Barrier systems for environmental contaminant containment and treatment

The USEPA’s Alternative Cover Assessment Program (ACAP) is also moni- toring the percolation rate from seven caps employing composite barrier layers consisting of a geomembrane underlain[r]

CONTAMINANT CONTAINMENT

and TREATMENT

A CRC title, part of the Taylor & Francis imprint, a member of the Taylor & Francis Group, the academic division of T&F Informa plc

Boca Raton London New York

ENVIRONMENTAL CONTAMINANT

CONTAINMENT

and TREATMENT

Edited by

Published in 2006 by CRC Press

Taylor & Francis Group

6000 Broken Sound Parkway NW, Suite 300 Boca Raton, FL 33487-2742

© 2006 by Taylor & Francis Group, LLC CRC Press is an imprint of Taylor & Francis Group No claim to original U.S Government works

Printed in the United States of America on acid-free paper 10

International Standard Book Number-10: 0-8493-4040-3 (Hardcover) International Standard Book Number-13: 978-0-8493-4040-6 (Hardcover) Library of Congress Card Number 2005047215

This book contains information obtained from authentic and highly regarded sources Reprinted material is have been made to publish reliable data and information, but the author and the publisher cannot assume responsibility for the validity of all materials or for the consequences of their use

No part of this book may be reprinted, reproduced, transmitted, or utilized in any form by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying, microfilming, and recording, or in any information storage or retrieval system, without written permission from the publishers

Danvers, MA 01923, 978-750-8400 CCC is a not-for-profit organization that provides licenses and registration for a variety of users For organizations that have been granted a photocopy license by the CCC, a separate system of payment has been arranged

Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation without intent to infringe

Library of Congress Cataloging-in-Publication Data

Barrier systems for environmental contaminant containment and treatment / contributing editors, Calvin C Chien, Hilary I Inyang, Lorne G Everett ; prepared under the auspices of U.S Department of Energy, U.S Environmental Protection Agency, DuPont

p cm

Includes bibliographical references and index ISBN 0-8493-4040-3 (alk paper)

1 In situ remediation Sealing (Technology) I Chien, Calvin C II Inyang, Hilary I III Everett, Lorne G IV United States Dept of Energy V United States Environmental Protection Agency VI E.I du Pont de Nemours & Company

TD192.8.B375 2005

628.5 dc22 2005047215

Visit the Taylor & Francis Web site at and the CRC Press Web site at

Taylor & Francis Group is the Academic Division of Informa plc

(http://www.copyright.com/) or contact the Copyright Clearance Center, Inc (CCC) 222 Rosewood Drive, quoted with permission, and sources are indicated A wide variety of references are listed Reasonable efforts

http://www.taylorandfrancis.com http://www.crcpress.com

For permission to photocopy or use material electronically from this work, please access www.copyright.com

Contributing Editors

Calvin C Chien, Ph.D., P.E. DuPont Fellow

DuPont

Wilmington, Delaware Hilary I Inyang, Ph.D.

Duke Energy Distinguished Professor and Director, Global Institute for Energy and Environmental Systems University of North Carolina, Charlotte, North Carolina Lorne G Everett, Ph.D., D.Sc.

President

L Everett and Associates, LLC Santa Barbara, California Prepared under the auspices of

U.S Department of Energy

U.S Environmental Protection Agency DuPont

With contributions by renowned experts on waste containment and waste treat-ment science and technology

2005

Technical Review Board

David E Daniel, Ph.D., Overall Book Reviewer

University of Illinois Urbana-Champaign, Illinois Skip Chamberlain U.S Department of Energy Washington, DC

Calvin C Chien, Ph.D., P.E. DuPont

Wilmington, Delaware Annette M Gatchett

U.S Environmental Protection Agency Washington, DC

Hilary I Inyang, Ph.D. University of North Carolina Charlotte, North Carolina

Lorne G Everett, Ph.D., D.Sc. L Everett and Associates, LLC Santa Barbara, California Brent E Sleep, Ph.D. University of Toronto Toronto, Ontario, Canada Craig H Benson, Ph.D., P.E. University of Wisconsin Madison, Wisconsin Ernest L Majer, Ph.D. Lawrence Berkeley Laboratory Berkeley, California

David J Borns, Ph.D. Sandia National Laboratories Albuquerque, New Mexico

Special Contributors

Jada M Kanak, Special Technical Assistant DuPont

Wilmington, Delaware

Kathy O Adams, Contract Technical Writer DuPont

Wilmington, Delaware

Introduction

Significant advances in subsurface containment technology occurred in the 1990s, both with the improvement of the technology and the broader acceptance and applications as a measure for environmental remediation Since 1995, the U.S Department of Energy (USDOE), U.S Environmental Protection Agency (USEPA), and DuPont have collaborated on a series of organized efforts to advance this technology In that year, these collaborators sponsored an international expert workshop that led to the publication of the first major book on containment technology Two international conferences were held by the same three partners in 1997 and 2001, with individuals from all over the world attending

Although subsurface containment technologies are becoming increasingly acceptable and popular in the environmental remediation field, questions remained on the prediction and verification of long-term barrier performance and this subject began to gain interest from the public, government agencies, and the U.S Congress With funding provided by USDOE, an executive committee, con-sisting of Skip Chamberlain (Chairperson, USDOE), Calvin C Chien (DuPont), and Annette M Gatchett (USEPA), was formed in October 2001 to plan and organize an expert workshop Sixty invited international experts participated The meeting was held between June 30 and July 2, 2002 in Baltimore, Maryland, and consisted of five discussion panels — three on prediction and two on verification Each panel was led by a panel leader and a co-leader to address particular technical topics in a designated area A designated graduate fellow, a graduate student whose research was related to these topics, recorded detailed notes for the panel discussions The graduate fellow group was coordinated and supervised by Jada M Kanak (DuPont) Each panel leader, assisted by the co-leader, was responsible for writing a chapter for this book, using the information generated from the panel discussions and the detailed notes recorded by the graduate fellows The prediction chapters were reviewed and edited by Hilary I Inyang, and Lorne G Everett reviewed and edited the verification chapters Calvin Chien had the responsibility for planning, coordinating, and editing the book, ensuring con-sistency and completeness, and resolving differences in opinions Skip Chamberlain provided technical input and crucial support in working with experts from the national laboratories on critical issues during the preparation of the book David E Daniel (University of Illinois) conducted an initial review of the first draft and provided high-level comments, which were useful in performing subsequent revisions Dr Daniel also wrote the preface for the book, which provides an outstanding introduction of containment technology history and book structure Relevant new information that became available during the period of preparation

and editing was identified, evaluated, and added to the book to ensure that the information is as up-to-date as possible

Preface

The containment of buried waste, contaminated soil or groundwater, refers to in situ (in place) management of contaminants in the subsurface Containment is achieved with individual barriers or control technologies that, together, provide a system of engineered control Containment is potentially applicable to any circumstance in which contaminants exist in the subsurface (e.g., uncontrolled landfills or dumps, chemical spills or leaks, pond or lagoon contaminant seepage) and can provide a safe and highly cost-effective mechanism for environmental control Containment is accomplished using physical, hydraulic, or chemical barriers that prevent or control the outward migration of contaminants

Containment has come full circle as an acceptable environmental control technology over the past 30 years Prior to the 1980s, containment was virtually the only technology available for managing subsurface contamination Although some wastes were exhumed and treated, more often than not, if the pollution problem was recognized at all, the problem was managed via containment During the 1980s, new environmental regulations emphasized treatment rather than con-tainment Research and development during this time dramatically expanded the portfolio of options available for treating or destroying contaminants at polluted sites Technologies such as vapor extraction, oxidation, bioremediation, surfactant flushing, and heat-induced treatment became viable, though often expensive, treatment alternatives

