropriate solution for managing the accumulated MBM stockpile should be found. MBM is a substance that is the result of thermal treatment of animal dead bodies and of waste of animal origin. MBM, the subject of this chapter, has been thermally processed by pressure sterilization of waste (after maceration into small pieces) at least 133°C for at least 20 minutes and at a water vapor pressure of 3 bar and higher. This procedure results in MBM particles with a size up to 5 mm. From Table 45.3, we can clearly see that MBM is more than half protein, with 10% fat and the rest inorganic substances. We must emphasize that in the process of MBM
ELSE_TSANG_CH045.qxd
8/23/2005
12:31 PM
Page 573
45.4 Disposal of Meat and Bonemeal (MBM)
Fig. 45.2. Benicanci-28 well construction.
573
ELSE_TSANG_CH045.qxd
8/23/2005
12:31 PM
574
Page 574
Disposal of Meat, Bonemeal, and Residual Ash by Injection Table 45.2. Injectivity test results Condition
Injected volume (m3)
Flow (l/min)
Injection pressure (bar)
1 2 3 4
10 20 30 40
200 400 600 800
0.7 3.4 10.3–13.8 20.7–34.5
Table 45.3. MBM substances Substance
Quantity (%)
Proteins Raw fat Raw ash Phosphorous Calcium
56 10 21 6.1 12
Table 45.4. Pollutants in MBM Substance
Concentration (mg/kg)
Lead Mercury Cadmium Chrome Copper Nickel Zinc Arsenic
4.25 0.18 0.43 2.6 12.0 3.1 110 0.3
incineration in Croatian cement kilns, excessive amounts of chloride and phosphates have been found in MBM samples. Note that most of the chloride in MBM is present as NaCl (common salt). The composition of the ash, which represents ⬃21% of MBM content, shows high levels of phosphorous and calcium. According to BSE analysis, it should be emphasized that there was no evidence of the BSE pathogen in MBM. (The pollutant content of various MBM samples is given in Table 45.4.) This analysis shows that MBM has a low pollutant content which, added to fact that there is no evidence of the BSE pathogen in MBM, indicates that MBM should not be classified as hazardous waste. Moreover, during the preparatory testing, MBM proved to have very good miscibility in organic acids and in concentrated HCl. Miscibility in different solvents is very important for deep-well injection, since in that way coagulation is inhibited and many problems within the well can be avoided.
45.5 DISPOSAL OF RESIDUAL ASH In the process of hazardous and nonhazardous waste incineration, smaller quantities of waste are generated with regard to the initial quantity. Brkic and Omrcen (2003). Specifically,
ELSE_TSANG_CH045.qxd
8/23/2005
12:31 PM
Page 575
45.5 Disposal of Residual Ash
575
two types of waste are thus generated: slag and residual ash, which has disposal restrictions because of its hazardous effect. Residual ash is a gray to dark gray powder, very light, of 0.365 g/cm3 density. Based on granulometric analysis, about 50% of the residual ash particles are smaller than 10 µm, while only 2% of particles are smaller than 2 µm. Chemical analysis of residual ash from hazardous waste incineration plants has shown that residual ash samples contain Na2CO3, namely sodium bicarbonate, NaCl, Na2SO4 and relatively large quantities of heavy metals. The residual ash analysis is shown in Table 45.5. The results of the residual ash analysis have shown that it is a hazardous waste, one that must not be disposed of at nonhazardous disposal sites (landfills). As concluded from further laboratory research, residual ash is best dissolved in water at a room temperature of 20°C. Such aqueous solutions have shown an alkaline reaction, and therefore they must not be discharged into the environment. The results of the research are provided in Table 45.6.
Table 45.5. The chemical analysis of residual ash Parameter
Results
Density Dry substance Si Al Fe Mn Ca Mg Na K S Cl Ash (800°C) TOC As Cu Zn Cd Cr Mn Ni Pb Fe Hg
0.357 kg/m3 99.06% 0.68 % 0.14% 5.26% 1.17% 0.32% 30.22% 0.21% 2.88% 11.35% 63.50% 0.14% 2.8 mg/kg 1621.0 mg/kg 4794.9 mg/kg 1 mg/kg 189.4 mg/kg 398.0 mg/kg 1851.3 mg/kg 1313.8 mg/kg 60,913.7 mg/kg 5.5 mg/kg
Table 45.6. Residual ash solubility in different solvents at 20°C Solvent
Solvent/residual ash ratio
Solubility (%)
10% HCl 10% HCl 3% HF Brinewater
2:1 3:1 5:1
49.3 79.2 96.2
ELSE_TSANG_CH045.qxd
8/23/2005
12:31 PM
576
Page 576
Disposal of Meat, Bonemeal, and Residual Ash by Injection
45.6 TRANSPORT MODELING AND RISK EVALUATION 45.6.1 Theory and Assumptions At the moment t 0 (see Fig. 45.3), an industrial fluid is injected into the injection zone (x 0). For t 0, a plume is formed and transported along the main flow path by advection (local dispersion can be neglected) toward the control plane (during analysis, it is wise to have several planes). The control plane is usually vertical with respect to the direction of average velocity; in this two-dimensional presentation, it is represented by a line. Because of fluctuations on a scale smaller or equal to the injection zone, the plume has changed its shape in a random, stochastic way. Velocity fluctuations on the scale greater than the injection zone advectively carry the plume as a whole (plume meandering). Hence, the main physical factor in the whole process is the velocity fluctuations and the scale on which they occur. While fluctuations on the smallest scale change the shape of the plume, at the same time, on the larger scale, the plume translation proceeds. This process has two mechanisms that can be observed separately, and thus transport must be defined in two ways: absolute, where both mechanisms exist; and relative, where the meandering effect is removed. Generally, the approach used here is based on the solute flux approach (Dagen et al., 1992) and not on the resident concentration. This approach has proved to be more practical and shows transport properties in a much more direct way than the classical approach based on resident concentration. There are two key random variables: the specific mass flux (q), defined as a solute mass passing through the unit area of the control plane over unit time (M/L2 T1); and solute discharge (Q), defined as a total mass passing through the control plane over unit time (M/T1). In this chapter, the focus is on the variable Q, described with the travel time τ, the time in which a mass particle travels from the injection zone to the control plane. The transport is described in the relative dispersion framework (Andricevic and
Fig. 45.3. The underground plume dispersion.
ELSE_TSANG_CH045.qxd
8/23/2005
12:31 PM
Page 577
45.6 Transport Modeling and Risk Evaluation
577
Cvetkovic, 1998), where each realization has the same travel time, equaling the average travel time from absolute dispersion. Solute discharge is calculated by Q (t; x) 兰 ρ0 (α)δ (tτ) dα,
(45.1)
A
where ρ0(α) represents the density along the injection zone (M/L), τ is the travel time from the injection zone point y α to the control plane chosen for a set x, dα is the area element, δ is Dirac’s function, and A is the injection zone. Moments for Q can be calculated as follows: 〈Q(t; x)〉 兰 ρ0(α) g1 (t; x, a) dα , A
(45.2)
where 〈 〉 represents the mathematical expectation operator and gl is the probability density function (pdf) of travel time for one particle, which is defined as the probability that the particle travels from y α and t 0 through the control plane in a set x and in time t. The second moment, variance, can be obtained by decomposition of variable Q to its average value and fluctuation around it (Q 〈Q〉 Q), so that the variance can be given as σ 20 〈Q2〉 〈Q2典 〈Q〉2. Since the second part of the equation can easily be obtained by Equation (45.2), only the first one remains to be defined: 〈Q 2 (t; x)〉 兰兰 ρ0 (α) ρ0 (α) g2 (t, t; x, α , α ) dα dα ,
(45.3)
A
where the function g2 presents the pdf of travel time for two particles, defined as the probability that one particle travels from y α and t 0 through the control plane in a set x and in time t, while the other travels from y α and t 0 through the control plane in a set x and in time t. The main problem in theoretical analysis is to define the pdf for both Equations (45.2) and (45.3). The usual approach is to assume a log-normal distribution (Cvetkovic et al., 1992) and then calculate the corresponding statistical moments. However, to define pdf and obtain the above-mentioned equations, we must define the first two moments for the travel time: 〈τ〉 x/U,
1 σ U2 2 τ
冕冕 C (εε, 0) dε dε , x x
u
0 0
1 σττ U2
冕冕 C (εε, αα) dε dε, x x
(45.4)
u
0 0
where Cu represents the longitudinal velocity covariance obtained according to the first-order theory (Rubin, 1990) and separation α–α ′ is the initial separation between two particles in the injection zone for t 0 and x 0. Therefore, for the given velocity covariance (obtained analytically or numerically), the pdf of travel time for one and two particles can be calculated by Equation (45.4). Solute discharge can be calculated by Equations (45.2) and (45.3). Relative dispersion (Andricevic and Cvetkovic, 1998) is a physical concept of transport modeling. It evaluates the spreading of the plume with regard to its center of mass and not to a set coordinate system. In this way, a plume is described that does not have meandering and that is the result of the indefiniteness of the average flow velocity. Transport modeling of a substance via relative dispersion enables us to have a better perspective of real physical dispersion of the plume and the maximal concentration that can occur in the specific time. That type of transport estimation should be used in risk analysis, because it gives us a realistic view of underground behavior. In the case of relative dispersion, plume meandering is
ELSE_TSANG_CH045.qxd
8/23/2005
12:31 PM
578
Page 578
Disposal of Meat, Bonemeal, and Residual Ash by Injection
removed, and only fluctuations on the scale equal to or smaller than the injection zone are taken into consideration. Equations (45.1)–(45.4) are the same for the relative dispersion, but plume meandering should be disregarded. In the modern world, interest in risk evaluation (Andricevic and Cvetkovic, 1996) is increasing every day. Through its basic definition, risk represents the possibility of hazardous occurrence. In this chapter, a risk quantification approach has been used that relates to the risk of exceeding a control point (plane or volume) defined in time or space. By applying the risk of exceedance concept to the transport of an injected waste in a deep geological formation, we are trying to evaluate the probability of the plume spreading into shallow water-bearing layers, which could be used as a water supply in the future. That is called the natural or inherent risk, or the excess risk. In other words, we are looking for risk as a probability that the pollution plume will reach the first or second hydrogeological zone. If L is the variable that represents the size of vertical migration of the injected waste, and C is the variable that represents the level of the geological horizon, which cannot come into contact with waste (that is the depth where shallow water-bearing layers end), then the risk can be given as the following equation: ∞
∞
R* P (C L) ⇒ R* 兰 [ 兰 f(L) dL] g (C) dC, ∞ C
(45.5)
where f (L) and g(C) stand for the distribution functions of L and C. The plume travels through the underground, advectively carried by the main flow and spreading due to the influence of mechanical dispersion, which arises primarily because of the velocity fluctuations on different scales. Since all these processes are very slow, at the beginning there is no risk for shallow water-bearing layers. With time, however, the plume spreads and, over the long time of transport simulation (10,000 years), the risk increases. For each control plane, the plume has the biggest spreading for the average travel time, in other words, in the instant of passing through the control plane. 45.6.2 Modeling Transport modeling of the underground is a very complex process. It requires evaluation of a large number of calculation steps, ending with quantifying the risk of threatening the shallow water-bearing layers. The most demanding step is gathering of input data from all available and often diverse sources. The most important input data are piezometric or pressure conditions in the injection zone and the confining zone or low permeability layers above that zone. Since the stochastic calculation method (Dagan, 1989) is being used, determination of those fields is quite difficult, because not only their values but their variability patterns are unknown. Usually, more types of exploratory work are used for determining the permeability field: measuring of electrical resistance, 3-D seismic, logging while drilling, laboratory measurements on cores, pressure drop down and pressure rise in the open hole. The permeability of the whole field can be determined by a combination of various types of measurements. Unfortunately, most of them are not done in deep wells, so some new, sophisticated methods are used that convert a small number of input data into a stochastic form. One of those very effective methods involves determining permeability from a combination of electrical resistance and permeability measurements, connected by analogy to Darcy’s and Archie’s experimental laws (Purvance and Andricevic, 2000a, b). A second commonly used method involves using geostatistics for determining the stochastic parameters of permeability fields and piezometric head (Deutsch and Journel, 1992). That approach is used in this chapter. In 1990, based on the three wells where the hydraulic
ELSE_TSANG_CH045.qxd
8/23/2005
12:31 PM
Page 579
45.6 Transport Modeling and Risk Evaluation
579
permeability was measured and the 12 wells where piezometric head was measured in 1990, a geostatistical analysis of the piezometric head was performed, using a universal kriging method. The results are given in the form of first two moments: average value (Fig. 45.4) and double standard deviation (Fig. 45.5). For each point, a confidence interval of 95% is given, within which the piezometric head is to be found. On the basis of these results, it can be concluded from Figure 45.4 that the hydraulic gradient is around 0.01 in the north–south direction. By correlating permeability and piezometric head (Kitanidis, 1988, 1997), the mean field of hydraulic permeability was estimated by a cokriging method (Fig. 45.6). The field has been obtained from optimal isotropic exponential variograms, given by the residual concept
Fig. 45.4. Piezometric head field, 1990, mean value.
Fig. 45.5. Piezometric head field, 1990, double standard deviation.
ELSE_TSANG_CH045.qxd
580
8/23/2005
12:31 PM
Page 580
Disposal of Meat, Bonemeal, and Residual Ash by Injection
Fig. 45.6. Field of hydraulic permeability, mean value.
(Kitanidis, 1997), and has the average value 〈K〉 1.85 106 (m/s), with a variance of 0.90 and a range from 3.57 105 (m/s) to 2.06 1010 (m/s), which shows that the range from 0.6 to 1600 103 µm2 that was obtained by measurement at the Benicanci Oil Field is contained in that estimated field. Both the variability of the mean field and the zones of low or high permeability are shown (Fig. 45.6). What is interesting is that all three wells where the measuring was done have relatively uniform values—from just these three measurements alone, we would never have been able to conclude that such a variability in permeability fields exists. The injection zone is located in the Benicanci Oil Field at a depth of 1900–2200 m, with intervals of dolomite breccia containing high secondary porosity and marl. A substantial loss of drilling fluid occurred in the drilling process. The waste was disposed of into the formation, where pore pressure is lower than the hydrostatic pressure. Using this method, the waste can be disposed of without further processing and grinding, and any injection of slurried solids will not result in fracturing of the receiving formation. The effective thickness of the injection zone is in some parts over 300 m, whereas the average effective thickness obtained by kriging has the value of 135 m. Technological fluid is injected under pressure (lower than formation fracture pressure), increasing the pore volume. The injection volume is about 77,000 m3. The density of the injected fluid is between 1006 and 1020 kg/m3, and the length of the injection zone is 135 m. Integral or correlational length is estimated as one-tenth of the observed area (Gelhar, 1993). Shallow water-bearing layers that are not to be endangered lie at a depth of 550 m. The calculation of plume transport is done after gathering and processing the input data. Horizontal flow is assumed in the injection zone, since there is no vertical hydraulic gradient. Transport is considered in absolute and relative forms. These two types of transport are complementary, with each providing data that is not given in the other form. In relative dispersion, we remove the meandering; each realization has the same travel time. That is why the mean plume is very similar to the ones from single realizations, which also results in a far more realistic maximal solute discharge than in the case of absolute transport.
ELSE_TSANG_CH045.qxd
8/23/2005
12:31 PM
Page 581
45.6 Transport Modeling and Risk Evaluation
581
On the other hand, relative dispersion does not give us information on all possible locations of the plume, because every realization is moved so as to correspond to the average travel time from absolute dispersion. That means that in relative dispersion (Andricevic and Cvetkovic, 1998), the real dimensions of the plume in time and space and the solute discharge are determined, while in absolute dispersion (Cvetkovic et al., 1992; Dagan et al., 1992) we determine the reach of the plume in the horizontal direction for the given time of 10,000 years. The calculation of absolute and relative dispersion for the control plane X 60 km is given in Figure 45.7. After 10,000 years, the maximum reach of the plume is 60 km, maximum solute discharge is 80,000 kg/year, and the standard deviation is 25,000 kg/year. Transverse spreading of the plume for the same control plane is shown in Figure 45.8. This first phase of risk calculation is given for transport under the condition that the entire ground above the injection zone has the properties of dolomite breccia. That condition is unrealistically conservative, very unfavorable for risk, for the sake of safety (upper limit). In the two-dimensional analysis, it can be seen that the largest transverse spreading corresponds with the moment the plume center transits the control plane. Therefore, Equation (45.5) defines the excess risk for every control plane and the corresponding moment when the plume center is in transit. The excess risk for six different control planes is given in Table 45.7. For planes close to the injection zone, there is no danger and the risk is very small, as expected. For remote planes, the excess risk increases, and for a maximum calculated time of 10,000 years, it would be 3.75 104. From the transverse results of plume spreading for the same period of 10,000 years, the solute discharge would decrease to 104 of its initial value. The second phase of risk calculation incorporates a far more realistic geological interpretation of the area above the injection zone. Generally, on the upper side, the injection zone is bounded by the confining formation, creating a low-permeability zone that reduces the transverse spreading of the plume and thus also reduces the risk. The input data for this calculation phase are: ● The ground above the injection zone has lower permeability than the ground within the injection zone, the roof formation consists mostly of clay and marl, and literature and experience show K 108 (m/s) and an effective porosity of 1%. ● The spreading in the vertical direction in the low-permeability roof formation is caused by the pressure difference between the injection zone and the roof formation; gradient is i 0.001. ● On the basis of relative dispersion calculations, we estimate that the mass entering the confining formation in the first 1000 years is about 20% of the total injected mass, and that the plume has spread under the confining formation over 20 integral lengths (ca. 4 km). It is assumed that the plume spreads toward shallow water-bearing formations with the above-mentioned gradient. ● The variance is the same as in the first phase of calculations (0.90), and the integral length has been reduced to 20 m, owing to the influence of anisotropy in the vertical direction. ● The distance between the injection zone and shallow water-bearing formations is about 1350 m. ● The calculation of absolute dispersion will be applied because of the determination of the reach of a maximum plume extension in the vertical direction within the next 9000 years. (Note: relative dispersion describes only the physical dispersion within the plume and does not account for a plume meandering.) The calculation of plume expansion in the vertical direction is given in Figure 45.9. If we consider the average value function from Figure 45.9 as the distribution function of
ELSE_TSANG_CH045.qxd
582
8/23/2005
12:31 PM
Page 582
Disposal of Meat, Bonemeal, and Residual Ash by Injection
Fig. 45.7. Analysis of absolute (a) and relative dispersion (b) for control plane X 60 (km).
transverse plume spreading f(L), then, using the function f(C) (here as Dirac’s function defined for t 9000 years) and Equation (45.5), we can calculate the excess risk for 10,000 years as 0.5 1022. This risk is very low in comparison with the first phase of the calculation. From this calculation, the risk can be considered totally acceptable and the waste management process entirely safe due to the extremely small risk of having the waste
ELSE_TSANG_CH045.qxd
8/23/2005
12:31 PM
Page 583
45.6 Transport Modeling and Risk Evaluation
583
Fig. 45.8. Two-dimensional transversal expansion of the plume.
appearing in the upper aquifers. It results directly from the influence of the low-permeability confining formation on plume spreading. Figure 45.9 shows that the plume practically has not reached the water-bearing formations after 9000 years. That is the reason for the reduced excess risk. It should also be mentioned that, besides this analytical way of calculating
ELSE_TSANG_CH045.qxd
8/23/2005
584
12:31 PM
Page 584
Disposal of Meat, Bonemeal, and Residual Ash by Injection Table 45.7. Exceeding risk in characteristic control planes Control plane (m)
Time (year)
Risk (dimensionless)
2000 5000 10,000 20,000 40,000 60,000
250 750 1600 3200 6200 10,000
4.22 1031 6.54 1023 1.84 1011 9.31 109 3.95 105 3.75 104
Fig. 45.9. Transverse plume spreading.
transport, there exist far more complex numerical calculations that employ the Monte Carlo method (Hassan et al., 1997, 2001).
45.7 CONCLUSION Numerous technological, financial, and ecological advantages with respect to existing waste disposal methods have been achieved through application of the procedure described herein for permanent waste disposal by deep-well injection. Our risk assessment modeling indicates that the use of deep-well injection into dry abandoned wells is also an appropriate method for the disposal of waste from other industries and incineration plants, as well as biosolids like MBM. Our company has a number of such wells at its disposal. Transport modeling of the injected waste is a very complex process supported by a stochastic approach that is based on mass flux rather than on the resident concentration. It requires evaluation of a large number of calculation steps, including the preparation of input data, geostatistical interpretation, and calculation of absolute and relative dispersion—and concludes with quantifying the exceedance risk of threatening the shallow water-bearing layers.
ELSE_TSANG_CH045.qxd
8/23/2005
References
12:31 PM
Page 585
585
International experience has been taken into account when developing the project documentation for waste disposal by deep-well injection. Furthermore, deep-well injection, for safe and permanent disposal into geologically appropriate formations, not only applies to waste generated in the petroleum industry, but also could apply to waste from other industries and incineration plants, as well as biosolids like MBM (if there is no evidence of the BSE pathogen). By using up-to-date injection-well equipment with permanent monitoring, the environment is fully protected. From an environmental perspective, the deep injection of waste is the only disposal option that effectively removes waste from the biosphere, minimizing its effect on human health. Disposal is complete, and there is no residual waste product that must be disposed of—future liabilities are thus minimized. All other forms of disposal place the waste either into the air; or into landfills located above the water table; or into rivers and streams that serve as recreation facilities, wildlife habitats, and sources of food and drinking water. Also, surface reclamation can be total, in comparison with the common remediation strategy of collecting and capping the waste on site, which may require perpetual monitoring and maintenance. Exceedance risk, which we have calculated in this chapter, is extremely small and does not represent a danger when the well is used for injection of various types of waste, such as residual ash and MBM. Since the risk limit of 106 is often taken as acceptable in risk assessment analysis, the risk we have calculated in this chapter is considerably smaller than that standard. That risk does not represent a danger when using the well for injection of various types of waste, such as residual ash and MBM. The risk assessment results have confirmed that deep-well injection is one of the more acceptable alternatives for disposal of various types of waste. In the near future, waste injection will be expected to compete favorably with other disposal technologies in the areas of economics, time, and public relations.
REFERENCES Andricevic, R. and Cvetkovic, V., 1996. Evaluation of risk from contaminants migrating by groundwater. Water Resour. Res., 32(3): 611–621. Andricevic, R. and Cvetkovic, V., 1998. Relative dispersion for solute flux in aquifers., J. Fluid Mech., 361: 145–174. Brkic, V. and Omrcen, I., 2003. Disposal of mercury-sulfide and residual ash by deep well injection. SPE 80579, SPE Exploration and Production Environmental Conference. San Antonio, TX. Brkic, V., Omrcen, B. and Loncaric, B., 2001. Waste disposal by injection into deep wells. Proc. 23rd ASME Energy Sources Technology Conference and Exposition, 2001. Houston, TX, USA. Cvetkovic, V., Shapiro, A.M. and Dagan, G., 1992. Solute flux approach to transport in heterogeneous formations, 2, Uncertainty analysis. Water Resour. Res., 28(5): 1377–1388. Dagan, G., 1989. Flow and Transport in Porous Media. Springer, New York. Dagan, G., Cvetkovic, V. and Shapiro, A.M., 1992. Solute flux approach to transport in heterogeneous formations, 1, The general framework. Water Resour. Res., 28(5): 1369–1376. Deutsch, C.V. and Journel, A.G., 1992. GSLIB, Geostatistical Software Library and User’s Guide. Oxford University Press, New York. Gelhar, L.W., 1993. Stochastic Subsurface Hydrology. Prentice–Hall, Englewood Cliffs, NY.
ELSE_TSANG_CH045.qxd
586
8/23/2005
12:31 PM
Page 586
Disposal of Meat, Bonemeal, and Residual Ash by Injection
Hassan, A.E., Andric evic´, R. and Cvetkovic, V., 2001. Computational issues in the determination of solute discharge moments and implications for comparison to analytical solutions. Adv. Water Resour., 24: 607–619. Hassan, A.E., Cushman, J.H. and Delleur, J.W., 1997. Monte–Carlo studies of flow and transport in fractal conductivity fields: comparison with stochastic perturbation theory. Water Resour. Res., 33(11): 2519–2534. Kitanidis, P.K., 1988. Prediction by the method of moments of transport in a heterogeneous formations. J. Hydrol., 102: 453–473. Kitanidis, P.K., 1997. Introduction to Geostatistics. Cambridge University Press, Cambridge. Purvance, D.T. and Andricevic, R., 2000a. On the electrical–hydraulic conductivity correlation in aquifers. Water Resour. Res., 36(10): 2905–2913. Purvance, D.T. and Andricevic, R., 2000b. Geoelectric characterization of the hydraulic conductivity field and its spatial structure at variable scales. Water Resour. Res., 36(10): 2915–2924. Rubin, Y., 1990. Stochastic analysis of macrodispersion in heterogeneous porous media. Water Resour. Res., 26(1): 133–141.
ELSE_TSANG_CH046.qxd
9/8/2005
4:48 PM
Page 587
Chapter 46
THERMAL TREATMENT, CARBON SEQUESTRATION, AND METHANE GENERATION THROUGH DEEP-WELL INJECTION OF BIOSOLIDS M.S. Brunoa, J.T. Younga, O. Moghaddamb, H. Wongb, and J.A. Appsc a
Terralog Technologies USA, Inc., Arcadia, CA, USA City of Los Angeles Department of Public Works, Los Angeles, CA, USA c Lawrence Berkeley National Laboratory, Berkeley, CA, USA b
46.1 INTRODUCTION Almost 10 million wet tons of municipal sewage sludge (biosolids) are generated each year in the United States. The volume and the costs associated with disposal and recycling of this material are steadily increasing nationwide. For example, most sanitation districts in Southern California are forced to truck their biosolids more than 100 miles to rural areas in Kern County and Riverside County at processing and transport costs of $40 per wet ton or more. The prevailing methods for biosolids management include land applications (to rangeland, forests, or public parks), composting (mixing biosolids with other organic wastes such as wood chips), and landfill disposal. Most biosolids generated in Southern California are currently applied to the surface or placed in landfills, where they degrade and are released into the atmosphere each year as several hundred thousand tons of carbon dioxide. A primary obstacle confronting biosolids land application is public perception and nuisances such as odor and wind-blown dust. Severe weather conditions can hamper land application for indefinite periods, leaving districts scrambling for alternative disposal techniques. Trucks hauling large volumes of waste on public roads and highways create additional risks and nuisance. In areas where biosolid material is disposed into municipal waste landfills, the capacities of these limited resources are also stretched. Greenhouse gas generation (primarily CO2 and CH4), surface land impairment, and potential groundwater impacts are also byproducts of land application and landfill practices. Occasional local opposition to landspreading, combined with increasing volume and costs, have pressured sanitation districts nationwide to investigate alternative management options for biosolids. Terralog Technologies and the City of Los Angeles have therefore proposed a new and innovative technology for biosolids management through deep-injection disposal. The ideal geologic target for such a disposal method would be a high-porosity and high-permeability sand formation with an effective overburden seal, such as a depleted oil and gas formation or similar geologic trap. In the elevated temperature environment of the subsurface, the biosolids should undergo a natural process of anaerobic biodegradation, which could fully sterilize the material within days, and within months, convert a significant portion the organic mass into methane and carbon dioxide. The carbon dioxide, which would otherwise be released into the atmosphere during land application, would preferentially
ELSE_TSANG_CH046.qxd
588
9/8/2005
4:48 PM
Page 588
Thermal Treatment, Carbon Sequestration, and Methane Generation
dissolve into the formation fluid, because it is about ten times more soluble than methane at the pressure and temperature conditions typical of the subsurface. Through a combination of solubility trapping and mineralization, the CO2 would be sequestered while relatively pure methane would accumulate for possible recovery and use as an energy source.
46.2 PROPOSED TECHNOLOGY Terralog Technologies and the City of Los Angeles, working in coordination with the U.S. Environmental Protection Agency, Region 9, propose to demonstrate the technology of biosolids injection through a four-year pilot project in Los Angeles County. New wells for the project would be drilled to a depth of about 1700 m in an isolated fault block near the Wilmington Oil Field in Long Beach. The injection–reservoir targets will be the Ranger and Upper Terminal zones, where the reservoir temperatures are on the order of 125–145°F. Extensive monitoring, sampling, and parallel laboratory research would be conducted to better quantify biodegradation rates, long-term carbon sequestration, and optimum injection parameters for enhanced methane generation. This technology holds a number of very significant environmental advantages over current long-distance transportation and land application options. These advantages include: ● More rapid and more thorough thermal treatment (sterilization) ● Greater protection for surface and groundwater ● Reduced truck traffic and associated emissions ● Reduced greenhouse gas (CH and CO ) release to atmosphere 4 2 ● Potential recovery and beneficial use of generated methane as a clean fuel. Placing material at least 1500 m below any usable source of groundwater, with thick and clearly defined permeability barriers that block upward flow, is inherently more protective of groundwater than placing material directly on the surface, where it can percolate unimpeded to groundwater only tens of feet below. Furthermore, the high-temperature (≈55°C) saline environment (on the order of 20,000 ppm total dissolved solids (TDS)) existing at depth is extremely hostile to pathogens. Deep-well injection of biosolids therefore substantially reduces any potential impact to the surface water and groundwater when compared with surface application. Deep-well injection would substantially decrease the current truck traffic associated with biosolids transport, and allow biosolids to be managed within Los Angeles County. After successful conclusion of the demonstration phase, new injection sites could be constructed adjacent to the current Hyperion and Terminal Island sewage treatment plants, both of which are situated near existing oil and gas reservoirs. A permit application for an experimental demonstration project titled “Converting Biosolids to Methane through Deep Injection and Biodegradation,” was submitted to the U.S. EPA in June 2001. The proposed experimental program includes both field measurements (monitoring and fluid and gas sampling) and supporting laboratory experiments to better interpret and optimize the field observations. The objectives of the laboratory experiments are to demonstrate and measure to what extent: ● The high temperature in the deep subsurface (about 55°C or 130°F) will thermally treat (sterilize) the biosolids. ● The extent to which biodegradation in the subsurface will convert the biomass into methane and carbon dioxide.
ELSE_TSANG_CH046.qxd
9/8/2005
4:48 PM
Page 589
46.3 Hyperion Anaerobic Mesophilic and Thermophilic Digestion Pilot Test
589
The potential for carbon dioxide to be permanently sequestered through solubility trapping in the formation brine. ● Whether methane could be generated that might eventually be recovered for beneficial use. Although a definitive technical validation can come only from field demonstration and measurement, we believe there is a strong likelihood of success, because full-scale digestion studies have been completed by the City of Los Angeles at the Hyperion treatment plant, El Segundo, California, demonstrating sterilization and biodegradation at high-temperature conditions consistent with the deep subsurface. Small-scale laboratory experiments at the University of California at Los Angeles also clearly demonstrate that biodegradation and methane generation will occur under high-temperature and high-pressure conditions expected in the subsurface. The remainder of this chapter is devoted to a review of both the Hyperion digester tests and the high-temperature, high-pressure, biosolids-degradation tests. The chapter concludes with a discussion of the feasibility of biosolids waste disposal through deep-injection disposal in the subsurface. ●
46.3 HYPERION ANAEROBIC MESOPHILIC AND THERMOPHILIC DIGESTION PILOT TEST The pilot test to evaluate treatment levels and gas generation under mesophilic and thermophilic conditions was conducted at the Hyperion treatment plant during 2001. Data for October 2001 are described in this chapter. During this period, one digester was operated under thermophilic conditions (about 130°F or 54°C) and another digester was operated under sustained mesophilic conditions (about 96°F or 36°C) at similar feed rates, thereby allowing a direct comparison between these two operating temperatures. 46.3.1 Description of Facilities Hyperion currently employs 20 new egg-shaped digesters and six conventional digesters to stabilize primary and waste-activated sludge. About 2 million gallons per day (MG/D) or 7.6 ML/day of Primary Sludge (PS) settling from the advanced primary treatment process is pumped and distributed equally into the 26 digesters through primary sludge lines. Another 0.8 MG/D Thickened Waste Activated Sludge (TWAS) discharge from the Waste Activated Sludge Thickening Facility is also pumped and distributed into all the digesters. The 20 egg-shaped digesters are grouped into three operational batteries called D1, D2, and E. Digester D1 was set at thermophilic conditions, and digester D2 was set at the mesophilic condition. A summary of their operational conditions and resulting process parameters is presented in Table 46.1. Steam was used to maintain optimal process temperature. The mesophilic digester required about 80,000 lbs (36,000 kg) of steam each day to maintain an average temperature of about 96°F (36°C) for an average retention time of about 18 days. The thermophilic digester required more than three times as much steam, about 269,000 lb (122,000 kg) each day, to maintain a temperature of about 130°F (54°C) for an average retention time of 19 days. Both digesters produced a substantial amount of biogas, containing methane and carbon dioxide. The biogas is collected and conveyed to on-site gas storage and compressor facilities, where it is either piped to the Scattergood Power Generating Station, or is used in onsite boilers. The gas storage and compressor facility is also capable of routing excess digester gas through a gas flare facility.
ELSE_TSANG_CH046.qxd
9/8/2005
590
4:48 PM
Page 590
Thermal Treatment, Carbon Sequestration, and Methane Generation
46.3.2 Test Results Figure 46.1 summarizes fecal coliform counts measured in digesters D1 and D2 during October 2001. Under thermophilic conditions, the fecal coliform count after discharge averaged about 88 MPN/g (most probable number per gram), thus meeting Class A pathogen requirements set by the U.S. EPA (less than 1000 MPN/g). Under mesophilic conditions, however, the fecal coliform count averaged about 290,000 MPN/g. The higher temperature conditions therefore provide enhanced biosolids treatment. Figure 46.2 summarizes methane and carbon dioxide generated in digesters D1 and D2 during October 2001. Average biogas production rates are also summarized in Table 46.1. Under thermophilic conditions, approximately 1.31 MSCF/day (399 L/s at STP) of methane and 0.71 MSCF/day (216 L/s at STP) of carbon dioxide were generated from an average feed rate of about 0.85 MG/D (37.2 L/s). Under mesophilic conditions, approximately 1.54 MSCF/day (469 L/s at STP) of methane and 0.83 MSCF/day (253 L/s at STP) of carbon
Table 46.1. Hyperion plant mesophilic and thermophilic test summary
Average temperature Retention time Average steam consumption Average sludge feed rate Average digested gas production Average CH4 produced Average CO2 produced Average fecal coliform count EPA pathogen requirement met
Mesophilic digester-D2
Thermophilic digester-D1
96°F (36°C) 18 days 80 k lb/day (0.42 kg/s) 1.03 MG/D (45.1 L/s) 2.37 MSCF/day (721 L/s at STP) 1.54 MSCF/day (469 L/s at STP) 0.83 MSCF/day (253 L/s at STP) 290,000 MPN/g Class B
130°F (54°C) 19 days 269 k lb/day (1.41 kg/s) 0.85 MG/D (37.2 L/s) 2.02 MSCF/day (615 L/s at STP) 1.31 MSCF/day (399 L/s at STP) 0.71 MSCF/day (216 L/s at STP) 88 MPN/g Class A
Fig. 46.1. Comparison of fecal coliform count under thermophilic and mesophilic conditions.
ELSE_TSANG_CH046.qxd
9/8/2005
4:48 PM
Page 591
46.4 Experimental Verification of Biodegradation and Methane Generation
591
14
Gas Produced (SCF/lb VSL)
12 10
D1 -CH4 prod @ 130 deg F D2 -CH4 prod @ 96 deg F D1 -CO2 prod @ 130 deg F D2 -CO2 prod @ 96 deg F
8 6
D1 - thermophilic D2 - mesophilic
4
0
1-Oct 3-Oct 5-Oct 7-Oct 9-Oct 11-Oct 13-Oct 15-Oct 17-Oct 19-Oct 21-Oct 23-Oct 25-Oct 27-Oct 29-Oct 31-Oct
2
Date
Fig. 46.2. Methane and carbon dioxide production from D1 and D2 digesters.
dioxide were generated from an average feed rate of about 1.03 MG/D (45.1 L/s). As illustrated in Figure 46.2, the pilot tests indicate that both thermophilic and mesophilic digestion will create biogas at roughly equal rates. In terms of volatile solids loading, about 6–8 scf (160–210 L at STP) of methane were produced and about 3–4 scf (80–100 L at STP) of carbon dioxide were produced per pound (0.5 kg) of volatile solids. The average quantity of volatile solids destroyed is about 58 wt%.