In the 1990s, a dose of reality swung the pendulum back toward containment It became apparent that it was not technically feasible to return contaminated sites to pristine condition Further, as a nation, the United States came to realize that it could not afford, nor did it need, the most sophisticated treatment technol-ogy available to manage pollution problems at every site effectively and safely In addition, further research clearly showed that the subsurface has advantages in addressing contamination problems — natural processes such as adsorption and biodegradation can serve to contain or degrade contaminants For certain materials such as radioactive wastes, it became apparent that the exposure risks associated with exhuming contaminants might be far greater than risks associated with managing the wastes in situ with containment Thus, for many reasons, interest in containment was revived in the 1990s Today, containment thrives as a viable environmental management technology, and is often the preferred choice for protecting human health and the environment

But a price was paid for putting containment “on hold” during the 1980s, when emphasis was placed on developing sophisticated treatment technologies: little research and development on containment technologies was achieved during

this time As interest shifted back toward containment in the 1990s, the industry found itself relying largely on pre-1980s technology Fortunately, in the past 10 years, important advances have occurred in several areas of containment, most notably in the area of permeable reactive barriers, which transform containment barriers into a passive treatment installation

In the early 1990s, the need to define the state of the art for containment was understood by three visionary organizations: DuPont, the U.S Environmental Protection Agency, and the U.S Department of Energy The DuPont Corporate Remediation Group (CRG) initiated the trio’s first collaborative effort in 1992 Experts from four nations experts were invited by DuPont to work with a team at the State University of New York at Buffalo to conduct a comprehensive review of the containment technology, the technology gaps, and future direction The product of the work, a 1993 internal report, was published in 1995 by John Wiley & Sons, New York, titled Barrier Containment Technologies for Environmental Remediation Applications, and edited by Ralph R Rumer and Michael E Ryan The principal chapters of the book focused on vertical barriers (walls), bottom barriers (floors), and surface barriers (caps) The three organizations joined again and organized an expert workshop on containment technology in 1995, inviting 115 international experts The book, Assessment of Barrier Containment Technol-ogies: A Comprehensive Treatment for Environmental Remediation Applications, was edited by Ralph R Rumer and James K Mitchell and was published the next year

With the rapidly increasing use of barrier technology in remediation, the need for better understanding, prediction, and monitoring of the performance of bar-riers emerged The trio organized another expert workshop on the topic in 2002, which led to the development of this book The workshop planning committee invited many of the world’s most knowledgeable researchers and practitioners to discuss the current state of the art and debate the appropriate applications and directions for containment The participants then went home and collectively created this book from their knowledge and exchanges This book is essentially a diary of those discussions and assessments, recast into the form of an easily readable, comprehensive book that is rich with discussion and references to literature, as well as further detail on specific topics of interest The first two

discussions in the first four chapters address caps, vertical walls, and permeable reactive barriers

how contaminants can get into the subsurface This is an important chapter, because one cannot understand how to contain something unless one knows how the contaminants got into the subsurface in the first place, and how they might spread and threaten the environment without containment This chapter not only describes pathways, but also introduces the essential concept of risk No control technology is without risk Ultimately, a low risk of adverse environmental impact chapters address prediction issues, Chapters and address monitoring tech-niques, and Chapter addresses the largely undeveloped field of verification The

draws from concepts in reliability of structures, and couples barrier structural failure to functional failure Relevant quantitative frameworks are presented for use in assessing the long-term performance of containment systems

basis for predicting the transport of water and contaminants through barrier components This chapter focuses on modeling the inflow of moisture to the buried waste (e.g., caps), or modeling the release of contaminants through subsurface barriers Fluid transport rate prediction is essential to the design process, because predictions can be integrated into the overall containment system details on the current state of the art for performance prediction, but also clearly delineates the limitations in modeling specific situations

in barriers, defining the properties of barrier materials and exploring how mate-rials perform in the field The matemate-rials used for barriers include a myriad of natural and man-made materials, such as natural soil, stones and cobbles, imper-meable plastic lining materials, man-made filter fabrics, and chemical agents designed to sorb or degrade contaminants that might come in contact with the material Factors such as clogging, deterioration, or alteration of physical, chem-ical, or hydraulic properties are explored, not only to define what is known about these materials, but also to provide a learned and balance sense of what is not known

a thorough description of the application of geophysical methods to subsurface barriers Geophysical methods have been used widely to assist in identifying potential mineral resources deep within the subsurface, and in more recent years, in the shallow environment, to help with identifying contaminant plumes and other anomalies When applied to subsurface barriers, geophysical methods are challenged beyond their traditional role of identifying gross features that might warrant more detailed exploration (e.g., via a borehole), toward identifying more subtle features, such as a leak in a subsurface barrier The techniques described in this chapter include both near- and far-field devices, spanning equipment deployed in aircraft flying above a site to devices placed on the ground surface that probe the subsurface directly with electromagnetic or other sources of energy The first half of this chapter describes the technologies that are available, and the second half addresses their applications to various types of barriers

the most challenging aspect of waste containment technology, i.e., validation of field performance Traditionally, monitoring has consisted of sampling of ground-water or soil gas from wells Although sampling soil, ground-water, and air can provide information about the general performance of a system, it does not provide imme-diate, specific information about how a particular barrier component is meeting its design goals Further, there is little to motivate stakeholders to spend money should be maintained in a way that uses resources as wisely as possible Chapter

Chapter 2, “Modeling of Fluid Transport through Barriers,” addresses the

performance assessment scheme presented in Chapter Chapter provides

Chapter 3, “Material Stability and Applications,” addresses the materials used

Chapter 4, “Airborne and Surface Geophysical Method Verification,” provides

The subject of Chapter 5, “Subsurface Barrier Verification,” tackles perhaps

for performance verification, unless required for compliance with regulations This chapter provides a comprehensive review of sensors and examples of how sensors can be used to document system performance, addressing the basic questions: where, what, how, and what-if? Ultimately, the performance verifica-tion scheme should be linked to the performance predicverifica-tion process It is perhaps this linkage that is our most important end point, and one that requires more work, particularly in terms of assessing reliability and risk associated with the use of waste containment as a technique for managing waste in the subsurface The two well-known case studies in the United States that are presented in this chapter provide particular value to this need

That which is buried in the subsurface, out of sight and out of mind, is that which in some respects is the most challenging Nature has placed geologic materials in the subsurface in rather unpredictable and unknowable locations, with properties that are difficult to discern Individual barriers are constructed in more controlled and documented ways, but still with considerable uncertainty in actual characteristics Systems comprised of multiple barriers enjoy considerable redundancy and tend not to rely on any single component for success Scientists and engineers strive to understand, predict, design, and verify safe containment schemes, both in terms of individual barriers and more complex containment systems This book provides a comprehensive report on the science and technol-ogy of waste containment, with a balanced presentation of what is and is not known Subsurface containment will continue to be a widely used environmental control technology in the years ahead This book will provide a valuable reference, helping to chart the way to successfully managing many contaminated sites

Editors

Calvin C Chien is a DuPont Fellow, one of only 13 individuals serving in this capacity in DuPont He has been working in the area of groundwater investigation and remediation since 1975 Since 1991, he has been responsible for evaluating and developing transport modeling and containment technologies As such, Dr Chien has played a leading role in improving the understanding of contain-ment technology for use in environcontain-mental remediation He orchestrated the First International Expert Workshop (1995) and the publication (based on the work-shop) of the first comprehensive containment book: Assessment of Barrier Con-tainment Technologies: A Comprehensive Treatment for Environmental Remedi-ation ApplicRemedi-ations (1996) In 1997, he spearheaded another effort to advance the technology: the First International Containment Technology Conference Through these efforts, he has been recognized as a leading contributor to improving the science of containment technology as well as its acceptance at the regulatory level He has authored and co-authored many technical papers for peer-reviewed journals and books Currently, Dr Chien provides technical environmental sup-port and oversight for existing and new DuPont operations in the Asia-Pacific region His contributions in the region led the Chinese Ministry of Science and Technology to invite him to evaluate candidates for the 2005 State Natural Science Award of the People’s Republic of China This award is the most prestigious award for scientists and engineers in China

Hilary I Inyang is the Duke Energy Distinguished Professor of Environmental Engineering and Science, Professor of Earth Science (GIEES), and Director of the Global Institute for Energy and Environmental Systems at the University of North Carolina–Charlotte From 1997 to 2001, he was the Chair of the Environ-mental Engineering Committee of the U.S EnvironEnviron-mental Protection Agency Science Advisory Board, and also served on the Effluent Guidelines Committee of the National Council for Environmental Policy and Technology He has authored and co-authored more than 170 research articles, book chapters, federal design manuals, and the textbook Geoenvironmental Engineering: Principles and Applications published by Marcel Dekker (ISBN: 0-8247-0045-7) Dr Inyang is an associate editor and editorial board member of 17 refereed international jour-nals, and contributing editor of three books, including the United Nations Ency-clopedia of Life Support Systems (Environmental Monitoring Section) He has served on more than 85 international, national, and state science/engineering panels and committees Since 1995, he has co-chaired several international con-ferences on waste management and related topics, and given more than 100 invited

speeches and presentations on a variety of technical and policy issues at institu-tions and agencies globally Professor Inyang holds a Ph.D with a double major in Geotechnical Engineering and Materials, and a minor in Mineral Resources from Iowa State University, Ames; a M.S and B.S in Civil Engineering from North Dakota State University, Fargo; and a B.Sc (Honors) in Geology from the University of Calabar, Nigeria His research has been sponsored by several agen-cies and corporations Dr Inyang’s research accomplishments and contributions to geoenvironmental science and engineering have been rewarded with honors by various national and international agencies among which are Fellow of the Geological Society of London; 2001 Swiss Forum Fellow selection by the Amer-ican Association for the Advancement of Science; 1991 Chancellor’s Medal for Distinguished Public Service awarded by the University of Massachusetts Lowell; and the 1992/93 Eisenhower Fellowship of the World Affairs Council to com-memorate the international achievements of the late U.S President Dwight Eisen-hower In 1999, Prof Inyang was appointed to Concurrent Professorship of Nanjing University, China and subsequently selected as an Honorary Professor of the China University of Mining and Technology, Jiangsu, China He is the President of the International Society of Environmental Geotechnology (ISEG) and the Global Alliance for Disaster Reduction (GADR)

best seller His book Groundwater Monitoring was endorsed by the U.S Envi-ronmental Protection Agency as establishing “the state-of-the-art used by industry today,” and is recommended by the World Health Organization for all developing countries

Table of Contents

1.2.1

1.2.2 Types of Performance Prediction Approaches 11

1.2.2.1 Empirical Prediction Approaches 11

1.2.2.2 Semi-Empirical Prediction Approaches 12

1.2.2.3 Less Empirical (Theoretical) Modeling Approach 14

1.3 Relationship of Structural Failure to Functional Failure 15

1.3.1 Economic or Pseudo-Economic Criteria 18

1.3.2 Regulatory Criteria 19

1.3.3 Prescriptive Design Criteria 19

1.3.4 Risk Criteria 20

1.3.5 Demonstrating Compliance: The Safety Case Concept 22

1.3.6 Mixed Criteria 23

1.3.7 Qualitative and Indexing Analyses 23

1.4 Quantification of Long-Term Damage Scenarios, Events, and Mechanisms 24

1.4.1 Categories of Degradation Mechanisms 24

1.4.1.1 Slow Physico-Chemical and Biological Processes 24

1.4.1.2 Intrusive Events 29

1.4.1.3 Transient Events 30

1.4.1.4 Cyclical Stressing Mechanisms 32

1.4.2 Quantitative Linkage of Contaminant Release Source Terms to Risk Assessment and Compliance Limits 37

1.4.3 Frameworks for Assessment of Event Consequences and Connectivities Among Causes of Failure 42

1.4.3.1 Fault Trees 42

1.4.3.2 Event Trees 42

1.4.4 Estimation of Long-Term Failure Probabilities 42

1.4.4.1 System Failure Probability 43

1.4.4.2 Component Failure Probability 44

1.4.4.3 Random Resistance 47

1.4.4.4 Simplifications of Theory 48

1.4.4.5 The Multi-Dimensional Case 51

1.4.5 Component and System Failure in Containing Contaminants 53

1.4.6 Relating Probable Contaminant Concentrations to Risks 54

4040_C000.fm Page xxi Friday, September 23, 2005 4:37 PM Chapter 1 Damage and System Performance Prediction