46.4 EXPERIMENTAL VERIFICATION OF BIODEGRADATION AND METHANE GENERATION UNDER SIMULATED DEEP SUBSURFACE CONDITIONS In contrast to the operational conditions of the Hyperion plant, the biosolids in the subsurface would be exposed to higher salinity fluids, to pressure, and to minerals of the host formation. However, it would be expected that biodegradation would still occur, although initially at slower rates while bacteria adapt to the ambient environment. To test this hypothesis, laboratory experiments were conducted to demonstrate methane and carbon dioxide production from digested biosolids in the presence of high-salinity brine and core materials. Two sets of experiments were conducted. In the first set, Phase I, the effect of salinity, temperature, and the presence of core material on methanogenesis at ambient pressure were investigated. In the second set, Phase 2, the effect of biosolids degradation in the presence of brine was investigated at elevated pressure and temperature. 46.4.1 Phase I: Effect of Salinity, Temperature, and the Presence of Core Material on Methanogenesis at Ambient Pressure A series of experiments were conducted to measure the impact of salinity, temperature, the effect of core material, and added organic nutrients on methanogenesis and carbon dioxide
ELSE_TSANG_CH046.qxd
9/8/2005
592
4:48 PM
Page 592
Thermal Treatment, Carbon Sequestration, and Methane Generation
formation. The amount of biogas generation was measured at discrete time intervals of 9, 20, and 55 days, and in one case 9, 27, and 59 days respectively. The materials preparation and conditions for the tests are described in the following paragraphs. Materials preparation Biosolid sample preparation. Dewatered mesophilic Class B sludge samples (biosolids) from the Hyperion wastewater treatment plant were obtained on three separate days during November 2000. A composite biosolid sample was prepared by using an equal weight of sludge from each sampling date. The composite sludge was then manually mixed and homogenized with 50% (w/w) of an aqueous solution in a glass mortar and pestle. Core sample preparation. The Southern California Gas Co. provided core samples from the Sesnon formation (Aliso Canyon, Porter 25R well; 164 slabs, approximately 300 g each; depth range: 7820–7958 ft.). Sesnon core samples are compacted sand and rocks. In order to transfer core material to the serum vials used for the test, it was necessary to manually crush the slabs to smaller pieces. Particles with a size diameter smaller than 12 mm and bigger than 4.7 mm were selected using sieve No. 4 (U.S. Standard). (Note: 12 mm is the diameter of the mouth of the vial, and 4.7 mm is the biggest particle size that can be sieved with commercially available sieves.) A composite of core samples was prepared by manually mixing crushed material from cores 5 ft depth apart from each other. (The whole sampling depth range was covered, and enough core material was obtained for all the ambient pressure studies.) Test vial preparation and analysis of gas produced. Thirty grams of crushed Sesnon core material were transferred to 120 mL serum vials and 30 g resuspended sludge (15 g of dewatered sludge) were then added. These amounts provided a 1:2 (w/w) biosolids to core material ratio, and approximately 60 mL headspace to collect the gas produced. Vials were sealed with butyl rubber stoppers and crimped with aluminum seals. Vials were manually shaken to blend the core material with the biosolids as much as possible. Air present in the headspace of the vials was flushed out with O2-free nitrogen for 5 min. Vials were then equilibrated for one hour at the test temperature, and the pressure zeroed to atmospheric pressure using a 60 mL syringe. Sterile controls to determine the abiotic contribution to methanogenesis were also prepared. Killed controls were made by twice autoclaving vials containing core material/sludge, sludge only, or core material/fluid at 130°C for 30 min. The volume of gas produced was measured at ambient pressure with a 60 mL syringe and then wasted. The pressure in the vials was zeroed again to atmospheric pressure after the gas volume had been determined. A 0.5 mL gas sample was taken using a gas-tight syringe at atmospheric pressure, and gas composition (CH4 and CO2) was analyzed with a gas chromatograph equipped with a thermal conductivity detector. A stainless-steel column packed with Carboxen 1000 (60/80 mesh, 8 ft length, 1/8 o.d.) was used. The oven, detector, and injector were maintained at 85°C. Helium was used as carrier gas at a flow equal to 30 mL/min. Methods of evaluation Effect of temperature. The effect of temperature on methanogenesis was tested in a series of three experiments at 45, 55, and 65°C (113, 131, and 149°F), respectively. Effect of additional carbon sources on methanogenesis. Although the use of additional carbon sources will not be practiced when injecting biosolids under actual field conditions, it
ELSE_TSANG_CH046.qxd
9/8/2005
4:48 PM
Page 593
46.5 Results and Discussion
593
was important to test whether the native bacteria present in the biosolids have the required metabolic activities to produce methane. Two substrates that can be easily used during the methanogenic process were added. In test series G, glucose, which is easily fermented to acetate and other volatile fatty acids, was added to the resuspended sludge with a final concentration of 5 g/L. In series A, acetate, which is the principal substrate for methane production, was added to the sludge with a final concentration of 50 mM. Effect of salinity on methanogenesis. In order to test whether the salinity present in the brine would have any effect on methanogenesis, three levels of salinity were tested. Brine from the Porter well was provided by the Southern California Gas Co. Three series of experiments were prepared, one for each salinity level tested: ● Series B, full-strength brine (pH 7.2) ● Series BW, brine diluted with tap water to 50% (v/v) of full strength ● Series W, tap water. Effect of core material on methanogenesis. The effect of the core material on methanogenesis was tested. Vials containing only sludge resuspended in brine and incubated at 55°C (131°F) were compared against vials containing the core material and sludge resuspended in brine. Effect of sample size. A 1.1 L bottle was filled with 300 g of reconstituted sludge, and 300 g of core material, and incubated at 55°C in order to address a potential effect of the size of the serum bottles (120 mL) used in this study on the CH4 production.
46.5 RESULTS AND DISCUSSION The experimental results are summarized in Tables 46.2–46.7, and illustrated in Figures 46.3–46.5. Experiments are designated where appropriate according to the following notation: the fluid used (B for brine), the additional carbon source (G for glucose and A for acetate), and the temperature tested (in °C). Figures 46.3 and 46.4 show the overall results of the effect of temperature and additional carbon sources on CH4 and CO2 production from biosolids resuspended in full-strength brine. Labels at the bottom of each set of bars indicate the fluid used (B for brine), the additional carbon source (G for glucose and A for acetate) and the temperature tested (in degrees Celsius). From Figures 46.3 and 46.4, it is clear that bacterial populations with the required metabolic activities (hydrolytic, fermentative, and methanogenic) were present in the biosolids. Also, it is clear that the enzyme systems of these microbes are nearly saturated, because vials with additional glucose only produced a slightly greater amount of CH4 and CO2 compared to those without glucose. Acetate addition did not produce any significant effect on CH4 and CO2 production. This result also indicates saturation of the enzyme systems related to the conversion of acetate to CH4 and CO2. It is also apparent from Figures 46.3 and 46.4 that temperature exerts a greater inhibitory effect on CH4 production than on CO2 production. Figure 46.5 was obtained using the same data as for Figures 46.3 and 46.4 for those vials without an additional carbon source. Figure 46.5 shows that at 45°C, methane was initially produced at a faster rate for the first 9 days than for the remainder of the incubation period. At 55 and 65°C, inhibition of CH4 production was observed immediately after incubation with no indicated adaptation. Figure 46.5 also shows that CO2 production was very similar for the three temperatures tested during the
ELSE_TSANG_CH046.qxd
9/8/2005
594
4:48 PM
Page 594
Thermal Treatment, Carbon Sequestration, and Methane Generation
Table 46.2. Methane production (in mL) from biosolids resuspended in brine with core material present Experiment
9 days
20 days
55 days
B45 BG45 BA45 B55 BG55 BA55 B65 BG65 BA65
48.2 58.9 53.3 19.3 32.5 16.9 13.6 12.7 3.5
77.2 84.8 86.0 25.5 40.6 33.0 17.3 18.9 10.6
147.3 177.5 144.3 46.6 51.0 44.9 26.8 26.7 17.7
Table 46.3. CO2 production (in mL) from biosolids resuspended in brine with core material present Experiment
9 days
20 days
55 days
B45 BG45 BA45 B55 BG55 BA55 B65 BG65 BA65
38.4 47.1 43.2 37.3 52.2 36.1 33.9 36.7 20.6
68.1 75.9 73.7 53.7 71.3 59.0 44.1 53.6 39.8
103.4 119.4 104.1 74.8 81.8 71.0 60.5 67.6 53.0
Table 46.4. Effect of temperature on CH4 and CO2 production from biosolids resuspended in brine and mixed with core material Days
Volume in mL 0 9 20 55 Volume in cf/ton biosolids 0 9 20 55
CH4
CO2
45°C
55°C
65°C
45°C
55°C
65°C
0.0 48.2 77.2 147.3
0.0 19.3 25.5 46.6
0.0 13.6 17.3 26.8
0.0 38.4 68.1 103.4
0.0 37.3 53.7 74.8
0.0 33.9 44.1 60.5
0.0 102.9 164.8 314.5
0.0 41.2 54.5 99.5
0.0 29.1 37.0 57.2
0.0 82.0 145.3 220.6
0.0 79.7 114.6 159.6
0.0 72.3 94.2 129.1
first 9 days of incubation, and inhibition was observed for the remainder of the incubation period. These observations can be explained by the hypothesis that the methanogenic bacterial populations are more sensitive to high temperatures than the fermentative bacterial populations. This difference in temperature sensitivity could produce an imbalance in the bacterial populations. Fermentative bacteria at any of the tested temperatures could initially produce a large amount of acetic acid and other volatile fatty acids (VFAs) and CO2. Methanogenic bacteria grown at 45°C could further metabolize these VFAs to CH4 and CO2. However, at 55 and 65°C methanogenic bacterial growth could be inhibited, and VFAs
ELSE_TSANG_CH046.qxd
9/8/2005
4:48 PM
Page 595
46.5 Results and Discussion
595
Table 46.5. Methane and carbon dioxide production from biosolids resuspended in brine and mixed with core material (55°C, 1 L bottle) Days
CH4 (mL)
CO2 (mL)
0 9 27 59
0 208.7 281.5 365.0
0 412.1 587.7 705.3
Table 46.6. Effect of salinity on CH4 and CO2 production from biosolids mixed with core material (55°C) Days
0 9 20 55
CH4 (mL)
CO2 (mL)
Brine
50% Brine
Water
Brine
50% Brine
Water
0.0 19.3 25.5 46.6
0.0 16.9 33.0 44.9
0.0 50.9 58.8 69.9
0.0 37.3 53.7 74.8
0.0 35.6 54.7 71.2
0.0 41.8 59.3 75.5
Table 46.7. Effect of core material on CH4 and CO2 production from biosolids resuspended in brine (55°C) Days
0 9 20 55
CH4 (mL)
CO2 (mL)
With
Without
With
Without
0 19.3 25.5 46.6
0 41.7 56.3 70.3
0 37.3 53.7 74.8
0 41.7 56.3 70.3
would not be further metabolized. As a consequence, acetate and other VFAs would accumulate, CH4 and CO2 production would decrease, and the pH would fall. Levels of VFAs and pH were measured in the sludge resuspended in brine, and in contact with core material after 55 days of incubation to obtain further support for the postulated hypothesis. Table 46.8 shows the concentration of VFAs and the pH values. The pH was similar for the three temperatures tested, but the higher the temperature, the higher the acetate level found. Levels of the other VFAs measured were very similar at the three temperatures tested. These results further indicate that high temperatures inhibit methanogenic bacteria that metabolize acetate to CH4 and CO2. The observation that the accumulation of acetate did not produce a decrease of the pH at high temperatures may be explained by the alkalinity present in the brine. Table 46.9 summarizes the effect of the temperature on the total amount of CH4 produced after 55 days of incubation. Methane produced is referred to the total organic carbon (TOC) present in the reconstituted sludge. Also, it is referred to kg of dewatered sludge. The cumulative amounts of CH4 and CO2 in a 1 L bottle are shown in Figure 46.6. Methane produced per unit of TOC and kg of dewatered sludge is also reported in Table 46.9. The data indicate that the size of the container used for the test has no significant effect. The effect of salinity on CH4 and CO2 production is shown in Figure 46.7. Methane production is initially inhibited by full-strength brine (Fig. 46.7a). However, after 55 days of
ELSE_TSANG_CH046.qxd
9/8/2005
4:48 PM
596
Page 596
Thermal Treatment, Carbon Sequestration, and Methane Generation 200 9 days 20 days 55 days
180
Cumulative CH4 Produced (mL)
160 140 120 100 80 60 40 20 0 B45
BG45
BA45
B55
BG55
BA55
B65
BG65
BA65
Sample Designation
Fig. 46.3. Production of methane from 15 g biosolid samples resuspended in 15 g of brine in the presence of 30 g core material. See text for sample designations.
140 9 days 20 days 55 days
Cumulative CO2 Produced (mL)
120
100
80
60
40
20
0 B45
BG45
BA45
B55
BG55
BA55
B65
BG65
BA65
Sample Designation
Fig. 46.4. Production of carbon dioxide from 15 g biosolid samples resuspended in 15 g of brine in the presence of 30 g core material. See text for sample designations.
ELSE_TSANG_CH046.qxd
9/8/2005
4:48 PM
Page 597
46.5 Results and discussion
597
Cum. Methane Production (cf/ton)
500
45°C 55°C
400
65°C
300
200
100
0 0
10
20
(a)
30
40
50
60
70
Time (days) 500
Cum. CO2 Production (cf/ton)
45°C 55°C
400
65°C 300
200
100
0 0 (b)
10
20
30
40
50
60
70
Time (days)
Fig. 46.5. Biogas production from resuspended biosolids in brine mixed with core material as a function of temperature: (a) methane; (b) carbon dioxide.
incubation, the levels of CH4 produced in the presence of water and full-strength brine become similar, indicating a gradual adaptation of methanogens to the high salinity. This observation is in agreement with reports of methanogenesis in other hypersaline environments (i.e., marine sediments). Inhibition of carbon dioxide production was not observed at any level of salinity. The core material initially inhibited CH4 production (Fig. 46.8a). However, after 55 days of incubation, partial recovery of the methanogenesis was observed. The reason for this inhibitory effect is unknown. A possible explanation is that residues of oil present in the core material may inhibit methanogenic bacteria. However, this needs further verification. No inhibitory effect by the core material was observed on CO2 production (Fig. 46.8b).
ELSE_TSANG_CH046.qxd
9/8/2005
4:48 PM
598
Page 598
Thermal Treatment, Carbon Sequestration, and Methane Generation Table 46.8. Effect of temperature on levels of volatile fatty acids (VFAs) and on pH
VFA
Concentration (mM)
Acetic Propionic i-Butyric Butyric i-Valeric Valeric Total
45°C
55°C
65°C
8.99 41.22 14.22 3.07 12.86 0.94 81.32
115.10 35.76 14.89 5.88 12.65 0.93 185.24
118.49 36.90 16.93 3.78 14.35 0.75 231.2
Table 46.9. Amount of methane produced at tested temperatures from biosolids resuspended in brine in the presence of core material Temperature (°C)
CH4 (ft3/ton biosolids)*
CH4 (L/kg TOC)
45 55 65
120 mL Bottle
1 L Bottle
120 mL Bottle
1 L Bottle
273 86 49
68
347 110 63
86
*
Metric tons of dewatered biosolids.
800
Cumulative Gas Produced (mL)
700 600
CH4 (ml) CO2 (ml)
500 400 300 200 100 0 0
10
20
30
40
50
60
70
Time (days)
Fig. 46.6. Biogas production from 300 g of resuspended biosolids with 300 g brine mixed with 300 g core material at 55°C.
After 55 days of incubation, methane production declined to very low values with core material in the presence of either water, 50% brine, or brine. Production of a few milliliters of carbon dioxide was observed in vials containing core material and brine, or 50% brine amended with glucose. This observation may indicate that fermentative bacteria present in the brine have a very slow growth rate under these conditions.
ELSE_TSANG_CH046.qxd
9/8/2005
4:48 PM
Page 599
46.5 Results and discussion
599
80.0
Cumulative CH4 Produced (mL)
70.0 60.0 50.0 40.0 30.0 Brine 50% Brine Water
20.0 10.0 0.0 0
10
20
(a)
30
40
50
60
Time (days) 80
Cumulative CO2 Produced (mL)
70 60 50 40 Brine 50% Brine Water
30 20 10 0 0
(b)
10
20
30
40
50
60
Time (days)
Fig. 46.7. Biogas production from resuspended biosolids as a function of brine concentration at 55°C: (a) methane (b) carbon dioxide.
No CH4 or CO2 production was observed from sterile controls, indicating that the gases produced in the described experiments are of biological origin. 46.5.1 Phase 2: The Effect of Pressure on the Anaerobic Digestion of Biosolids in the Presence of Saline Fluid In contrast to the Phase 1 experiments in which mesophilic Class B biosolids were used, the Phase 2 experiments employed Class A thermophilic biosolids.
ELSE_TSANG_CH046.qxd
9/8/2005
600
4:49 PM
Page 600
Thermal Treatment, Carbon Sequestration, and Methane Generation 80
Cumulative CH4 Produced (mL)
70 60 50 40 30 20
With core material Without core material
10 0 0
10
20
30
40
50
60
Time (days)
(a) 80
Cumulative CO2 Produced (mL)
70 60 50 40 30 With core material Without core material
20 10 0 0
(b)
10
20
30 Time (days)
40
50
60
Fig. 46.8. Biogas production from resuspended biosolids in brine with or without core material at 55°C: (a) methane (b) carbon dioxide.
Equipment used The experiment was conducted using a 250 cc stainless-steel Tempco pressure vessel able to compress a mixture of biosolids to high pressure, as might be expected to occur within a fracture in the subsurface injection zone. The experimental arrangement is illustrated in Figure 46.9. The pressure vessel was wrapped with Thermodyne/BriskHeat heating tape, and surrounded by a 1.5 in. thick cylinder of insulation. A Fisherbrand digital temperature controller and indicator were used to maintain a constant temperature as recorded with a resistance temperature detector (RTD) type contact sensor placed on the outer vessel wall. The controller set
ELSE_TSANG_CH046.qxd
9/8/2005
4:49 PM
Page 601
46.5 Results and discussion
601
Fig. 46.9. Generalized schematic for high-pressure biodegradation test apparatus.
output power to the heating tape, maintaining a constant set temperature. The sensor had been previously calibrated to take into account small differences in temperature between the interior and exterior of the vessel wall. An external Enerpac hydraulic pump, connected to the downstream piston chamber, was used to maintain a constant pressure during the testing period. Experimental procedures The basic laboratory test procedure may be described as follows: 1. 135.5 gm of Class A biosolids (wet cake) from the thermophilic digester of the Hyperion treatment plant was combined with 25 mL of brine and placed into a 250 cc stainless-steel Tempco pressure vessel. The initial biosolids wet cake contained 30.2% total solids and 69.8% water by weight. The salinity of the brine mixture was 10,000 mg/L. The vessel was placed vertically, with the biosolids mixture placed above the piston. 2. Air was displaced from the vessel by slowly moving the piston upwards until only liquid could be seen exiting the upper vessel isolation valve. This valve was then closed. 3. Vessel temperature was increased to 50°C. 4. Piston pressure on the sample was increased to 2200 psia (15.17 MPa). 5. The sample was then left to biodegrade for a total of about 85 days. The vessel assembly was checked daily to verify that temperature and pressure remained at the set points. 6. Gas samples were collectzed and analyzed after 34, 56, and 85 days, in accordance with the procedures described in the following section, and transferred to the California Institute of Technology Petroleum Energy and Environmental Research (PEER) laboratory for gas component analysis. 7. At the conclusion of the test period, the remaining gas was purged, and the remaining biosolids mixture was removed and analyzed for volatile solids content. Sample analysis Identification and quantification of individual hydrocarbons and non-hydrocarbon gas components was carried out using a two-channel Hewlett-Packard 6890 Series Gas
ELSE_TSANG_CH046.qxd
602
9/8/2005
4:49 PM
Page 602
Thermal Treatment, Carbon Sequestration, and Methane Generation
Chromatograph (GC), which was custom-configured by Wasson ECE Instrumentation. This gas chromatograph (HP 6890) is designed to analyze small volumes of gas samples (a few mL of gas at atmospheric pressure) for light hydrocarbons (up to hexane), carbon dioxide, carbon monoxide, hydrogen sulfide, hydrogen, helium, nitrogen, and oxygen/argon. The GC was fitted with a capillary column, five-packed columns, a flame ionization detector (FID), and two thermal conductivity detectors (TCD). For hydrocarbon analysis, the GC oven was programmed from 85°C (5 min hold time) to 180°C at 15°C/min (10 min hold time) with FID detection on Channel 1. The non-hydrocarbon analysis was run on Channel 2 with the packed column array and dual TCDs for detection. Detector responses were calibrated using certified gas standards from Scott Specialty Gases (precision to within 1 mol% for each compound). The chromatogram was integrated using PE Nelson Turbochrom 4 software. The sample arrived in an assembly that had a valve at either end of a piece of tubing. After connecting the sample assembly into the system, one valve was opened into a custommade glass vacuum line with a residual pressure of 0.001 mbar. After isolation from the vacuum pump, the product gases were volatilized into the glass vacuum line. The heavier gaseous components were cold-trapped with liquid nitrogen (−196°C), while the lighter gases were concentrated into a precalibrated volume using a mercury Toepler pump. Replacing the liquid nitrogen with a mixture of dry ice and acetone (−80°C) released the other volatile species, excluding water and organic compounds heavier than C5. These gases were drawn into the same calibrated volume in order to determine the moles of gas. The number of moles of gas was calculated via the ideal–gas relationship. Experimental results The gas-component analyses reported by the Caltech PEER Center verified that biodegradation did in fact occur within the high pressure and high temperature reaction vessel within 30 days, and continued during the entire test period. Table 46.10 presents a summary of the current laboratory test results. Figure 46.10 presents a plot of methane and carbon dioxide content versus time. At the conclusion of the test, methane content is about 70 vol% and carbon dioxide content is about 27 vol% of the generated biogas. These percentages of methane and carbon dioxide reported are generally consistent with large-scale, high temperature, anaerobic digestion tests previously conducted by the City of Los Angeles as discussed above.
46.6 SUMMARY AND CONCLUSIONS The pilot plant tests, using the Hyperion treatment plant digesters under both mesophilic and thermophilic conditions, i.e., at 36°C and 54°C respectively, show that both methane and carbon dioxide were generated in significant quantities in a ratio of approximately 2:1. The average quantity of volatile solids destroyed was about 60 wt% after 18–19 days. Although the thermophilic conditions resulted in incrementally less methane and carbon dioxide production, the coliform bacterial count was drastically reduced to 88 MPN/g, thereby meeting Class A pathogen requirements set by the U.S. EPA. When dewatered Hyperion mesophilic sludge samples were subjected under laboratory conditions to elevated temperatures between 45 and 65°C in the presence of brine and core material, it was found that methane production is strongly inhibited at 55 and 65°C, i.e., elevated temperature is the most inhibitory parameter for methane production during anaerobic digestion in the presence of brine and core material. Carbon dioxide production was also curtailed,
ELSE_TSANG_CH046.qxd
9/8/2005
4:49 PM
Page 603
46.6 Summary and Conclusions
603
Table 46.10. Summary of laboratory results for high-pressure and high-temperature biodegradation tests. Test conditions: temperature (°C): 50; pressure (psi g.): 2200 Materials load in the vessel
HTP Class A biosolids (12/18/02) Brine solutuion (NaCI w/digester effluent) At end of test (03/13/03)
Weight (g)/ vol (mL)
%Total solid (%) salinity (mg/L)
% Volatile solids
Fecal coliform (MPN/dry g)
135.5 g
30.20
61.30
⬍6.9 (under detection limit)
25 mL
10,000 56
⬍6.9 (under detection limit)
Test results reported by Gas sampling date 1/21/2003 Sampling time 34 after initial test (days) Volume (%) Methane CH4 53.815 Carbon dioxide CO2 15.353 Nitrogen N2 29.19 Oxygen/Argon 0.425 Hydrogen sulfide H2S N.D.* Total (%) 98.783 Volume (mL at STD) Collected for analysis 5.302 Vent out and waste 5.302 Gas collected 10.604 Total gas collected (mL at STD)
2/12/2003 56
3/13/2003 85
71 26.1 2.98 0.07 N.D. 100.15
69.622 26.809 3.5 0.248 N.D. 100.179
37.29 37.29 74.58 ⬎183
17.76 >80 >97.76
Nondetectable.
*
100 90 80 Normalized %
70 60 50 40
CH4 % CO2 %
30 20 10 0 1/11/2003 1/21/2003 1/31/2003 2/10/2003 2/20/2003 3/2/2003 3/12/2003 3/22/2003 Sampling Date
Fig. 46.10. Methane and carbon dioxide content of sampled gas at 2200 psi pressure.
ELSE_TSANG_CH046.qxd
9/8/2005
604
4:49 PM
Page 604
Thermal Treatment, Carbon Sequestration, and Methane Generation
but to a lesser extent. The methanogenic bacteria that metabolize acetate to CH4 and CO2 (aceticlastic methanogens) seem to be the most sensitive to inhibition at elevated temperature. The high salinity present in the brine also inhibited the methane production. However, after 55 days of incubation, a gradual acclimation of methanogens was observed. This may indicate that longer incubation times will relieve the inhibitory effect caused by the brine. Also, the use of water to pump the dewatered biosolids may improve the CH4 production rate. Core material also inhibited methane production, although the reason for this effect is unknown. However, methanogenesis was observed to recover partially after 55 days of incubation. The Phase 1 studies suggest that the presence of brine, core material, and high temperature adversely affect the anaerobic digestion of wastewater biosolids. However, the process still is capable of transforming organic matter into CH4 and CO2 at a lower rate. After 55 days of incubation, 63 ft3of CH4 per metric ton of dewatered biosolids were produced under the most unfavorable conditions (65°C, full strength of brine and in presence of core material). Over time, the microbial populations present in the injected biosolids would be expected to adapt to the conditions found in the deep well. The use of biosolids digested under thermophilic conditions (55°C) could solve the inhibitory effect observed on sludge that was treated under mesophilic conditions (35°C), as was demonstrated in the Phase 2 laboratory tests. In these tests, conducted at 2200 psia (15.17 MPa) at 50°C using biosolids from the thermophilic digester in the presence of brine, gas production was similar to that observed during the thermophilic digestion at the Hyperion plant. In the Phase 2 tests, biogas was clearly generated with relatively high methane content (about 70 vol%). Although the conditions encountered in the subsurface can never be precisely duplicated, a similar biodegradation process to that observed in the Hyperion and laboratory-scale tests should occur when biosolids are injected into a deep subsurface reservoir. The underground in situ temperatures would autoclave and digest the biosolids, thereby sterilizing and converting the sludge into benign materials. The biosolids would be exposed indefinitely in situ to natural thermophilic conditions, which would be similarly anaerobic (lacking oxygen). Therefore, it is reasonable to assume that the biosolids will biodegrade into methane and carbon dioxide in a similar manner. The observations from all of the tests conducted to date provide some insight on expected methane generation volumes in the field. In the Phase 2 tests, almost 20% of the available volatile solids were destroyed within 90 days. We can therefore expect that in the deep subsurface within the four-year demonstration period, most of the volatile solids content injected during the first year would be destroyed, and converted to approximately 70% methane and 30% carbon dioxide. Furthermore, the tests have demonstrated that methane can be generated even from highly digested Class A biosolids. The Phase 2 tests have demonstrated that biodegradation of municipal biosolids can occur at the high pressure and high-temperature conditions expected in the subsurface at about 1500 m depth at Terminal Island, providing an incentive to proceed with field experimentation and demonstration. In the field experiment, it is planned to inject class B material with a higher organic volatile solids content, which should lead to additional gas generation. ACKNOWLEDGMENTS Some of the laboratory experiments described in this chapter were conducted by Felipe Alatriste-Mondragon and Birgitte Ahring at the University of California, Los Angeles. Gas sampling was completed by Mark Haught and Yongchun Tang at the Caltech PEER Center.
ELSE_TSANG_CH047.qxd
8/23/2005
8:45 AM
Page 605
VII. CO2 INJECTION
605
This page intentionally left blank
606
ELSE_TSANG_CH047.qxd
8/23/2005
8:45 AM
Page 607
Chapter 47
THE POTENTIAL FOR CO2 SEQUESTRATION IN LARGE AQUIFER STRUCTURES IN NORTHEASTERN GERMANY M. Stöwer, W. Gilch, and J. Zemke Untergrundspeicher- und Geotechnologie-Systeme GmbH Mittenwalde, Mittenwalde, Germany
47.1 INTRODUCTION By the late 1990s, the interrelationship between the emissions of greenhouse gases and global warming was generally accepted by most international governments. Efforts were made for an agreement on reducing greenhouse gases and, as a result, the protocol of Kyoto was drafted in December 1997. Within this framework, Germany decided to reduce its emissions substantially. The current intention of the German government is to reduce the emissions of greenhouse gases by 21% in the period from 2008 to 2012, based on the data of 1990 (Bundesministerium, 2000). Furthermore, since an increasing atmospheric content of CO2 is regarded as a major factor in climate change (IPCC, 2001), Germany planned, within a national climate protection program, to reduce CO2 emissions 25% by 2005 (relative to 1990) (Bundesministerium, 2000). The setting of this ambitious target was influenced by the development of CO2 emissions in Germany since 1990. Data about emissions are published regularly as total emissions, for different fuel types and for different sectors of emittants (Ziesing, 2003). As shown in Figure 47.1, by 1995, emissions were reduced by more than 100 Mt, or more than 10%. Thereafter, emission reduction decreased, seemingly approaching a kind of steady state at around 850 Mt/a, which is significantly above the target of 760 Mt/a. The decline in CO2 emissions was initiated by the changing political and economical environment in Germany associated with the reunification. By comparing the data for single components with the total emissions (Fig. 47.1), it is clear that the changing of fuel types in Eastern Germany, i.e., the decreasing amount of lignite being used, caused this initial lowering. The still-existing difference of ⬃90 Mt (temperature-corrected) between current and target emissions requires a reduction rate of 30 Mt/a, a rate that was reached only in the very early 1990s. According to Ziesing (2003), it is therefore impossible to meet the target of a 25% reduction by 2005 through conventional methods (such as more efficient energy use or an increased use of renewable energy sources). Although no solution exists for reducing the emissions to the needed amount by 2005, sequestration of CO2 is one option for meeting the target in the next decade. Combustion of fossil fuels is the main source for CO2 emissions, and any concept of sequestration has to deal with this source. Because sequestration is limited to immobile emission sources, and power generation is the main source of CO2 in Germany, we must obviously consider power generation for any demonstration project.
ELSE_TSANG_CH047.qxd
608
8/23/2005
8:45 AM
Page 608
The Potential for CO2 Sequestration in Large Aquifer Structures in Northeastern Germany
Fig. 47.1. Temperature corrected CO2 emissions of Germany for 1990–2002 after Ziesing (2003). Data for 2000–2002 are preliminary.
The data published here result from studies carried out primarily by UGS GmbH (UGS) and secondarily by VNG AG (VNG, a large gas supplier in East Germany). Other work, directed more toward a pilot project, is expected to follow. 47.2 SEQUESTRATION IN GERMANY The idea of CO2 sequestration in Germany is not new. Sequestration technology has similarities in its technological and geological background to underground storage of natural gas. The first attempts at CO2 sequestration were made by UGS, specialists in construction and operation of underground storage facilities, who, in the early 1990s, attempted to find industrial partners for this task. However, because at that time there was little political or economic reason to pursue this endeavor, no project was initiated. Currently, with the political and economic framework having changed, there are a number of European Union (EU) funded projects (NASCENT, GESTCO, SACS, CO2STORE), in which the Federal Institute for Geosciences and Natural Resources is the main German participant. There are still just a few projects financed and carried out by private companies in Germany, but interest is likely to grow in such projects as discussion about taxes and emission trading intensifies. The potential contribution of CO2 sequestration to emission lowering in Germany is strongly related to regional and local considerations. Geological conditions differ considerably throughout the country (e.g., hydrocarbon deposits are concentrated mainly in the north and the very south). An overview of sequestration options in Germany and their qualitative and quantitative ranking are shown in Table 47.1 (Zemke et al., 2003). Because of the relatively limited sequestration potential and possible safety problems, the option to use salt mines, coal mines, and oil fields is suitable only in rare instances. With respect to deep coal seams, there is uncertainty about the capacity of such sites as well as the technology to be used. So, as of now, the most promising options for CO2 sequestration in Germany are the use of saline aquifers and gas fields. Only a few gas fields exist in northeastern Germany, and their capacity for CO2 sequestration is rather small. On the other hand, the geological setting is optimal for the use of
ELSE_TSANG_CH047.qxd
8/23/2005
8:45 AM
Page 609
47.3 Geological Conditions in Northeastern Germany
609
Table 47.1. Sequestration options and ranking according to qualitative criteria Option
Capacity (109 t CO2)
Permeability Pore fluid
Cap rock integrity
Security Enhanced Conflicts recovery of interest
Gas fields Saline aquifers Deep coal seams Oil fields Coal mines Salt mines
2.6 23–43 0.34–1.7 0.11 0.78 0.04
⫹
⫹ ⫹ ⫺ ⫹ ⫺ ⫹⫹
⫺ ⫹ ⫹ ⫺ ⫺⫺ ⫺⫺
⫺ ⫹ ⫹⫹
⫹ ⫺ ⫺ ⫺ ⫹⫹ ⫹⫹
⫹⫹ ⫹ ⫹
⫺ ⫺ ⫺ ⫺⫺ ⫺⫺
Prepared by the Federal Institute for Geosciences and Natural Resources published in Zemke et al. (2003). Rankings go from “very good” (⫹ ⫹) to “good” (⫹) to “poor” (⫺) to “very poor” (⫺ ⫺)
saline aquifers. Four decades of hydrocarbon/geothermal exploration and experience with underground storage of natural gas provide a good base of geological data. These data form a starting point for further regional studies and perhaps for the development of a demonstration plant.
47.3 GEOLOGICAL CONDITIONS IN NORTHEASTERN GERMANY The area of interest is located in northeastern Germany and includes the federal states of Mecklenburg-Vorpommern in the north, Brandenburg and Berlin in the south, and some parts of Sachsen-Anhalt in the southwest (Fig. 47.2). The area is limited by the extent of the North German Basin and by the available data. Although the basement structures are important for later development (because they set the initial conditions and were moreover partly reactivated later), the first strata relevant to this study are of Permian age. The clastic early Permian (Rotliegend) contains several gas fields in northern Germany, but for aquifer structures, only the evaporitic late Permian (Zechstein) is relevant, as the main driving force for the anticline buildup. During the Zechstein period, the area was situated in the basinward position of the southern Permian Basin (Ziegler, 1989). Cyclic transgressions through a northern connection to the ocean were caused by tectonic activities in the North Atlantic domain and glacio-eustatic sea-level changes. These transgressions and activities led, in combination with a generally arid climate and paleogeographic setting, to the sedimentation of the Zechstein evaporites. There are seven Zechstein cycles known in the Permian Basin, but only the Z2–Z4 (Staßfurt, Leine, Aller) play a major role in subsequent development, when salt accumulation in diapirs and pillows took place. The initial thickness of the salt is estimated to be ⬃1100–1500 m. Generally, middle and upper Mesozoic strata are most suitable for sequestration in northeastern Germany. Figure 47.3 shows a typical standard profile for one area. The lower Mesozoic is formed by the Triassic, which starts with the Buntsandstein, showing a mainly clastic succession in the lower and middle parts; whereas the middle Buntsandstein consists of typically alternated bedding of sandstones and silty claystones. These are in some places the stratigraphically deepest horizons suitable for sequestration. The upper parts of the Buntsandstein are characterized by pelitic, carbonatic, and partly evaporitic sequences. The overlying series of the Muschelkalk consists of different carbonate rocks and subdominant evaporitic layers, with only minor relevance for sequestration.
ELSE_TSANG_CH047.qxd
610
8/23/2005
8:45 AM
Page 610
The Potential for CO2 Sequestration in Large Aquifer Structures in Northeastern Germany
Fig. 47.2. Study area in northeastern Germany.