Hilary I Inyang and Steven J Piet 1.1 Overview

Concepts and Analytical Framework

1.5 Use of Barrier Damage and Performance Models for Temporal

Scaling of Monitoring and Maintenance Needs 59

1.5.1 Updating 59

1.5.2 Effect of Updating on System Management 60

1.6 Life-Cycle Decision Approach and Management 61

References 62

Brent E Sleep, Charles D Shackelford, and Jack C Parker 2.1 Overview 71

2.2 Caps 72

2.2.1 Features, Events, and Processes Affecting Performance of Caps 72

2.2.1.1 Hydrologic Cycle 72

2.2.1.2 Layers and Features 74

2.2.2 Current State of Practice for Modeling Performance of Caps 75

2.2.2.1 Water Balance Method 75

2.2.2.2 HELP 81

2.2.2.3 UNSAT-H 82

2.2.2.4 SoilCover 82

2.2.2.5 HYDRUS-2D 83

2.2.2.6 VADOSE/W 84

2.2.2.7 TOUGH2 84

2.2.2.8 FEHM 85

2.2.2.9 RAECOM 85

2.2.3 Modeling Limitations and Research Needs for Caps 86

2.2.3.1 Role of Modeling 86

2.2.3.2 Data Needs 86

2.2.3.3 Code Quality Assurance and Quality Control 87

2.2.3.4 Verification, Validation, and Calibration 88

2.2.4 Unresolved Modeling Challenges 89

2.2.4.1 Time-Varying Material Properties and Processes 89

2.2.4.2 Infiltration at Arid Sites 90

2.2.4.3 Role of Heterogeneities 90

2.3 PRBs 90

2.3.1 Features, Events, and Processes Affecting Performance of PRBs 91

2.3.1.1 Groundwater Hydraulics 91

2.3.1.2 Geochemical Processes 92

2.3.1.3 Reaction Kinetics 98

2.3.2 Impacts on Downgradient Biodegradation Processes 98

2.3.2.1 Enhancement of Geochemical Conditions Conducive to Anaerobic Biodegradation 98

2.3.2.2 Overall Contaminant Concentration Reduction 99

Chapter 2 Modeling of Fluid Transport through Barriers 71

2.3.2.3 Production of Hydrogen 99

2.3.2.4 Electron Donor Production 100

2.3.2.5 Direct Addition of Dissolved Organic Carbon 100

2.3.3 PRB System Dynamics 101

2.3.4 Geochemical Modeling 104

2.3.4.1 Speciation Modeling 105

2.3.4.2 Reaction Path Modeling 106

2.3.4.3 Reactive Transport Modeling 107

2.3.4.4 Inverse Modeling 108

2.3.5 Modeling Limitations and Research Needs of PRBs 109

2.4 Walls and Floors 110

2.4.1 Vertical Barriers 110

2.4.2 Horizontal Barriers 110

2.4.3 Current State of Practice for Modeling Performance of Walls and Floors 111

2.4.4 Contaminant Transport Processes 112

2.4.4.1 Aqueous-Phase Transport 112

2.4.4.2 Coupled Solute Transport 117

2.4.4.3 Modeling Water Flow through Barriers 119

2.4.4.4 Analytical Models 120

2.4.5 Modeling Limitations and Research Needs of Walls and Floors 123

2.4.5.1 Input Parameters and Measurement Accuracy 123

2.4.5.2 Time-Varying Properties and Processes 125

2.4.5.3 Influence of Coupled Solute Transport 125

2.4.5.4 Membrane Behavior in Clay Soils 126

2.5 Complicating Factors 128

2.5.1 Constant Seepage Velocity Assumption 128

2.5.2 Constant Volumetric Water Content Assumption 128

2.5.3 Anion Exclusion and Effective Porosity 129

2.5.4 Nonlinear Sorption 129

2.5.5 Rate-Dependent Sorption 130

2.5.6 Anion Exchange 130

2.5.7 Complexation 131

2.5.8 Organic Contaminant Biodegradation 131

2.5.9 Temperature Effects 132

References 132

Craig H Benson and Stephan F Dwyer 3.1 Overview 143

3.1.1 The Role of Barrier Material Mineralogy and Mix Composition on Performance 144

3.1.2 Approaches to Material Evaluation and Selection 147

3.1.3 Geosynthetics and their Durability in Barrier Systems 149

3.2 Material Performance Factors in Caps 153

3.2.1 Material Performance Factors in Composite Barriers 155

3.2.2 Material Performance Factors in Water Balance Designs 160

3.2.3 Coupling of Vegetation and Material Performance Factors 163

3.3 Material Performance Factors in PRBs 167

3.3.1 Approach to Selection of PRB Materials 168

3.3.2 Evaluation of Field Performance Using Pilot Testing 170

3.3.3 Effects of Hydraulic Considerations on Reactive Material Performance 172

3.3.4 Structural Stability Factors in Performance 178

3.3.5 Material Durability Factors 183

3.3.5.1 Effect of Mineral Precipitation on Porosity and Hydraulic Conductivity 185

3.3.5.2 Effect of Mineral Precipitation on Reactivity 186

3.3.6 Applications of Geochemical Models in Reaction Tracking 187

3.4 Material Performance Factors in Cutoff Walls 191

3.4.1 In Situ Hydraulic Conductivity 193

3.4.2 Design Configuration 196

3.4.3 Geosynthetics in Vertical Cutoff Walls 198

3.4.4 Permeant Interaction Effects 199

References 201

Ernest L Majer 4.1 Geophysical Method Application and Use 209

4.1.1 Characterization and Geophysics 210

4.1.2 Performance Monitoring and Geophysics 212

4.1.3 Geophysical Methods for Site Characterization and Monitoring of Subsurface Processes 214

4.1.3.1 Seismic 214

4.1.3.2 Electrical and Electromagnetic 214

4.1.3.3 Natural Field and Magnetic 215

4.1.3.4 Remote Sensing 216

4.2 Specific Methods 216

4.2.1 Seismic Methods 216

4.2.1.1 Conventional and Advanced Ray and Waveform Tomography 220

4.2.1.2 Guided/Channel Waves 221

4.2.1.3 Scattered and Reflected Energy 221

4.2.1.4 Cross-Well/VSP/Single Well Imaging 222

4.2.1.5 Summary 224

4.2.2 Electrical and Electromagnetic Methods 224

4.2.3 Natural Field and Magnetic Methods 227

4.2.4 Airborne Geophysical Methods 228

Chapter 4 Airborne and Surface Geophysical Method Verification 209

4.2.5 State-of-the-Practice Remote Sensing Methods 231

4.2.5.1 Aerial Photography 232

4.2.5.2 Multi-Spectral Scanners 232

4.2.5.3 Thermal Scanners 233

4.2.6 State-of-the-Art Remote Sensing Technologies 233

4.2.6.1 Hyperspectral Imaging Sensors 234

4.2.6.2 LIDAR Systems 235

4.2.6.3 Laser-Induced Fluorescence (LIF) 236

4.2.6.4 Radar Systems 237

4.2.6.5 Fused Sensor Systems/Data Streams 238

4.3 PRBs 239

4.3.1 Requirements, Site Characterization, Design Verification, and Monitoring 239

4.3.1.1 Site Characterization 240

4.3.1.2 PRB Construction Verification 241

4.3.1.3 Short-Term Monitoring 242

4.3.1.4 Long-Term Monitoring 242

4.3.2 Case Histories 243

4.3.2.1 Electrical Imaging of PRB Construction and Installation (Kansas City, Missouri) 243

4.3.2.2 Cross-Hole GPR Investigations (Massachusetts Military Reservation, Massachusetts) 245

4.4 Vertical Barriers 246

4.4.1 Requirements, Site Characterization, Design Verification, and Monitoring 249

4.4.1.1 Design 249

4.4.1.2 Installation/Verification 249

4.4.1.3 Short-Term Monitoring 254

4.4.1.4 Long-Term Monitoring 254

4.4.2 Case Studies 254

4.4.2.1 Cross-Hole GPR 255

4.4.2.2 Seismic 259

4.4.2.3 ERT 260

4.5 Caps and Covers 261

4.5.1 Requirements, Site Characterization, Design Verification, and Monitoring 262

4.5.2 Case Histories 263

4.5.2.1 EMI and GPR 263

4.5.2.2 Apparent Conductivity Maps 267

4.5.2.3 Electromagnetic Radar for Monitoring Moisture Content 269

4.5.2.4 Aerial Photography 272

4.5.2.5 Multi-Spectral Scanners 273

4.5.2.6 Thermal Scanners 273

4.6 Summary 274

4.6.1 Primary Needs for Advancement 275

4.6.1.1 Integration 275

4.6.1.2 Processing and Interpretation 275

4.6.1.3 Code Development 276

4.6.1.4 Instrumentation 276

4.6.2 Future Developments 276

References 278

David J Borns, Carol Eddy-Dilek, John D Koutsandreas, and Lorne G Everett 5.1 Overview 287

5.2 Goals 288

5.3 Verification Monitoring 289

5.3.1 Methods 292

5.3.1.1 Moisture Change Monitoring Methods 292

5.3.1.2 Moisture Sampling Methods 294

5.3.1.3 Vadose Zone Monitoring Considerations 295

5.4 Verification System Design 296

5.5 Moving from State of the Practice to State of the Art 297

5.5.1 System Approach 298

5.5.1.1 Links to Modeling and Prediction 298

5.5.1.2 Optimization 299

5.5.1.3 Decision and Uncertainty Analysis 299

5.5.2 Smart Structures 300

5.5.2.1 Long-Term, Post-Closure Radiation Monitoring Systems (LPRMS) 302

5.5.2.2 Environmental Systems Management, Analysis, and Reporting (E-SMART™) Network 304

5.5.2.3 Direct Push Technologies 305

5.5.2.4 Nanotechnology Sensors 307

5.5.3 Advanced Environmental Monitoring System (AEMS) 307

5.5.4 A New DOE Barrier Design Code 308

5.6 Drivers for Implementation of New Approaches 309

5.6.1 Costs 309

5.6.2 Enabling Desired End States 309

5.7 Covers 310

5.7.1 Moving from State of the Practice to State of the Art 310

5.7.1.1 Methods 310

5.7.1.2 Verification Measurement Systems 311

5.7.1.3 Barrier Cap Monitoring 311

5.7.2 Case History: Mixed Waste Landfill 312

5.7.2.1 Cover Infiltration Monitoring 313

5.7.2.2 Neutron Moisture Monitoring 313

Chapter 5 Subsurface Barrier Verification 287

5.7.2.3 Fiber Optics Distributed Temperature Moisture

Monitoring 314 5.7.2.4 Shallow Vadose Zone Moisture Monitoring 314 5.7.3 Case History: Fernald On-Site Disposal Facility 315 5.7.4 Verification Needs 318 5.7.4.1 Optimization and Trend Analysis 319 5.7.4.2 Sensors and Other Hardware 320 5.8 PRBS 321 5.8.1 Regulatory Framework 324 5.8.2 Moving from State of the Practice to State of the Art 325 5.8.2.1 Flow Characterization and Monitoring 325 5.8.2.2 Verification of Geochemical Gradients and Zones 327 5.8.3 Case History: Subsurface Monitoring 329 5.8.4 Verification Needs 329

5.8.4.1 Spatial and Temporal Flow Monitoring

Considerations 330 5.8.4.2 Geochemical and Hydrological Process Monitoring

Considerations 331 5.8.4.3 Acoustic Wave Devices 331 5.9 Walls and Floors 332 5.9.1 Moving from State of the Practice to State of the Art 337 5.9.1.1 Neutron Well Logging 337 5.9.1.2 Perfluorocarbon Tracer (PFT) Monitoring/

Verification 338 5.9.2 Case History: Colloidal Silica Demonstration 341 5.9.3 Case History: Barrier Monitoring at the Environmental

Restoration Disposal Facility (ERDF) 343 5.9.3.1 Study Conclusions 345 5.9.3.2 Study Recommendations 345 5.9.4 Verification Needs 346 5.9.4.1 Adequacy of the Containment Region 347 5.9.4.2 Long-Term Performance of the Containment 347 5.10 Conclusions 348 References 349

1

1 Damage and System

Performance Prediction

Prepared by* Hilary I Inyang

University of North Carolina at Charlotte, Charlotte, North Carolina

Steven J Piet

Idaho National Engineering and Environmental Laboratory, Idaho Falls, Idaho

1.1 OVERVIEW

Long-term hazardous waste containment using physical barriers such as caps demands the estimation of system reliability, system deterioration rate, and con-sequences of system failure The use of a systematic approach and a set of sequential analytical steps, such as those included in Figure1.1, enables opportunities for system improvement to be identified Barrier systems for buried waste or migrating contaminants are subjected to various physical, physico-chemical, and biological phenomena The synergistic action of these phenomena ultimately damages bar-rier systems and produces or enlarges flow channels through which pollutants can escape The observation that the degradation of constructed facilities increases with service life is not unique to containment systems: deterioration characterizes all constructed facilities, from roadways to pyramids Current uncertainties per-tain to the establishment of reasonably valid deterioration rates for various barrier designs, waste types, management systems, climatic and geohydrologic environ-ments, site stability, and barrier construction materials for time frames that range from hundreds to thousands of years

The diversity of waste types and desirable service lives for facilities under various regulatory programs are summarized in Tables1.1 and 1.2, respectively

* With contributions by James H Clarke, Vanderbilt University, Nashville, Tennessee; John B Gladden, Westinghouse Savannah River Company, Aiken, South Carolina; Horace K Moo-Young, Villanova University, Villanova, Pennsylvania; Priyantha W Jayawickrama, Texas Tech University, Lubbock, Texas; W Barnes Johnson, U.S Environmental Protection Agency, Washington, DC; Robert E Melchers, University of Newcastle, Callaghan, NSW, Australia; Mark L Mercer, U.S Environ-mental Protection Agency, Washington, DC; V Rajaram, Black and Veatch Corporation, Overland Park, Kansas; and, Paul R Wachsmuth, University of North Carolina at Charlotte, Charlotte, North Carolina

2 Barrier Systems for Environmental Contaminant Containment & Treatment

Estimation of the long-term deterioration pattern of barriers is necessary to improve the reliability of estimates of long-term contaminant release source terms for input into human health and ecological risk assessments, as well as facility monitoring and maintenance planning Monitoring of barrier performance pro-vides useful but inadequate data for performance predictions, because of limited field experience with barriers of various configurations in many environments, and because epochal events such as floods and earthquakes produce transient effects that cause deviations from performance patterns

The majority of quantitative methods that are currently used to estimate long-term barrier performance have time-invariant material characteristics and load/fluid application rates The use of these fate and transport models, most of

FIGURE 1.1 Flow chart for risk-based decision making (From Stewart, M.G and

Melchers, R.E., 1997. Probabilistic Risk Assessment of Engineering Systems, Chapman & Hall, London With permission.)

Define Context

social, individual, organizational, political, technological

Define System Hazard Scenario Analysis

• what can go wrong? • how can it happen? • what controls exist?

Estimate Probability

(of occurrence of consequences)

Estimate Consequences

(magnitude)

Define Risk Scenarios Sensitivity Analysis Risk Assessment

compare risks against criteria

Risk Treatment

• avoidance • reduction • transfer • acceptance

Damage and System Performance Prediction 3

TABLE 1.1

Types of Hazardous Materials

Type Typically Found in Nature? Importance of Chemical Form to Toxicity Does Hazard Decay Naturally?

Do We Know How to Destroy Hazard?

Radioactive isotopes

Yesa Can affect the

level of exposure to the hazard by altering the ingestion or inhalation uptake of isotopes

Natural decay is fixed for each isotope Negligible prospects for in situ destruction or treatment Ex situ treatment

may be practical to separate long-lived isotopes from short-lived isotopes Toxic organic compoundsb

No Affects ingestion

and inhalation uptake Decay generally slow (years, decades) and often dependent on specific chemical environment, e.g., trichloroethylene

In situ decay may be deliberately enhanced by microbes Determines toxicity level Ex situ destruction generally possible, but the associated risks and costs of transportation and destruction are high

Toxic metals Yes, although

sometimes not in the more hazardous chemical forms Can affect ingestion or inhalation uptake

Metals won’t decay, but the chemical form may naturally change into less toxic forms

Destruction is not practical

Generally affects toxicity

4 Barrier Systems for Environmental Contaminant Containment & Treatment

which are based on one-dimensional differential equations, for describing contam-inant advection-dispersion has simplified barrier performance analyses but does not address long-term barrier system performance adequately As the performance

TABLE 1.1 (continued) Types of Hazardous Materials

Type

Typically Found in Nature?

Importance of Chemical Form

to Toxicity

Does Hazard Decay Naturally?

Do We Know How to Destroy Hazard?