The upper Triassic (Keuper) is formed by varying pelitic and clastic series in which, particularly in the uppermost parts (Rhaetian Sandstones), some target horizons for CO2 injection exist. The Schilfsandstein of the middle Keuper contains a few places of interest as well. Within the Keuper, some of the old basement structures were reactivated (Beutler, 1995) and, in combination with the lithostatic pressure and some heterogeneities of the Rotliegend surface, caused a mobilization of the Zechstein salt. The salt accumulation in flat pillows led to the development of primary sinks in adjacent areas, influencing the sedimentary environment. The Jurassic is mainly formed by pelitic and clastic series, both of which contain sandstones suitable for sequestration (Hettangian, Toarcian, Aalenian) and seals of regional importance. Owing to intensified salt movements (e.g., the first diapirs begin rising in the Jurassic), the thickness of the series varies strongly, from 50 m in an erosional environment at a salt top to more than 600 m in peripheral sinks. The Cretaceous is lithologically characterized by clastic, pelitic, and subdominant carbonate layers in the lower part and chalk, marl, and other carbonate strata in the upper part. Whereas some clastic layers of the lower Cretaceous may locally form target layers, the marls of the upper Cretaceous are of regional importance as a seal. Structural development in the Cretaceous was dominated by two mechanisms: (1) most of the salt accumulations reached their peak activity and therefore strongly influenced the sedimentation; and (2) an inversion of the former basin led to structural uplift (Baldschuhn et al., 2001). Combined with the transgression in the early upper Cretaceous, strong erosion influenced the local sedimentary succession. Tertiary sediments are poorly lithified and consist mainly of a clastic series with moderately high coal content. Sedimentation is still influenced by the ongoing salt movement. A specific clay layer of the Oligocene is important as a regional seal. In an uneroded succession, it is this clay that separates the strongly mineralized deep water from the shallower fresh water. As a result of the generally shallow Tertiary series, injection of CO2 is not possible in this unit. The development of the Quaternary is dominated by the different ice ages in northern Europe. Processes related to a glacial environment impacted the local geology in the studied area in two ways. First, the glacio-fluvial erosion cut deep locally and eroded the Tertiary seals. These stream channels are known to reach a depth exceeding 300 m. Because of the cyclically returning ice ages (three are most significant for northeastern Germany), there are locally several generations of channels. Second, the subglacial sediments may act as a local upper seal, especially for the fresh water horizons used for water supply.
ELSE_TSANG_CH047.qxd
8/23/2005
8:45 AM
Page 611
47.3 Geological Conditions in Northeastern Germany
611
Fig. 47.3. Location standard profile, showing the succession of potential target horizons and seals.
The following geological features of the studied area are important for any evaluation of sequestration potential: ● The sedimentary succession is characterized by several clastic reservoirs and seals, varying in occurrence and thickness, depending on the local setting. ● The aquifer structures are generally anticlines, built by a salt accumulation in the deeper underground. ● The structural development of the anticlines went through various stages in different periods for each anticline. Depending on the amount of salt to be accumulated and the
ELSE_TSANG_CH047.qxd
612
●
8/23/2005
8:45 AM
Page 612
The Potential for CO2 Sequestration in Large Aquifer Structures in Northeastern Germany
characteristics of the uplifted overburden, each anticline shows individual fault configurations at the top. Erosion of seals occurred individually for each anticline. The cross section in Figure 47.4 shows some examples of these features.
47.4 EVALUATING THE STRUCTURES There are more than 40 structures in the studied area (Figs. 47.5 and 47.6). For selecting appropriate anticlines for CO2 sequestration, we split the evaluation into different themes: 1. Reservoir characteristics: Based on core data, seismic features, and regional trends, the characteristics of the potential reservoir rocks were defined. The focus was on depth, areal extension, thickness, lithology, porosity, permeability, lateral development, and extent of facies. 2. Seal characteristics: Based on the same data used for the reservoir rocks, nearly the same parameters were analyzed. Since the seals were not targets of the previous explorations, the data pool was limited, and the comparison of regional trends became more significant. 3. Structure: Specifically, data about faults and their characteristics and the structural amplitude of the reservoir strata were analyzed. As a first step, an overview incorporating these three themes was created for candidate structures, and then, secondarily, the most suitable structures were selected for a closer analysis. For the selected structures, we performed static volumetric calculations with consideration of pore space, spill point, reservoir pressure, and reservoir temperature. If there is further need, these data could be transformed for dynamic simulations using the ECLIPSE simulator. The distances to existing CO2 point sources were not considered, because it was possible to construct a demonstration plant at a new location, and there was no need to use existing power stations as a CO2 source. Furthermore, the question of how to economically optimize the location for sequestration (with regard to pipeline length) has still not been answered. As a result of this effort, we developed an inventory of potentially suitable structures, providing a useful base for any further work.
Fig. 47.4. Schematic cross section of the study area. Note the varying thickness of the Mesozoic strata and the erosional discordances caused by diapirs and the Quarternary. Modified after Kleinmachnow (1997).
ELSE_TSANG_CH047.qxd
8/23/2005
8:45 AM
Page 613
47.5 Parameter of Structures and Reservoirs
613
47.5 PARAMETER OF STRUCTURES AND RESERVOIRS Although there are more than 60 structures in the studied area, only 45 were selected for a closer analysis (after a quick-look judgment). Then, considering the available data pool and qualitative criteria, only 16 anticlines were selected for volumetric calculations (Fig. 47.5). The most common reason for abandoning structures was eroded or faulted seals, mainly at the top of the structure. Related to the Mesozoic succession of alternating reservoirs and seals, once the caprock tightness was proven, up to four target horizons in different depth intervals were analyzed. Owing to the different reservoir conditions and separating seals, each horizon was regarded as an individual reservoir. The depth of the reservoir tops varied from 465 to 1490 m below mean sea level (bmsl), with most of the variance occurring in an interval of 700–1200 m (Fig. 47.6). One of the preliminary suitability criteria was that for both financial reasons (drilling costs)
Baltic Sea
Baltic Sea
Poland
ROSTOCK
SCHWERIN
Highway Border
BERLIN
Structure, analyzed for the study Scale 0
10
Fig. 47.5. Map of analyzed structures for CO2 sequestration.
20
30 km
ELSE_TSANG_CH047.qxd
614
8/23/2005
8:45 AM
Page 614
The Potential for CO2 Sequestration in Large Aquifer Structures in Northeastern Germany
Fig. 47.6. Histogram showing the spread of depth (top) of suitable horizons.
and geological reasons (decreasing porosities and permeabilities at greater depth), a prospective site should not exceed a depth of ⬃1500 m. The amplitude between top and spill-point of the target layer varied strongly between the structures over a range of 100–400 m; the differences between single horizons in one structure are not that relevant. For volumetric calculations, the safest—not always the deepest—possible spill-point was chosen. Formation temperature and pressure in the studied area depended mainly on the depth of the target layer. The geothermal gradient was relatively constant, ranging only slightly, from 3.3 to 3.5 K/100 m. Formation pressures (final average storage pressures) were calculated with the idea of avoiding a migration of injected CO2 below the spill-point. Depending on reservoir properties, the pore fluid would be displaced by the CO2 and move laterally and vertically within the target horizon. If we compare the temperature/pressure data of the reservoirs (Table 47.2) with the phase diagram of CO2 (Fig. 47.7), it is clear that in most cases the gas would be in a supercritical state within the reservoir, or would soon reach a supercritical state, resulting from the increased pressures during sequestration. Because the analyzed anticlines are currently not used for storing natural gas, no mixing or layering phenomena, as described in Oldenburg (2003), have to be considered. As described in the geological background information, potential target horizons in the studied area can be made up of very different strata. The reservoir properties of the strata vary, depending on the initial sedimentary environment, the diagenesis or cementation, and structural movement. Thus, the reservoir parameters (i.e., porosity and permeability) show a significant spread. As expected, the spread is mainly controlled by facies heterogeneities and diagenetic processes and only slightly by depth (Fig. 47.8). The data pool for the reservoir properties was also quite heterogeneous, because the data were obtained from diverse sources such as core samples, wire-line loggings, and well tests. If we add the uncertainties regarding structural features, further investigations are clearly necessary to verify (or adjust) the basic parameters before any further planning. The amount of required exploration differs from one anticline to another.
ELSE_TSANG_CH047.qxd
8/23/2005
8:45 AM
Page 615
47.5 Parameter of Structures and Reservoirs
615
Table 47.2. Spread of formation temperature and pressure for suitable reservoirs. Not the initial pressure, but the maximum pressure near the spill-point is indicated Reservoir temperature (°C)
Number of reservoirs
(a) Formation temperature ⬍40 40–50 50–60 ⬎60
7 8 11 9
Reservoir pressure (bar)
Number of reservoirs
(b) Pressure ⬍100 100–125 125–150 ⬎150
9 6 8 11
Fig. 47.7. Phase diagram of CO2.
As a result, a potential capacity for CO2 sequestration was calculated for each target horizon of the 16 structures. As shown in Table 47.3, 45% of all reservoirs have a capacity below 50 ⫻ 106 t CO2, the minimum being around 10 ⫻ 106 t CO2. Although dissolution of CO2 in the reservoir fluid was not accounted for (according to Johnson et al. (2002), about 15% could be added), the calculated capacities are sufficient for a demonstration plant, but for large-scale, long-term industrial sequestration they are too small. For large-scale sequestration, only five reservoirs, with a capacity of more than 150 ⫻ 106 t CO2, were regarded as suitable. The situation changes if all potential reservoirs within an individual structure are summarized, i.e., if the simultaneous use of separated target horizons in one structure is confirmed to be suitable under technical, geological, and (most importantly) legal conditions. There were eight structures exceeding a capacity of 150 ⫻ 106 t CO2; among these, there
ELSE_TSANG_CH047.qxd
616
8/23/2005
8:45 AM
Page 616
The Potential for CO2 Sequestration in Large Aquifer Structures in Northeastern Germany
Fig. 47.8. Porosities and permeabilities of reservoirs related to the depth of top structure.
Table 47.3. Capacity for CO2 sequestration in analyzed reservoirs Capacity for CO2 (106 t)
Number of single reservoirs
Number of structures
⬍50 50–100 100–150 150–200 200–300 ⬎300
16 7 7 3 2 —
7 1 — 3 3 2
was one structure with five different target horizons within an interval of 640–1410 m bmsl, with a total capacity of more than 800 ⫻ 106 t CO2. As a first result, this regional study indicated good geological conditions for CO2 sequestration within aquifer structures in northeastern Germany. Because the high-capacity anticlines are not concentrated in one part of the study area, options (besides a long pipeline) would have to be considered for how to connect the major potential locations to a sequestration site.
47.6 CASE STUDY AT KETZIN A detailed study was conducted of the Ketzin anticline, ⬃20 km west of Berlin. The structure is characterized by an accumulation of Zechstein salt in the deeper subsurface. Jurassic sandstones are currently used in Ketzin for underground storage of natural gas. This storage facility is now shut down and will be completely abandoned in 2006.
ELSE_TSANG_CH047.qxd
8/23/2005
8:45 AM
47.6 Case Study at Ketzin
Page 617
617
For legal and technical reasons, using the Jurassic sandstones for CO2 sequestration is only possible after finishing the abandonment. Therefore, other strata were analyzed for sequestration potential. During the construction of the underground storage facility, data on deeper sandstones, geologically separated from the storage reservoir, were gathered that indicated a potentially useable reservoir. 47.6.1 Buildup of the Anticline The accumulation of the Zechstein salt started in the Triassic, with later periods of uplift during the Cretaceous and Tertiary. Currently, the salt shows a thickness of ⬃1600 m. Strata have been lifted several hundred meters. The spill-point of the target horizon, the Schilfsandstein, is expected to be at 710 m bmsl. The anticline shows a SSW–NNE orientation, with gently dipping flanks of ⬃15° (Figs. 47.9 and 47.10). The extensional uplift regime in this region caused a graben associated with antithetic faults to form in the structural top region. This graben is characterized by faults with low vertical displacements. During the last uplift, the Jurassic layer on top partly eroded, so that there is discordant contact between the uppermost seal (the Tertiary Rupel Clay) and the Mesozoic. Additionally, glacio-fluvial channels cut through the Tertiary and parts of the Mesozoic in the Pleistocene, but these are at a lateral distance from the reservoir. Apart from the discordances mentioned above, the stratigraphic buildup is generally undisturbed. Related to the structural position, the thicknesses of the strata are variable. (The standard profile is shown in Table 47.4.)
Fig. 47.9. Isobaths of the target horizon for the Ketzin anticline. Most drillings in the eastern part of the structure do not reach this horizon.
ELSE_TSANG_CH047.qxd
618
8/23/2005
8:45 AM
Page 618
The Potential for CO2 Sequestration in Large Aquifer Structures in Northeastern Germany
Fig. 47.10. Cross section of the Ketzin anticline. Compare with Table 47.4 for the single horizons. Table 47.4. Standard profile at the top of the Ketzin anticline Stratigraphy System
Formation
Quarternary Tertiary
Pleistocene Miocene Oligocene (Rupel) Eocene Toarcian Pliensbachian Lotharing Hettangian/Sinemurian Rhaetian Dolomitmergelkeuper Upper Gipskeuper Schilfsandstein
Jurassic
Triassic (Keuper)
*
Depth
Thickness Reservoir
(m bgl)*
(m)
properties
75
75
150 155
75 5
210 230 315 360 515 600 690
55 20 85 45 155 85 90
Fresh water aquifer — Regional seal Local seal — — Local seal Reservoir (natural gas storage) Regional seal Regional seal Regional seal Reservoir (CO2 storage)
bgl—below ground level.
47.6.2 Reservoir Properties The Schilfsandstein is known for its varying lithology, changing between the silt-dominated standard facies of a low energetic environment and the sandy reservoir facies of a high energetic environment. Both lateral and vertical interbeddings of the facies types are known. The net thickness of the Ketzin sandstones is ⬃35 m.
ELSE_TSANG_CH047.qxd
8/23/2005
8:45 AM
47.6 Case Study at Ketzin
Page 619
619
The drilled cores showed poor lithification of the reservoir and were only partly analyzed. The grain-size analysis found silty sandstone with porosities in the range of 21–27%. The formation was tested, and permeabilities of 0.5–1.0 D were calculated. Because of the poor areal data distribution, these locally measured permeability values were not representative of the whole reservoir. In addition to the heterogeneities of the lithology, there is evidence of a secondary pore space cementation in some areas. The following calculations under “Reservoir Capacity,” below, are based on conservative parameters, with a porosity of 24% and a permeability of 0.5 D: 47.6.3 Reservoir Seal There are several seals at varying depths for the Schilfsandstein. The lowermost seal is formed by the succession of Upper Gipskeuper and the Dolomitmergelkeuper, with a total thickness of ⬎200 m for the whole anticline. The described faults in the top region show only little vertical displacement and therefore do not influence reservoir tightness. This succession separates the gas storage reservoirs from the CO2 storage reservoir. The properties of the aforementioned seal were analyzed in the western part of the anticline. Because of the regionally homogeneous lithology distribution in these layers, the results are applied to the main part of the structure. The detected porosities were in the range of 0.5–1.9 %, and the capillary displacement pressures were high. Pore sizes were small and show a statistical maximum in the range of the smallest pore sizes. It is concluded that the reservoir tightness is given. The need for observation wells could be determined from the results of the detailed exploration (and may in any case be demanded by mining authorities). Additionally, the geological suitability of additional, overlying seals, preventing gas migration to near-surface levels, have already been confirmed by several decades of underground storage operations. 47.6.4 Reservoir Capacity The calculation of reservoir capacity started with estimating the pore volume of the Schilfsandstein, considering the structural features and the lithology. The following parameters were averaged for the entire reservoir: ● Top of structure: ⬃610 m bmsl ● Spill point: ⬃710 m bmsl ● Net thickness: ⬃35 m ● Porosity: ⬃24% This leads to a pore space in the range of 200 ⫻ 106 m3. To prevent CO2 migration below the spill-point, a vertical safety distance was used; hence, the pore volume usable for CO2 sequestration was reduced to ⬎130 ⫻ 106 m3. At the initial reservoir conditions at the start of sequestration, CO2 would not be in a supercritical state. But it is expected to enter a supercritical state after relatively small amounts of CO2 are injected and the reservoir pressure increases. The reservoir conditions during sequestration were: ● Average pressure: ⬃75 bar ● Average temperature: ⬃35°C ● Average gas saturation: ⬃50% ● Average z factor for CO : ⬃0.4 2
ELSE_TSANG_CH047.qxd
620
8/23/2005
8:45 AM
Page 620
The Potential for CO2 Sequestration in Large Aquifer Structures in Northeastern Germany
We calculated a storage potential for CO2 of 10.5 ⫻ 109 m3 or 21 ⫻ 106 t under static conditions in the Ketzin anticline, without accounting for dissolution of CO2 in the reservoir fluid. Using this volume, we determined that the Ketzin anticline could store the CO2 emission of a 300 MW gas-fired power plant for an operational period of 25 years, without any CO2 being emitted into the atmosphere. 47.6.5 Well Construction The wells initially drilled for exploration purposes were planned to be integrated into the injection process at a later stage. The well at the top region could be used as the operation well, with the wells in the flanks acting as observation wells for pressure development in the reservoir during injection. Well construction is comparable to the underground storage of natural gas, with some adaptations related to the local conditions. The injection well should be completed with a perforated 7 in. production string and a coated 4½ in. tubing over the entire length of the well. For the exploration wells, which in this stage of operation would be used as observation wells, a smaller diameter would be chosen. Both types of wells should be equipped with a packer installation above the injection interval. The tightness of the underground storage facility during the drilling operations through the reservoir has to be guaranteed by using special casings and cements. We also plan to have subsurface installation packers combined with corresponding landing nipples for setting plugs or, if needed, for installing instruments for surveying the injection. The 4½ in. tubing would be connected via seal units with the packer and installed up to the well head. To avoid corrosion, the tubings would be coated. At the surface, the well head would be installed at the 13 3/8 in. anchor string during drilling, to avoid gas flow out of the Jurassic sandstones (by controlling all annuluses). This is also a precondition for controlling the annulus during the injection of CO2. According to the temperature profile during the injection, the tubings are pulled in tension or sealed in the well head with a travel joint. All fittings in contact with the CO2 must be coated. 47.6.6 Material for Casings and Installations An examination of the equipment used during 100 years of CO2 production in the Vorderrhön (central Germany) showed that there is no (or only minor) evidence for corrosion in the pipelines. These pipelines are made of ST35b steel or (in the casings) of a ferritic, perlitic grain comparable to grade D steel (according to the GOST 632–64 standard), or the grades J 55 and K 55 (according to API). Installations and casings should thus be made of the steel grades mentioned above. 47.7 CONCLUSIONS This study confirms the suitability and high capacity for CO2 sequestration in northeastern Germany. Using existing data, several questions, according to Gale (2002), have been answered, but because we lack detailed and recent information, we still have a need for exploration and modeling at the start of future sequestration projects. If the integrity of seals is proven and if, as is typically the case in northeastern Germany, there are multiple seals, the use of anticlinal aquifers would prevent CO2 migration in the overburden or uncontrolled lateral migration. Although an adjustable monitoring system is required for each sequestration project, it should concentrate on only a few points.
ELSE_TSANG_CH047.qxd
8/23/2005
References
8:45 AM
Page 621
621
In general, the geological structures of northeastern Germany offer favorable conditions for CO2 sequestration. Although several specific problems are still not solved (e.g., the legal status of sequestration and the cost optimization of pipelines versus new plants), the state of knowledge about the studied area enables us to start planning a demonstration project for CO2 sequestration. The amount of injected CO2 should correspond to the reservoir capacity for producing detectable and certain effects, i.e., the amount of 1 Mt/a as proposed by the U.S. Department of Energy (DOE, 2002) is regarded as a minimum. Furthermore, the scale and technical aspects of the demonstration project should allow for a methodical transfer to other locations.
REFERENCES Baldschuhn, R., Binot, F., Fleig, S., and Kockel, F., 2001. Geotektonischer Atlas von Nordwest-Deutschland und der deutschen Nordsee—Strukturen, Strukturentwicklung, Paläogeographie. Geologisches Jahrbuch, A153, 88 S., 3 CD. Beutler, G., 1995. Der Einfluss der Mitteldeutschen Hauptabbrüche auf die Mächtigkeitsentwicklung der Trias., Berliner geowiss. Abh., A 168: Berlin, S. 3–42. Bundesministerium für Umwelt, Naturschutz und Reaktorsicherheit. Nationales Klimaschutzprogramm. Umwelt, 11, Berlin. Gale, J., 2002. Geological storage of CO2, What’s known, where are the gaps and what more needs to be done. GHGT 6 Abstracts, Kyoto, Paper No. A1-1. IPCC. Climate change 2001, The scientific basis (Contribution of Working Group I), in The Third Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge, UK, 2001. Johnson, J.W., Nitao, J.J., Steefel, C.I., and Knauss, K.G., 2002. Reactive transport modeling of geologic CO2 sequestration. GSA Meeting Abstracts, Denver, Paper No. 174-1. Kleinmachnow, S., 1997. Landesamt für Geowissenschaften und Rohstoffe Brandenburg. Atlas zur Geologie von Brandenburg-13. Oldenburg, C.M., 2003. Carbon dioxide as cushion gas for natural gas storage. Energy Fuels, 17: 240–246. U.S. Department of Energy, 2002. CO2 Capture and Storage in Geologic Formations. A white paper prepared for the NCCTI. Zemke, J., Stöwer, M., Arnold, C., Becker, W., May, F., Gerling, P., and Krull, P., 2003. CO2 sequestration in Germany—General conditions and first field studies. WGC 2003 Abstracts. Ziegler, P.A., 1989. Geological Atlas of Western and Central Europe. Shell Internationale Petroleum Maatschappij B.V., textbook. Ziesing, H.J., 2003. Nur schwacher Rückgang der CO2 Emissionen im Jahre 2002. DIW Wochenbericht, 8, 14 S.
This page intentionally left blank
622
ELSE_TSANG_CH048.qxd
8/23/2005
8:48 AM
Page 623
Chapter 48
DEEP INJECTION OF ACID GAS IN WESTERN CANADA S. Bachu, K. Haug, K. Michael, B.E. Buschkuehle, and J.J. Adams Alberta Energy and Utilities Board, Edmonton, Alberta, Canada
48.1 INTRODUCTION A significant portion of produced gas in the Alberta basin in western Canada contains varying percentages of hydrogen sulfide (H2S). To meet transportation requirements, H2S is removed from the sour gas in a process that generates acid gas (a mixture of H2S and CO2). Since 1989, regulatory agencies in western Canada have required that gas plants with a sulfur throughput of more than 1 t/d recover the sulfur from the gas stream rather than burn it in flare stacks or incinerators, as previously done if sulfur-recovery technology could not economically remove the sulfur. Significant volumes of acid gas are generated from the produced natural gas, and acid gas desulfurization is uneconomic in a weak sulfur market dominated by recovered sulfur. Therefore, more operators in the Alberta basin are turning to acid-gas disposal through on-site injection into deep formations (Connock, 2001). Although the purpose of the acid-gas injection operations is to dispose of H2S, significant quantities of CO2 (actually more than H2S) are being injected at the same time because of the cost involved in separating the two gases. More than 2.5 Mt CO2 and 2 Mt H2S had been injected into deep formations in western Canada by the end of 2003. Forty-two acid-gas injection operations had been approved in western Canada by the end of 2003, of which 39 were active. One operation, although approved, was never implemented, and two others were rescinded after an active period, one because the injection volume reached the approved limit, and the other because the gas plant producing the acid gas had been decommissioned. Most operations inject acid gas at a single site (single well), but a few inject or have injected at several sites, such that in total there are 48 injection sites for the 42 operations. At three operations, the acid gas is dissolved in water prior to injection, and the resulting weak acidic solution (“sour water”) is injected. At seven other sites, of which three have been rescinded, wet acid gas (i.e., acid gas with free water present) is injected. Dry acid gas (i.e., with no free water present) is injected at all other sites. Except for the cases of injection of acid gas in solution, the acid gas is injected as a dense fluid (liquid or supercritical). The average composition of the injected acid gas varies from 84% H2S and 14% CO2 to 2% H2S and 98% CO2. Acid-gas injection is a mature technology, and applications of this process are growing in number and size. Currently, there are close to 20 acid-gas injection operations in the United States, and the experience gained so far in North America can be applied elsewhere in the world. As gas resources become more depleted, more sour gas that requires desulfurization will be produced. Since a CO2 stream with no H2S is less corrosive and hazardous, the acidgas injection technology can successfully be expanded to large-scale operations that will reduce CO2 emissions into the atmosphere from large CO2 point sources through CO2 geological sequestration. Thus, understanding the technology and characteristics of these acid-gas
ELSE_TSANG_CH048.qxd
8/23/2005
8:48 AM
624
Page 624
Deep Injection of Acid Gas in Western Canada
injection operations will help in expanding the application of this method for safe disposal of acid and greenhouse gases. 48.2 SURFACE OPERATIONS After separation of any gas liquids, the produced sour gas is passed through a contactor or absorber tower in a one- or two-stage process, in which the sour gas typically comes in contact with an amine or amine-derivative solution. The amine, or its derivative, 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. The gas-saturated amine solvent is then 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 35–70 kPa; it must be cooled, then compressed, for deep injection. To optimize disposal and minimize risk, the acid gas needs to be injected (1) into a densefluid phase, to increase capacity and decrease buoyancy (hence decreasing the potential for migration and leakage); (2) at bottomhole pressures greater than the formation pressure, for injectivity; (3) at temperatures in the system generally greater than 35°C, to avoid hydrate formation, which could plug the pipeline and disposal well; and (4) with a water content lower than the saturation limit, to avoid corrosion. Thus, the requirements for the selection of a disposal site, and for facility design and operation, depend on acid-gas properties, mainly phase behavior, water content and solubility, and hydrate formation (Carroll and Lui, 1997; Ng et al., 1999). In their pure state, CO2 and H2S have similar phase equilibria (Fig. 48.1), with CO2 condensing at lower temperatures than H2S (Carroll, 1998a). The critical points are T ⫽ 31.1°C and P ⫽ 7380 kPa for CO2, and T ⫽ 100.2°C and P ⫽ 8963 kPa for H2S. The phase behavior of the acid-gas binary system is represented by a continuous series of two-phase envelopes separating the liquid and gas phases, located between the unary bounding systems in the pressure–temperature space (Fig. 48.1). The bottomhole pressure needed for injection
Fig. 48.1. Phase diagrams for methane (CH4), carbon dioxide (CO2) and hydrogen sulfide (H2S), and hydrate conditions for CO2 and H2S.
ELSE_TSANG_CH048.qxd
8/23/2005
48.2 Surface Operations
8:48 AM
Page 625
625
is provided by the pressure at the wellhead and the hydrostatic weight of the acid gas in the well, minus friction losses downhole and across perforations. The wellhead pressure is achieved by compressing the acid gas that leaves the amine regeneration unit, typically in four compression and cooling stages (Fig. 48.2). To avoid cavitation, the acid gas must not enter the two-phase region during compression. 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 (Fig. 48.1; Carroll and Lui, 1997). The hydrate-forming temperature at any pressure of an acid gas increases with the increasing content of H2S. If there is too little water, it 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, 1998b), thus operating above the hydrate-forming temperature is desirable. Compression temperatures maintained above 35°C will avoid hydrates forming (Fig. 48.2), thus preventing compressor breakdown and plugging. Similarly, maintaining the acid-gas temperature above 35°C during transport and injection will avoid plugging of the pipeline and injection well. Acid gas obtained after the removal of H2S and CO2 from the sour gas is usually saturated with water vapor in the range of 2–6%. In general, the ability of acid gas to hold water increases with temperature and decreases with the addition of small amounts of methane. Unlike the case of hydrocarbon gases, the solubility of water in both H2S and CO2 (and hence in acid gas) decreases as pressure increases up to 3–8 MPa, depending on temperature, after which it dramatically increases (Fig. 48.3). The solubility minimum reflects the pressure at which the acid-gas mixture passes into the dense liquid phase, where the solubility of water can increase substantially with increasing pressure, owing to the molecular attraction between these polar compounds (Wichert and Royan, 1996, 1997). This property of the acidgas mixture is used in naturally dewatering the acid gas, by compressing the gas from about 100 kPa to around 8–10 MPa for injection, and therefore reducing the water content to less than 0.5 mol%, to avoid pipe and well corrosion (Wichert and Royan, 1996, 1997). 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
Fig. 48.2. Four-stage compression cycle for acid gas with a 50% CO2 and 50% H2S composition.
ELSE_TSANG_CH048.qxd
626
8/23/2005
8:48 AM
Page 626
Deep Injection of Acid Gas in Western Canada
Fig. 48.3. Solubility of water in acid gas as a function of pressure for: (a) different acid gas compositions (CO2 and H2S) at 30°C, and (b) different temperatures for an acid gas with a composition of 49% CO2, 49% H2S and 2% CH4 (see Lock, 1997; Wichert and Royan, 1996, 1997).
is needed. By the fourth stage in a cycle (Fig. 48.2), compression at pressures between 3 and 5 MPa will tend to dewater the acid gas if there are no hydrocarbon impurities present (Fig. 48.3). Further compressing the acid gas increases the solubility of water in the acid gas, such that residual excess water dissolves into it, and more than counteracts the decrease in solubility caused by interstage cooling. Hence, a separate dehydration step is usually not needed because no free water is present in the system (Clark et al., 1998). By collecting the condensed water during each compression stage and keeping the temperature a few degrees above the hydrate stability temperature, the probability of hydrate formation and corrosion are minimized, and less expensive steels can be used for transporting and injecting the acid gas into the disposal formation. Higher temperatures, although required to avoid hydrate formation and plugging, lead to accelerated corrosion, and both plugging and corrosion can lead to containment failure.
ELSE_TSANG_CH048.qxd
8/23/2005
8:48 AM
Page 627
48.3 Injection Well and Subsurface Requirements
627
Thus, additional safety precautions involve the 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 corrosion-resistant steels. Corrosion monitoring is normally done to assess the effectiveness of these treatments. Although a number of safety valves are always installed in the well and in the surface facilities, to isolate the containment lines for the acid-gas injection system into small volumes, 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. An emergency planning zone (EPZ) (i.e., area which may be impacted by the release of H2S) is defined around the sour gas facility. To have an effective ERP, the area surrounding the injection site is equipped with H2S detection and alarm systems, windsocks, self-contained breathing apparatus, and remote unit and plant shutdown stations. Accidents, if they occur, have to be reported according to the ERP; however, no accidents have been reported to date. The operators are fully aware of the dangers and possible consequences of H2S escape, and are paying attention to their operations.
48.3 INJECTION WELL AND SUBSURFACE REQUIREMENTS Acid-gas injection currently takes place in the Alberta basin in western Canada, with most of the operations being located in Alberta and several in northeastern British Columbia (Fig. 48.4). Disposal wells in western Canada are grouped into four classes, depending on the nature of the injected fluid (AEUB, 1994). Class Ia 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 other gases. Class IV wells are used for injection of potable water or steam. This classification system serves to define well completion requirements. Because of the nature of acid-gas injection operations, the wells are considered 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, and include: 1. Identification of all geological zones using logs and/or cores. 2. Hydraulic isolation by cement of all potential hydrocarbon-bearing zones and of shallow potable groundwater aquifers, confirmed by a full-length casing log. 3. Injection through tubing, and filling of the annulus with a corrosion-inhibiting fluid. 4. Installation of safety devices 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. According to regulations, an acid-gas injection well consists of a central steel tubing string with an outer annulus bounded by a steel casing 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 annulus between tubing and casing is filled with a corrosive-resistant fluid. Several safety features are generally 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
ELSE_TSANG_CH048.qxd
628
8/23/2005
8:48 AM
Page 628
Deep Injection of Acid Gas in Western Canada
Fig. 48.4. Location of acid-gas injection sites in the Alberta basin, Canada.
subsurface disposal formation, which is pressure tested for integrity once a year. A downhole safety valve or a check valve is incorporated in the tubing string, so that if equipment fails at surface, the back-flow of acid gas from the formation to the surface is prevented. The wellhead of the injection well is similarly protected with valves. In Alberta, applications for acid-gas disposal must conform to the specific requirements listed in Chapter 4.2 of Guide 65, which deals with applications for conventional oil and gas reservoirs (AEUB, 2000). Requirements in British Columbia are modeled after those in Alberta. The selection of an acid-gas injection site needs to address various considerations that 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 specific location of the acid-gas injection well is based on a general assessment of the local and regional geology and hydrogeology, which is designed to evaluate the potential for leakage (Longworth et al., 1996) and includes: 1. Size of the injection zone, to confirm that it 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.
ELSE_TSANG_CH048.qxd
8/23/2005
8:48 AM
Page 629
48.4 Characteristics of Acid-Gas Injection
629
3. Location and extent of the underlying or lateral bounding formations. 4. Folding or faulting in the area, and an assessment of seismic risk. 5. Rate and direction of the natural fluid 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. Analyses of core from the disposal zone and caprock (if available). 10. A complete and accurate drilling history of offsetting wells within several kilometers of the injection well, to identify any wells or zones that may be impacted by the injected acid gas. In addition, the regulatory agencies require that environmental concerns must be addressed, such as injection-formation suitability, wellbore integrity, operating parameters (to ensure formation and well integrity), and optimization of the injection space, considered to be a limited resource. Assessment of the integrity of the acid-gas disposal zone is critical for approving applications to dispose of acid gas by deep injection. To avoid gas migration through the caprock pore space, the difference between the pressure at the top of the disposal formation and the pressure 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 saturating 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 (BHIP) is set by regulatory agencies in western Canada at 90% of the rock fracturing pressure, which has to be obtained through tests. In the absence of site-specific tests, BHIPs are limited by pressure-depth correlations, which are based on basin-wide statistical data for the Alberta basin. If acid gas is injected into a depleted oil or gas reservoir, the maximum BHIP is often set at the initial reservoir pressure or lower (90% of it). From this point of view, injection into a depleted oil or gas reservoir has the advantages of injection pressures being low and of wells and pipelines being already in place (Keushnig, 1995).
48.4 CHARACTERISTICS OF ACID-GAS INJECTION 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). This operation experienced elemental sulfur deposition, which plugged the injection well, but this problem was subsequently resolved (Longworth et al., 1996). In 1993, Renaissance Energy Ltd. applied, and was approved, for injecting 20% H2S and 80% CO2 into a watersaturated carbonate reef in central Alberta, and operations started in 1994. In 1995, PanCanadian Petroleum Ltd. (now EnCana), started mixing, at surface, acid gas with produced water at elevated pressure, and injected the solution into a water-saturated sandstone zone underlying an oil pool at Hansman Lake and into a deep, dolomitized carbonate-shelf aquifer at Thompson Lake in eastern Alberta (Kopperson et al., 1998a, 1998b). The Hansman Lake injection site was abandoned in 1997 for economic reasons. Also in 1995, Pennzoil Canada Inc. started injecting a mixture of 20% H2S and 80% CO2 at 120,000 m3/d
ELSE_TSANG_CH048.qxd
630
8/23/2005
8:48 AM
Page 630
Deep Injection of Acid Gas in Western Canada
into a depleted oil reservoir in a pinnacle carbonate reef at Zama in northern Alberta. Because the acid gas was miscible with oil in that reservoir, at one point during the operation, it was used in conjunction with a miscible-flood scheme to enhance the oil recovery (Davison et al., 1999). This site was subsequently suspended by AEUB because the operator greatly exceeded the injection pressure limit. The largest operation in the Alberta basin at Sukunka in British Columbia (with a licensed capacity of 900,000 m3/d, but operating only at 500,000 m3/d) was also the largest in the world until 2004, when it was possibly overtaken by Exxon’s La Barge operation in Wyoming. Acid-gas injection in western Canada occurs over a wide range of aquifer and reservoir characteristics, acid-gas compositions, and operating conditions. At three sites, acid gas is dissolved in water at the surface, resulting in sour water, prior to injection at depths less than 1000 m. Dry acid gas, with minor hydrocarbons and traces of water dissolved in the acid gas, is injected at all but six other operations. At these six operations, water is present in free phase. One operation, although approved, has never been built by the operator. Three sites were rescinded after several years of operations, because the gas plant was decommissioned or the reservoir volume was filled up and the operator had to find another injection site. Three other sites, all at the same operation, were successively suspended by the regulatory agency because the operator greatly surpassed the approved maximum wellhead injection pressure and pressured up the injection reservoirs. Currently, there are 41 injection sites at 39 active injection operations (at several operations injection occurs through several wells and/or into different disposal targets). This review henceforth covers all the current and past acid-gas injection operations in western Canada. At 26 sites, the acid gas is/was injected into deep saline aquifers. At 18 sites, injection took or takes place in depleted oil or gas reservoirs, and at 4 sites the acid gas is injected into the underlying water leg of depleted oil and gas reservoirs. The amount of H2S licensed for injection varies between 2 and 97%, with the balance comprising mainly CO2, and minor amounts of C⫹ gases. The in situ temperature and licensed injection pressure position the injected acid gas in the P–T space between the supercritical points for CO2 and H2S (Fig. 48.5). Table 48.1 shows the characteristics of the licensed acid-gas operations and of the injected acid gases (averaged since injection start until the end of 2003). Based on the estimated total injection volume and aquifer or reservoir capacity, the acid-gas injection sites are licensed to
Fig. 48.5. In situ acid-gas pressure and temperature for injection sites in the Alberta basin, Canada.
ELSE_TSANG_CH048.qxd
8/23/2005
8:48 AM
Page 631
48.4 Characteristics of Acid-Gas Injection
631
Table 48.1. Operating ranges of acid-gas injection schemes in western Canada Characteristic
Minimum value
Maximum value
Licensed H2S (mol fraction) Actual injected H2S (mol fraction) Actual injected CO2 (mol fraction) In situ acid gas density (kg/m3) In situ acid gas viscosity (mPa s) Maximum well head pressure (kPa) Maximum injection rate (103 m3/day) Actual average injection rate (103 m3/day) Maximum injection volume (106 m3)
0.05 0.02 0.14 204.8 0.02 3750 4.2 1.0 6
0.97 0.84 0.98 728.3 0.09 19,000 900 500 1876
Fig. 48.6. Size of approved acid-gas injection operations in the Alberta basin, Canada.
operate for a period of 10–25 years. Although one may initially expect an increase in operation size as knowledge and technology advances, the lack of a historical trend (Fig. 48.6) is more a reflection of economic and geological considerations than of technological knowhow. The actual injection rates and volumes since 1989 are significantly below the approved limits. In 2003, the cumulative injection rates for CO2 and H2S reached 0.45 and 0.55 Mt/yr, respectively. A total of approximately 2.5 Mt CO2 and 2 Mt H2S have been injected into deep saline aquifers and depleted reservoirs in western Canada to the end of 2003. The depth of surface casing in the injection wells range from 128 to 550 m, with diameters ranging from 219 to 340 mm. Production casing depths range from 856 to 3334 m, the diameter ranges from 140 to 244 mm, while the tubing diameter ranges from 60 to 178 mm. The annulus fluid is sealed by a packer and is filled with a corrosion-inhibiting fluid. The 29 carbonate and 19 sandstone injection formations are overlain mostly by thick, competent shales, although in several cases thick evaporites (anhydrite and salt) and tight limestones form the caprock. The range for various characteristics of the injection reservoirs and aquifers is shown in Table 48.2, while the characteristics of the disposal zones are
ELSE_TSANG_CH048.qxd
8/23/2005
632
8:48 AM
Page 632
Deep Injection of Acid Gas in Western Canada
described in more detail by Bachu et al. (2003). Net pay represents the more porous and permeable interval in a formation in which injection takes place. The density and viscosity of formation water at in situ conditions were calculated on the basis of water temperature, pressure, and salinity (Bachu and Adams, 2002, 2004). The original formation pressure in the disposal zones is generally subhydrostatic with respect to freshwater, which is characteristic of the Alberta basin. Acid gas dissolved in water is injected in relatively shallow reservoirs or aquifers (843–948 m depth). Otherwise, acid gas is injected in aquifers or reservoirs whose initial pressure varies between 6636 and 35,860 kPa (Fig. 48.7). The cases where pressure seems to be slightly above hydrostatic correspond to high-salinity formation waters. If the real formation-water density is taken into account, then pressures at all but three sites are hydrostatic or less. Two cases of abovehydrostatic pressures correspond to reefal gas reservoirs. The only overpressured case is that of injecting into a deep structural trap in the thrust and fold belt of the Rocky Mountains. In the case of acid-gas injection into depleted oil or gas reservoirs, the original reservoir pressure has been drawn down as a result of production, such that formation pressure at the start of acid-gas injection was less than the original formation pressure, sometimes significantly, reaching values as low as 1170 kPa. Production-induced drawdown also occurred in cases where waste was injected into an aquifer underlying an oil pool, and into an aquifer located very close to an oil pool. There were 10 occurrences where formation pressure at injection startup was below the critical pressure of CO2. Vertical stresses, SV, at the disposal sites were evaluated from density logs, and they increase with depth, with an approximate gradient of 23.8 kPa/m (Fig. 48.7a). Minimum horizontal stresses, SHmin, were evaluated from microfrac, minifrac tests, leak-off tests, and breakdown pressures in wells near the injection sites, since no tests were performed in the injection wells, and they increase with an average gradient of 16.7 kPa/m (Fig. 48.7a). Minimum horizontal stresses are smaller than the vertical ones, indicating that fractures are, or will be, vertical. Fracturing pressures, Pf, at the injection sites were provided by the operators, according to regulatory requirements, and they generally show an increase with depth, with a gradient of 19 kPa/m (Fig. 48.7b). The maximum BHIP was calculated on the basis of the maximum wellhead pressure (WHIP) and of the hydrostatic weight of the acid-gas column in the injection well, and reaches up to 39,000 kPa. The maximum BHIP is always lower
Table 48.2. Characteristics of the aquifers and oil or gas reservoirs used for acid-gas injection in western Canada Characteristic
Minimum value
Maximum value
Average injection depth (m) Formation thickness (m) Net pay (m) Porosity (%) Permeability (mD) Formation pressure (kPa) Formation temperature (°C) Water salinity (mg/L) Brine density (kg/m3) Brine viscosity (mPa s) Oil gravity (°API) Gas specific gravity
824 4 3 4 5 5915 34 19,740 998 0.36 16 0.573
3432 276 100 30 4250 35,860 110 341,430 1273 1.32 68 1.121
ELSE_TSANG_CH048.qxd
8/23/2005
48.5 Conclusions
8:48 AM
Page 633
633
Fig. 48.7. In situ characteristics of acid-gas injection operations in the Alberta basin, Canada: (a) vertical and minimum horizontal stresses (SV and SHmin, respectively), and (b) formation pressure, BHIP and rock fracturing pressure Pf.
than the fracturing pressure, on average at 62% of Pf, although in a few cases it is as high as 0.9 Pf, and in three cases it is actually lower than the original formation pressure (Fig. 48.7b). 48.5 CONCLUSIONS Currently, acid-gas injection in the Alberta basin in western Canada occurs over a wide range of aquifer and reservoir characteristics, acid-gas compositions, and operating
ELSE_TSANG_CH048.qxd
8/23/2005
634
8:48 AM
Page 634
Deep Injection of Acid Gas in Western Canada
conditions. To the end of 2003, approximately 2.5 Mt CO2 and 2 Mt H2S have been successfully injected into deep hydrocarbon reservoirs and saline aquifers in Canada alone. Together with similar acid-gas injection operations in the United States, these acid-gas injection operations indicate that acid-gas injection is a mature and safe technology that can be applied elsewhere in the world as increasingly more sour gas is produced from deep gas reservoirs. Furthermore, these acid-gas injection operations constitute a commercial-scale analogue for future large-scale CO2 geological sequestration efforts to reduce CO2 emissions into the atmosphere from large CO2 point sources. The technology and engineering experience developed at these acid-gas injection operations (i.e., design, materials, leakage prevention, and safety) can be easily adopted for large-scale CO2 geological sequestration operations, since a CO2 stream with no H2S is less corrosive and hazardous. However, as these operations are scaled up, the leakage risk moves from the near well to the surrounding area, where the uncertainty in geology and reservoir or aquifer characteristics is greater.
REFERENCES 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/g51-1994.pdf. AEUB (Alberta Energy and Utilities Board), 2000. Guide 65: Resources Applications for Conventional Oil & Gas Reservoirs. Alberta Energy and Utilities Board, Calgary, AB, 113–136. DOI eub.gov.ab.ca/bbs/products/guides/g65.pdf. Bachu, S. and Adams, J.J., 2002. Equations of state for basin geofluids: Algorithm review and intercomparison for brines. Geofluids., 2(1): 257–271. Bachu, S. and Adams, J.J., 2004. Equations of state for basin geofluids: Algorithm review and intercomparison for brines, Erratum. Geofluids., 4(3): 250. Bachu, S., Adams, J.J., Michael, K. and Buschkuehle, B.E., 2003. Acid gas injection in the Alberta basin: A commercial-scale analogue for CO2 geological sequestration in sedimentary basins. Proceedings of the Second Annual Conference of Carbon Dioxide Sequestration, Alexandria, VA, May 5–8. Carroll, J.J., 1998a. Phase diagrams reveal acid-gas injection subtleties. Oil Gas J., 96(9): 92–96. Carroll, J.J., 1998b. Acid gas injection encounters diverse H2S, water phase changes. Oil Gas J., 96(10): 57–59. Carroll, J.J. and Lui, D.W., 1997. Density, phase behavior keys to acid gas injection. Oil Gas J., 95(25): 63–72. Clark, M.A., Syrek, W.Y., Monnery, W.D., Jamaluddin, A.K.M., Bennion, D.B., Thomas, F.B., Wichert, E., Reed, A.E. and Johnson, D.J., 1998. Designing an optimized injection strategy for acid gas disposal without dehydration. Proceedings of the Seventy-Seventh Gas Processors Association Annual Convention. Tulsa, OK, pp. 49–56. Connock, L., 2001. Acid gas injection reduces sulphur burden. Sulphur., 272: 35–41. Davison, R.J., Mayder, A., Hladiuk, D.W. and Jarrell, J., 1999. Zama acid gas disposal/miscible flood implementation and results. J. Can. Petrol. Technol., 38(2): 45–54. Keushnig, H., 1995. Hydrogen sulphide–If you don’t like it, put it back. J. Can. Petrol. Technol., 34(6): 18–20. Kopperson, D., Horne, S., Kohn, G., Romansky, D., Chan, C. and Dugworth, G.L., 1998a. Injecting acid gas with water creates new disposal option. Oil Gas J., 96(31): 33–37.
ELSE_TSANG_CH048.qxd
References
8/23/2005
8:48 AM
Page 635
635
Kopperson, D., Horne, S., Kohn, G., Romansky, D., Chan, C. and Dugworth, G.L., 1998b. Two cases illustrate acid gas/water injection scheme. Oil Gas J., 96(32): 64–70. Lock, B.W., 1997. Acid gas disposal—A field perspective. Proceedings of the Seventy-Sixth Gas Processors Association Annual Convention. San Antonio, TX, March 10–12, 1999, Tulsa, OK, pp. 161–170. Longworth, H.L., Dunn, G.C. and Semchuk, M., 1996. Underground disposal of acid gas in Alberta, Canada: regulatory concerns and case histories, SPE Paper 35584. Proceedings of the Gas Technology Symposium. Calgary, AB, Canada, 28 April–1 May 1996, SPE, pp. 181–192. Ng, H.-J., Carroll, J.J. and Maddocks, J.R., 1999. Impact of thermophysical properties research on acid-gas injection process design. Proceedings of the Seventy-Eighth Gas Processors Association Annual Convention. Nashville, TN, March 1–3, 1999, Tulsa, OK, pp. 114–120. Wichert, E. and Royan, T., 1996. Sulfur disposal by acid gas injection, SPE Paper 35585. Proceedings of the Gas Technology Symposium. Calgary, AB, Canada, 28 April–1 May 1996, SPE, pp. 193–200. Wichert, E. and Royan, T., 1997. Acid gas injection eliminates sulfur recovery expense. Oil Gas J., 95(17): 67–72.
This page intentionally left blank
636
ELSE_TSANG_CH049.qxd
8/23/2005
9:08 AM
Page 637
Chapter 49
UNDERGROUND INJECTION OF CARBON DIOXIDE IN SALT BEDS S. Bachua and M.B. Dusseaultb a
Alberta Energy and Utilities Board, Edmonton, Alberta, Canada University of Waterloo, Waterloo, Ontario, Canada
b
49.1 INTRODUCTION Geological sequestration of carbon dioxide (CO2) is an option that is immediately available and technologically feasible for reducing anthropogenic emissions of CO2 into the atmosphere. Although CO2 sequestration in salt caverns is generally the least-considered option because of high associated costs, cavern sequestration of CO2 may be viable in places that lack other sequestration options. This may be the case for the giant tar sands and heavy oil operations in the eastern-northeastern part of Alberta in western Canada. Currently, three mines for bitumen extraction are in operation in northeastern Alberta, and several in situ thermal operations are either in production or in the pilot-demonstration stage. Carbon dioxide emissions from these plants are on the order of several Mt/yr each. These tar sands and heavy oil deposits are located in a region of the Alberta Basin where the sediments are thin (1500 m), close to the basin edge at the Canadian Precambrian Shield. Options for CO2 geological storage, such as oil and gas reservoirs, coal beds and deep saline aquifers, are very limited or nonexistent, but extensive, thick salt beds are present (Bachu and Stewart, 2002). Thus, injection and sequestration of CO2 into salt caverns may provide a short-tomedium-term solution for reducing CO2 emissions into the atmosphere from existing and future plants in the area. This option may be attractive in other regions in the world where salt is mined for other purposes, such as in the Gulf Coast Basin in Texas and Louisiana, and in the Michigan Basin, thereby significantly reducing the cost of CO2 sequestration. Cavern sequestration of CO2 is attractive for several more reasons. Sequestration of CO2 in salt caverns allows a significantly higher sequestration efficiency (by at least one order of magnitude) than geological sequestration of CO2 by other means. For example, the ultimate capacity for CO2 sequestration in solution in formation water is of the order of 2–15 kg CO2/m3 rock (Bachu et al., 2003; Bachu and Adams, 2003), and through mineral immobilization is on the order of 2–5 kg CO2/m3 rock (Xu et al., 2003). These are achieved after tens to thousands of years. In contrast, the CO2 sequestration capacity of a salt cavern is of the order of 600–900 kg/m3, being controlled only by cavern temperature and pressure, and can be achieved almost immediately. In addition, the rate of cavern filling is not limited by well injectivity, as in the case of CO2 sequestration in hydrocarbon reservoirs, coal beds, or deep saline aquifers, but only by the flow capacity of the CO2 delivery system. Finally, salt caverns can be used for the temporary storage of large volumes of CO2, rather than for permanent sequestration, to serve either as a buffer for other, slower technologies, helping thus to dampen out
ELSE_TSANG_CH049.qxd
8/23/2005
9:08 AM
638
Page 638
Underground Injection of Carbon dioxide in Salt Beds
variations in CO2 emissions, or until other sequestration sites become available, such as oil and gas reservoirs that have to be produced first before being used for CO2 sequestration. Geological security of a storage cavern in bedded salt is high. Not only is the salt impermeable, but the bounding strata usually have pores plugged by crystalline salt for a considerable thickness beyond the salt bed, as in the case of the strata that underlie and overlie the salt beds in western Canada. The salt plugging of these strata significantly reduces their porosity and permeability, thus limiting possible CO2 leakage mechanisms to wellbore impairment and to fracturing of the cavern roof. Furthermore, because salt is naturally self-sealing by virtue of creep behavior, fissures or fractures created by drilling and dissolution processes will heal, and a permeability seal will be reestablished in a relatively short time frame. In western Canada and elsewhere, the technology for cavern solution mining has already been developed and used for underground hydrocarbon storage (e.g., CH4, C2H6), nonaqueous fluids (e.g., ethylene glycol), oilfield wastes and compressed air, or for salt extraction (e.g., Tek, 1989; Bradley et al., 1991; Crossley, 1998). Currently, single storage caverns in bedded salt can be up to about 0.5 106 m3 in volume, and in some cases have been permitted to store fluids at pressures up to 80% of the fracturing threshold. However, sequestration of CO2 in salt caverns differs from compressed air and natural gas storage in terms of sequestration permanency. The former are stored for relatively short periods of time, to be retrieved when demand goes up, and the cavern is periodically “filled and emptied” (cyclical cavern pressuring). In contrast, cavern-emplaced CO2 may be considered to be permanently sequestered, the objective being to avoid any CO2 return to the atmosphere for several hundreds to thousands of years, implying slow but continuous pressure buildup in the cavern as the salt slowly creeps. The sequestration of CO2 is also different from the permanent disposal of oilfield and other liquid wastes, because CO2 is a much more compressible fluid than these wastes; thus, cavern pressurization will be much slower and the reduction in cavern volume due to salt creep much larger than for incompressible and slightly compressible liquids. Cavern creation and filling require a number of steps that relate broadly to two processes: solution mining under hydrostatic pressure conditions, and cavern filling, perhaps with overpressuring conditions, to reduce the volumetric closure potential. Full-scale creep and integrity tests can be performed at the end of the mining stage, when the cavern is still filled with solution brine. Filling the cavern with CO2 may take a few years, depending on CO2 sources, during which time additional security measures must be taken to prevent CO2 escape. After cavern closure by appropriate borehole sealing, cavern closure behavior should be monitored by continuous pressure measurements that will help also to confirm the long-term constitutive creep behavior of the salt. Casing monitoring is also needed to ensure that CO2 leakage does not occur within or behind the well casing.
49.2 CAVERN CONSTRUCTION AND BEHAVIOR 49.2.1 Salt Characteristics The Middle Devonian Elk Point Group in the Alberta Basin in western Canada was deposited in a shallow-marine restricted environment that led to the deposition of several extensive, thick, and high-purity salt beds (Fig. 49.1). Because of the southwesterly dip of the sedimentary strata, these salt beds are found at depths that range from several hundred meters to more than 2000 m. The salt beds range in thickness from several tens to a few
8/23/2005 9:08 AM
Fig. 49.1. Location of Middle Devonian salt beds, and of major oil sands and heavy-oil CO2 producers in the Alberta Basin, Canada.
ELSE_TSANG_CH049.qxd Page 639
49.2 Cavern Construction and Behavior 639
ELSE_TSANG_CH049.qxd
8/23/2005
9:08 AM
Page 640
640
Underground Injection of Carbon dioxide in Salt Beds
hundred meters. The vertical stress (weight of the overburden) at the top of these salt beds ranges from ~10 to ~50 MPa, and temperatures range from 20 to 70oC. Stresses are assumed to be isotropic and equal to the lithostatic (overburden) within the salt beds, because salt viscoplastic properties and slow creep tend to dissipate shear stresses. Vertical stress gradients at the top and bottom of the salt beds are in the 24–25 kPa/m range, and differ by only 0.1–0.3 kPa/m between the two, indicating that they are basically constant across the entire rock package containing the salt beds. The average orientation of the minimum horizontal stress determined from borehole breakouts in the overlying carbonate and shale strata is 283 13°, the principal horizontal stresses being parallel and perpendicular to the Rocky Mountain deformation front. Most of the lowermost salt bed is recrystallized evaporitic salt of exceptional purity (90%), with none of the clay and anhydrite seams that are usually found in bedded salts. The degree of lateral continuity of the salt beds and of the overlying strata is large. These salt beds are located at the thin, eastern edge of the basin where no tectonic activity has taken place; consequently, there are no large natural fractures, faults, or folds in the overlying rocks. Furthermore, existing joints and fissures have not opened through flexure. No direct geomechanical tests exist for these salts, but on the basis of geophysical logs and by comparison with other published data (Hansen et al., 1984), Young’s modulus and Poisson’s ratio are estimated to be 40 GPa and 0.35, respectively. As with all natural, laterally continuous salt beds, owing to their extremely low permeability (10−21–10 −20 m2; Bredehoeft, 1988; Beauheim and Roberts, 2002), it can be safely assumed that they are impermeable. If, for some reason, a CO2-filled cavern were breached, the escaped CO2 would be confined by the overlying regional-scale aquitards and aquicludes in the sedimentary succession. 49.2.2 Cavern Construction and Filling Cavern construction through solution mining usually takes 2 to 3 years. Typically, a 12¼ in. borehole is drilled to the bottom of the groundwater-protection zone, cased with 10½ in. steel casing and cemented to surface using a dense, nonshrinking cement to avoid CO2 leakage outside casing (Dusseault et al., 2000). A 9 in. hole is drilled through the shoe to a depth of several tens of meters below the top of the salt bed, using a salt-saturated drilling fluid to maintain borehole geometry in the rock salt. The hole is cased with a 7 in. casing and cemented to the surface with nonshrinking cement placed with the highest-possible quality control. Special packer systems should be placed to increase the security against leakage behind the casing. The casing should be installed with pressure gauges and microseismic transducers strapped to the casing exterior and connected to surface with armored cable. Either special concentric tubing strings (e.g., 3½ in. tubing inside a 5½ in. tubing) or two side-by-side strings hanging in the hole are used for the salt dissolution process. The cavern is dissolved from the bottom to the top using accepted practice to generate the desired shape (Fig. 49.2). The process brine can be used or disposed of by injection into deep saline aquifers. Extensive coring and logging of the hole should provide the needed information regarding the properties of rock salt and overlying strata to permit rigorous analysis and interpretation of cavern tests that may be performed. Knowing the small compressibility of brine (~0.4 10−6 kPa−1), a series of pressure tests should be conducted on the brine-filled cavern at the end of the mining stage to determine salt and cavern behavior, particularly for assessment of salt creep and cavern integrity. These tests can take from a few days to several months. After testing and instrumentation of the hanging tubing strings with bottom hole pressure gauges, CO2 would be pumped into the cavern through the upper tubing, displacing brine into the lower tubing, which is placed as close to
ELSE_TSANG_CH049.qxd
8/23/2005
9:08 AM
49.2 Cavern Construction and Behavior
Page 641
641
Fig. 49.2. Schematic diagram of a salt cavern and injection well, and the setting of a cavern array.
the cavern bottom as possible (Fig. 49.2). Because of the lower density of CO2, a wellhead pressure of about 8–12 MPa would be required during cavern filling to overcome the hydrostatic head difference with the brine. The filling time depends on the amount of available CO2 from major point sources. The technology for CO2 compression and injection is well developed and has been applied to the injection of acid gas (a mixture of H2S and CO2) at approximately 40 operations in western Canada (Bachu et al., 2005). If the cavern is filled to above-hydrostatic pressures to reduce cavern shrinkage and strains in the overburden, then more robust surface facilities are necessary. Because of CO2 rapid decompression, at no time during cavern filling and sealing can the pressure in the cavern be allowed to drop, and the system must be equipped with safety valves to prevent backflow. After cavern filling, the access borehole will have to be sealed, in a manner similar to the sealing of a nuclear repository (Hansen et al., 1993, 1997). However, the pressure will be comparatively much higher in the case of supercritical CO2 in a cavern approaching lithostatic conditions than in a nuclear repository at hydrostatic conditions. In addition, in a nuclear repository the buoyancy forces caused by water heating as a result of radioactive decay are significantly smaller than the buoyancy of decompressing CO2. Sealing of the solution-mining and cavern-filling well can be done with a viscous substance, possibly recompacted granular salt (Fordham et al., 1988), so that permeable paths will gradually disappear as stress-induced creep compaction continues. Because of the lateral continuity of the strata, a cavern array can be designed to take advantage of the economies of jointly used facilities. On the right-hand side of Figure 49.2, a possible cavern array is shown, with caverns having approximately 26% areal coverage. Accounting for the shape of the caverns and the salt security barriers at the top and bottom (Fig. 49.2, left-hand side), this coverage would require dissolving and removing approximately 11% of the salt bed. Depending on cavern diameter, it is feasible to place about 20 caverns/km2, each cavern having a storage capacity approaching 1 million m3. Given the large area where salt beds are present (tens to hundreds of thousand km2), the CO2 sequestration potential in salt caverns is large in the area of Alberta under discussion.
ELSE_TSANG_CH049.qxd
8/23/2005
9:08 AM
642
Page 642
Underground Injection of Carbon dioxide in Salt Beds
49.2.3Cavern erm Long-T Behavior If a salt cavern is filled with a fluid at an initial pressure, in the long term the pressure inside the cavern will change as a result of any of: (1) salt creep, (2) leakage along the well bore, (3) thermal expansion of the cavern fluid, (4) flow of the fluid out of the cavern into the adjacent strata, or (5) additional salt solution or precipitation in the cavern (Berest et al., 2000). Since salt is insoluble in CO2 and salt beds are practically impermeable, the last two situations do not apply for the case of CO2 sequestration in salt caverns. Because of the intermediate compressibility of supercritical CO2, cavern pressurization will be gradual and much slower than for a liquid such as brine, but much faster than for a gas such as methane (Ehgartner, 1994). Furthermore, because salt conducts heat relatively rapidly and because of the thermal capacity of the Earth at depth, the temperature in the cavern will equilibrate rapidly by comparison with the slow cavern pressurization. In any case, cold placement of fluids in salt caverns is not recommended (Schalge and Swartz, 1998); therefore, for all practical purposes, it can be assumed that the thermal field inside and outside the cavern will equalize rapidly, rendering the pressuring process approximately isothermal. As a result, the only significant mechanisms for pressure change in the cavern are salt creep and potential leakage along the well bore. If the mining and filling well is properly completed and sealed, no leakage should occur, so great care should be paid to well completion and pressure monitoring. The salt surrounding a properly managed cavern will be exposed to a deviatoric stress (a shear stress) under moderate temperatures. Owing to its creep properties, the salt will slowly flow over periods of thousands of years, closing in and reducing the cavern volume until the CO2 pressure inside the cavern approaches the isotropic overburden stress acting in the salt bed. Salt is a viscoplastic substance, and slow creep tends to dissipate shear stresses over times of several hundreds to thousands of years. Hence, both the initial and the ultimate stress states in salt are isotropic (σvσHminσHmax, where v and H are the vertical and horizontal stresses, respectively). Isothermal steady-state salt creep is assumed to be a power function of the shear stress σ of the form . . σ n ε ε0 , σ0
冢 冣
(49.1)
where σ σvσH can be taken to be an approximation of the shear stress, σ0 is a reference . . stress, ε is the shear strain rate, and ε 0 is the shear strain rate at σ0 The power creep law holds only for a limited stress range (Munson and Dawson, 1982), but mine-back analysis . and long-term laboratory data suggest the values of n 3 and ε 0 0.001–0.002 yr−1 for 10 MPa (Rothenburg et al., 1993, 2002a). A few years after cavern sealing, will certainly be 10 MPa, and use of the power-law exponent of 3 for salt creep is appropriate. In practice, a salt cavern would be dissolved approximately as a sphere or a prolate (flattened) spheroid or ellipsoid of revolution with a vertical axis. The isothermal increase of CO2 pressure in a spherical salt cavern can be expressed analytically by (Rothenburg et al., 2002b, Bachu and Rothenburg, 2003) Cp dp 3 . 3 σ∞p ε , p dt 2 0 n σ0
冢
冣
(49.2)
ELSE_TSANG_CH049.qxd
8/23/2005
9:08 AM
Page 643
49.2 Cavern Construction and Behavior
643
where p is the pressure, Cp the CO2 compressibility coefficient expressed in terms of reduced temperature and pressure (i.e., temperature and pressure relative, respectively, to the critical temperature and pressure of the stored gas), the initial in situ stress in the salt bed, and t the time. The pressure increase in the salt cavern filled with CO2 depends on the ratio of initial in situ stress to critical gas pressure, σ /pc, on the ratio of initial cavern pressure and initial in situ stress, p0 /σ , and on the gas reduced temperature, Tr, through the compressibility coefficient Cp. Using these parameters and the dimensionless variables 3 . 3 σ∞ p ε0 p~ and ~t 2 σ∞ n σ0
冢
n
冣 t,
(49.3)
the pressure increase in the salt cavern can be represented though a set of generalized curves (Rothenburg et al., 2002b; Bachu and Rothenburg, 2003). Analysis of the behavior of a spherical cavern in salt shows that cavern pressure buildup has an asymptotic behavior, achieving ~90% of the stress in the surrounding salt bed within 5% of the time needed for 98% equalization (Bachu and Rothenburg, 2003). For illustration, Figure 49.3 shows the pressure buildup inside a cavern, and the reduction in cavern volume (compression) in time, for an initial cavern pressure equal to half of the lithostatic ( /pc~ 2), which corresponds to filling the cavern under brine hydrostatic conditions (brine ~ 1200 kg/m3), and for conditions characteristic of the salt beds in the Alberta Basin (overburden mean density ~ 2400 kg/m3). The time factor is ~ t 0.063t, i.e., ~16 years real time. The dimensionless volume reduction is defined as the reduction in volume in relation to the current cavity volume, i.e., (V0V)/V. The pressure buildup is relatively rapid at the beginning ( t~ 60), and becomes very slow after the pressure in the cavern reaches ~90% of the virgin in situ stress in the salt bed. The effect of initial pressure at the time of cavern sealing, p0 , on pressure inside the cavern is lost for large times (t~90) (Fig. 49.3a), but it is significant for the final volume reduction magnitude (Fig. 49.3b); therefore, it also affects ground subsidence and the bending strains in the elastic overburden rocks. For example, the cavern shrinks to approximately a fifth of its initial size for p0 / 0.4, to approximately half of its original size for p0 / 0.6, but only by ~14% for p0 / 0.8 (Fig. 49.3b). Volumetric changes are significant (and relatively “fast”) in the early life of the cavern (t~30). After that, the cavern volume stabilizes, with negligible rates of change (closure) over time, in part because supercritical CO2 is far less compressible than gaseous CO2. Cavern closure will occur during cavern mining, filling with CO2, and after sealing. During solution mining and cavern filling, the pressure in the cavern can be assumed to be constant and equal to the hydrostatic pressure of the brine (11.8 kPa/m). Pressure buildup occurs only after cavern sealing, or if the cavern is deliberately pressurized. If a salt cavern of 100 m in diameter is mined over a 3-year period, assuming a constant volumetric rate of solution mining at a depth of ~1200 m, in a salt bed where the in situ stress and temperature are ~28 MPa and T 37°C – then the decrease in cavern volume during mining, filling with CO2 and after sealing at the hydrostatic brine pressure of 14.2 MPa bottom-hole pressure is ~18% of the original cavern volume. The maximum long-term ground subsidence caused by this reduction in cavern volume is negligible, on the order of 5 mm (Bachu and Rothenburg, 2003). The cavern will sequester ~500 kt CO2 (0.5 Mt) at supercritical conditions and density of ~900 kg/m3. The cavern closure will be very slow, pressurizing the
ELSE_TSANG_CH049.qxd
644
8/23/2005
9:08 AM
Page 644
Underground Injection of Carbon dioxide in Salt Beds
Fig. 49.3. Cavern behavior in time for different initial pressures at sealing: (a) pressure buildup; and (b) volume reduction (compared to final cavern volume).
CO2 to ~94% of the initial stress in the salt bed in about 4000 years. Actually, it seems to be desirable to fill the cavern to a pressure higher than hydrostatic, closer to the final equilibrium stress in the salt mass, because this increases capacity and minimizes deformations in the overlying strata. Because a CO2-sequestration cavern will not be pressure cycled, because the internal pressure will slowly approach lithostatic, because the cavern roof can be properly shaped, and because the pressure gradient is strongly outward—roof collapse should not be an issue if the overall closure strains are modest. Roof and floor salt security zones will retain sealing characteristics as long as they remain under compressive stress. Unlike a mine, where the internal cavern pressure is zero, the internal pressure in the cavern (hydrostatic and higher) will counteract any tendency for brittle fracture in the walls of the cavern and of the access borehole. Furthermore, construction of an array of caverns (Fig. 49.1) to increase CO2 sequestration capacity will reduce the potential for roof collapse and fracturing in the overlying strata, by diminishing local bending strains.
ELSE_TSANG_CH049.qxd
8/23/2005
9:08 AM
Page 645
49.3 CO2 Leakage and Monitoring
49.3
645
CO2 LEAKAGE AND MONITORING
Although CO2 is not as hazardous as methane and other natural gases commonly stored in salt caverns, CO2 leakage should nevertheless be avoided. Given the extremely low permeability and creep properties of salt, to which one may add the confining nature of overlying aquitards, the potential for CO2 leakage from a salt cavern, through diffuse flow into adjacent formations and/or through fractures, is very low, almost nonexistent, except for the well bore. The 1% porosity characteristic of recrystallized salt is usually in the form of occluded porosity, and an interconnected porosity network may not even exist. The low permeability of halite (from less than 10–21–10–20 m2 — Bredehoeft, 1988; Beauheim and Roberts, 2002) may actually be partly allocated to slow solution/precipitation processes controlled by salt/brine diffusion rather than classical Darcy flow in porous media (Spiers et al., 1988). As stated before, adjacent strata usually have the pore space plugged with salt, adding an extra barrier to the flow of CO2. If microfissures develop in the cavern walls and along the access borehole, they will anneal over time as a result of pressure solution in interstitial brine, since the CO2 will not be able to displace the brine from these microfissures because of capillary effects and a zero far-field permeability. If roof collapse is avoided, then no fracture will develop through which CO2 would leak from the cavern into overlying strata that have sufficient permeability to allow CO2 migration and leakage to shallower strata, to groundwater aquifers, and eventually to the atmosphere. Even if fracture conditions are reached, the release of a small amount of CO2 will lead to a pressure drop inside the cavern and to fracture closure. It seems that the only potential path of any significance for CO2 leakage is along casing and/or through the well cement. Leakage along the wellbore would lead to a straight, vertical conduit for CO2 flow to the surface. Because the supercritical CO2 will decompress as it flows upward, its density will decrease, its buoyancy will increase, and the flow will accelerate as a result of this gradient selfenhancement process. Thus, the borehole is the weakest link and the only real risk in the use of salt caverns for CO2 sequestration. However, given the very low cement permeability combined with relative permeability effects of CO2 flow through a water-saturated cement, the risk of CO2 leakage through competent cement around the well bore is extremely small. On the other hand, this risk may increase significantly if the well is improperly completed and sealed (Dusseault et al., 2000). The steel casing is cemented in place, and, although it is surrounded by salt, there is a strain incompatibility interface between the rigid cement and the viscoplastic salt. In addition, the cement exposed to salt and to brines in overlying formations for an extremely long time may degrade over time and lose its properties (Powers, 2000). These appear not to be serious problems, but they must be considered if this technology is adopted. After cavern sealing, the monitoring strategy should take into account that surface displacements are very small and develop extremely slowly. The small strains and modest volumetric closures mean that active seismic monitoring (reflection, refraction, VSP, crosshole) is not useful, and there is no value in measuring microgravity changes or ground deformation (which is expected to be of the order of a few millimeters over decades). Microseismic monitoring in the rocks immediately overlying the salt bed could be used to track the response of the rock mass and to verify all assumptions used in cavern design and construction, although the data may simply show that nothing of consequence is happening. The only monitoring method that has detection capabilities and is also relevant for leakage assessment is casing and annulus pressure monitoring. Furthermore, given the long times
ELSE_TSANG_CH049.qxd
8/23/2005
646
9:08 AM
Page 646
Underground Injection of Carbon dioxide in Salt Beds
involved (hundreds of years) and the life span of electronic devices (tens of years), a purely mechanical and robust pressure measuring system seems advisable.
49.4 CONCLUSIONS Geological sequestration of CO2 is a mitigation option for significantly reducing CO2 emissions into the atmosphere. This technology is immediately available and technologically feasible. Injection technologies have been developed for the storage of petroleum products and natural gas, and for the disposal of hazardous wastes. Various operations in North America inject CO2 into deep formations for enhanced oil recovery and for acid-gas disposal, while various liquid wastes are disposed of by injection into permeable formations and salt caverns. Although perhaps the least economic of the various means of CO2 geological sequestration, injection and sealing of CO2 in salt caverns is a method that may be the only alternative in regions that lack other sequestration means, particularly in areas where the sedimentary succession is too thin to allow injection of CO2 as a dense-fluid phase into permeable formations. Such a region is in northeastern Alberta, where bitumen and oil production from tar sands located at the shallow edge of the Alberta Basin results in very large CO2 emissions that will significantly increase in the future. Salt caverns have the advantage of an extremely high volumetric capacity, of the order of 600–900 kg CO2/m3 of cavern, compared with 2–15 kg CO2/m3 of rock for solubility or mineral trapping. Furthermore, this capacity is achieved immediately, whereas dissolution and mineralization processes take tens to thousands of years. In addition, CO2 storage in salt caverns may serve as a temporary solution until other sequestration sites and facilities, such as depleted oil and gas reservoirs and pipelines, become available. Owing to the creep properties of salt, a cavern filled with supercritical CO2 will close in, thus reducing its volume, until the pressure inside the cavern equalizes the external stress in the salt bed. At the beginning, the pressure buildup in the cavern is rapid, reaching close to 90% of the stress in the surrounding salt. However, after a relatively short period of time (tens to hundreds of years, depending on depth and temperature), the pressure buildup becomes extremely slow, and it no longer depends on the initial cavern pressure. The pressurizing process then continues for hundreds and thousands of years, approaching the pressure equivalent to the overburden stress in an asymptotic manner. Similarly, the reduction of cavern volume is significant at the beginning, after which it becomes negligible. The cavern closure volume depends strongly on the initial pressure in the cavern at sealing, indicating that it is desirable to “overfill” the cavern by injecting CO2 under high pressure (greater than hydrostatic), close to the in situ stress, before cavern sealing. The CO2 sequestration capacity is increased while the magnitude of flexural stresses in the overlying strata is reduced, therefore enhancing the safety of the injection well and the integrity of the overlying elastic strata after cavern sealing. The only realistic pathway for CO2 leakage from the cavern appears to be along the wellbore, and therefore special care should be taken during well completion and sealing. Pressure monitoring of the cavern and wellbore should detect any leakage that might occur, allowing mitigative measures to be taken. Calculations show that, for conditions specific to salt beds in northeastern Alberta, a single cavern of 100 m in diameter may hold 0.5 Mt of CO2. A single cavern may not satisfy the needs of large CO2 emitters, but arrays of such caverns can be built as a regional repository in the extensive and thick salt beds of the Alberta Basin and elsewhere with similar conditions, without impairing general security of the repository.