Toxic metals Ex situ destruction

generally possible, but with associated risks and costs during transportation and destruction a However, the specific radioactive isotopes are typically are not the specific isotopes found in nature

bThere are also some toxic compounds that are neither organic nor metals, e.g., asbestos. Source: INEEL (2000) Environmental Laboratory Report INEEL/EXT-2000-01094; Piet et al (2001) INEEL technical report INEEL/EXT-2001-01485

TABLE 1.2

Time frames for Waste Containment Performance under Various U.S Regulatory Programs

Time Frame Regulatory Program

10,000 years Nuclear Regulatory Commission and EPA regulations for high-level and transuranic waste (10 CFR 60, 10 CFR 63, 40 CFR 191, 40 CFR 197)

1000 years EPA regulations for near-surface uranium and thorium mill tailings (40 CFR192)

and DOE policy for new land burial (DOE M 435.1)

500 years NRC regulations for near-surface burial of low-level waste (10CFR61)

30 years Baseline EPA RCRA time period for near-surface burial chemical hazards

(40 CFR264); EPA can increase or decrease this value for each case

Indefinite Baseline EPA CERCLA time period for residual hazards (CERCLA requires a

Damage and System Performance Prediction 5 analyses time frames extend from one or two decades to hundreds of years, changes in barrier material characteristics; cyclic changes, waning, or growth of stressing events; and possible exhaustion of initially present parent contaminants and/or generation of daughter contaminants combine to decrease the reliability of contaminant release estimates

Long-term performance modeling of waste containment systems and individ-ual barriers within such systems require identifying possible damage mechanisms and assessing the system resistance in all possible ways in which the system might fail Various techniques have been developed in practice (in different industries) and, hence, with different names, including the following:

• Preliminary hazard analysis (PHA) (nuclear industry)

• Walk-down analysis consisting of on-site visual inspection, particularly of pipe work (nuclear industry)

• Failure modes and effects analysis (FMEA), which uses generic terms as prompts (various applications)

• Failure modes, effects, and criticality analysis (FMECA), which also assesses criticality of consequences

• Hazard and operability studies (HAZOP), which uses guide words as prompts (primarily chemical industry)

• Incident data banks, which contain data such as accident data and near-miss data

For the range of barrier applications available now and in the future, there is a need for improved capacity to predict containment barrier damage and system performance Damage and system performance models must be:

• Responsive to the needs of a diverse set of decision makers (i.e., designers of new barriers, managers of barriers in service, regu-lators, funding agencies, and the public)

• Integrative of the most important mechanisms of failure (i.e., both spatially uniform degradation and localized degradation; both contin-uously acting and discrete in time)

• Comprehensive with regard to the range of performance measures relevant to a given barrier design that solves a particular problem at a particular location

• Stochastic to allow evaluation of the sensitivities of parameter uncer-tainties compared to performance measures

• Probabilistic in consideration of failure scenarios and mechanisms that may or may not occur during the service life

• Validated by data to the extent practical

6 Barrier Systems for Environmental Contaminant Containment & Treatment • Informative regarding barrier degradation to guide barrier surveillance

and maintenance, justification for reduction of surveillance and main-tenance, barrier lifetime extension while in service, and future barrier designs

• Graded in its implementation according to the severity and longevity of the associated risks (barriers with lower severity or shorter duration hazards not need all of the above)

Attempts to provide satisfactory system performance demands that one or more criteria be available against which to measure system performance The setting or derivation of performance criteria is a problem with a fascinating and complex history, much of it based originally on issues associated with the nuclear industry This history includes some deep philosophical questions, including “Who is to bear what level of risk, who is to benefit from risk-taking, and who is to pay? Where should the line be drawn between risks that are to be managed by the state and those that are to be managed by individuals, groups, or corpo-rations? Who evaluates success or failure in risk management and how? Who decides what should be the desired trade-off between different risks?” (Hood et al., 1992) The decisions about these matters are influenced by judgments about the following (Stewart and Melchers, 1997):

• Anticipation of system failure and resilience against unexpected catas-trophe

• Assumptions used to compute a numerical estimate of system risks • Size of uncertainties in estimating system risks

• Organizational vulnerabilities to system failure • Cost of risk reduction

• Size and composition of groups involved in decision-making processes • Aggregation of individual preferences (i.e., distribution of benefits and

risks)

• Counter-risks (i.e., alternatives may have other societal risks)

Damage and System Performance Prediction 7

exposure generally depends on the five factors listed below and illustrated in Figure 1.2 Eventually, hazard either decays (with some half-life) or escapes

1 Hazardous half-life

2 Mobilization rate/year (e.g., leaching, diffusion in the absence of a barrier)

3 Time at which barrier begins to degrade Barrier degradation rate/year

5 Transport time of escaped materials between barrier and recipients Factors 2, 3, and control when and how fast the hazard escapes Factor controls how much time (with additional hazard decay) will elapse before the escaped hazard impacts human health and the environment With reference to the range of time horizons in various regulations, there is no systematic connection between the hazard timescales and regulatory timescales that are summarized in Table1.2 Different regulations were established at different times by different legislation in response to different issues Thus, the appropriate framework for predicting barrier system performance is not always clear: the time frames can differ greatly and the appropriate assumptions on how long to monitor and manage the barrier system can also differ

1.2 LONG-TERM PERFORMANCE ANALYSIS FRAMEWORK

It is necessary to formulate a long-term performance analysis framework that enables the consideration of factors that are significant for a given class of containment systems The failure states of the constructed system in terms of both structural failure and functional failure need to be defined Also, the performance assessment

FIGURE 1.2 Simplistic illustration of processes that influence exposure to individuals

Hazard halflife

Hazard inventory

Escaping inventory

Time barrier starts to degrade Rate barrier degrades per year

Rates inventory mobilizes per year

Delay time in transport

8 Barrier Systems for Environmental Contaminant Containment & Treatment framework should incorporate nodes to which pre-failure performance models can be linked

1.2.1 CONCEPTSAND ANALYTICAL FRAMEWORK

Several concepts and analytical frameworks have been proposed for use in assess-ing the long-term performance of containment systems The concepts pertain to the performance pattern of containment systems during service lives and post-closure time frames that can range from 30 years to thousands of years The focus of the analyses is the formulation and use of performance prediction models that are capable of determining contaminant release rates as a function of estimated, measured, or designed magnitudes of containment system design parameters, waste characteristics, stressing events and processes, and site/hydrological con-ditions The factors that need to be considered are numerous, as exemplified by the case of a near-surface barrier illustrated in Figure 1.3

Several attempts have been made to establish the expected general pattern of barrier performance over long service lives Figure1.4a shows the containment system performance model that is implicit to current practice The facility is assumed to provide a constant level of service, or to be structurally sound until external monitoring data indicate the release of contaminants at unacceptable

FIGURE 1.3 An illustration of the interaction among various processes and parameters

that influence the long-term performance of near-surface containment systems Plugging and

surface tension

Output: Contaminant flow to the vadose zone Surface ecology

(especially evapo-transpiration

barriers)

Interfacial ecology (especially

capillary barriers)

Waste zone

Hydrology (including micropores, capillaries)

Structure Natural boundary conditions

(weather, climate, biota)

Engineered boundary conditions (design, maintenance, repair)

Plants Dimensions,

materials configuration Temperature

Precipitation Plant/animal intrusion

Soil type and thickness

Erosion

Subsidence Compaction

Wind/water erosion

Biochemical changes? Ecological

Damage and System Performance Prediction 9

concentrations Figure 1.4b shows a more realistic performance pattern in which the performance degrades gradually during the immediate post-implementation period and then decays abruptly After abrupt decay, the performance decreases much more gradually in a period that is characterized by large uncertainties The reader should note that system damage vs time plots have configurations that are the reverse of those of system performance (or effectiveness) vs time plots Thus, Figure1.5 shows an increase in the risk of containment system failure with time It should be noted that although the system deterioration pattern may be represented by a smooth curve, the performance pattern of a particular component of the containment system could exhibit temporal fluctuations in response to transient stressing mechanisms, the passage of contaminant fronts, and mainte-nance activity In developing the conceptual framework for estimating the long-term performance pattern of containment systems, Inyang (1994) identified the various stages illustrated in Figure1.6 Curve shows the barrier degrading via continuous deterioration mechanisms The branching to Curve shows a barrier suffering from a discrete (in time) negative perturbation, such as a flood or an earthquake The branching to Curve reflects a barrier being upgraded or repaired In the illustration, following Curve 1, the containment system effective-ness decays from an initial level of Eto, to a minimum acceptable level of Etr at time, tr Etr corresponds to the functional performance level that is typically