ELSE_TSANG_CH049.qxd
8/23/2005
References
9:08 AM
Page 647
647
REFERENCES Bachu, S. and 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, Energ. Convers. and Manage., 44(20): 3151–3175. Bachu, S., Haug, K, Michael, K., Buschkuehle, B.E. and Adams, J.J., 2005. Deep injection of acid gas in western Canada. Underground Injection Science and Technology (this publication), Elsevier, U.S.A. Bachu, S., Michael, K. and Adams, J.J., 2003. Effects of in situ conditions on aquifer capacity for CO2 sequestration in solution, Proceedings of the 2nd Annual Conference on Carbon Dioxide Sequestration, Alexandria, Virginia, May 5–8, 2003. Bachu, S. and Rothenburg, L., 2003. Carbon dioxide sequestration in salt caverns: Capacity and long term fate, Proceedings of the 2nd Annual Conference on Carbon Dioxide Sequestration, Alexandria, VA, May 5–8, 2003. Bachu, S. and Stewart, S., 2002. Geological sequestration of anthropogenic carbon dioxide in the Western Canada Sedimentary Basin, J. Can. Petrol. Tech., 41(2): 32–40. Beauheim, R.L. and Roberts, R.M., 2002. Hydrology and hydraulic properties of a bedded evaporate formation. J. Hydrol., 259: 66–88. Berest, P., Brouard, B. and Durup, J.G., 2000. Shut-in pressure tests: Case studies. Proceedings Solution Mining Research Institute 2000 Fall Technical Meeting, San Antonio, TX. Bradley, R.A., Watts, E.C. and Williams, E.R. (Eds), 1991. Limiting net greenhouse gas emissions in the United States, Vols. I and II, U.S. Department of Energy, Office of Environmental Analysis, Washington, DC. Bredehoeft, J.D., 1988. Will salt repositories be dry? EOS, Trans. AGU, 69(9): 121. Crossley, N.G., 1998. Conversion of LPG salt caverns to natural gas storage: “A transgas experience,” J. Can. Petrol. Tech., 37(12): 37–47. Dusseault, M.B., Gray, M.N. and Nawrocki, P.A., 2000. Why oil wells leak: Cement behavior and long-term consequences, Proceedings of the SPE International Oil and Gas Exhibition, SPE Paper 64733, Beijing, China. Ehgartner, B.L., 1994. Long-term sealing analyses for US strategic petroleum reserve (SPR) Caverns. Report SAND92-2891, Sandia National Laboratories, Albuquerque, NM. Fordham, C.J., Dusseault, M.B. and Mraz, D., 1988. Strength development in halite backfill. Proceedings of the 41st Canadian Geotechnical Conference, Waterloo, ON, pp. 192–198. Hansen, F.D., Ahrens, E.H., Dennis, A.W., Hurtado, L.D., Knowles, M.K., Tillerson, J.R., Thompson, T.W. and Galbraith, D., 1997. A shaft seal system for the waste isolation pilot plant, Proceedings of the Solution Mining Research Institute Fall Meeting, El Paso, TX, Oct. 5–8. Hansen, F.D., Callahan, G.D. and Sambeek, L.L., 1993. Reconsolidation of salt as applied to permanent seals for the waste isolation pilot plant, Proceedings of the 3rd Conference on the Mechanical Behavior of Salt, Transtech Publications, Palaiseau, France, pp. 323–335. Hansen, F.D., Mellegard, K.D. and Senseny, P.E., 1984. Elasticity and strength of ten natural rock salts, Proceedings of the 1st Conference on the Mechanical Behavior of Salt, Transtech Publications, University Park, PA, pp. 71–83. Munson, D.E. and Dawson, P.R., 1982. A transient creep model for salt during loading and unloading. Report SAND82-0962, Sandia National Laboratories, Albuquerque, NM. Powers, D.W., 2000. Evaporites, casing requirements, water-floods, and out-of-formation waters: Potential for sinkhole developments, Proceedings of the Solution Mining Research Institute Fall Meeting Technical Session, San Antonio, TX, pp. 186–195.
ELSE_TSANG_CH049.qxd
648
8/23/2005
9:08 AM
Page 648
Underground Injection of Carbon dioxide in Salt Beds
Rothenburg, L., Dusseault, M.B. and Bachu, S., 2002b. Closure of a spherical cavity in salt: An infinite medium solution for long-term gas sequestration studies, In: R. Hammah, W. Bawden, J. Curran and M. Telesnicki, (Eds), Mining and Tunneling Innovation and Opportunity Vol. 2. Proceedings of the 5th North American Rock Mechanics Symposium (NARMS) and 17th Tunnelling Association of Canada (TAC) Conference, University Toronto Press, Toronto, Ontario, July 7–10, 2002, pp. 1259–1266. Rothenburg, L., Dusseault, M.B. and Mraz, D.Z., 2002a. On the third-power creep law for salt in mine conditions, In: N.D. Cristescu, R.R. Hardy Jr. and R.O. Simionescu, (Eds), Basic and Applied Salt Mechanics: Proceedings of the 5th Conference on Mechanical Behavior of Salt, MECASALT V, Bucharest, Romania, 9–11 August 1999, Balkema, Lisse (Netherlands), pp. 171–176. Rothenburg, L., Frayne, M.A. and Mraz, D.Z., 1993. Application of two and three dimensional numerical models for intact salt rock. In: W.F. Bawden and J.F. Archibald, (Eds), Innovative Mine Design for the 21st Century Balkema, pp. 609–620. Schalge, R. and Swartz, W., 1998. Chilling natural gas to increase salt cavern storage capacity. Proceedings of the Solution Mining Research Institute Fall Meeting, Rome, Italy, pp. 417–429. Spiers, C.J., Urai, J.L. and Lister, G.S., 1988. The effect of brine (inherent or added) on rheology and deformation mechanisms in salt rock. Proceedings of the 2nd Conference on the Mechanical Behavior of Salt, Transtech Publications, Hannover, Germany, pp. 89–102. Tek, M.R. (Ed), 1989. Underground storage of natural gas: Theory and practice, NATO ASI Series E, Appl. Sci., Kluwer, Boston, MA, 171pp. Xu, T., Apps, J.A. and Preuss, K., 2003. Reactive geochemical transport simulation to study mineral trapping for CO2 disposal in deep arenaceous formations. J. Geophys. Res., 108(B2): 2071–2083.
ELSE_TSANG_CH050.qxd
9/15/2005
1:06 PM
Page 649
Chapter 50
COUPLED HYDROMECHANICAL EFFECTS OF CO2 INJECTION J. Rutqvist and C.-F. Tsang Earth Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, California, USA
50.1 INTRODUCTION Underground storage of carbon dioxide (CO2) in permeable formations, such as deep saline aquifers, depleted oil and gas reservoirs, and coal seams, has been suggested as an important potential method for reducing the emission of greenhouse gases to the atmosphere (DOE 1999). The injection would take place at a depth below 800 m, so that the CO2 would be within the temperature and pressure range of a supercritical fluid. As a supercritical fluid, CO2 behaves like a gas with low viscosity but with a liquid-like density of 200–900 kg/m3, depending on pressure and temperature. Because supercritical CO2 is less dense than water, deep underground disposal requires a sufficiently impermeable caprock above an underground storage zone to trap the injected CO2. Caprock integrity and reservoir leakage is a key issue for both short- and long-term performance of geological CO2 storage. In the short term, leakage is an important safety issue during active CO2 injection. In the long term, leakage impacts the sequestration effectiveness of the once-injected CO2. In general, two kinds of leakage mechanisms can be identified (Yamamoto and Takahashi, 2004): 1. Steady or slow leakage processes of buoyancy-driven gas flow at a rate that depends on formation permeability and fluid capillarity. 2. Dynamic or rapid leakage processes along fluid paths created by interaction between formation and injected CO2. For the slow-leakage mechanism, the fluid travels through the rock matrix, fractures, and abandoned boreholes present in the storage volume. The fluid path can be regarded as fixed in the short term, but may change slowly over the long term. Because of the limited toxicity of CO2, a slow leakage of CO2 is allowable from the viewpoint of short-term safety. However, such leakage may reduce long-term usefulness of CO2 sequestration. The second, rapid leakage mechanism has been observed in nature, caused by external forces such as earthquakes. Such rapid changes could include a breach in a caprock caused by mechanical changes such as hydraulic fracturing or fault slip. For CO2 sequestration, a rapid change in the geologic system may increase CO2 leakage so significantly that it may be detrimental to both short-term safety and long-term environmental conservation. In considering a site for CO2 sequestration, it will be important to evaluate the effects of CO2 storage on the formation, so as to minimize the risk of a breach occurring in the system. First, injection of CO2 will result in an increase in formation fluid pressure, especially around the injection source. Such a fluid pressure increase will cause local changes in the stress field, which, in turn, will induce mechanical deformations and possible irreversible mechanical failure in the caprock. This mechanical failure may involve shear along many of
ELSE_TSANG_CH050.qxd
650
9/15/2005
1:06 PM
Page 650
Coupled Hydromechanical Effects of CO2 Injection
the existing fractures or creation of new fractures that reduce the sealing properties of the caprock system. Second, replacing the native formation fluid with CO2 may cause changes in rock mechanical properties through chemomechanical interactions between the CO2 and the host rock, or through desiccation fractures. The evaluation of the breaching risk in a caprock/reservoir system involves prediction of complex coupled thermal–hydrological–mechanical-chemical (THMC) processes over a long period of time. Such predictions can be accomplished with appropriate numerical models and input data determined from laboratory and field experiments. Such evaluation should, if possible, be combined with in situ monitoring of system mechanical responses to the CO2 injection. For example, 4D (time lapse) seismic monitoring and acoustic tomography can monitor fluid-induced changes in elastodynamic formation constants, from which the formation of a new leakage path through a caprock might be detected. A rapid opening of a new leakage path in an otherwise homogeneous caprock is a dynamic change that could be followed by local fluid pressure changes, leading to CO2 leakage. Inversely, any changes in fluid pressure will be accompanied by mechanical deformations, leading to potential changes in local elastodynamic properties. If those changes can be detected with seismic monitoring or other geophysical methods, they can be used as detectable precursors to a breach in a caprock and subsequent CO2 leakage. This paper presents an approach for analysis of coupled hydraulic and mechanical effects on caprock integrity and potential reservoir leakage during CO2 injection. First, the fundamentals of relevant rock mechanical and hydromechanical (HM) phenomena in fractured porous rock are introduced. Then, a number of natural and industrial analogues relevant to caprock integrity and reservoir leakage are described. Following this, a numerical tool for the evaluation of coupled thermal-hydrological-mechanical (THM) processes during underground CO2 injection is presented. Finally, we demonstrate the use of this numerical tool for evaluation of caprock integrity and reservoir leakage for hypothetical but realistic examples.
50.2 FUNDAMENTALS OF HYDROMECHANICAL INTERACTIONS IN FRACTURED ROCK The term “HM coupling” refers to the interaction between hydraulic and mechanical processes. Various potential modes of mechanical responses in fractured and porous geological media (such as caprock) are schematically presented in Figure 50.1. The mechanical processes are coupled to hydraulic processes because geological media contain pores and fractures that can be fluid-filled and deformable. According to Rutqvist and Stephansson (2003), these couplings can be divided into “direct” HM couplings, occurring through deformation and pore-fluid pressure changes, or “indirect” HM coupling, occurring through changes in hydraulic and mechanical properties (Fig. 50.2). In any porous geological material, both direct and indirect coupled HM processes will occur, and both are relevant for evaluation of the long-term performance of a CO2 injection site. 50.2.1 Direct HM Coupling and the Theory of Poroelasticity Direct HM couplings occur through deformation and pore-fluid interactions, and include two basic phenomena (Wang, 2000): 1. A solid-to-fluid coupling that occurs when a change in applied stress produces a change in fluid pressure or fluid mass.
ELSE_TSANG_CH050.qxd
9/15/2005
1:06 PM
Page 651
50.2 Fundamentals of Hydromechanical Interactions in Fractured Rock
(d) Effective Normal Stress
(a) Effective Confining Pressure
n ’n = n - p
’ = - p
Contacts p
p
(b) Volumetric Deformation
Fractured Porous Rock
(e) Fracture Normal Deformation n
p’
p
un
K p
651
1 v
’n kn 1
p
(c) Shear Deformation s
(f) Fracture Shear Deformation us
s
un
s
s ks
G 1
1 2s
us
Fig. 50.1. Schematic overview of rock-mechanical processes in a fractured geological media, relevant for caprock integrity (from Rutqvist and Stephansson, 2003).
2.
A fluid-to-solid coupling that occurs when a change in fluid pressure or fluid mass produces a change in the volume of the porous medium. These two cases of direct HM couplings are shown schematically in Figure 50.2 and are labeled (i) and (ii). The second coupling is considered in Terzaghi’s (1923) concept of effective (intergranular) stress. Terzaghi (1923) defines the effective stress, σ ⬘zz, in a geological medium, as the total vertical stress, σzz, less the pore fluid pressure, i.e.:
σ ⬘zz ⫽ σ ⬘zz ⫺ p
(50.1)
with compressive stress positive. However, the direct couplings—(i) and (ii) in Figure 50.2—are fully described in Biot’s (1941) general theory of three-dimensional consolidation. Biot’s equations for isotropic linear elastic porous media can be written in a “mixed stiffness form” (Wang, 2000), as
m ⫽ Kv ⫹ α p,
(50.2)
1 ξ ⫽ αεv ⫹ ᎏ p, M
(50.3)
where σm is the total mean stress (positive for compression), K the usual (drained) bulk modulus, εv the volumetric strain (positive for contraction), α the Biot–Willis’ (1957) coefficient, ξ the increment of fluid content (positive for “gain” of fluid), and M the Biot’s modulus. The introduction of the Biot–Willis coefficient as a factor multiplied to fluid pressure in Equation (50.2) signifies a modification and generalization of Terzaghi’s effective stress law to
σ m⬘ ⫽ σ m ⫺ α p.
(50.4)
ELSE_TSANG_CH050.qxd
9/15/2005
1:06 PM
Page 652
652
Coupled Hydromechanical Effects in CO2 Injection Mechanical Process Solid Stress and Strain (ii)
(iv)
Mechanical Properties
Pore Volume Change Hydraulic Properties (iii)
(i) Hydraulic Process Fluid Pressure and Mass
Fig. 50.2. HM couplings in geological media. (i) and (ii) are direct couplings through pore volume interactions, while (iii) and (iv) are indirect couplings through changes in material properties (from Rutqvist and Stephansson, 2003).
The coefficient α, which usually ranges between 0 and 1, has been measured in laboratory experiments (e.g., Nur and Byerlee, 1971) for a range of geological materials. Equation (50.2) governs the elastic responses of the pore structure; Equation (50.3) governs pore-fluid responses. The two equations are coupled through the volumetric strain and fluid-pressure terms. Since the theory describes interaction between pore fluid and elastic responses, it has been called the theory of poroelasticity. Porosity variation with mechanical deformation and fluid pressure is implicitly included in Equations (50.2) and (50.3), since α and M depend on porosity, and εv can be linked to porosity variation. However, the macroscopic quantities α and M are frequently determined in laboratory experiments directly, without knowledge of the detailed pore-volume response at the microscopic level. Direct coupled processes occur in all types of porous, deformable geological media, but some poroelastic phenomena tend to be more important in soft and low-permeability media. Because caprocks, such as mudstone, at CO2 storage sites would need to have permeability as low as 10 nD, it is expected that direct HM couplings are very relevant during CO2 injection. 50.2.2 Indirect HM Couplings and Property Changes The direct couplings that occur through pore volume changes will also be accompanied by indirect couplings in the form of changes in mechanical and hydraulic properties. For example, a mechanically induced reduction in pore volume leads to a reduction of fluid flow capacity (hydraulic permeability), thereby impacting hydraulic processes. On the other hand, a reduction of pore volume may result in a stiffer material, as more contacts occur between neighboring grains. These changes in material properties can be considered indirect HM couplings. More generally, two basic phenomena of indirect HM coupling may be considered (Rutqvist and Stephansson, 2003): 1. A solid-to-fluid coupling that occurs when an applied stress produces a change in hydraulic properties. 2. A fluid-to-solid coupling that occurs when a change in fluid pressure produces a change in mechanical properties.
ELSE_TSANG_CH050.qxd
9/15/2005
1:06 PM
Page 653
50.2 Fundamentals of Hydromechanical Interactions in Fractured Rock
653
These phenomena are labeled (iii) and (iv) in Figure 50.2. Both direct and indirect coupled processes may be fully reversible. However, inelastic responses, including yielding, fracturing, or fault slip, are also possible, and these cause irreversible changes in porous and fractured media. There are a large number of constitutive equations for describing changes in hydraulic properties with effective stress or deformation in geological media (Rutqvist and Stephansson, 2003). Such changes in a caprock may produce slight changes in diffuse leakage through a caprock system. However, without mechanical failure, the impact of permeability changes in intact rock is expected to be relatively small, for both the short- and long-term behavior of a caprock/reservoir system. It is likely that only the HM mechanisms that induce deformation of pre-existing fractures or creation of new fractures—thus significantly changing hydraulic permeability and water retention properties—are of importance to analysis caprock integrity associated with CO2 injection. 50.2.3 Deformations in Pre-existing Fractures and the Creation of New Fractures Deformation of a pre-existing fracture can be induced by a change in the stress field acting on the fracture. The most basic responses of a fracture include normal and shear displacement, caused by a change in normal and shear stresses, respectively. This is formulated according to Goodman (1970) as ∆σ n⬘/kn,
(50.5)
∆σs /ks.
(50.6)
In Equation (50.5) ∆un is the normal deformation of the fracture caused by a change in effective normal stress, ∆σ ⬘n, with the magnitude of opening or closing dependent on the fracture normal stiffness kn (Fig. 50.1e). Likewise, Equation (50.6) describes the shear displacement, ∆us, which depends on the shear stiffness, ks, and the change in shear stress, ∆σs (Fig. 50.1f). In a real fracture, the normal deformation is typically nonlinear, as shown in Figure 50.3a. The fracture stiffness typically increases as normal stress increases. One common feature of fracture deformation is the hysteresis effect during stress loading and unloading (not shown in Fig. 50.3a), which is caused by processes arising from surface mismatch, sampling disturbances, or crushing of asperities (Barton et al., 1985). If the shear stress acting on a fracture surface is sufficiently high, shear slip may be induced in the plane of the fracture. The most fundamental criterion for fault slip can be derived from the effective stress law and the Coulomb criterion, rewritten as
σsc ⫽ σsc0 ⫹ µs(σn ⫺ p),
(50.7)
where σsc is the critical shear stress, σsc0 cohesion, µs coefficient of shear friction, and σn the normal stress (Scholz, 1990). In a real fracture, the typical shear stress and shear displacements involve rapid increase in shear stress up to a peak, followed by a loss in load-carrying capacity. The shear displacement is accompanied by a shear dilation, as shown in the ∆un curves of Figure 50.3c. For a portion of the stress/deformation curve corresponding to elastic deformation of the fracture, there is minimum dilation. The onset of rapid dilation occurs when asperities begin to slide against each other. Rate of dilation (slope of ∆un curve in Fig. 50.3c) increases and reaches a maximum at the peak shear stress (Barton et al., 1985).
ELSE_TSANG_CH050.qxd
9/15/2005
1:06 PM
Page 654
654
Coupled Hydromechanical Effects in CO2 Injection Drill Core
Drill Core
In Situ (1 m2)
’n
’n
In Situ (1 m2) 2-3 orders of magnitude
kn
un (a) sPeak
In Situ (1 m2) ks
Log T
Drill Core 100
s un
un
1 mm
us
Log T/T0
s
(c)
Tr
(b)
(d)
Drill Core 10
In Situ (1 m2)
1 mm
us
Fig. 50.3. Typical mechanical and HM fracture responses under normal closure (a,b) and shear (c,d). Effects of sample size is indicated with the laboratory sample response (dashed lines) compared to in situ fracture response (1 m2 size) (modified from Rutqvist and Stephansson, 2003).
The shear strength of a fracture depends on normal stress, with a higher peak shear stress for a higher normal stress. This was examined by Byerlee (1978) for normal stresses up to 100 MPa, who derived the relationship
σscPeak ᎏ ⫽0.85. σn
(50.8)
Barton and Choubey (1977) studied shear behavior at engineering stress levels and developed empirical relationships that could be related to basic fracture characteristics, such as joint roughness or compressive strength of the rock walls. Using such empirical relationships, the amount of dilation (∆un versus ∆us) can be calculated by the following expression: ∆un ⫽ ∆us tan dmob,
(50.9)
where dmob is the mobilized dilation angle, which in turn depends on rock-wall compressive strength and normal stress across the fracture. The fundamental criterion for tensile failure (or hydraulic fracturing) is that incipient fracture propagation will occur when the fluid pressure exceeds the least principal stress by an amount equal to or greater than the tensile strength of the rock: pct ⱖ σ ⫹ σt
(50.10)
ELSE_TSANG_CH050.qxd
9/15/2005
1:06 PM
Page 655
50.2 Fundamentals of Hydromechanical Interactions in Fractured Rock
655
More detailed analysis of fracturing may involve linear or nonlinear fracture mechanics, which have been applied in the analysis of propagation and control of hydraulic fracturing stimulations in petroleum engineering (Perkins and Kern, 1961; Geertsma and Deklerk, 1969). 50.2.4 Fracture Fluid Flow As a first approximation, the fluid flow along a fracture can be analyzed using the parallel plate flow concept, in which the fracture flow (Snow, 1965; Louis and Maini, 1970) between two plates with constant separation or aperture b is given by Qf ᎏ ⵜh
b3ρ f g w, ⫽ᎏ 12µ f
(50.11)
where Qf is volume flow rate for parallel plate (or fracture) of width w, ∇h head gradient, ρf and µf fluid density and viscosity, respectively, and g the acceleration of gravity. For a real fracture, hydraulic or effective aperture, bh, can be defined as the parallel plate aperture b value that produces the same relationship between Q and ∆h. Hence, fracture transmissivity is given by bh3 ρfg T⫽ ᎏ . 12µ f
(50.12)
The aperture bh can therefore be back-calculated from fracture transmissivity, which can be determined in a flow test. 50.2.5 Fracture HM Behavior as a Function of Normal Stress Experimental results typically show a decrease in fracture transmissivity with normal stress (Fig. 50.3b), but with an apparent residual transmissivity, Tr, at high stress when the fracture appears to reach its compression limit. Figures 50.3a and c also show a size effect for normal closure. The size effect has been observed in experiments (e.g., Witherspoon et al., 1979; Barton and Bakhtar, 1982) and confirmed in theoretical studies (e.g., Neuzil and Tracy, 1981; Swan, 1983). Neuzil and Tracy (1981) attributed this size effect to a truncation of the aperture frequency distribution, implying that fewer of the largest, least-frequent flow channels would be included in a smaller sample. Witherspoon et al. (1980) developed a modified cubic law, which they validated against laboratory experiments on artificial tension fractures in samples of granite and marble. They considered a general flow law, C Q ᎏ ⫽ ᎏ (b)n, ∆h f
(50.13)
where f is a friction factor that accounts for the roughness of the fracture surface, b an apparent physical aperture and C a constant depending on the flow-domain geometry and the properties of the fluid (e.g., C ⫽ ρfgw/12µf for parallel flow, see Equation (50.11). If n ⫽ 3, this is a cubic law or modified cubic law, and then the apparent physical aperture is related to the hydraulic aperture as: b ⫽ f 1/3bh.
(50.14)
ELSE_TSANG_CH050.qxd
9/15/2005
656
1:06 PM
Page 656
Coupled Hydromechanical Effects in CO2 Injection
A complete relationship between fracture transmissivity and effective normal stress can be derived by combining a fracture flow law (Equation 50.12) with an equation for fracture normal closure (e.g., Equation 50.5). Using this approach scientists have developed a number of empirical and theoretical models, which include logarithmic, hyperbolic, and exponential functions for describing the nonlinear relationship between hydraulic aperture and the stress normal to the fracture (e.g., Walsh 1981; Bandis et al., 1983). 50.2.6 Fracture HM Behavior during Shear Slip Laboratory measurements of HM behavior during shear have been rare because of a lack of specialized test equipment. The first comprehensive experimental study of permeability changes caused by shear was conducted by Makurat et al. (1990), who developed a coupled shear-flow test apparatus that could apply bi-axial stress. They concluded that whether the conductivity increases or decreases with shear depends on both the joint and rock properties, as well as the exact nature of the stress applied. Makurat et al. (1990) determined that decreases in hydraulic conductivity during shearing were a result of gouge production, which tended to block flow paths. The cases for increased hydraulic conductivity were modeled using a Barton–Bandis joint model, which predicted an increasing aperture with shear dilation, according to bE ⫽ bE i ⫹ ∆un,
(50.15)
where ∆un is the dilation of the fracture obtained from Equation (50.9) and bE i is the initial mechanical aperture before shear. This equation overpredicted the increases in fracture permeability during shear, because it does not correct for the formation of gouge material in the fractures. Gutierreze et al. (2000) conducted tests of stress-dependent permeability of a de-mineralized fracture in shale. The tests were conducted under varying normal stress as well as for shearing up to 4 mm. The results showed a typical nonlinear normal stress versus permeability relationship in which the permeability may change one or two orders of magnitude between maximum and minimum normal stress. The results indicated that fractures would never completely close even under normal stresses close to or higher than the unconfined compressive strength of intact shale. On the other hand, shearing of the fracture at a constant normal stress lower than the unconfined compressive strength of the shale caused dilation of the fracture and an order of magnitude increase in fracture permeability. Shearing at a constant normal stress higher than the unconfined compressive strength caused a negative dilation of the fracture and about a six-orders-of-magnitude reduction in fracture permeability. The reduced permeability results from the shear-induced gouge formation blocking the fracture aperture. However, despite being close to a six-orders-of-magnitude reduction, the fracture permeability was still about three orders of magnitude larger than the intact shale permeability. 50.2.7 Summary of Indirect Couplings and Current State-of-the-Art In summary, indirect coupling, involving changes in properties associated with fracturing and/or deformation of pre-existing fractures in an otherwise low-permeability shale, is probably the most relevant process to the study of caprock integrity and reservoir leakage during CO2 injection. Great progress has been made in the field of rock mechanics for developing constitutive models for coupled HM responses in rock fractures. The empirical work on
ELSE_TSANG_CH050.qxd
9/15/2005
1:06 PM
Page 657
50.3 Natural and Industrial Analogs Related to Study of Caprock Integrity
657
constitutive models by Barton and Bandis (Barton and Choubey, 1977; Bandis et al., 1983; Barton et al., 1985) has been especially important for practical applications. There is, however, a lack of data on the HM behavior of fractured shale, especially related to permeability evolution during shear under various normal stresses.
50.3 NATURAL AND INDUSTRIAL ANALOGS RELATED TO STUDY OF CAPROCK INTEGRITY AND RESERVOIR LEAKAGE Many natural and industrial analogs exist for CO2 sequestration. For studying caprock integrity and potential reservoir leakage during CO2 injection, long-term containment of pressurized gas and mechanical damage to shale-like formations are of particular interest. Below are brief discussions of these analogues and related studies. 50.3.1 Correlation between Shear Stress and Permeability in Fractured Formations In the last decade, investigations of the active lithospheric-plate boundary in California have shown that fractures favorably oriented for shear slip, the so-called critically stressed fractures, tend to be active groundwater flow paths (Barton et al., 1995, 1998; Ferrill et al., 1999). The rationale for bulk permeability being dominated by critically stressed fractures is that most fractures in the bedrock are cemented because of water/rock chemical reactions. If shear slip occurs on a critically stressed fracture, it can raise the permeability of the fracture through several mechanisms, including brecciation, shear dilation, and breakdown of seals (Barton et al., 1995). Lately, similar correlations have been found at the KTB Scientific Drill Hole in Germany down to 7 km (Ito and Zoback, 2000), and also at Äspö, Sweden, in the Precambrian rocks of the Baltic Shield deep in the Eurasian plate (Talbot and Sirat, 2001). On the other hand, experience from injection experiments at the hot-dry rock geothermal sites in Soultz, France, and Rosemanowes, UK, suggests that a pore-pressure increase of 5–6 MPa over ambient is needed to stimulate significant microseismicity (Evans et al., 1999). This indicates that under undisturbed stress and pressure conditions, the fractures would not be at the verge of shear failure. However, the observations about a possible correlation between maximum shear stress and permeability and experiences during injection at hot-dry rock sites can be useful for developing safe geological CO2 sequestration techniques. 50.3.2 Overpressured Sediments Geological containment of pressurized gases over long time periods can be studied in the evolution of overpressured sediments and gas reservoirs (Poston and Berg, 1997). Mechanisms of pressure generation include rapid sedimentation and compaction of sediments, hydrocarbon generation, and sediment diagenesis. In overpressured sediments, hydraulic fracturing through shale plays an important role. Field data indicate that overpressured sediments can experience episodic fluid expulsion into overlying layers during the evolution of the basin. In the case of tight sediments such as shale-rich materials with very low permeability, the most probable path of fluid leakage is by way of fractures through the tight formation (Gutierreze et al., 2000). The decrease in pore pressure as the fluids are expelled through newly opened fractures may facilitate the precipitation of minerals along the fracture wall, which may eventually seal the fracture. The sealing of fractures can then start a new episode of overpressuring and fluid leakage during continued pressure buildup
ELSE_TSANG_CH050.qxd
9/15/2005
1:06 PM
Page 658
658
Coupled Hydromechanical Effects in CO2 Injection
in the sediment. Given the evidence of leakage from overpressured reservoirs, evaluation of the potential for hydraulic fracturing, reopening of sealed fractures, or reactivation of faults associated with CO2 injection is essential. 50.3.3 Geological Storage of Natural Gas Underground gas storage in reservoirs and rock caverns is a mature technology that has been practiced for decades. In North America, it is a typical practice to operate gas-storage reservoirs at or below the original reservoir pressure, out of concerns about caprock integrity, fracturing, faulting, and gas loss. As pointed out by Bruno et al. (1998), the maximum safe operating pressure depends on several geomechanical factors, including in situ stresses, stresses induced by local and global changes in the reservoir, and the mechanical and hydraulic properties of the reservoir and overburden. Nagelhout and Roest (1997) performed numerical modeling of a generic underground natural gas storage facility to investigate the possibilities of fault slip in some of the new gas storage facilities in The Netherlands. This result showed that fault slip of up to 5 cm occurred during the initial depletion of fluid pressure from 300 to 115 bar. On the other hand, no inelastic fault slip occurred during the following simulated gas storage operations. However, the fluid pressure did not exceed the initial reservoir pressure by more than 14%. Generally, experience from underground gas storage is an important asset for development of a safe and efficient injection technology for CO2. However, it is likely that underground injection of CO2 will involve injection pressures that are considerably higher than the ambient formation fluid pressure, which is significantly higher than what is typically used in underground gas-storage projects. 50.3.4 EDZ in Tunnels through Argillaceous Rock Formations Damage in “shale-like formations” has been studied in the context of geological storage of nuclear waste in argillaceous formations. Damage in the “excavation disturbed zones” around tunnels caused by the formation of fracturing has been shown to induce significant anisotropy in ultrasonic velocity (Martin et al., 2003). In these damage zones, two failure modes can be observed: extensional fractures parallel to tunnel walls (perpendicular to the least principal stress) and a combined tensile and shear failure of bedding planes (Alheid, 2003). The opening of fractures is always associated with an increase in permeability parallel to fractures. It is possible that a similar type of damage could occur in the caprock system during underground CO2 injection. If the fluid pressure is increased sufficiently, hydraulic fractures may be created or shear-slip may occur on pre-existing fractures. Such fracturing and shear-slip may induce changes in the elastodynamic properties that could be detectable with seismic monitoring. The increased research efforts associated with nuclear waste disposal in deep tunnels in argillaceous formations could well lead to new insights into the HM behavior of those rocks, which are likely candidates for caprock at CO2 sequestration sites.
50.4 THE TOUGH-FLAC THM SIMULATOR Coupled HM processes in CO2 injection are complex and nonlinear, as discussed above. Their study requires the use of numerical modeling. In this section, we shall describe a recently developed numerical simulator used for CO2 injection studies.
ELSE_TSANG_CH050.qxd
9/15/2005
1:06 PM
Page 659
50.4 The TOUGH-FLAC THM Simulator
659
The TOUGH-FLAC simulator (Rutqvist et al., 2002) is based on the coupling of the two existing computer codes TOUGH2 (Pruess et al., 1999) and FLAC3D (Itasca Consulting Group, 1997). TOUGH2 is a well-established code for geohydrological analysis with multiphase, multicomponent fluid flow and heat transport, while FLAC3D is a widely used commercial code designed for rock and soil mechanics. For analysis of coupled THM problems, the TOUGH2 and FLAC3D are executed on compatible numerical grids and linked through external coupling modules, which serve to pass relevant information between the field equations solved in the respective codes (Fig. 50.4). A TOUGH-to-FLAC link takes multiphase pressures, saturation, and temperature from the TOUGH2 simulation and provides updated temperature and pore-pressure information to FLAC3D (Fig. 50.4). Because the TOUGH2 mesh uses one gridpoint within each element, and FLAC3D nodes are located in element corners, data must be interpolated from mid-element (TOUGH2) to corner locations (FLAC3D). After data transfer, FLAC3D internally calculates thermal expansion and effective stress according to: ∆ε T ⫽ ⌱β T ∆T,
(50.16)
σ ⬘ ⫽ σ ⫺ ⌱α P,
(50.17)
where ε T is thermal strain, β T the linear thermal expansion coefficient, I the unit tensor, T the temperature, σ ⬘ the effective stress, σ the total stress, α the Biot effective stress parameter, and P the pore fluid pressure. In a multiphase flow calculation, the value of P transferred to FLAC3D could represent an average pore pressure calculated from the pressures of the various phases (Rutqvist et al., 2002).
EOS Fluids
PVT functions for water, air, CO2, etc.
TOUGH2 Reservoir Simulator P, T, S
THM Loop
Multiphase fluid and heat transport analysis
, k, Pc
FLAC3D Soil & Rock Mechanics
Mechanical Model = H( , T, / tt)
TOUGH-FLAC Coupling Modules
Geomechanical Stressstrain analysis with HM and TM coupling Constitutive model for visco-elastic, poroelasto-plastic material
Fig. 50.4. Schematic links between TOUGH2 and FLAC3D for a coupled THM simulation within each time step.
ELSE_TSANG_CH050.qxd
9/15/2005
1:06 PM
660
Page 660
Coupled Hydromechanical Effects in CO2 Injection
A FLAC to TOUGH link takes element stress and deformation from FLAC3D and updates element porosity, permeability, and capillary pressure to be used by TOUGH2, according to the following general expressions:
φ ⫽ φ (σ ⬘,ε),
(50.18)
k ⫽ k(σ ⬘,ε),
(50.19)
Pc ⫽ Pc(σ ⬘,ε).
(50.20)
No interpolation in space is required for this data transfer because stress and strain are defined in FLAC3D elements, which are identical to TOUGH2 elements. A TOUGH-FLAC coupling module for this link should calculate the hydraulic property changes, based on material-specific theoretical or empirical functions. A separate batch program controls the coupling and execution of TOUGH2 and FLAC3D for the linked TOUGH-FLAC simulator. It was done within the FLAC3D input file using the FLAC-FISH programming language (Itasca Consulting Group, 1997). The calculation is then stepped forward with the transient TH analysis in TOUGH2, by conducting, at each time step or at the TOUGH2 Newton iteration level, a quasi-static mechanical analysis with FLAC3D, to calculate stress-induced changes in porosity and intrinsic permeability. Because of small strain conditions, there is no change in mesh dimension during the simulations. 50.5 APPLICATION OF THE TOUGH-FLAC CODE TO CO2 INJECTION The TOUGH-FLAC code is applied to simulate an injection operation for disposal of CO2 into a permeable brine formation, which is overlain by a semi-permeable caprock (Fig. 50.5). Coupled HM interactions in a caprock/reservoir system are studied with a view toward their role on the integrity of the caprock system and reservoir leakage. This analysis is an extension of a recently published analysis of coupled HM changes in a brine-aquifer/caprock system during CO2 injection (Rutqvist and Tsang, 2002). Our present study utilizes the same model for the brine-aquifer/caprock system, but is focused on the evolution of stresses during the CO2 injection. In particular, we are studying the role of the initial stress regimes on the potential for faulting and fracturing in a pre-existing fracture zone and along existing sealing faults. In our model, we inject CO2 at a constant rate over a 30-year period at a depth of 1300–1500 m. The injection zone is overlain by a 100 m thick caprock, located at 1200–1300 m. Three cases of brine-aquifer/caprock systems are studied (Fig. 50.5): 1. A brine-aquifer caprock system with no lateral confinement 2. A caprock intersected by a fracture zone which is initial hydraulically inactive 3. A brine-aquifer/caprock system with two sealing faults. All simulations are conducted in two-dimensional vertical section models, which are extended far enough in the lateral directions to be “infinite-acting” (Fig. 50.5). The twodimensional geometry implies that injection takes place in a well field that consists of a long line of injection wells. 50.5.1 Material Properties The material properties are given in Table 50.1. They correspond to a sandstone aquifer with a caprock of shale. Functions are developed or discussed for relative permeability, capillary
ELSE_TSANG_CH050.qxd
9/15/2005
1:06 PM
Page 661
50.5 Application of the TOUGH-FLAC Code to CO2 Injection
661
CO2
Faults Fracture Zone 1.5 km Cap rock 100 m thick
Cap Rock
CO2 Injection zone 200 m thick
Large Lateral Extension
Fig. 50.5. Geometry of simulation cases for CO2 injection.
pressure, and porosity-stress and porosity-permeability correlation (Rutqvist and Tsang, 2002). Isotropic HM rock properties are represented by a porosity-mean stress and a permeability-porosity relationship. The porosity, φ, is related to the mean effective stress as
φ ⫽ (φ0 ⫺ φr) exp(⫺5 ⫻ 10⫺8σ ⬘M) ⫹ φ r,
(50.21)
where φ0 is porosity at zero stress, φr the residual porosity at high stress, and the mean effective stress (in Pa) is defined from the principal stresses as 1 σ ⬘M ⫽ ᎏ (σ ⬘1 ⫹ σ ⬘2 ⫹ σ ⬘3). 3
(50.22)
The effective principal stresses (with compression positive) are calculated (Rutqvist et al., 2002) as follows: – σ ⬘1 ⫽ σ 1 ⫺ αP ,
(50.23)
– σ ⬘2 ⫽ σ 2 ⫺ αP ,
(50.24)
– σ ⬘3 ⫽ σ 3 ⫺ αP ,
(50.25)
– where α is Biot’s effective stress parameter and P an average pore pressure defined for an unsaturated system as: – P ⫽ S l Pl ⫹ (1 ⫺ Sl )Pg.
(50.26)
ELSE_TSANG_CH050.qxd
9/15/2005
1:06 PM
Page 662
662
Coupled Hydromechanical Effects in CO2 Injection Table 50.1. Material properties used in TOUGH-FLAC simulations
Property Young’s modulus, E (GPa) Poisson’s ratio, ν (dimensionless) Biot’s parameter, α (dimensionless) Saturated rock density, ρs (kg/m3) Zero stress (flowing) porosity, φ0 (dimensionless) Residual (flowing) porosity, φr (dimensionless) Zero stress permeability, k0 (m2) Corey (1954) irreducible gas saturation, Srg (dimensionless) Corey (1954) irreducible liquid saturation, Srl van Genuchten (1980), P0 (kPa) (at zero stress) van Genuchten (1980) exponent, m
Upper
Cap
Aquifer
Basement
Fracture zone
5 0.25
5 0.25
5 0.25
5 0.25
2.5 0.25
1
1
1
1
1
2260
2260
2260
2260
2260
0.1
0.01
0.1
0.01
0.1
0.09
0.009
0.09
0.009
0.05
1 × 10⫺17
1 × 10⫺12
1 × 10⫺15
1 × 10⫺17
1 × 10⫺13
0.05
0.05
0.05
0.05
0.05
0.3
0.3
0.3
0.3
0.3
196 0.457
3100 0.457
19.6 0.457
3100 0.457
1 0.457
In the current calculations, we have put α ⫽ 1, which is a reasonable value. The permeability is correlated to the porosity according to the following exponential function (modified from Davis and Davis, 1999): k ⫽ k0 exp[22.2(φ /φ0 ⫺ 1)].
(50.27)
where k0 is the zero-stress permeability. The coefficients in the functions for porosity and permeability changes—Equations (50.27) and (50.25)—are obtained by matching with laboratory measurements on sandstone presented by Davis and Davis (1999). Their experimental data show a one-order-of-magnitude reduction in permeability with effective stress increasing from 0 to 30 MPa. In addition to the two coupling functions in Equations (50.21) and (50.27), the capillary pressure is also modified according to a function due to Leverett (1941): k0苶 /φ苶0 兹苶 Pc ⫽ Pc0(Sl) ᎏ . 兹苶k苶 /φ
(50.28)
Thus, porosity, permeability, and capillary pressure are all directly or indirectly dependent on the mean effective stress. In this analysis, the relationships for porosity as a function of stress and for permeability as a function of porosity—derived for sandstone—are also used for the shale. However, for the shale, the flowing porosity and permeability are assumed to be two to four orders of magnitude lower than for sandstone. Note that total porosity of shale can be much greater than the porosity given in Table 50.1, which is the connected porosity available for fluid flow.
ELSE_TSANG_CH050.qxd
9/15/2005
1:06 PM
Page 663
50.5 Application of the TOUGH-FLAC Code to CO2 Injection
663
In our modeling, we will analyze the possibilities of a failure of the caprock by looking at the critical pressure that could induce hydraulic fracturing or the critical pressure that could induce shear slip of pre-existing faults. A conservative assumption is that a hydraulic fracture could develop as soon as the fluid pressure exceeds the least compressive principal stress, and hence, the critical pressure for fracturing derived from Equation (50.10) is Pfc ⫽ σ 3
(50.29)
A conservative assumption for the onset of fault slip is that it is the condition under which a fault could exist at any point with an arbitrary direction. For such a case, the Coulomb failure criterion can be written in the following form (Jaeger and Cook, 1979): |τ m2| ⫽⫺(σm2 ⫺ Psc)sinϕ ⫹ S0cosϕ,
(50.30)
where τm2 and σm2 are the two-dimensional maximum shear stress and mean stress in the plane σ1, σ3, defined as 1 τ m2 ⫽ ᎏ (σ1 ⫺ σ3) 2
(50.31)
1 σ m2 ⫽ ᎏ (σ 1 ⫹ σ 3), 2
(50.32)
and S0 and ϕ are the faults coefficient of internal cohesion and angle of internal friction, respectively. For faults, a zero cohesion may be assumed, and a typical range for ϕ is 25–35° (Goodman, 1989). In the following calculations, we test for slip using a zero cohesion (S0 ⫽ 0) and a friction angle of 30°. 50.5.2 Initial and Boundary Conditions At the start of the CO2 injection operation, the initial temperature and pressure at the injection point (about 1500 m depth), are T ⫽ 47.5°C and P ⫽ 15.1 MPa, respectively (see Table 50.2). This is well within the range for assuring a supercritical CO2. At this depth, the initial stress is σv ⫽ σh ⫽ 33.2 MPa (for isotropic initial stress case), and the initial mean effective stress is 18.1 MPa, leading to a permeability of about 0.3 ⫻ 10⫺13 m2 (~30 mD) in the aquifer and 0.3 ⫻ 10⫺17 m2 (~3 nD) in the caprock. Table 50.2. Initial conditions in the most important geological formations before CO2 injection Parameter
Caprock
Aquifer
Fracture zone (closed)
Temperature, T (°C) Pressure, P (MPa) Mean stress, σ’M (MPa) Porosity, φ (dimensionless) Permeability, k (m2) Van Genuchten’s airentry pressure, P0 (kPa)
40–42.5 12.1–13.1 26.6–28.8 0.0094 0.3×10–17 5800
42.5–47.5 13.1–15.1 28.8–33.2 0.094 0.3×10–13 36.8
40–42.5 12.1–13.1 26.6–28.8 0.045–0.047 0.3–0.4×10⫺17 5800
Values are given as a range because they vary with depth in each formation.
ELSE_TSANG_CH050.qxd
9/15/2005
664
1:06 PM
Page 664
Coupled Hydromechanical Effects in CO2 Injection
As mentioned above, the lateral boundaries are placed far away from the injection points, and therefore the model can be considered as infinite in its lateral direction. At the bottom, roller boundaries (no displacement normal to the boundary) and a constant temperature of 85°C is maintained throughout the simulation. This simulation is conducted in an isothermal mode, which implies that the thermal gradient is maintained according to the initial conditions throughout the simulation. 50.5.3
CO2 Injection
Compressed CO2 is injected at a constant rate of 0.05 kg/s per meter (normal to the twodimensional model). With such an injection rate, the aquifer pressure increases substantially with time, and after several years of injection, the pressure would exceed the lithostatic stress of 33.2 MPa. Considering the amount of CO2 that would have to be disposed from a single coal-fired power plant, a high pressure increase would be expected. For example, a standard size 1000 MW coal-fired power plant that produces CO2 at a rate ~350 kg/s (Hitchon, 1996) would require a well field about 7 km wide (0.05 kg/s/m ⫻ 7000 m ⫽ 350 kg/s). An axisymmetric simulation conducted by Pruess et al. (2001) for this injection rate (350 kg/s) into a 1 km diameter well field leads to a pressure increase of about 17 MPa, which is very similar to what we obtain in our two-dimensional modeling with an injection rate of 0.05 kg/s/m. 50.5.4 Evolution of Fluid Pressure and Spread of CO2 Figure 50.6a presents the calculated injection pressure during a 30-year injection period, with or without consideration of the stress-dependent rock-mass permeability. The difference in injection pressure is explained by permeability in the injection zone increasing, in the former case, because of a general reduction in effective stresses. However, these changes in permeability are moderate (less than a factor of 2) because of a rather insensitive stresspermeability relationship for porous sandstone. Figure 50.6b presents the spread of CO2 fluid and fluid pressure within the aquifer/caprock system after 10 years of injection for the HM calculation. The figure shows that the CO2 has spread under the cap over 4 km and has penetrated upward into the caprock by about 10 m. At this time, injection pressure has increased to 33 MPa, which is slightly less than the lithostatic stress at the injection point (Fig. 50.6a). Figure 50.7 presents the evolution of fluid pressure, CO2 gas saturation, and permeability along a vertical profile at the injection point (x ⫽ 0). The figure shows that although fluid pressure is approximately equal to the lithostatic stress at the injection point (at about 1450 m), the fluid pressure actually exceeds the lithostatic stress at the lower parts of the caprock. However, even though the fluid pressure is close to or even exceeds the lithostatic stress, changes in permeability are relatively small. This is because these changes in permeability depend on changes in the effective mean stress, which in turn depend on stress changes in both the horizontal and vertical direction. As will be described in more detail below, poroelastic compressive stresses can build up and provide increased confining stresses, especially in the horizontal direction. These injection-induced increases in confining stresses tend to prevent large changes in effective stress, which in turn prevent large permeability changes. 50.5.5 Evolution of Injection-Induced (Poroelastic) Stress Changes Figures 50.8 and 50.9 show how hydraulic injection causes changes in both confining and effective stresses in the caprock/reservoir system. These local changes in the stress field are
ELSE_TSANG_CH050.qxd
9/15/2005
1:06 PM
Page 665
INJECTION PRESSURE, P (MPa)
50.5 Application of the TOUGH-FLAC Code to CO2 Injection 45
665
H Calculation with k = k0
40 Lithostatic pressure
35 30
HM Calculationwith k = k(')
25 20
Current study
15 0
5
10
(a)
15
20
25
TIME (Years)
Injection Pressure 33 MPa
P (MPa) 5
DEPTH (m)
500 1000 Cap 1500
Upper Aquifer
10
CO2 25
Injection Zone
20
Base
20
2000
25
2500 -10000 (b)
-5000
0
5000
10000
DISTANCE FROM INJECTION POINT(m)
Fig. 50.6. TOUGH-FLAC simulation of CO2 injection into a caprock/reservoir system with a homogeneous caprock and no laterally sealing faults: (a) injection pressure versus time, and (b) spread of CO2 (white line) and fluid pressure distribution after 10 years of injection.
caused by complex poroelastic responses as CO2 fluid is added to the system. Figure 50.8 indicates that the vertical and horizontal in situ stresses (total stresses) increase near the injection point. The stresses increase as a result of poroelastic stresses that occur when the porous rock attempts to expand in a confined rock mass. Both vertical and horizontal stresses increase in the injection aquifer. However, in the caprock, just above the injection interval, the horizontal stresses increase much more than the vertical stresses. Figure 50.9 presents changes in vertical and horizontal effective stresses near the CO2 plume. The figure shows that the effective stresses are reduced mostly at the interface between the injection aquifer and the caprock, and the vertical effective stress is reduced more than the horizontal one. 50.5.6 Potential for Shear Slip and Fracturing at Different Stress Regimes The injection-induced incremental stresses caused by poroelastic responses (shown in Figs. 50.8 and 50.9) are independent of the initial stress field as long as the rock behaves elastically. However, if the induced stress changes are sufficiently large, or if the fluid pressure exceeds the confining stresses, inelastic mechanical yielding failure may occur. In this case,
ELSE_TSANG_CH050.qxd
9/15/2005
1:06 PM
Page 666
666
Coupled Hydromechanical Effects in CO2 Injection 0
1000
1 Year
500
1000
10 Years
1 Year
Z (m)
Cap Inj.zone Base
1500
0
Until 10 years the injected CO2 is contained within the injection zone and maximum CO2 gas saturation is 68%
500
Z (m)
Z (m)
1000
ss re St tic ta e s ur os res th cP Li tati ro s Hyd
500
0 Fluid pressure exceeds lithostatic stress at lower part of caprock
1500
Relatively small changes in permeability of intact rock 10 Years
1500 1 Year
10 Years 2000
2000
2000
2500
2500
2500
3000
0
10
20
30
FLUID PRESSURE (MPa)
40
3000
0
0.5 CO2 GAS SATURATION
1
3000 -18 10 10-17 10-16 10-15 10-14 10-13 PERMEABILITY (m2)
Fig. 50.7 Vertical profiles of fluid pressure, CO2 gas saturation, and permeability at t = 0, 1, and 10 years.
Fig. 50.8. Calculated changes in total in situ stresses after 10 years of injection. Extent of CO2 is shown as a white line.
ELSE_TSANG_CH050.qxd
9/15/2005
1:06 PM
Page 667
50.5 Application of the TOUGH-FLAC Code to CO2 Injection
667
Fig. 50.9. Calculated changes in effective stress in the aquifer-cap rock system after 10 years of CO2 injection. Light pressure contours indicates, where highest stress reduction occurs and the extent of CO2 is shown as a white line.
the potential for rock-mass failure is governed by the evolution of the magnitude and directions of maximum and minimum principal effective stresses and their relation to critical stresses, defined in the adopted failure criteria (Equations 50.29 and 50.30). Figure 50.10 presents vertical profiles of the evolution of fluid pressure, and vertical confining (total) and effective stresses. The vertical effective stresses are reduced most at the interface between the injection zone and the caprock (at 1300 m). At this location there is a relatively large increase in fluid pressure, but a relatively small increase in confining stress. However, note that even though the fluid pressure exceeded the lithostatic stress in the lower part of the caprock and over most of the injection zone, the effective vertical stress is still in compression. As a result, no hydraulic fracturing is expected. Figures 50.11 and 50.12 present vertical profiles of the evolution of horizontal confining and effective stresses for two different stress regimes: 1. An isotropic stress regime (σh ⫽ σv), 2. A reverse fault stress regime (σh ⫽ 0.7σv). Figure 50.11 indicates that the horizontal effective stress is well into the compression at all times, and therefore formation of vertical hydraulic fractures is highly unlikely. Comparing the evolution of horizontal stresses in Figure 50.11 with the evolution of vertical
ELSE_TSANG_CH050.qxd
9/15/2005
1:06 PM
Page 668
668
Coupled Hydromechanical Effects in CO2 Injection 0 o th Li re
Cap Inj. zone
Z (m)
Base
1 Year
1500
Base
1 Year 10 Years
10 Years
Z (m)
ssu
1500
1000
1000
Inj. zone
Cap Inj. zone
1500
2000
2000
2500
2500
2500
3000
3000 0
10
20
30
40
Base
10 Years 1 Year
2000
3000
Vertical effective compressive stress in compression but close to tension
500
Vertical stress increases slightly in cap
ss
Pre
Cap
Z (m)
500
tre
c tati ros
1000
S ic at st
Hyd
500
0
0
Fluid pressure exceeds lithostatic stress at lower part of caprock
0
-10 -20 -30 -40 -50 -60
0
10
zz (MPa)
FLUID PRESSURE (MPa)
20
30
'zz (MPa)
Fig. 50.10. Vertical profiles of fluid pressure, total in situ stress, and effective vertical stress at t = 0, 1, and 10 years.
0
0
tic St
ss
Inj. zone Base
Z (m)
e
Z (m)
1000 Cap
1 Year
Inj. zone Base 1 Year
1500
10 Years
2000
2500
2500
0
10
20
30
Cap Inj. zone
1500
40
FLUID PRESSURE (MPa)
3000
Base
10 Years 1 Year
10 Years
2000
3000
Horizontal effective stress in compression at cap and hence no vertical fracturing expected
500
1000
sur
1500
500
res cP
tati
Cap
Poroelastic stress increases horizontal stress in injection zone
re
ros
1000
0
Z (m)
Hyd
ta
os
th
Li
500
Fluid pressure exceeds lithostatic stress at lower part of caprock
2000 Slight unloading of horizontal stress away from injection zone
0
10
20
30
40
xx (MPa)
2500
50
60
3000
0
10
20
30
' xx (MPa)
Fig. 50.11. Vertical profiles of fluid pressure, total in situ stress, and effective horizontal stress at t = 0, 1, and 10 years for isotropic stress regimes (σxxi ⫽ σzzi).
stress in Figure 50.10 again illustrates that the vertical effective stresses change much more than the horizontal effective stresses. As discussed above, the reason is that poroelastic stresses develop more strongly in the horizontal direction than the vertical direction, which helps to offset the effects of increasing fluid pressure. Comparing the vertical profiles of effective stress evolution in Figure 50.11 (for isotropic stress regime) with those in Figure 50.12 (for reverse faulting stress regime) demonstrates the great importance of the initial (pre-injection) horizontal stress magnitude. Figure 50.12 shows that the local horizontal
ELSE_TSANG_CH050.qxd
9/15/2005
1:06 PM
Page 669
50.5 Application of the TOUGH-FLAC Code to CO2 Injection 0
0
0
500
s
e sur res
1000
1000
Z (m)
Z (m)
Cap
Inj. zone
1500
Base
Horizontal effective stress close to tensile in lower part of cap
500
Poroelastic stress increase
s re St
cP tati ros
ic at st
Hyd
o th
1000
Fluid pressure exceeds lithostatic stress at lower part of caprock
1 Year
Cap Inj. zone
1500
Base 10 Years
1 Year
Z (m)
Li
500
669
Cap Inj. zone
1500 10 Years
Base
1 Year
10 Years
2000
2000
2000
2500
2500
2500
3000 0
10
20
30
40
FLUID PRESSURE (MPa)
3000 0
3000 10
20
30
40
xx (MPa)
50
60
0
10
20
30
xx ' (MPa)
Fig. 50.12. Vertical profiles of fluid pressure, total in situ stress, and effective horizontal stress at t = 0, 1, and 10 years for reverse faulting stress regimes (σxxi ⫽ 0.7σzzi).
effective stresses become very close to tensile at the interface between the caprock and injection zone (1300 m). This implies that for the case of a reverse faulting stress regime, there is an increased likelihood for formation of vertical fractures at this location. Figures 50.13 and 50.14 show the zones of possible fault slip and hydraulic fracturing for the two cases of stress regime. The potential for hydraulic fracturing and fault slip was evaluated using Equations (50.29) and (50.30). Figure 50.13 indicates that for the isotropic stress regime, mechanical failure would most likely initiate at the interface between the caprock and injection the zone. In this zone, the reduction in vertical effective stress can lead to the formation of horizontal hydraulic fractures (Fig. 50.13b). Furthermore, a larger zone of possible slip on pre-existing fractures occurs at the upper and lower part of the injection zone. This finding implies that an unfavorably oriented fault could be reactivated with accompanying micro-seismicity and possible permeability change. However, even if fracturing or fault reactivation would take place in the lower parts of the caprock, the simulation indicates that it would be contained within the lower part of the cap and would not propagate through the upper part of the cap. Figure 50.14 shows that, for the reverse faulting stress regime, the zones of potential shear slip and hydraulic fracturing are more extensive. This is because in this case the initial, far-field horizontal stress is lower. Importantly, in the case of a reverse faulting stress regime, shear slip and hydraulic fracturing would preferentially occur on vertical fractures. Further, the shear slip could occur in the entire caprock and not just in its lower part. Thus, there would be a potential for slip of a pre-existing fault crossing the entire caprock with a potential for leakage. 50.5.7 Effect of a Vertical Weakness Zone in the Cap A weakness is introduced into the caprock in the form of a 10 m wide vertical fracture zone (Fig. 50.15). The fracture zone is simulated as a porous medium, more porous than the surrounding caprock, with a much more sensitive porosity–stress relationship (Table 50.1). The fracture zone is initially assumed to have the same permeability as the surrounding caprock
ELSE_TSANG_CH050.qxd
9/15/2005
1:06 PM
Page 670
670
Coupled Hydromechanical Effects in CO2 Injection 1100
Zone of possible slip along subhorizontal fractures
Upper Aquifer
DEPTH (m)
1200 Cap 1300
CO2+Brine Water
1400
Brine Water Injection Zone
1500 Base -2000
(a)
-1000
0
1100
2000
Zone of possible subhorizontal hydraulic fracturing
Upper Aquifer
1200 DEPTH (m)
1000
DISTANCE FROM INJECTION POINT (m)
Cap 1300 1400
CO2 +Brine Water Brine Water
Injection Zone
1500 Base -2000
(b)
-1000
0
1000
2000
DISTANCE FROM INJECTION POINT (m)
Fig. 50.13. Calculated zones (shaded) of possible shear slip and hydraulic fracturing after 10 years of injection for the case of isotropic stress regime (σh ⫽ σv). The black line shows extent of CO2.
(Table 50.2). Thus, the fracture zone could be envisioned as consisting of closely spaced fractures that have been healed and are completely closed (from the hydraulic point of view). The fractures in this zone have no tensile strength and can thus open up as a result of a reduction in effective stress when fluid pressure increases. Using the assumed sensitive stress-versus-permeability function, the fault permeability increases as the fluid pressure increases within the caprock. As a result, CO2 starts to leak at the top of the caprock after about 6 years of injection (Fig. 50.15). As the pressure increases during the CO2 injection, the effective stresses are reduced in the fault, and consequently the fault tends to open for more flow, according to the assigned stress-porositypermeability relationships (Equations 50.21 and 50.27 and Table 50.1). In this simulation, the permeability of the fault increases gradually up to two orders of magnitude during the first 10-year period of the injection. Consequently, the brine could more easily be displaced upwards through the fault by the less-dense CO2 fluid. At 10 years, the CO2 has broken through the cap and is slowly leaking and migrating upward by buoyancy (Fig. 50.15, left). However, at 10 years, the increase in the leakage rate is only about 6% of the injected CO2, and therefore the leakage cannot be detected by monitoring the evolution of injection pressure. As described by Rutqvist and Tsang (2002), the leakage rate through the fracture zone accelerates at about 5.7 years (Fig. 50.15, right) due to several contributing factors, including changes in relative permeability, fluid viscosity, and fluid pressure-induced fracture opening, as the fracture zone is invaded by CO2.
ELSE_TSANG_CH050.qxd
9/15/2005
1:06 PM
Page 671
50.5 Application of the TOUGH-FLAC Code to CO2 Injection 1100
Upper Aquifer
1200 DEPTH (m)
671
Zone of possible slip along sub-vertical faults
Cap 1300
CO2+Brine Water
1400
Brine Water Injection Zone
1500 Base -2000
-1000
0
1000
2000
DISTANCE FROM INJECTION POINT (m) Zone of possible subvertical hydraulic fracturing
1100 Upper Aquifer
DEPTH (m)
1200 Cap 1300
CO2+Brine Water
1400
Brine Water
Injection Zone 1500 Base -2000
-1000
0
1000
2000
DISTANCE FROM INJECTION POINT (m)
Fig. 50.14. Calculated zones (shaded) of possible shear slip and hydraulic fracturing after 10 years of injection for the case of reverse faulting stress regime (σh ⫽ 0.7σv). The black line shows extent of CO2.
The evolution of the stress field in the aquifer/caprock system is similar to that of the homogeneous caprock case (Figs. 50.8 and 50.9), except for the evolution of stresses in the fracture zone itself. This is illustrated in Figure 50.16, which shows the evolution of fluid pressure, confining stress, and effective stresses along two vertical profiles through the caprock: 1. A vertical profile through the caprock along the vertical fracture zone at x ⫽⫹75 m 2. A vertical profile through intact caprock at x ⫽⫺75 m. The figure indicates that fluid pressure in the vertical fracture zone (in the upper part of the caprock, at 1200 m) significantly increases compared to the case of an intact caprock. The higher permeability in the fracture zone implies that fluid pressure can penetrate upward more rapidly than in the surrounding low-permeability intact rock. Furthermore, almost no difference exists in the confining stress within the fracture zone. This is explained by the fact that poroelastic stresses in the caprock result from the overall fluid pressure increases over a wide lateral dimension, whereas a local fluid pressure change within a narrow fracture zone will not give rise to significant poroelastic stresses. Instead, the invaded fluid pressure will tend to reduce the effective stresses within the fracture zone, with the effect of a local expansion and opening of pre-existing fractures in the fracture zone. The simulation results for the case of a reverse stress regime indicates that the fluid-pressure increases in the fault would be sufficiently high to change the horizontal effective stresses to tension. A reduction of horizontal effective stresses along the fracture zone implies that pre-existing fractures
ELSE_TSANG_CH050.qxd
9/15/2005
1:06 PM
Page 672
672
Coupled Hydromechanical Effects in CO2 Injection 0 P (MPa)
Pinj = 33 MPa
5
DEPTH (m)
500
10
1000
15
30
1500
25
2000 2500
-2000
-1000
0
1000
2000
DISTANCE FROM INJECTION POINT (m)
FlUX OF CO2 (% of injection)
6.0 5.0 4.0 3.0 2.0 Influx into fault Outflux from fault
1.0 0.0
0
1
2
3
4
5
6
7
8
9
10
TIME (Years)
Fig. 50.15. Fluid pressure distribution (left) after 10 years of injection for the case of vertical fracture zone in caprock, and CO2 mass flux into the bottom and out of top of the fracture zone (right) as a function of time.
could be completely unloaded and propagated to provide additional connectivity for fluid flow through the fracture zone (Fig. 50.17). 50.5.8 Injection into a Reservoir/Caprock System Bounded by Two Sealing Faults The TOUGH-FLAC code can simulate slip explicitly along major faults through constitutive mechanical models and slip line elements available in FLAC3D. Figure 50.18 shows an example of fault reactivation analysis using this model. In this example, CO2 is injected within a permeable injection zone laterally confined between two sealing (low permeability) faults. The material properties of the injection zone and caprock are those given in Table 50.1. An internal friction angle of 25° is assumed for the faults. In this case, CO2 was injected at high pressure until slip was triggered along the two bounding faults. A maximum fault slip of about 5 cm was predicted along fault sections intersecting the injection zone (Fig. 50.18, right). However, analysis shows that shear failure is limited to a zone of substantially increased fluid pressure and does not propagate further than about 100 m above and below the injection zone. In general, the stress evolution around the faults and the injection zone is more complex in this case than for
9/15/2005 1:06 PM
Fig. 50.16. Vertical profiles of fluid pressure, total in situ stress, and effective horizontal stress at t ⫽ 0, 1, and 10 years across intact caprock (at x ⫽⫺75 m) and along vertical fracture zone (at x ⫽⫹75 m) for istotropic and reverse faulting stress regimes.
ELSE_TSANG_CH050.qxd Page 673
50.5 Application of the TOUGH-FLAC Code to CO2 Injection 673
ELSE_TSANG_CH050.qxd
9/15/2005
1:06 PM
Page 674
674
Coupled Hydromechanical Effects in CO2 Injection 1100
Zone of possible subvertical hydraulic fracturing
Upper Aquifer
DEPTH (m)
1200 Cap 1300 CO2+Brine Water 1400
Brine Water
Injection Zone 1500 Base -2000
-1000
0
1000
2000
DISTANCE FROM INJECTION POINT (m)
1150 1175
Zone of high potential for vertical hydraulic fracturing
UPPER FORMATION
DEPTH (m)
1200 1225
CAP
1250 1275 1300 1325 1350
INJECTION ZONE
-100
0
100
200
DISTANCE FROM INJECTION POINT (m)
Fig. 50.17. Calculated zone of possible hydraulic fracturing for the case of a pre-existing fracture zone and a reverse faulting stress regime (σh ⫽ 0.7σv). Right figure shows a zoomed in version of the left figure.
0
U
LT
500
FA
FA
U
LT
-500
2
1
0
DEPTH (m)
Z (m)
-1000 -1500
1000 FAULT 1
1500
-2000
2000
-2500
2500
-3000
2000
3000
4000 X (m)
5000
FAULT 2
3000 -0.05
0 0
0.05
SHEAR SLIP (m)
Fig. 50.18. TOUGH-FLAC simulation of fault slip during CO2 injection into a brine formation sealed by two faults: (a) model with calculated pressure contours, where darker contours represent higher pressure and (b) calculated shear slip along the two faults.
ELSE_TSANG_CH050.qxd
9/15/2005
1:06 PM
Page 675
50.6 Discussion and Conclusions
675
the homogeneous caprock case. Localized concentration of stresses (including shear stresses) as well as localized stress releases are more likely, and this could lead to additional damage, particularly in areas where the fault intersects the caprock. Such damage may induce increased permeability along the fault. However, this possibility was not considered in this initial study. In general, slip on pre-existing faults and other discontinuities, which intersect the caprock, are viewed as a likely scenario for generation of possible leakage paths for CO2. However, further analysis is required to evaluate whether fluid flow will occur in conjunction with the slip. Shear test on single fractures in shale indicates that permeability can increase or decrease depending on the current stress normal to the fractures (Gutierrez et al., 2000). At high normal stresses, shear slip is accompanied with significant gouge production, and the permeability can actually decrease by several orders of magnitude. However, geological studies indicate that local stresses and the presence of faults control containment and release of deep overpressured fluids. Further research is needed for a realistic modeling of complex fault structures and for modeling of potential changes in fault permeability and mechanical properties.
50.6 DISCUSSION AND CONCLUSIONS Given that this is a general presentation of coupled HM changes during a CO2 disposal operation, the analysis is somewhat simplified. Because of a lack of field data, some of the properties used in our simulations have been estimated. As such, results of this study should be taken only qualitatively, since the quantitative results are very sensitive to the assumed rock properties. At a real injection site, the parameters for these empirical relationships should be calibrated against in situ measurements at an appropriate scale, and over an appropriate range of values. The analysis provided in this report focuses on the evolution of injection-induced stresses in the caprock and the potential for inducing mechanical failure that could be detrimental to the performance of a CO2 injection operation. A simplified failure analysis is conducted, based on the evolution of induced effective stress changes for three cases of initial in situ stresses. Results indicate that the most important process in HM behavior of the caprock is a general reduction of the mean effective stress, caused by the high-pressure injection of CO2. The largest reduction of mean stress was found at the interface between the injection zone and caprock, and in the lower parts of the caprock. This reduction of mean stress is not only important for permeability changes and hydraulic fracturing, but is also very important for potential initiation of shear slip. It results in a lower shear strength at the same time when the shear stress is somewhat increased. In combination, these effects increase the possibility of shear failure. The analysis also shows that it is very likely that an onset of shear failure would occur prior to any hydraulic fracturing. This is partly a consequence of the slow increase in fluid pressure over the 10-year injection period. When the pressure increases slowly, fluid has the time to diffuse into the neighboring rock formation, which then expands and locally increases the total stress. Because of the geometry of the extensive horizontal aquifer, total stresses increase more in the horizontal than in the vertical direction. This has two consequences: First, it will prevent reduction of effective stress in the horizontal direction and thereby prevent the formation of vertical hydraulic fractures. Second, because the total stresses increase more in the horizontal direction, the maximum principal stress, σ1, will tend to become horizontal, and the difference between σ1 and σ3 (a measure of shear stress) will increase. An important result of this analysis is the observation that the induced hydraulic fracturing and shear reactivation could be contained within the lower portion of the caprock. Thus,
ELSE_TSANG_CH050.qxd
9/15/2005
676
1:06 PM
Page 676
Coupled Hydromechanical Effects in CO2 Injection
the sealing mechanism of the caprock may still be functioning in the upper part. The reason for this continued functioning is that the mean stress does not change very much in the upper part of the caprock, because the fluid pressure at that location is mostly controlled by the constant hydrostatic pressure in the upper permeable formation above the caprock. In the case of a vertical fault, it can also be observed that most of its permeability changes occur in the lower part of the caprock, while the opening of the fault was much smaller in the upper part. All these observations indicate that the upper part of the caprock can act as a bottleneck for CO2 migration through the caprock. Once the CO2 reaches the upper exit of the fault, the upward CO2 migration will accelerate because of both hydraulic and HM changes. Further, if a more rapid invasion of CO2 and fluid pressure occurs through a permeable fracture zone, the effective stresses would change more dramatically in this zone compared to the surroundings. The local reduction in effective stress may reopen fractures within this fracture zone, and if pressure is sufficiently high, fracturing could occur. The analysis also shows that the magnitude and anisotropy of the initial stress field is an important factor in determining when and how failure could occur. In the case of an isotropic stress field, with both stresses equal to the weight of the overburden, shear slip along low-angle faults and the formation of horizontal hydraulic fractures are the most likely failure modes. In the case of relatively low horizontal stress (which might be the most common case in these types of formations), shear slip along steep faults and formation of vertical fractures are the most likely failure modes. In summary, the following points on coupled HM effects in CO2 injection may be highlighted: ● A general reduction in the effective mean stress induces strongly coupled HM changes in the lower part of the caprock. Therefore, the strongest HM changes and the greatest risk of rock failure occur in the lower part of the caprock. ● Because the aquifer pressure slowly increases during the injection period, fluid has time to diffuse into the rock and create poroelastic stresses. These events will decrease the probability of fracturing and shear, but also make the shear reactivation more likely. Thus, shear reactivation of existing fractures is the primary failure mode of concern in CO2 injection. ● The analysis indicates that shear reactivation in the lower part of the caprock could take place at an injection pressure well below the lithostatic pressure. However, depending on the initial in situ stress field, this fault slip reactivation may or may not be confined to the lower parts of the caprock. ● The type of stress regime (e.g., isotropic or reverse fault types) is a key parameter that determines whether fracturing and shear slip are likely to take place along subhorizontal or subvertical fractures. For a common reverse fault type of stress regime, fracture slip would preferentially take place along subvertical fractures, so that hydraulic fracturing would be vertical. ● Once the CO fluid reaches the upper part of the caprock (for example, through a perme2 able fault) the upward CO2 migration is accelerated because of the combined effects of relative permeability and viscosity changes, as well as changes in intrinsic permeability caused by pressure-induced HM effects.
ACKNOWLEDGMENTS Review and comments by Dr. Christine Doughty and Dr. Kurt Nihei of Lawrence Berkeley National Laboratory are much appreciated. We gratefully acknowledge the support by the
ELSE_TSANG_CH050.qxd
9/15/2005
References
1:06 PM
Page 677
677
Office of Science, Office of Basic Energy Sciences, Chemical Sciences and Geosciences Division of the U.S. Department of Energy, under Contract No. DE-AC03-76SF00098.