FIGURE 1.4 Conceptual pattern of long-term performance of containment systems

(a) abrupt failure pattern implicit to current practice (b) gradual degradation pattern that is more realistic

Current Methodology

Pe

rf

o

rmanc

e

Realistic Performance

Pe

rf

o

rmanc

e

Uncertain how to manage barriers & resistance to new materials and designs

Design Life

Detect only after failure (leakage through barrier)

Time

Uncertain long-term performance

(a)

10 Barrier Systems for Environmental Contaminant Containment & Treatment

specified by regulators or other authorities If the facility is repaired at a time, tm, the effectiveness can abruptly increase to Etm so that an improved performance (described by Curve 3) results Essentially, repairs postpone the attainment of Etr

FIGURE 1.5 Conceptual degradation-time function of a containment system (Illustrated

by Melchers, R.E., 2001 Reliability Engineering and System Safety, 71(2), 201–208 With permission.)

FIGURE 1.6 A conceptual long-term deterioration pattern and maintenance scheme for

waste containment system (From Inyang, H.I., 1994 Proceedings of the First International Congress on Environmental Geotechnics, Calgary, Canada, pp 273–278 With permission.)

Ri

sk of f

ailur

e

Particular structure deterioration

Acceptable risk level

Expected deterioration

Time (age of structure)

Sy

st

em eff

ec

tivene

ss

, E (f

rac

tion)

Eto

Etm

E1

E2

Etg

Etr

to

Time, t (years) Curve

Curve

Curve

Damage and System Performance Prediction 11 by slowing down the deterioration of the repaired component(s) and, hence, the system The system can also degrade abruptly, as at tg, such that its effectiveness falls to Etgand system performance follows Curve to failure at t2 (much sooner than would result from the regular deterioration pattern)

1.2.2 TYPESOF PERFORMANCE PREDICTION APPROACHES

In order to serve practical purposes, performance patterns need to be quantified, requiring the development of rating systems and models Approaches to perfor-mance prediction can be categorized as empirical, semi-empirical, and less empirical (theoretical modeling)

1.2.2.1 EmpiricalPredictionApproaches

Empirical prediction approaches involve the extrapolation of current knowledge of system behavior and/or similar system behavior to long-term system behavior Such knowledge can also be acquired through accelerated testing in intensified environments Another example of an empirical approach is performance index-ing In most cases, indexing criteria not explicitly include time functions with performance factors Table 1.3 shows the ratings of single components and com-posite configurations of barriers (Piet et al., 2001) In general, the scores on

TABLE 1.3

Overall Benefit of Each Barrier Configuration of Cover/Liner Materials

Design

Alternate Description

Overall Benefit

Estimated Cost (dollars/ft2)

Benefit/Cost Ratio

Ranking in Group One-Barrier Layer

A CCL 36 0.70 51

B GM 64 0.70 91

C GCL 46 0.70 66

Two-Barrier Layer

D GM/CCL 58 1.40 41

E GM/GCL 66 1.40 47

Three-Barrier Layer

F GM/CCL/GM 71 2.10 34

G GM/GCL/GM 77 2.10 37

CCL, single compacted clay liner; GM, single geomembrane; GCL, single geosynthetic clay liner; GM/CCL, two-component composite; GM/GCL, two-component composite; GM/CCL/GM, three-component composite liner; GM/GCL/GM, three-component composite liner

12 Barrier Systems for Environmental Contaminant Containment & Treatment

overall benefit or utility of a particular design increase with the number of components

Inyang and Tomassoni (1992) indexed the long-term performance pattern of waste covers for use in regulatory impact analysis The scores are presented in Table 1.4 The reader should note that these scores are general indices and are not precise estimates of the performance of the components scored Other researchers exemplified by Hagemeister et al (1996) developed detailed perfor-mance indexing systems that incorporate ratings of barrier components, contam-inant transport pathway factors, and human exposure potential

1.2.2.2 Semi-EmpiricalPredictionApproaches

These approaches involve the use of semi-empirical models to estimate the damage time functions or deterioration pattern of containment systems or specific containment system components Using adaptations from product reliability anal-yses, a parameter that is generically referred to as the “failure rate” is used to quantitatively describe the effectiveness or reliability of a barrier or containment system with time The magnitude of the failure rate is the significant determinant of the barrier degradation rate in the absence of transient events It is tempting

TABLE 1.4

Estimated Long-Term Effectiveness of Selected Waste Containment Measures

Effectiveness, Et (%)

Indexing Time Increments (t years)

t0 t10 t30 t100

Clay cap 80 75 60 20 (85a)

Synthetic cap 90 85 75 15 (90b)

Clay plus synthetic cap 95 92 80 35 (98c)

RCRA C composite liner system 98 95 85 60

Clay liner 70 60 40

Synthetic liner 85 75 35

HDPE wall 65 60 50 25 (65d)

Slurry wall 70 60 20 (70e) 0

a Assumes addition of new clay cap at 100 years. bAssumes addition of new synthetic cap at 100 years.

cAssumes addition of new composite clay and synthetic cap at 100 years.

dAssumes addition of new HDPE at 100 years.

e Assumes addition of new slurry wall at 30 years.

Damage and System Performance Prediction 13 to erroneously assume that failure rates for containment systems are constant In practice, the failure rates of most engineered systems are not constant with time Generally,

(1.1) where λ(t) is the time-variable failure rate of the containment system; λ0 is the

initial failure rate of the containment system; β is an exponent that describes the variation (usually decay) of the failure rate with time, t Equation (1.1) represents the general exponential form of the decay equation The linear and Weibull forms of the equation are presented below as Equations (1.2) and (1.3), respectively The parameters are as defined for Equation (1.1) The time parameter, t0, is the

time corresponding to the origin of the initial failure, λ0

(1.2)

(1.3)

For Equations (1.1) through (1.3), the value of the constant β determines the shape of the failure rate function The failure rate is increasing with time if β > 0, it is constant if β = 0, and it is decreasing if β < For more information, the reader is referred to Wolford et al (1992), who used this approach to estimate the aging pattern of nuclear power plant equipment Such techniques have already been successful in extending the license of 10 United States nuclear power plants by 20 years Inyang (1994) observed that the Weibull format of failure analysis provides the curve geometries that match the expected deterioration pattern of most containment systems and proposed the use of Equation (1.4) with shape parameters ranging from to The use of Equation (1.4) enables long-term performance to be addressed within the context of system reliability

(1.4)

where Rt is the reliability of the containment system at a future time of reference, t is the future time of reference, and n is the scale or normalization parameter that corresponds to the time duration at which the failure probability is 0.632 Generally, the larger the magnitude of β, the greater the deterioration rate

Considering that there is a complimentary relationship between the probabil-ity of failure, Pt, and reliability, Rt, of a component or system as indicated by Equation (1.5), initial values of reliability can be established

λ( )t =λ0exp(βt)

λ( )t =λ0(1+βt)

λ λ

β ( )t t

t

= 

  

0

R t t

n

t = − −

   

  

   

exp

(1.5) The damage functions for each system component can be generated from current knowledge, testing, and extrapolations, and can be used to determine the probability that barrier characteristics will meet specified standards at specified future times