REFERENCES Alheid, H.-J., 2003. Lessons learned in indurated clays. In: Impact of Excavation Disturbed or Damaged Zone (EDZ) on the Performance of Radioactive Waste Geological Repositories. Proceedings of the European Commission CLUSTER Conference. Luxembourg, November 3–5. Bandis, S., Lumsden, A.C. and Barton, N.R., 1983. Fundamentals of rock joint deformation. Int. J. Rock Mech. Min. Sci. Geomech. Abstr., 20: 249–268. Barton, N. and Choubey, V., 1977. The shear strength of rock joints in theory and practice. Rock Mech., 10: 1–54. Barton, N.R., Bandis, S. and Bakhtar, K., 1985. Strength, deformation and conductivity coupling of rock joints. Int. J. Rock Mech. Min. Sci. Geomech. Abstr. 22: 121–140. Barton, N.R. and Bakhtar, K., 1982. Rock joint description and modeling for the hydrothermomechanical design of nuclear waste repositories. Technical Report 83-10. TerraTek Engineering, Salt Lake City, Utah. Barton, C.A., Zoback, M.D. and Moos, D., 1995. Fluid flow along potentially active faults in crystalline rock. Geology, 23: 683–686. Barton, C.A., Hickman, S.H., Morin, R. and Zoback, M.D., 1998. Reservoir-scale fracture permeability in the Dixie Valley, Nevada, Geothermal fields. Society of Petroleum Engineers. Richardson, TX, SPE Paper No. 47371. Biot, M.A., 1941. A general theory of three dimensional consolidation. J. Appl. Phys., 12: 155–164. Biot, M.A. and Willis, D.G., 1957. The elastic coefficients of the theory of consolidation. J. Appl. Mech., 24: 594–601. Bruno, M.S., Srinivasa, M.M., Danyluk, P.M., Hejl, K.A. and Anduisa, J.P., 1998. Fracture injection of solid oilfield wastes at the West Coyote field in California. In: EUROCK ’98, Rock Mechanics in Petroleum Engineering. Society of Petroleum Engineers. SPE Paper No. 47371. Byerlee, J.D., 1978. Friction in rocks. J. Pure Appl. Geophys., 73: 615–626. Corey, A.T., 1954. The interrelation between oil and gas relative permeabilities. Producers Monthly, Nov., Vol. 19. pp. 38–41. Davis, J.P. and Davis, D.K., 1999. Stress-dependent permeability: Characterization and modeling, Society of Petroleum Engineers. SPE Paper No. 56813. Evans, K.F., Cornet, F.H., Hashida, T., Hayashi, K., Ito, T., Matsuki, K., Wallroth, T., 1999. Stress and rock mechanics issues of relevance to HDR/WDR engineered geothermal systems: Review of developments during the past 15 years. Geothermics, 28: 455–474. Ferrill, D., Winterle, J., Wittmeyer, G., Sims, D., Colton, S. and Armstrong, A., 1999. Stressed rock strains groundwater at Yucca Mountain, Nevada, GSA Today. Geological Society of America, May. Geertsma, J. and Deklerk, F., 1969. A rapid method of predicting width and extent of hydraulically induced fractures. J. Pet. Tech., 21: 1571–1581. Goodman, R.E., 1970. Deformability of joints, in Symposium on Determination of the InSitu Modulus of Deformation of Rock. ASTM, S T P 477. pp. 174–196. Goodman, R.E., 1989. Introduction to Rock Mechanics, 2nd edn. Wiley, New York, U.S.
ELSE_TSANG_CH050.qxd
678
9/15/2005
1:06 PM
Page 678
Coupled Hydromechanical Effects in CO2 Injection
Gutierreze, M., Oino, L.E. and Nygård, R., 2000. Stress-dependent permeability of a demineralised fracture in shale. Marine Petrol. Geol., 17: 895–907. Hitchon, B., 1996. Aquifer Disposal of Carbon Dioxide. Geoscience Publishing Ltd., Sherwood Park, Alberta, Canada. Itasca Consulting Group, 1997. FLAC 3D. Fast Lagrangian Analysis of Continua in 3 Dimensions, Version 2.0, 5 vols. Itasca Consulting Group, Minneapolis, MN. Ito, T. and Zoback, M.D., 2000. Fracture permeability and in situ stress to 7 km depth in the KTB scientific drillhole. Geophys. Res. Lett., 27: 1045–1048. Jaeger, J.C. and Cook, N.G.W., 1979. Fundamentals of Rock Mechanics. Chapman & Hall, London. Leverett, M.C., 1941. Capillary behavior in porous media. Trans. AIME, 142: 341–358. Louis, C. and Maini, Y., 1970. Determination of in situ hydraulic parameters in jointed rock. Proc. 2nd International Congress on Rock Mechanics. Vol. I. pp. 235–245. Makurat, A., Barton, N. and Rad, N.S., 1990. Joint conductivity variation due to normal and shear deformation. In: N. Barton and O. Stephansson (Eds), Rock Joints, pp. 535–540. Martin, C.D., Lanyon, G.W., Blumling, P. and Mayor, J.C., 2003. The excavation disturbed zone around a test tunnel in the Opalinus Clay. In: Impact of Excavation Disturbed or Damaged Zone (EDZ) on the Performance of Radioactive Waste Geological Repositories, Proc. European Commission CLUSTER Conference, Luxembourg, November 3–5. Nagelhout, A.C.G. and Roest, J.P.A., 1997. Investigating fault slip in a model of underground gas storage facility. Int. J. Rock Mech. Min. Sci. Geomech. Abstr. 34: Paper No. 212. Neuzil, C.E. and Tracy, J.V., 1981. Flow through fractures. Water Resourc. Res., 17: 191–199. Nur, A. and Byerlee, J.D., 1971. An exact effective stress law for elastic deformation of rock with fluids. J. Geophys. Res., 76: 6414–6419. Perkins, T.K. and Kern, L.R., 1961. Widths of hydraulic fractures. J. Pet. Tech., Sept., Vol.13 937–949. Poston, S.W. and Berg, R.R., 1997. Overpressured gas reservoirs. Society of Petroleum Engineers, Richardson, TX, 138. Pruess, K., Oldenburg, C. and Moridis, G., 1999. TOUGH2 User’s Guide, Version 2.0. Report LBLN-43134, Lawrence Berkeley National Laboratory, Berkeley, CA. Pruess, K., Xu, T., Apps, J. and Garcia, J., 2001. Numerical modeling of aquifer disposal of CO2. Society of Petroleum Engineers, SPE Paper No. 66537. Ruqvist, J. and Stephansson, O., 2003. The role of hydromechanical coupling in fractured rock engineering. Hydrogeol. J., 11: 7–40. Rutqvist, J. and Tsang, C.-F., 2002. A study of caprock hydromechanical changes associated with CO2 injection into a brine aquifer. Environment. Geol., 42: 296–305. Rutqvist, J., Wu, Y.-S., Tsang, C.-F. and Bodvarsson, G., 2002. A modeling approach for analysis of coupled multiphase fluid flow, heat transfer, and deformation in fractured porous rock. Int. J. Rock Mech. Min. Sci., 39: 429–442. Scholz, C.H., 1990. The Mechanics of Earthquakes and Faulting. Cambridge University Press, New York. Snow, D.T., 1965. A parallel plate model of fractured permeable media. Ph.D. Thesis. University of California, Berkeley. Swan, G., 1983. Determination of stiffness and other properties from roughness measurements. Rock Mech. Rock Eng., 16: 19–38. Talbot, C.J. and Sirat, M., 2001. Stress control of hydraulic conductivity in fractured-saturated Swedish bedrock. Eng. Geol., 61: 145–153.
ELSE_TSANG_CH050.qxd
References
9/15/2005
1:06 PM
Page 679
679
Terzaghi, K., 1923. Die Berechnung der Durchlässigkeitziffer des Tonesaus dem Verlauf der hydrodynamischen Spannungserscheinungen, Akad. Der Wissenschaften in Wien, Sitzungsberichte. Mathematisch-naturwissenschafttliche Klasse. Part IIa 142(3/4), pp. 125–138. U.S. Department of Energy (DOE), 1999. In: D. Reichle et al., (Eds), Carbon Sequestration Research and Development. DOE Report DOE/SC/FE-1, Washington, DC. van Genuchten, M.T., 1980. A closed-form equation for predicting the hydraulic conductivity of unsaturated soils. Soil Sci. Soc. Am. J., 44: 892–898. Wang, H.F., 2000. Theory of Linear Poroelasticity. Princeton Univ. Press, Princeton. Walsh, J.B., 1981. Effects of pore pressure and confining pressure on fracture permeability. Int. J. Rock Mech. Min. Sci. Geomech. Abstr., 18: 429–435. Witherspoon, P.A., Amick, C.H., Gale, J.E. and Iwai, K., 1979. Observations of a potential size effect in experimental determination of the hydraulic properties of fractures. Water Resourc. Res., 15: 1142–1146. Witherspoon, P.A., Wang, J.S.Y., Iwai, K. and Gale, J.E., 1980. Validity of cubic law for fluid flow in a deformable rock fracture. Water Resourc. Res., 16: 1016–1024. Yamamoto, K. and Takahashi, K., 2004. Importance of geomechanics for the safety of CO2 geological sequestration. In: Y. Ohnishi and K. Aoki (Eds), Proc. ISRM International Symposium: Third Asian Rock Mechanics Symposium, Kyoto, Japan, Nov. 30–Dec. 2. pp. 467–472.
This page intentionally left blank
680
ELSE_TSANG_CH051.qxd
8/23/2005
12:36 PM
Page 681
Chapter 51
SUBSURFACE PROPERTY RIGHTS: IMPLICATIONS FOR GEOLOGIC CO2 SEQUESTRATION* E.J. Wilson Hubert H. Humphrey Institute of Public Affairs, University of Minnesota, Minneapolis, MN, USA
51.1 INTRODUCTION Geologic sequestration of CO2 (GS) presents a unique set of challenges and opportunities for making deep cuts in atmospheric emissions of CO2. GS projects inject captured CO2 into deep (greater than ~1 km) geologic formations for the explicit purpose of avoiding atmospheric emissions of CO2. Because of the large volumes of CO2 to be injected annually (millions to billions of tonnes), the long storage time frames required for GS (hundreds to thousands of years), and the monitoring and verification needs for injected CO2, GS presents a novel set of demands on the current legal regime for subsurface property rights. If GS is to play a significant role in reducing atmospheric emissions of CO2, the quantities to be injected need to be substantial—a one GWe capacity coal-fired power plant generates approximately 30,000 tonnes of CO2 per day. Furthermore, during the plant’s 30-year lifetime, injected CO2 could increase formation pressures over an area of more than 100 km2, assuming a 100-m-thick injection zone.1 In 2001, coal-fired power plants generated approximately half of all U.S. net electricity generation, producing 1856 Gt of CO2.2 The clarification of property rights as they relate to GS is important from both regulatory and liability perspectives, because both can have a significant impact on the future cost, acceptability, and feasibility of GS projects. Sequestration of CO2 in the subsurface poses several legal questions: Could the right to store CO2 in the pore space be transferred to another party? Should rights to use subsurface pore space be truncated as for adjacent airspace and be granted to another interest, separating them from ownership of the surface property? Who owns the CO2 stored in the pore space? How can the sequestration of CO2 in the
* This is an overview of the potential legal issues surrounding geologic sequestration. I am not a trained lawyer and the goal of this article is to highlight potential legal considerations that could influence the eventual deployment of geologic sequestration. 1 Tsang, C.-F. et al., 2002. Scientific considerations related to regulation development for CO2 sequestration in brine formations. Environ. Geol. 42: 275–281 at 275; Pruess, K. et al., 2001. Numerical modeling of aquifer disposal of CO2. Society of Petroleum Engineers. SPE 66537, at 2. 2
1904 billion kilowatt-hours, or 51% in 2001, U.S. DOE, Electric Power Annual 2001. 2003. Energy Information Administration, U.S. Department of Energy, Washington, DC; U.S. EPA, Inventory of U.S Greenhouse Gas Emissions and Sinks: 1990–2001. 2003. U.S. Environmental Protection Agency, Washington DC at 29.
ELSE_TSANG_CH051.qxd
682
8/23/2005
12:36 PM
Page 682
Subsurface Property Rights: Implications for Geologic CO2 Sequestration
pore space be managed so as to assure minimal damage to other property rights (mineral resources, water resources) sharing the same space? Will GS be treated as “disposal” or “storage” activity? In this chapter, I briefly outline the historical context within which subsurface property rights have evolved and examine the gradual decomposition of property rights. Next, I outline relevant case law that has shaped subsurface property rights in the oil and gas industry and injecting industries (hazardous waste and natural gas storage case law). Finally, I explore the implications of case law for GS and future research concerns and directions.
51.2 HISTORY OF U.S. PROPERTY RIGHTS The property rights system in the United States (U.S.) is based on English common law, which in turn is based upon more ancient property rights regimes. The historic principle of land ownership, Cujus est solum, ejus est usque ad coelum et ad inferos (“To whomsoever the soil belongs, he owns also to the sky and to the depths. The owner of a piece of land owns everything above and below it to an infinite extent.”3) has been tempered by case law and the subsequent decomposition, or separating out, of property rights. The Fifth Amendment to the U.S. Constitution guarantees the right to keep private property (“. . . nor shall private property be taken for public use without just compensation” 4) and much case law has been devoted to defining private and common rights of property ownership. In United States v. Causby, the Supreme Court truncated the ancient doctrine of property ownership and declared that it “has no place in the modern world,” with respect to air traffic.5 Property ownership does not extend to the superadjacent airspace, which hasn’t been occupied by structures or otherwise. In the same case, however, the Court affirmed that if such flights were to make the property uninhabitable, it would be considered a taking that would be compensable under the Fifth Amendment. The rules governing property rights are largely set by the states and affect property access, management, and exclusion, with property rights being divided spatially, temporally, and according to use.6 The courts have determined that a taking has occurred if the government divides property along spatial or temporal dimensions; however, government division along the rights of use is less obvious. Takings depend both on the degree of financial importance and whether the “regulation deprives the owner of an ongoing use.”7 Claims of unjust takings appear in much of the case law presented here. Property and mineral rights are framed in various ways. Under the English common-law system in the U.S., land can be divided into surface and mineral estates. Mineral estates can be conveyed separately from the surface estate, and in the majority of states, such conveyance will include oil, gas and other petroleum products.8 Further, differentiations are made
3
Black, H.C., 1990. Black’s Law Dictionary, 6th edn. West Group, at 378.
4
The Constitution of the United States, The Constitution of the United States, in Amendment 5.
5
United States v. Causby, 328 U.S. 256; S. Ct. 1062; 90 L. Ed. 1206; (1946 U.S.).
6
Stake, J.E., 2000. Decomposition of Property Rights, 1300. In: B. Bouckaert and G. De Geest (Eds), Encyclopedia of Law and Economics. Edward Elgar, Cheltenham, at 33–34, Thomas, W.A., 1979. Ownership of Subterranean Space. Underground Space. 3(4): 155–163 at 160. 7
Ibid. at 34.
8
Hemingway, R.W., 1991. The Law of Oil and Gas, 3rd edn. St. Paul, MN, West Publishing at 1.
ELSE_TSANG_CH051.qxd
8/23/2005
12:36 PM
Page 683
51.3 The Negative Rule of Capture and Secondary Recovery
683
between resource stocks and resource flows.9 Stocks can be easily controlled by the possessor (fields, hard rock minerals, buildings), whereas flows are not controllable by the possessor and must be captured to be possessed (fish, groundwater, water in streams, wild game). At the turn of the century, several state courts viewed capture of oil and gas as similar to the capture of wild animals, ferae naturae. Oil and gas were considered “fugitive,” and like wild game, the landowner did not possess the gas or oil until it was captured. Modified versions of this “non-ownership theory” of oil and gas resources are still practiced in roughly half of the states, with states recognizing that the owner possesses the exclusive rights of mineral exploration and exploitation, and is able to protect the reservoir from operations that might be harmful.10 The remaining states adhere to the “ownership in place” theory, where the mineral interests are severed from the surface estate and the mineral rights holder owns the oil and gas beneath his land, much as it is with solid minerals.11 Owners have a “possessory estate” in the oil and gas in place, giving them the right to explore and produce oil and gas (and exclude others) subject to regulatory and legal restrictions, which may limit operations and protect adjacent mineral rights owners. Both of these broad frameworks are subject to the rule of capture, where the legitimate producing party “gains title to all oil and gas produced, regardless of drainage.”12 Statutes governing the conservation of oil and gas and the prevention of waste have been adopted in all fossil fuel producing states.13
51.3 THE NEGATIVE RULE OF CAPTURE AND SECONDARY RECOVERY As technology evolves, so does the law, and case law surrounding secondary recovery operations has served to further refine property rights and liabilities of oil and gas holdings. Secondary recovery operations have revitalized oil production in the U.S. By injecting fluid into the producing reservoir, secondary recovery operations repressurize the reservoir and increase oil and gas recovery. These operations have raised questions surrounding the operator’s liability vis-à-vis reducing the amount of recoverable oil and gas from an adjacent mineral rights holder.14 An operator has the right to a fair share of the oil and gas in place and a duty to protect the common source of supply. As water injected for secondary recovery migrates through the subsurface, it can affect a neighboring rights-holder’s source of oil or gas supply and affect his or her ability to recover the resource. The legal ramifications of secondary recovery projects are thus significant. One of the first cases to deal with the implications of secondary recovery operations was Railroad Commission of Texas v. Manziel.15 Manziel was the mineral rights owner in a property adjacent to a secondary recovery project approved by the Railroad Commission of
9 Bouckaert, B., 2000. Original assignment of private property. In: B. Bouckaert and G. De Geest (Eds), Encyclopedia of Law and Economics, Edward Elgar, Cheltenham at 3. 10
Hemingway, R.W., supra note 6 at 27; Anderson, O. Kuntz Chair in Oil, Gas and Natural Resources, University of Oklahoma, personal communication 11/17/2003. 11
Ibid. at 27.
12
Anderson, O. supra note 8.
13
Schepens, A.B., 1999. Prospecting for oil at the courthouse: Recovery drainage caused by secondary recovery operations. Alabama Law Rev., 50(2): 603–622 at 606. 14
Ibid. at 608.
15
Railroad Commission of Texas v. Manziel, 361 S.W.2d 560; (1962 Tex.).
ELSE_TSANG_CH051.qxd
8/23/2005
684
12:36 PM
Page 684
Subsurface Property Rights: Implications for Geologic CO2 Sequestration
Texas. Because injected fluids may migrate across property lines, Manziel took the Commission to court for trespass of his subsurface property. In the case, the Texas Supreme Court noted that secondary recovery increased total oil and gas recovery, and decided that technical rules of trespass could not defeat a valid secondary recovery project.16 The resulting rule of nonliability has come to be known as the “negative rule of capture” and essentially states that less valuable substances can migrate through the subsurface and replace more valuable substances without incurred liability.17 Most secondary recovery activities will only take place in a field that has been unitized. With “field unitization” (or the combination of oil or gas field leases for resource development, creating a field-wide operation18), liability is removed as a driving concern, because production and profits are shared by all unit members and the field is managed in order to optimize resource recovery. However, many oil and gas fields are not unitized, and Williams and Meyers note that liability has been imposed upon the operator in several subsequent cases in Oklahoma (nuisance), Indiana (nuisance), Arkansas (trespass), California (trespass and compensatory damages), Nebraska (compensatory damages), and Kansas (compensatory damages).19 They also note that the courts have been far from consistent in their interpretation and awarding of damages.20 However, the power of state regulatory boards to grant permits for secondary recovery operations, or in some states, forced unitization, has consistently been upheld.21 This discretionary power by the state oil and gas boards is seen as necessary to ensure the greatest recovery and least waste of a valuable resource. In a recent Alabama case, Phillips Petroleum Co. v. Stryker,22 damages of $26.9 million were awarded to Stryker by a jury in the lower court on claims of trespass, negligence, fraud, nuisance, and punitive damages for draining the plaintiff’s oil and gas reserves. After being upheld by the circuit court, Phillips appealed to the Alabama Supreme Court, which reversed the judgment of the circuit court, finding that the plaintiffs, in accordance with Alabama Code, should have petitioned for inclusion in the unit within 30 days after the unitization order had been filed and that to hold Phillips liable was against the state’s policy on secondary recovery.23 The Court reversed the judgment of the lower courts and no damages were awarded, stating “an owner of interests outside a unit should not be entitled to damages from the operator of the unit if the circumstances are such that he can protect himself by engaging in an independent operation, or if he has been extended a fair opportunity to participate in the unit.”24 16 Schepens, A.B., supra note 10 at 613–614. Williams, H.R. and Meyers, C.J. (Eds), 2000. Oil and Gas Law. Vol. 1. Matthew Bender and Company, New York, at §204.5. Interestingly, a later Railroad Commission of Texas Oil and Gas Proposal for Decision and Orders Application of LASMO Energy Corporation for an exception to convert the following producing wells to injection wells: J.D. Hansborough Unit No. A-1, W.W. Wingfield Unit No. 2, and the J.R. Lusk Unit No. 1 in the Alabama Ferry (Glenrose “D” Field, Leon and Houston Counties, Texas, in Office of General Counsel. 1990, Railroad Commission of Texas: Austin, Texas. p. 9 speculates that the result of Railroad Commission of Texas v. Manziel may have been different if the secondary recovery project had “drowned” one of Manziel’s producing wells, thus increasing the total damage of the secondary recovery operation. 17
Williams, H.R. and C.J. Meyers, supra note 13 at §204.5, Schepens, A.B., supra note 10 at 617.
18
Hemingway, R.W., supra note 6 at 427.
19
Williams, H.R. and C.J. Meyers, supra note 13 at §204.5.
20
Ibid. at §204.5.
21
Schepens, A.B., supra note 10 at 614, 616.
22
Phillips Petroleum Co. v. Stryker, 723 So. 2d 585; (1998 Ala.).
23
Schepens, A.B., supra note 10 at 620, Phillips Petroleum Co. v. Stryker, 723 So. 2d 585; (1998 Ala.) at 590.
24
Phillips Petroleum Co. v. Stryker, 723 So. 2d 585; (1998 Ala.) at 591
ELSE_TSANG_CH051.qxd
8/23/2005
12:36 PM
Page 685
51.4 Injecting Industries
51.3.1
685
Implications for GS
While clear on the importance of secondary recovery for field maintenance and the right of state oil and gas boards to approve it, the courts have not been consistent in the resulting liability rulings. Liability for resulting mineral damage has been found in many cases, and fields that are not fully unitized could be especially vulnerable. If CO2 were injected for sequestration, the operator could, theoretically, be held liable if neighboring mineral rights that were not part of the same producing unit were harmed. Many secondary recovery operations only occur in fully unitized fields because of this potential liability. Injecting for GS within a fully unitized field could help to avoid potential liabilities associated with mineral owners resulting from the operation. It should be noted that all of the case law discussed above dealt specifically with the injection of water for secondary recovery, not other fluids, like CO2,25 which are used in subsequent recovery projects. Overall, considerations of other mineral and surface rights holdings should be thoroughly reviewed to avoid compromising resource production and potentially litigious situations resulting from GS.
51.4 INJECTING INDUSTRIES While the experience with secondary oil recovery provides a rich body of case law, the goals of secondary recovery are inherently different from those of GS. While many early GS projects will probably be linked to CO2-enhanced oil recovery (tertiary recovery projects that begin after secondary operations), the fundamental goal of enhanced oil recovery is not to store CO2, and a fair amount of injected CO2 is produced with the pumped oil. In contrast, the goal of GS is to sequester the injected CO2 underground for time periods that extend thousands of years, well beyond the 25–35-year time frame of tertiary oil recovery projects. Additionally, GS injection into saline aquifers without hydrocarbon resources may not be as influenced by the laws, property rights, statutes, and regulations that specifically govern oil and gas production and the defined legal interests in the subsurface could conceivably be smaller. As discussed below, a small body of case law has emerged surrounding hazardous waste injection and natural gas storage. Whereas hazardous waste is injected for disposal, natural gas is injected expressly for the purpose of storage; depending on how GS is interpreted, some of the rulings and underlying logic may be applicable to GS. However, the direct applicability of this guidance is limited, due both to the small amounts involved in hazardous waste injection and to the seasonal nature of natural gas storage. 51.4.1 Hazardous Waste Injection Case Law There are approximately 120 hazardous waste wells operating in 19 states, most injecting at depths of ~1400 m (4500 ft) and disposing of approximately 50% of all liquid hazardous wastes produced.26 The following is a discussion of several significant cases that deal directly with hazardous waste injection. In all of them, neighboring citizens sued the operator of a hazardous waste injection well; in none of the cases did the court uphold the plaintiffs’ claims. 25 While CO2 is gaseous at ambient temperatures, CO2 is injected as a supercritical fluid (Pcrit ⫽73.82 bar, Tcrit ⫽31.05°C). 26 Wilson, E.J., Johnson, T.L. and Keith, D.W., 2003, Regulating the ultimate sink: Managing the risks of geologic CO2 storage, Environ. Sci. Technol., 37: 3476–3483 at 3478.
ELSE_TSANG_CH051.qxd
8/23/2005
686
12:36 PM
Page 686
Subsurface Property Rights: Implications for Geologic CO2 Sequestration
51.4.1.1 Curtailing Subsurface Ownership Rights: Chance v. BP Chemical, Inc. In their class action lawsuit against BP Chemical, plaintiffs-appellants charged that injectate from BP’s injection well had laterally migrated below the plaintiff’s property, violating their rights as property owners.27 The plaintiffs sought recovery for trespass, nuisance, negligence, and fraudulent concealment, and claimed damage to their substrata for other uses. Since underground injection is less costly than other ways of handling hazardous waste, plaintiffs also claimed unjust enrichment on the basis that BP Chemical was able to dispose of toxins below their subsurface at a lower cost. The plaintiffs tried to amend their initial complaint by requesting a judgment “declaring that the appellants owned everything below the surface of their properties, including the geologic formations into which the injectate was allegedly going, and further declaring that they had the right to exclude the appellee from using their properties.”28 The motion to amend was denied by the trial court. At trial, the court also excluded plaintiffs’ evidence involving other problems at other deep-well injection sites, Ohio’s property disclosure law, and information on the Lima, Ohio, housing market, among other things.29 The trial court essentially limited the case to a claim of trespass.30 BP Chemical tried to cite Railroad Commission of Texas v. Manziel31 and the “rule of negative capture,” but the court found that this situation was not analogous because the case at hand was not related to oil and gas extraction or storage.32 For that same reason, the Court also rejected Columbia Gas Transmission Corporation v. Exclusive Natural Gas Storage Easement,33 which outlined various compensation schemes for surface property owners overlying natural gas storage projects. (This will be discussed in more detail in the next section.) Both sides presented expert hydrologist/geological engineer witnesses with hydrogeologic models that proved the respective side’s point and discredited their opponent’s model. Finally, the plaintiffs-appellants “sought a ruling that a trespass had occurred and that damages could be presumed from the act of trespassing.”34 The trial court denied this motion, and based upon the evidence presented, the jury found in favor of BP Chemical. The Court of Appeals affirmed for BP Chemical. While the court found that even though BP Chemical “operates the wells pursuant to the permits, that fact in and of itself does not insulate [BP Chemical] from liability,”35 the Court held that the plaintiffs had the burden of proving that a trespass had occurred.36 The Court in Chance v. BP Chemical, Inc. employed United States v. Causby,37 to address the question of property ownership. Reasoning that “. . . absolute ownership of air 27 Chance v. BP Chemicals, Inc., 77 Ohio St. 3d 17, 670 N.E. 2d 985; (1996 Ohio); 670 N.E. 2d 985, 1996 Ohio Lexis 1664. 28
Ibid. at 6.
29
Ibid. at 9. Included, presumably to demonstrate the depreciation of the fair market value caused by the underground injection activity. 30
Ibid. at 10.
31
Railroad Commission of Texas v. Manziel, 361 S.W.2d 560; (1962 Tex.).
32
Chance v. BP Chemicals, Inc, 77 Ohio St. 3d 17, 670 N.E. 2d 985; (1996 Ohio) at 17.
33
Columbia Gas Transm. Corp. v. An Exclusive Natural Gas Storage Easement, 67 Ohio St. 3d. 463; (1993 Ohio). 34
Ibid. at 11–12.
35
Ibid. at 15.
36
Ibid. at 16.
37
United States v. Causby, 328 U.S. 256; S. Ct. 1062; 90 L. Ed. 1206; (1946 U.S.).
ELSE_TSANG_CH051.qxd
8/23/2005
12:36 PM
Page 687
51.4 Injecting Industries
687
rights is a doctrine which ‘has no place in the modern world,’ to apply as well to ownership of subsurface rights,” the Court effectively truncated subsurface property ownership, though no specific depth was cited.38 In this interpretation, subsurface rights to exclude invasions are only valid as long as the invasions actually interfere with “reasonable and foreseeable” use of the subsurface.39 Thus, physical damage or interference with use must be shown to be associated with any alleged trespass. The plaintiffs’/appellants’ trespass case was found to be too “speculative.”40 The Court did speculate that there was possibly a valid trespass claim against BP Chemicals, Inc., for one class member who had supposedly abandoned plans to drill for natural gas and thus was “prevented from enjoying the reasonable and foreseeable use of its property” by the injection operations.41 Thus, ownership of mineral rights might have conferred a claim of trespass (and damages) that mere surface property ownership did not. Justice Pfeifer dissented in part, believing that Columbia Gas Transmission Corporation v. Exclusive Natural Gas Storage Easement42 (discussed later) was relevant, and the rental value of plaintiff’s properties should have been determined by the jury. 51.4.1.2 Liability and Full Subsurface Property Rights: Mongrue v. Monsanto Co. In Mongrue v. Monsanto Co.,43 the 5th Court of Appeals confirmed the decision of the Louisiana District Court that wastewater injected by Monsanto onto Monsanto property, but possibly migrating under Mongrue’s subsurface property, did not constitute a taking without just compensation. The plaintiffs/appellants had originally charged trespass as well, but later dropped this charge in the appeal, even though the judge confirmed that under Louisiana law,44 “the ownership of a tract of land carries with it the ownership of everything that is directly above or under it.”45 Both the District Court and the Appeals Court assumed, for the sake of argument, that Monsanto’s injected wastewater had migrated into the plaintiff’s subsurface property, and that the plaintiff did have ownership rights to the strata, although Monsanto filed an affidavit disputing the physical presence of its injectate under the plaintiff’s property. The Courts found that “appellants may recover under a state unlawful trespass claim against Monsanto regardless of the permit allowing for injection.”46 Because plaintiffs dismissed their trespass claim, the Court did not rule on this issue. The appeal for this case consisted only of the unconstitutional takings, a charge that was not valid under the Louisiana Constitution. Because the plaintiffs never specified that their case also included a federal claim in addition to the stated Louisiana claim, the federal claim was not included. Under Louisiana law, only private agents authorized by the government can be held liable for an unconstitutional taking from the expropriation of property. Since Monsanto had received no such delegation of authority, they could not be implicated in an unconstitutional taking. Whereas the Louisiana Commissioner of the Office of Conservation 38
Chance v. BP Chemicals, Inc, 77 Ohio St. 3d 17, 670 N.E. 2d 985; (1996 Ohio) at 20.
39
Ibid. at 21.
40
Ibid. at 28.
41
Ibid. at 29.
42
Columbia Gas Transm. Corp. v. An Exclusive Natural Gas Storage Easement, 67 Ohio St. 3d. 463; (1993 Ohio). Discussed in further detail in the next section. 43
Mongrue v. Monsanto Co., 249 F.3d 422; (2001 US. App.).
44
Unless otherwise limited by statues affecting unitization or pooling or other property limitations.
45
Ibid. at 432.
46
Ibid. at 432.
ELSE_TSANG_CH051.qxd
688
8/23/2005
12:36 PM
Page 688
Subsurface Property Rights: Implications for Geologic CO2 Sequestration
has the power to delegate permits for injection and issue unitization orders affecting property rights, there was no evidence in this case that the rights of the property owners could be “redefined or limited.”47 51.4.2
Implications for GS
In all cases, holding a valid permit to inject did not exempt the operator from liability; however, the burden of proving damages is on the plaintiff. In Chance v. BP Chemical, Inc., the subsurface depth of surface property holder’s interest was truncated, yet in Mongrue v. Monsanto, the court affirmed the subsurface ownership by the surface property holder. While it might have been possible to prove trespass in the latter case, it is unlikely that any damages would have been awarded, because no harm to existing or future interests could be proven. The implications for GS are significant, as surface property owners have not been found to have a legally defensible interest in the subsurface in one case, and have not been found to have one that resulted in damage awards in the other, when hazardous waste was injected and had possibly migrated below the surface of their property. These cases highlight the difficulty in proving trespass (especially when no monitoring wells are required) by the plaintiff and subsequently underline the lack of material interest that surface property holders have been determined to have in the subsurface. However, unlike hazardous waste injection, the quantities that will be injected for GS projects are large, and subsurface trespass could be more easily proven. The cases were fought out, in large part, by experts with dueling hydrogeologic models. Much of the testimony focused on the validity of the specific model parameters and underlying assumptions. Uncertainties in the subsurface geology and the ability of the specific various models to capture these were the focus of much of the argument. Because of the large quantities and large area influenced by GS injection, modeling the heterogeneous subsurface features over such a large area would be a challenge. 51.4.3 Natural Gas Storage Case Law Natural gas storage provides another relevant analog for understanding the evolution of subsurface property rights. Natural gas is stored underground in depleted oil and gas reservoirs, salt caverns, or suitable natural aquifers to provide for the increased market demand during the winter months. U.S. natural gas storage capacity is approximately 230 billion m3 (8.1 trillion ft3).48 McGrew provides an insightful discussion of property rights, trespass, and valuation methods for natural gas storage operations.49 McKinnon has written a summary of case law surrounding rights for natural gas storage, and while the article focuses mainly on the ownership of the stored natural gas and the subsequent exploitation of minerals below the storage site (and doesn’t directly address subsurface property rights), it is still a useful summary of issues presented in the case law.50
47
Ibid. at 432.
48
Heinrich, J.J., Herzog H.J. and Reiner, D.M. 2003. Environmental Assessment of Geologic Storage of CO2, MIT LFEE 2003-002 at 21. 49 McGrew, S.D., 2000. Selected issues in federal condemnations for underground natural gas storage rights: Valuation methods, inverse condemnation, and trespass. Case West. R. L. Rev., 2000. 51(Fall): 131–185. 50 McKinnon, R.J., 1998. The interplay between production and underground storage rights in Alberta. The Alberta L. Rev., 36(400): 18.
ELSE_TSANG_CH051.qxd
8/23/2005
12:36 PM
Page 689
51.4 Injecting Industries
689
51.4.3.1 Storage Space Property Rights A large number of cases have upheld that after the removal of underground minerals, oil, or gas, the surface owner retains the right to use the remaining space for storage. This is different from the English rule (also practiced in much of Canada) where the mineral owner owns the subsurface space, even after the minerals have been removed.51 A list of cases and relevant findings are presented in Table 51.1. While surface owners retain the right to the storage space, mineral rights holders have also been found to have a continued property, interest, and right “after the underground storage facility . . . has been created,” in an “ownership in place” state like Texas.52 Because administrative law and case law is different in each jurisdiction, legal opinions will vary from state to state. Other parties that may have an interest in the subsurface storage of natural gas include mineral lessees (who have leased the mineral rights for a defined term and might need to be compensated for the taking of the exploratory rights) and future interest owners (those having a vested right in the future estate).53 McKinnon highlights the case law interpretation of differences between storage rights and mineral rights.54 Several cases have found that they are fundamentally different, and storage rights do not preclude exploration or production of mineral rights. It is McKinnon’s view that it would not be possible for the party storing natural gas to stop production leases for zones beneath the storage formation from being issued.55 While the storing party would be able to observe the production to ensure that it did not harm her operation, she would not Table 51.1. A sample of cases citing surface property holder’s rights to remaining underground storage space (LexisNexis search, October 31, 2003) Case
Citation
Finding
Dept. of Transportation v. Goike Southern Natural Gas Co v. Sutton
220 Mich. App. 614; 560 N.W. 2d 365
“A surface owner possesses the right to the storage space created after the evacuation of underground minerals or gas,” at 617. “Surface ownership includes the right to use the reservoir underlying the surface for storage purposes,” at 671. “The mineral owner cannot be considered to have ownership of the subsurface strata containing the spaces where the minerals are found,” at 1046. Decided that the right and power to use a depleted reservoir for gas storage purposes is vested in the surface owner.
406 So.2d 669
United States v. 43.42 Acres of Land
520 F. Supp. 1042 (WD LA 1981)
Emeny v. United States
Ct. Cl. 1042, 412 F.2d 1319 (1969)
51 Stamm, A., 1957–1958. Legal problems in the underground storage of natural gas. Texas L. Rev., 36 p. 161–185 at 168; Lyndon, J.L., 1955–1956. The legal aspects of underground storage of natural gas—should legislation be considered before the problem arises? The Alberta L. Rev., 1: 543–548. 52
Mapco, Inc. v. Carter, 786 S.W.2d 386; (1989 Tex. App.).
53
Scott, R.R., 1966. Underground storage of natural gas: A study of legal problems. Oklahoma L. Rev., 47: 27–94 at 62. 54
McKinnon, R.J., supra note 47 at 8–10.
55
Ibid. at 10.
ELSE_TSANG_CH051.qxd
690
8/23/2005
12:36 PM
Page 690
Subsurface Property Rights: Implications for Geologic CO2 Sequestration
be able to stop it. However, natural gas storage operators usually contract with mineral owners so storage integrity is not compromised.56 51.4.3.2 Ownership of Injected Gas One of the first cases decided in 1934 was Hammonds v. Central Kentucky Natural Gas Co.57 It determined that gas injected for storage lost its title, and therefore trespass did not occur because the injected gas had no owner. This rule was widely criticized and is not currently followed in the majority of the United States.58 In subsequent cases59 the courts held that the injecting company held title to the re-injected natural gas and determined that third parties were forbidden to produce the stored gas, even if it had migrated to production wells that were on the third party’s land not within the storage area. In two of these cases, stored gas had migrated to other parts of the receiving reservoir (where the gas storage company had not acquired storage rights) or adjoining formations that were not known to transmit with the storage formation. The injector of gas must therefore obtain control of storage rights for the full area of the reservoir, dealing with all parties that have mineral or surface rights.60 51.4.3.3 Access to Storage Rights The Natural Gas Act of 1938 granted the power of eminent domain to private companies for the construction of interstate natural gas transportation. The act was later clarified to include the construction of underground storage facilities.61 However, most storage operations are not interstate, and state legislation grants the power to establish storage operations. Rights for storage are voluntarily contracted directly between the natural gas storage company and the affected property owners, paying, for example, an estimated $4–5 per year/acre in Ohio in 1993 to surface rights owners.62 No documentation on payments to mineral rights owners was found in the literature, but since the mineral rights owners often own a “cushion” of remaining gas in the formation, they must be compensated.63 When a property owner overlying a natural gas storage project does not voluntarily enter into a contract with the storage company, these remaining properties are termed “windows” in the storage field.64 If the owner of a “window” property threatened to drill into the storage formation or because of geologic uncertainties wasn’t included in the original project boundary, the storage company could file a condemnation action in court.65 The property owner could then counterclaim for trespass under state law.66 While condemnation of the
56 Anderson, O. supra note 8. In Kansas, an administrative work-around solution has essentially ensured stored gas remains the property of the injecting party, although the law has not been overturned. 57
Hammonds v. Central Kentucky Natural Gas Co., 255 KY 685; 75 S. W.2d 204 (1934 Ky.).
58
McKinnon, R.J., supra note 47 at 18; in Kansas, the rule is qualified a bit, Anderson, O. supra note 8.
59
White v. New York State Natural Gas Corporation White v. New York State Natural Gas Corporation, 190 F. Supp. 342; (1960 U.S. Dist.), Lone Star Gas Co. v. Murchison Lone Star Gas Co. v. Murchison, 353 S.W.2d 870; (1962 Tex. App.), and Oklahoma Natural Gas Company v. Mahan & Rowsey, Inc. Oklahoma Natural Gas Company v. Mahan & Rowsey, Inc., 786 F.2d 1004; (1986 U.S. App.). 60
McKinnon, R.J., supra note 47 at 8.
61
McGrew, S.D., supra note 46 at 138 and 140. Ibid. at 142 and 144. 63 Anderson, O. supra note 8. 64 McGrew, S.D., supra note 46 at 142. 65 Ibid. at 142. 66 Ibid. at 142. 62
ELSE_TSANG_CH051.qxd
8/23/2005
12:36 PM
51.4 Injecting Industries
Page 691
691
property owner can “obtain compensation as of the date of the taking,” a claim of trespass allows for compensation as well as punitive damages.67 Formulas for granting compensation for storage of natural gas under land were explored in Columbia Gas Transm. Corp. v. An Exclusive Natural Gas Storage Easement.68 In an effort to clarify Ohio State Law on compensation for natural gas storage projects, the Sixth District Court asked the Ohio State Supreme Court to clarify “a measure of just compensation for the appropriation of an underground gas storage easement.”69 The Court held that the methods outlined by U.S. District Court Justice Dowd, Jr. were appropriate for examining this issue. The commission that Justice Dowd appointed under the Federal Rules of Civil Procedure governing the condemnation of property70 outlined several proper ways to determine the value of a natural gas storage easement. The commission’s analysis suggested that it is necessary to consider “fair market value,” which could be based on one of the following methods: 1. Comparable sales of easements for natural gas in the particular formation. 2. Present value calculation (if sufficient natural gas exists for commercial recovery) of the “foreseeable net income flow from the property for its foreseeable life.”71 3. Capitalization of rental income for the right to store gas, calculated by multiplying the area to be rented with the value of comparable storage rights. 4. Calculation of the depreciation of the entire tract from taking of the easement used for storage, calculating the difference of the market value of the property before and after the taking. 5. Mineral lease value. 6. Viewpoint of value: The value calculated from the point of the view of the landowner, “The yardstick is what the landowner has lost, not what Columbia has gained.”72 Whereas Ohio law is now clear at both state and federal levels, the same cannot be said for other jurisdictions. The federal law in the Sixth Circuit is clear, but state law in Michigan, Kentucky, and Tennessee is not as explicit, and in other jurisdictions outside of the Sixth Circuit, the issue is largely undecided.73 51.4.4
Implications for GS
Unlike Chance v. BP Chemicals Inc., which involved hazardous waste injection, case law surrounding natural gas storage affirms that the surface estate owner also owns the subsurface storage pore space. Mineral owners, however, could also have a substantial interest, even after minerals and gas have been removed. In developing natural gas storage projects, both surface and mineral rights holders are traditionally included.74 This is in marked
67
Ibid. at 149 and 151.
68
Columbia Gas Transm. Corp. v. An Exclusive Natural Gas Storage Easement, 67 Ohio St. 3d. 463; (1993 Ohio). 69
Ibid. at 1.
70
Fed.R.Civ. P. 71A(h).
71
Columbia Gas Transm. Corp. v. An Exclusive Natural Gas Storage Easement, 67 Ohio St. 3d. 463; (1993 Ohio) at 2. 72
Ibid. at 3.
73
McGrew, S.D., supra note 46 at 146.
74
Anderson, O., supra note 8.
ELSE_TSANG_CH051.qxd
692
8/23/2005
12:36 PM
Page 692
Subsurface Property Rights: Implications for Geologic CO2 Sequestration
contrast with English and most of Canadian law, where the mineral rights owner is also considered the owner of the pore space remaining after mineral exploitation. Recent case law varies but has largely confirmed that ownership of injected natural gas rests with the storage operator. If the same principle applies for GS, injected CO2 would remain the property of the operator injecting, as would the associated liability that this implies. However, stored natural gas is a valuable commodity that is injected and then recovered on an annual basis, while injected CO2 will essentially be abandoned underground for hundreds to thousands of years, making it more akin to a disposal activity. Natural gas storage rights were also found to be secondary to those of mineral production. Possible mineral extraction below GS storage strata however, could conceivably lead to compromising the integrity of the storage reservoir. Additionally, as many of the cases found extensive lateral movement of stored natural gas, ensuring that GS projects characterize the reservoirs as thoroughly as possible and contact property interests will be a key requirement of any GS project. This will be especially challenging given the long time frames for GS projects and thus the greater importance of future interest holders. Finally, with ownership of the injectate comes liability. Trespass is tolerated for natural gas storage projects because the gas is a valuable commodity and the activity is considered necessary for the common good. Because no power of eminent domain exists for potential GS projects, it will be interesting to see how “window properties” overlying the storage reservoir are handled. Because GS is to be stored underground for long time periods, the implications of different types of compensation methods outlined in Columbia Gas Transm. Corp. v. An Exclusive Natural Gas Storage Easement should be further explored for GS.
51.5 CONCLUSIONS Case law builds upon itself; principles cited in cases are used as building blocks for other situations and arguments. The goal of this chapter is to highlight the historic legal interpretations and begin to understand the context within which general principles may affect GS projects. The preponderance of case law presented here suggests that surface and mineral owners will have a legitimate claim on subsurface strata used for GS projects. Many actors will have real interests in GS projects, including the injector, owner of injected material, surface property owner, mineral owner, mineral lessee, neighboring surface and mineral owners, and neighboring mineral lessees. Since property rights are governed by state laws and interpreted by state courts, any legal opinion on GS projects will be influenced by jurisdictional differences. Several issues that could directly affect GS projects are outlined here: subsurface property ownership, potential liability, ownership of injected CO2, and methods for evaluating potential compensation for utilization of the subsurface. In most of the case law explored here, the ownership of the subsurface pore space seems to rest with the owner of the surface estate. This is different from English and most Canadian law, where the pore space ownership remains with the mineral estate owner. Though it has been theoretically explored in several texts,75 the one case that truncated subsurface rights of surface owners was Chance v. BP Chemicals, Inc., which concerns hazardous waste injection
75
See Thomas, W.A., supra note 4, for an overview of the history of subsurface ownership.
ELSE_TSANG_CH051.qxd
8/23/2005
Acknowledgments
12:36 PM
Page 693
693
wells. Efforts at initiating GS projects in the U.S. need to be aware of surface estates as well as mineral estates, because both have a stake in any injection or disposal project. Much as is done with natural gas storage, explicit “storage rights” could be granted for GS projects. Both surface and mineral estates will need to be involved in this process. Additionally, compensation for subsurface use and royalties for mineral rights owners has regularly been paid for natural gas storage projects. Further study on the impacts of compensation schemes on GS costs should be undertaken. GS project time frame is an important consideration for leasing versus purchasing subsurface interests. Theoretically, competing GS projects could create a market for storage strata. The title of injected natural gas largely remains with the storage operator, and this has been affirmed in subsequent cases. This is true even if the gas migrates out of the defined boundaries of the storage area. However, natural gas is a valuable commodity that will be recovered and sold. For other fluids, this is not as certain, and no case law deals directly with injected CO2. It seems that CO2 injected under a GS scheme could also remain the property of the injecting party, but further clarification would help, especially if injected CO2 has no economic value. Because of the large size of many proposed GS projects, care needs to be taken to ensure that adjacent mineral rights owner’s holdings are not compromised. Because mineral rights have been found to trump both storage uses and surface holdings, any GS project needs to carefully examine mineral, water, and surface uses of the land that will be influenced by an injection project. Mineral extraction in adjacent strata could compromise the formation’s storage integrity. Likewise, pressure increases in the substrata could affect lateral movement of waters in the subsurface and affect groundwater quality. The advantages of some type of unitization framework for GS should be further explored. Such a framework could combine mineral interests and coordinate storage activities to ensure efficient and safe GS projects. Each of these issues should be examined in more detail to fully understand their ramifications on GS. Litigation is costly and time consuming, and substantial efforts should be made to understand the legal framework and implications of legal precedents on GS projects to avoid costly future litigation. The legal framework is likely to vary significantly between states and among specific project sites, as will the science. Note that several dimensions of import to GS have not been explored in this paper: scale related issues; the impact of injecting large quantities of CO2 on the areal extent of lateral migration, such as displacement and the effect on groundwater resources; and legal process considerations, such as the need to prove causality for damage and the implications and liability when GS is linked to an international carbon accounting structure. Finally, the structure of GS projects, whether they occur in a large centralized operation or through smaller, dispersed projects, will also influence overall liability. These are fruitful areas for future research.
ACKNOWLEDGMENTS A special thanks to Maria Savasta-Kennedy from the University of North Carolina Law School and Tony Beyer of the U.S. EPA for their legal editing, and to Tim Johnson and David Keith for their careful eyes and insightful comments. I also thank Malcolm Wilson, Wolfgang Heidug, and Stefan Bachu, who planted the seed and helped bring this work to fruition. A very special thanks also goes to Owen Anderson from the University of Oklahoma Law School for his insight into oil and gas law and to Myron Sereda for an understanding of differences between U.S. and Canadian systems.
This page intentionally left blank
694
695
AUTHOR INDEX
A Adams, J.J. 623 Ake, J. 361 Apps, J.A. 587 Arthur, J.D. 327 B Bachu, S. 623, 637 Bakshevskay, V.A. 151, 203 Balakhonov, B.G. 481, 491 Baydariko, E.A. 501 Behr, A. 167, 243 Block, L. 361 Bloetscher, F. 29, 65 Bonura, D.K. 3, 157, 451 Borgmeier, M. 403 Brkic, V. 569 Brown, C. 51 Bruno, M.S. 587 Bryson, W. 549 Bukvic, S. 569 Bundy, J. 361 Buschkuehle, B.E. 623 C Champollion, Y. 419 Clark, J.E. 3, 51, 157, 235, 451, 459 Coudray, J. 441 Cowart, J.B. 327 Cutler, T. 79, 557 D Dabous, A.A. 327 Danilov, V.V. 203, 487, 511 Darskaya, E.N. 491 Davis, K.E. 177 Dekking, M. 193 Dusseault, M.B. 539, 637 Dyer, J.A. 257
Gilch, W. 607 Gleixner, M.R. 419 Gorbatenko, B.P. 413 Gotovac, H. 569 H Haug, K.
623
I Istomin, A.D. 481, 491 J Join, J.-L. 441 K Kaimin, E.P. 271, 413, 511 Kamnev, E.N. 13 Kessler, A.G. 491 Khafizov, R.R. 13 Knape, B. 21 Kobelski, B.J. 39 Konosavsky, P.K. 271 Korotkevich, V.M. 13 Kuprienko, V.I. 473 Kurochkin, V.M. 13 Kurockin, V.M. 151 L Ladzin, A.S. 473 Larkin, R.G. 235, 313
F Faust, C.R. 51 Fischer, F.T. 451 Fritzler, B.C. 257
M MacFarlane, W.D. 419 Mahrer, K. 361 Makarova, O.V. 511 Martial, J.-S. 441 Matyukha, V.A. 481 McDonald, L.K. 177 McGowen, R.R. 157 Mercer, J.W. 51 Michael, K. 623 Miller, C. Jr., 459 Miller, C. 451 Mironov, V.V. 473 Mironova, A.V. 271 Moghaddam, O. 587 Mtchedlishvili, G. 243 Muniz, A. 29, 65
G Gerrish, H. 139,377 Ghose, S. 431
N Nedelkova, M. 521 Nieto, A. 377
E Egorov, G.F. 481, 491 Elfeki, A. 193
696
Author Index
Nopper, R.W. 459 Noskov, M.D. 481, 491
Stöwer, M. 403, 607 Sukhorukov, V.A. 151, 487, 491
O O’Connell, D. 361 Omrcen, B. 569 Omrcen, I. 569
T Thurston, D. 79 Tobon, M. 29, 65 Trivedi, P. 257 Tsang, C.-F. 139, 203, 649 Turkovskiy, A.D. 413
P Pankina, E.B. 271 Papadeas, P.W. 157 Park, E. 193 Pimenov, M.K. 13, 413 Platt, S. 45 Pozdniakov, S.P. 151, 203 Puder, M.G. 549
U Ulyushkin, A.M.
473, 501
V Van Voorhees, R.F. 3 Veil, J.A. 539, 549
R Radeva, G. 521 Rectenwald, D. 45 Redkin, E.A. 151 Rish, W.R. 93 Rumynin, V.G. 271 Rutqvist, J. 649 Ryabov, A.S. 487 Rybalchenko, A.I. 13, 203, 413, 487, 501
W Whitehurst, A.L. 39 Wilson, E.J. 681 Wong, H. 587 Wozniewicz, J. 419 Wu, Y.-S. 221
S Sanders, S.J. 257 Scrivner, N.C. 257 Selenska-Pobell, S. 521 Shestsakov, V.M. 151 Silin, D. 139 Sindalovskiy, L.N. 271 Skulski, L. 419 Smith, R.E. 39 Sparks, D.L. 257 Spycher, N.F. 313
Z Zagvozkin, A.L. 501 Zakharova, E.V. 271, 413, 491, 511 Zelic, M. 569 Zemke, J. 403, 607 Zhiganov, A.N. 481, 491 Ziegenbalg, G. 341 Zinin, A.I. 501 Zinina, G.A. 501 Zubkov, A.A. 13, 151, 203, 271, 481, 487, 491, 511 Zykov, A.I. 151
Y Young, J.T.
587
697
SUBJECT INDEX
A Aalen Formation, Germany 407 absolute permeability 223 absorbed components 482 absorbing capacity 511 accessible environment 95 acetic acid 512 acetonitrile 56 acid gas 623 – injection of 623 acid stimulation 367 acidic lead, aqueous waste stream containing 258 acidic waste 511 acrylamide 56, 341 acrylonitrile 51 Actinobacteria 529 Adsorption – barriers 352 – causing deactivation of inhibitors 344 – constants 285 – hysteresis in 304 – isotherms 274 – parameters 302 adsorption/desorption isotherm 275 aftershocks 379 Akron magnetic boundary 378 Alabama MADE Test Site 196 alarm failures 97 Alaska North Slope 8, 542 Alaska Oil and Gas Conservation Commission (AOGCC) 85, 540, 558 Alberta Basin 623, 638 Alberta Energy and Utilities Board (EUB) 423 Alberta Environment (AENV) 423 Alpine Field, Alaska 541 ambient reservoir pressure 142 American Water Works Association (AWWA) 69 amine 624 Anahuac Shale, Texas 54, 158 anhydrite 41 – precipitation 364 aniline 51 animal and food processing 25 anisotropy, hydraulic 212 annular injection 540, 554, 565 annulus 100 – fluid pressure loss 97 – pressure 86 Appalachian Foreland Basin 378 aqueous activity coefficient model 258 aqueous phase, immobile 482 aqueous waste, zinc-containing 258
aquifer 151 – saline 609, 630, 637, 649 – sand 273 Aquifer Storage and Recovery (ASR) 65, 327 aquitard 151 Arctic environment 79, 557 Area of Review (AOR) 45, 98 arsenic 34, 327 artificial penetrations (APs) 56 Asarco Copper Refinery 24 Asgard platform, North Sea 541 ash, residual 569 atmospheric pressure, elimination of effect on water levels 154 Avon Park Formation, Florida 68 B bacteria, halophilic 188 bacterial communities 521 Badami Field, Alaska 558 barite 82, 341 Beaumont, Texas 51 Bell Canyon Formation, New Mexico 435, 437 Benicanci Oil Field, Croatia 570 bentonite 82 bioaccumulation 521 biodegradation 589 biogas 589 biofilm 187 biomineralization 521 biosolids 587 Biot’s poroelastic parameter 388 biotransformations 521 Biscayne Aquifer, Florida 30 bitumen 419 – extraction 637 Boolean algebra 102 Boolean logic 97 borax pulse neutron tracer 87 Boulder Zone, Floridan Aquifer, Florida 31 Bovine Spongiform Encephalopathy (BSE) 569 Boynton Beach ASR, Florida 70 brackish water 68 brine – calcium-chloride 391 – disposal 403, 407 – process 640 brines, formation 413 Brooks-Corey relationship 247 Buckley-Leverett equation 246 Buntsandstein Formation, Germany 609 buoyancy effects 235
698 C calcite 341 capillary forces 222 capillary pressure 247, 249 capillary pressure functions 221 caprock 649 caprock breach, by CO2 649 carbon dioxide (CO2) 180, 590, 607, 623, 625, 637, 649 – capacity for injection 615 – corrosion 182 – emissions 607 – geological sequestration of 26, 607, 623, 637, 649, 681 – leakage 645 – leakage rate 670 Carter-Tracy method 58 casing annulus 563 casing corrosion 158 casings 98 cavern closure 643 cavern solution mining 638 cavity, formation of 459 cavity geometry 459 Cedar Keys Formation, Florida 68 Celanese Chemical Company 21 cement 17, 82, 87 cement microannulus failure 112 chemical conditioning 407 chemical fate 58, 157 chemical inhibitors 627 chemical speciation 318 Chepetsk Mechanical Plant (Glasov) 501 chlorine 180 chlorinity 181 cholera 29 Class I and Class V program 85 Class I wells 3 Class II wells 3, 85 Class III wells 3 Class IV wells 3 Class V wells 3 clay layers 16 clogging, of formations 425 CO2 See also carbon dioxide CO2-rich phase 451 coal beds 637 coal mines 555, 608 coal seams 649 coal-fired power plants 681 Cocoa Beach ASR, Florida 72 colloidal silica 341 Colorado River 361 Comprehensive Everglades Restoration Plan (CERP) 70, 328 compressibility 57, 170 Conasauga Sandstone, Ohio 389
Subject Index Concentration Reduction Factor (CRF) 59 conductivity, effective vertical 212 cone of influence 73 confining zones 98, 116 contaminants, immobilization of soluble 352 Cook Inlet, Alaska 542, 552, 560 Copper Ridge Dolomite, Kentucky 451, 459 correlation scales 212 corrosion 620 – electrochemical 178 – localized 178 – failure 177 – inhibitor 37 corrosive bacteria 190 Coupled Markov Chain Model 194 Courant number 60 critical stresses 667 Cryptosporidium 34 effluents 441, 137, 508 cyanide 56 Cytophaga/Flavobacterium/Bacterioides (CFB) 528 D Darcy’s law 221 decomposition 512 decomposition rate 56 degassing effects 427 degradation of injectate 58 Delaware Basin, New Mexico 431 Delta Alpine Field, Alaska 558 density changes in well plumes 235 Department of Energy (DOE) 431 depleted oil 630, 649 depleted gas reservoirs 177 desalination 26 desorption constants 277, 285 desorption parameters 273 detergent 203 digestion of biosolids 589 dimensionless variables 224 Dimitrovgrad, Russia 15, 507 dispersivity, longitudinal 213 disposal of meat and bone meal 569 disposal-well completion 406 dissolved organic carbon 441 distilleries, distillery slop 441 distribution coefficient, effective 281 DNA Extraction 522 Dolores River, Colorado 361 domestic wastewater reuse 33 drill cuttings 83, 539, 544, 549 drilling – fluids 82, 539, 549 – muds 540 – wastes 79, 539 drinking water sources 39
Subject Index dual-continuum methodology 223 dual porosity approximation 482 dual-porosity assumption 494 DuPont Beaumont Works, Texas 51 dysentery 29 E earth tides 154, 469 earthquake focal mechanism solutions 370 earthquakes 367, 377, 379 Eau Claire Formation, Ohio 141 elastic storage 210 electric power generation 25 electric power plants 25 Elk Point Group, Alberta 638 embrittlement of steel 184 endangerment of USDWs 39 Engineered or Designed Underground Repository (EUR site) 271, 288 enhanced recovery 47 environmental remediation 25 episodic fluid expulsion 657 epoxides 341 event tree 101 excavation disturbed zones 658 expert information 97 exposure studies 95 extended zone of discharge 75 F Failure – analysis 460 – criteria 467 – modes 101 Failure Modes and Effects Analysis (FMEA) 101 fall-off curves 145 fate and transport modeling 95 fault slip 672 fault trees 102 Fe oxides 327 fecal coliform counts 590 ferrihydrite 258 ferrihydrite, 2-line 260 finite-difference grid 58 Finland Gulf 273 Florida Department of Environmental Protection (FDEP) 74, 328 Florida Geological Survey (FGS) 328 Floridan Aquifer System (FAS) 30, 327 flow-rate data 139 flow velocity 184 fluid escape 565 Forchheimer equation 221 Formation – porosity 316 – transmissivity 140
699 formation-fluid density 238 Formation waters – analyses of 314 – chemical characterization of, prior to waste injection 313 Fort Lauderdale, Florida 70 Fracture – deformation 653 – development 243 – dilation 653 – fluid flow 655 – propagation 245 Frasch-process sulfur mining 21 Freundlich 274 Freundlich adsorption 277 Frio Formation, Texas 26, 158 G galvanic corrosion 178 Gamma logging 477, 496 gas hydrate formation 625 gas reservoirs 243, 630, 649 gas yield 489 Gauss–Newton method 169 geostatistical interpolation 285 geostatistical simulations 193 glycol 627 gradient method 169 Grand Rapids Formation, Alberta 421 gravitational constant vector 222 Great Lakes Region 93 greenhouse gas 588 ground subsidence 643 Ground Water Protection Council 101 grout curtains 341 grouts 341 Gulf Coast 22, 26 Gulf Coast Basin 637 Gulf of Mexico 541 Gulf States 93 Gulf Stream 32 gypsum 341 H haloacetic acids (HAAs) 74 Hawthorn Group, Florida 30 Hazardous and Solid Waste Amendments (HSWA) hazardous disposal well plumes 235 hazardous waste 24 – injection restrictions 85 health impacts 34 Health-Based Limits (HBLs) 53 heavy oil operations 637 heterogeneity, stochastic 271 heterogeneity, 3-D model for 204 history matching 170, 410
700 Hoek–Brown criterion 385 horizontal stresses 466 horizontal well 419 Horner plot 139 hydraulic conductivity 210 – effective 212 – horizontal 212 hydraulic fracture injections 365 hydraulic fracturing 16, 243, 433, 663 hydraulic properties 139 hydrocarbons 422 hydrochloric acid 459 hydrodynamic dispersion 57, 494 hydrogen-induced cracking 184 hydrogen gas 184 hydrogen sulfide (H2S) 180, 183, 422, 623, 625 hydrogen sulphide corrosion 183 hydrogeologic models 686 hydrolysis of injectate 57 hydromechanical coupling 650 I immobilization 354 impermeable caprock 649 in situ immobilization 352 in situ monitoring 650 in situ uranium mines 25 incompatible waste 118 indicator variograms 193 induced crystallization 347 induced earthquakes 368 induced seismicity 361, 378 industrial and municipal waste 21 industrial solid waste 26 inhibitor deactivation 343 injectate density 235, 238 injected effluents 489 injection – formation compressibility 47 – interval 100 – into deep aquifers 403 – of pond water 25 – pressure 16, 361, 544 – rate 47 injection tube 100 – leak 112 – tube failure, major 112 injection well failures 94 injection zone 8, 98, 481–485, 488–490, 491, 492, 494, 496–499, 506, 512, 516–519 injection zone extraction 118 injection-induced seismicity 377 inorganic species 258 inorganic suspended solid 257 interference testing 456 ion-exchange 413
Subject Index J Joli Fou Formation, Alberta 421 K Kalinin Nuclear Power Plant 414 Ketzin anticline 616 Keuper Formation, Germany 610 Kjeldahl nitrogen 34 Knox Dolomite, Ohio 142 Koenigstein uranium mine, Germany 341 Kozeni–Carman equation 210 Kraak Salt Dome, Germany 404 Kraak storage site, Germany 412 Krasnoyarsk-26, Russia 13, 271, 297 Kuparuk Field, Alaska 541, 543, 557 Kyoto Protocol 607 L LaBiche Formation, Alberta 421 Lagarto Formation, Texas 158 Lagarto Shale, Texas 54 Lake Karachai, Russia 271, 301 Lake Okeechobee ASR, Florida 70 Land Disposal Program Flexibility Act of 1996 9 land ownership 682 land-ban provision 7 Latin Hypercube Sampling (LHS) 102 lawsuits 686 leaching, in situ 352 lead sorption 258 leakoff coefficient 243 lepidocrocite 274 Levenberg–Marquardt method 169 liquid industrial waste 413 liquid radioactive waste (LRW) 13, 473, 481, 487, 491, 501, 511 – acidic 491, 498 – alkaline 481 – disposal 473 – intermediate-level 301 – low-alkalinity 487 – neutral 487 liquid waste disposal 413 liquid-organic radioactive waste (LORW) 481 Lomonosovsky Aquifer, Russia 273 Long Lake Project 419 low-permeability rock 167 M macrodispersivity 213 mad cow disease 569 Markov’s chain 206 Markov’s processes 206 matrix permeability 224 Maximum Contaminant Level (MCL) 328, 334 Mayak Production Association 14, 105, 271, 511 McMurray Formation, Alberta 419, 421
Subject Index mechanical deformations 649 meat and bone meal (MBM) 569 mechanical failure 649 mechanical integrity 86 Mechanical Integrity Testing (MIT) 86, 100 membrane facilities 67 membrane water treatment plants 33 metastable systems 347 methane 441, 588, 590 methanogenesis 592 methanol 627 Miami-Dade ASR, Florida 70 Michigan Basin 637 Microbiologically Influenced Corrosion (MIC) 186 microearthquake detection 365 microfiltration 407 Middle Run Formation, Ohio 141 Mining and Chemical Combine 13 mixing zone between injected waste and native water 73 model calibration 210 model calibration, surface complexation 260 modeling – flow and transport 203 – inverse 167 Mohr circle 385 Mohr–Coulomb failure criterion 467 monitoring, brine disposal 407 monitoring wells 47, 151 Monsanto Chemical Company 21 Monte Carlo analysis 96, 97, 102, 121 Monte Carlo simulation 196 Mt. Simon Formation, Ohio 141, 389 muds 82, 539 municipal sewage sludge 587 N National Oceanic and Atmospheric Administration (NOAA) 32 National Pollutant Discharge Elimination System (NPDES) 552 natural gas 608, 623 Natural Gas Act of 1938 690 New Mexico Oil Conservation Division (NMOCD) 434 nitrate ions 203 nitric acid 512 N-nitrosodimethylamine 34 no-migration demonstration 7 no-migration petition 8, 95, 98, 235 noise logging 100 non-Darcy flow 221 non-Darcy flow coefficient, effective 223 nonhazardous waste 24 normal stress 653 North American Craton 378 North Slope 51, 557, 541
701 Northstar Field, Alaska 558 Northwestern Center of Nuclear Energy (NWCNE) 271 n-paraffin 483 nuclear power plants 413 nucleation 343 nuclei growth 343 O Oakville Formation, Texas 158, 313 observation wells 16, 416, 444, 476 Ocala Limestone, Florida 68 ocean outfalls 29 Operational Data Analysis (ODA) 139 offshore platforms 80 Ohio Environmental Protection Agency, Ohio (EPA) 390 Ohm’s law 178 oil and gas – depletion 433, 434 – production 432 – reservoirs 637 oil fields 608 oil sands 419 oil-bearing horizons 431 oilfield waste 79, 80 Oldsmar Formation, Florida 68 Operations Data Analysis 139 Orange Salt Dome, Texas 157 organic phase, immobile 482 organic species 258 overburden pressure 174 overpressured sediments 657 oxidation 521 oxygen – activation 100 – activation tracer 87 – corrosion 178 – scavengers 179 – solubility 180 P Packer – failure 100, 106 – leak 106 Paradox Valley, Colorado 361 Peace River ASR, Florida 327 Peace River Project 72 Peclet number 60 Pelican Formation, Alberta 421 permafrost 86, 557 permitting wells 35 petroleum industry 3 pH conditions 327 phase boundary (PB) 482 physical properties 258 pipelines 620
702 Piper diagrams 315 plugging 555 plugging, aquifer 445 Point Thomson Field, Alaska 558 Polyamerase Chain Reaction (PCR) amplification 522 polyurethanes 341 poroelastic stresses 665 porosity 170 Port Neches Dome, Texas 54 potash ore 431 poultry processing 25 Powell method 169 precipitation inhibitors 342 pressure 139 pressure buildup 45 primacy, state 85 primary recovery 433 Probabilistic Risk Assessment (PRA) 96 property rights 682 Proteobacteria 523 Prudhoe Bay, Alaska 557, 558 Punta Gorda ASR, Florida 327 Pyrite – arsenian 327 – euhedral diagenetic 336 – framboidal 336 Q quantitative analysis 157 R radial flow 47 radioactive decay 482, 498 radioactive isotopes 15, 16 – 241Am 271, 487, 300, 511 – 140Ba 478 – 134,137Cs 487, 491 – 137Cs 271, 293, 478, 483, 508 – 144Ce 483, 487, 491 – 95Nb 483, 487, 491 – 237Np 487, 511 – 239Pu 271, 487, 511 – 106Ru 483, 487, 491 – 90Sr 271, 300, 483, 478, 487, 491, 508, 511 – 90Sr, adsorption and desorption of 291 – 238U 511 – 95Zr 483, 487, 491 Radioactive Tracer (RAT) 36, 87, 100 radioactive waste 203, 271, 431 radiolysis 489, 512 radionuclides. See also radioactive isotopes radionuclides 498, 511 – absorption 498 – adsorption 273 – adsorption/desorption 271 – contaminated sites 521 – migration 511
Subject Index Rät Formation, Germany 407 reaction-path simulations 317 redox 327 reducing conditions 327 reduction 521 regulations, enforcement of 86 relative permeability 223, 249 reservoir 614 – leakage 649 – pressure 47 – properties 614 – simulator 410 Resource Conservation and Recovery Act (RCRA) 551, 558, 710 Restriction Fragment Length Polymorphism (RFLP) typing 522 reuse 33 reverse osmosis 71 risk issues 34 risk analysis 3 risk calculation 581 rock caverns 658 rock elastic properties 462 rock permeability 244 rock salt 167 Rocky Mountain Arsenal, Colorado 384 Rome Avenue ASR, Florida 327 rotavirus 34 Rotliegend Formation, Germany 609 Rüdersdorf storage site, Germany 412 Russian Federation 413 Russian Federation Research Institute of Atomic Reactors (RIAR) 473, 501 Russian Ministry of Public Health 18 Russian Platform 413 S Safe Drinking Water Act (SDWA) 3, 39, 85, 550, 558 Sakhalin 79 Salado Formation, New Mexico 437 salinity, effect on corrosion 180 salt – cavern disposal 555 – caverns 26, 167, 403, 539, 637 – salt creep 642 – diapirs 26 – dissolution 26 – domes 26 – mines 608 salt pillow of Rüdersdorf 404 salt-tolerant organisms 188 saltwater 413 – disposal 435 sampling, chemical-radiochemical 477 Sandia Waste-Isolation Flow and Transport Model (SWIFT) 51, 55
Subject Index sealing of fractures 657 sealing of porous or fractured rock 347 seawater 26 – composition of 315 secondary recovery 434, 683 sedimentary rocks 17 seismic – activity 379 – events 377 – fracturing 97 – monitoring 645, 650, 658 – observations 17 seismicity 16, 378, 416 seismometers 364 Sequence Analysis 523 Sequential Indicator Simulation (SIS) 193 Sesnon Formation, California 592 Seversk, Siberian Chemical Combine, Russia 13 Shear – displacement 653 – failures 370 – stress 653 shear-slip 658 Siberian Chemical Combine 13, 151, 203, 271, 487 simulation program 407 skin factor 140 skin value 427 slurry injection 539, 549 sodium chloride 422 solid materials 539 solidified wastes 26 solution cavity 459, 460 solution-mined salt caverns 177 solution mining 433, 435 Solution Mining Research Institute (SMRI) 167 sorption 58, 293 – data 260 – experiments 492 – properties of the rock 476 – ion exchange 482 sour gas 623 sour water 623 South Liberty Field, Texas 26 Southeast Florida Outfall Experiment 32 speciation of native and waste fluids 317 Spindletop Dome, Texas 54, 508 standard annulus pressure tests (SAPT) 87 State Scientific Center of Russian Federation 13 Steam Assisted Gravity Drainage (SAGD) 419 sterilization 588 Stiff diagrams 315 storage in salt caverns 403 storativity 140 stress, effective 174 stress, principal effective 667 stress analysis 460
703 stress state 167 subarctic 79 subfracture injection 554 subsurface ownership 688 subsurface property rights 681 subsurface radionuclide transport 271 sulfate 203 sulfate reduction 316 sulfate-reducing bacteria (SRB) 186 sulfides 441 sulfur deposition 629 supercritical state 614 supersaturated solutions 343 Surface-Complexation Model (SCM), equilibrium 257 Survey Waste Injection Program (SWIP) 56 Suwannee Limestone, Florida 68, 327 SWIFT code 51 T Tampa Limestone, Florida 68 tar sands 637 tensile failure 654 Texas Administrative Code 22 Texas Commission on Environmental Quality 21 Texas Department of Water Resources 21 Texas Injection Well Act 21 Textin smelter 25 thermodynamic variables 258 thermodynamic data 317 thermodynamic modeling 511 thermolysis 512 tight gas formations 243 Tomsk-7, Russia 13, 203, 271, 290, 521 total dissolved solids (TDS) 5 trace metals 257, 327 tracers 15, 34 transient permeability tests 167 transient-pressure well testing 139 transmissivity 427 transport properties 212 transuranic elements (TRU) 431, 487 tributylphosphate (TBP) 483 trihalomethanes 74 typhoid fever 29 U uranium mobilization 327 underground gas storage 658 Underground Injection Control (UIC) 39, 22, 557 – rules 3 Underground Injection Control, Alaska 79 Underground Sources of Drinking Water (USDW) 95, 551, 561 underground storage 66, 608, 649, 690 United States Environmental Protection Agency (EPA) 3, 39, 85, 431, 550, 558
704 unplugged well 97 unprocessed waste 203 Upper Oakville Sand, Texas 51 uranium 336 uranium mining wastes 521 V vertical channels 35 vertical effective conductivity 212 vertical stress 466 vertical transport 214 viscoelastic deformation 56 volatile fatty acids 593 Vorderrhön District, Germany 620 W waste 13, 203, 459 – attenuation factor 95 – effluents, alkaline 511 – high-level 13 – low-level 13, 203 – medium level 13 – management 18 – migration 34 – treatment 416 Waste Isolation Pilot Plant (WIPP) 431
Subject Index wastewater 413 wastewater discharges 257 water glass 341 water management 65 water quality monitoring 40 water softening 413 water treatment 83, 413 waterflooding 434 water–rock interaction 328 well abandonment 555 well bore, leakage along 642 well integrity testing 87 West Siberian Platform 487 Western Canada Sedimentary Basin 419 Western Siberian Artesian Basin 151, 297 Wilmington Oil Field 588 Winona 25 Y Yenisei River, Siberia 297 Z Zechstein Formation, Germany 609 Zeleznogorsk Mining and Chemical Combine, Russia 13 Zone of Endangering Influence (ZEI) 45