1.2.2.3 Less Empirical (Theoretical) Modeling Approach

This approach involves modeling the stresses, deterioration processes, waste transformations and release, barrier material durability, and flaw evolution for a barrier component or system In this approach, the failure probabilities of system components and the system itself are modeled Interactions among various param-eters that promote or negate effects are considered Considering that various stressors and their impacts have different probabilities of occurrence within dif-ferent timescales, the challenge of deciphering the interactions among parameters is quite great Therefore, an innovation within this modeling approach is the use of modeling frameworks that enable the incorporation of various models and the establishment of dynamic linkages among them This technique is nested in the subdiscipline of system dynamics

System dynamics is the study of dynamic feedback systems using computer modeling and simulation (Forrester, 1961) Unlike other scientists, who study the world by breaking it up into smaller and smaller pieces, system dynamicists look at things as a whole The central concept of system dynamics is understanding how all objects in a system interact with one another Visualization of the system is one of the assets of this modeling technique However, beneath the visual exterior is a series of differential equations that define the behavior of the system over time An example of software that can be used in this modeling exercise is Stella Research (Stella, 2001) The calculations are performed using numerical integration Although the interface makes the modeling look superficial and almost trivial, a sophisticated mathematical engine performs the calculations Using this modeling technique, it is possible to model complicated systems A thorough understanding of the structure of these complex systems can lead to an explanation of their performance, both over time and in response to internal and external perturbations By understanding the underlying system structure, predic-tions can be made relative to how the system will react to change System dynamics models are descriptive in nature The elements in the models must correspond to actual entities in the real world The decision rules in the models must conform to actual practice and real-world phenomena A new project at the Idaho National Engineering and Environmental Laboratory (INEEL) is addressing barrier degradation dynamics (Piet and Breckenridge, 2002) One component of the effort is the use of relatively simple but flexible system dynamics models to explore possible interactions of processes These models provide a tool to explore uncertaintiesinscenarios andmechanisms, whereas more sophisticated models are tools for exploring sensitivities to parameter uncertainties

To illustrate the necessity of addressing interactions among various parame-ters, the effects of the burrowing of covers by animals on evapo-transpiration are considered During the summer months, more water is lost from plots with animal burrows than from plots where no animal burrows are present During the winter months, both the plots with animal burrows and the control plots gain water In addition, water does not infiltrate below approximately meter (m), even though burrow depths always exceed approximately 1.2 m The lack of significant water infiltration at depth and the overall water loss in the lysimeter plots are occurring despite the following worst-case conditions:

• No vegetative cover (no water loss through transpiration) • No water run off (all precipitation is contained)

• Burrow densities in lysimeters greater than those in natural settings • Extreme rainfall events applied frequently (i.e., three 100-year storm

events in three months)

• Animals burrowing deeper in the lysimeters than in natural settings As part of the conclusion of the study described in the preceding paragraph, the investigators noted that “the overall water loss from soils with small-small burrows appears to be enhanced by a com-bination of soil turnover and subsequent drying, ventilation effects from open burrows, and high ambient temperatures” (Gee and Ward, 1997) Thus, in this case, animal intrusion had a net positive effect Indeed, earlier work shows that soils were more dry beneath burrows than elsewhere (Cadwell et al., 1989; Link et al., 1995) Link et al (1995) report that the increased moisture in burrows facilitated vegetation response that increased plant transpiration as plants took advantage of the moisture and sent roots to use it, leading to dry zones under the burrows Indeed, Link et al (1995) note that “eco-logically, it is expected that a local abundance of a limiting resource, in this case moisture, would be rapidly and therefore depleted.”

1.3 RELATIONSHIP OF STRUCTURAL FAILURE TO FUNCTIONAL FAILURE

In real-world situations, defining satisfactory system performance can be difficult It is a vector with many components, governed by different criteria, and driven by different and perhaps interacting processes These processes may not be well understood and, hence, can be represented analytically only with considerable uncertainty This situation is not too different from that in other spheres and disciplines

the consequences usually are the critical outcome(s) of the system because the larger community seldom has particular interest in the structural system itself

The foregoing discussion leads to the need to examine the performance factors necessary to evaluate containment systems These factors are divided into the following two categories:

• Total system (parameters that define functional performance) • Concentration of hazardous materials in surface/aquifer water • Exposure to humans (e.g., water, air, intrusion pathways) • Risk to humans

• Risk to ecologies

• Barrier and barrier subsystems (parameters that define structural per-formance)

• Resistance to human intrusion • Water flux through barrier • Gas flux through barrier

• Hazardous material flux through barrier

• Measures of individual degradation mechanisms (e.g., erosion, subsidence)

The satisfaction of both functional and structural design functions of the composite containment system requires that the various system components meet specific design functions that contribute to overall system performance The variability in the combination of various containment system components implies that long-term performance under a given set of applied stresses will also be different Inyang (1999) suggested the following nonexclusive criteria as indices of containment system and component performance:

• Ability of the system to reduce the concentrations of aqueous phase contaminants to acceptable levels through one or more contaminant attenuation processes (e.g., sorption, precipitation)

• Ability of the system to reduce the volume of contaminants that is released into protected media to acceptable levels

• Ability of the system to reduce the leaching of bound contaminants from stabilized media to acceptable levels

• Ability of near-surface system components to attenuate radiation to nondamaging levels

TABLE 1.5

Summary of Performance Objectives Applicable to the Monticello Mill Tailings Repository Media Standard Point of Compliance Period of Compliance Regulation

All pathways <100 mrem/year

Effective Dose Equivalent from all routine DOE activities

To a member of the public

Not defined DOE Order

5400.5 II 1.a

Atmosphere <10 mrem/year

Effective Dose Equivalent, excluding Rn

To a member of the public

Not defined 40 CFR 61.92

Atmosphere Average flux of

Rn-222 <20 pCi/m2/s or (see next row)

In air above landfill, averaged over entire landfill

1000 years if reasonably achievable and, in any case, for at least 200 years

40 CFR 192.02(a) and 40 CFR 192(b)(1)

Atmosphere Annual average

concentration of Rn-222 in air <0.5 pCi/L

At or above any location outside the landfill

1000 years if reasonably achievable and, in any case, for at least 200 years

40 CFR 192.02(a) and 40 CFR 192(b)(2)

Groundwater Arsenic

<0.05 mg/La,b Chromium

<0.05 mg/La,b Lead <0.05 mg/La,b Molybdenum

<0.01 mg/La,b Selenium

<0.01 mg/La,b Combined Ra-226 and

Ra-228 <5 pCi/La,b Combined U-234 and

U-238 <30 pCi/La,b,c Gross alpha-particle

activity, excluding Rn and U <15 pCi/La,b

Intersection of vertical plane with uppermost aquifer at downgradient limit of disposal area plus area taken by dike or other waste barrier

1000 years if reasonably achievable and, in any case, for at least 200 years

40 CFR 192.02(a) and 40 CFR 192.02(c)(4), and Table to Subpart A of 40 CFR 192

Groundwater Beta particles, and

photons made from manmade radionuclides <4 mrem/year In community water supply systems

1.3.1 ECONOMICOR PSEUDO-ECONOMIC CRITERIA

Economic evaluation has the advantage (and disadvantage) of forcing all parties to evaluate their objectives in monetary terms Pseudo-economic criteria, such as utility analysis, require a similar approach but in terms of a different unit of measurement In principle, the maximum expected net present value criterion can be stated as follows:

(1.6)

where k is the alternative or system configuration being considered, i is the state of the system (e.g., normal operation, one or other mode of system failure), pi is the probability of occurrence for each such state of nature, M is the number of such states, j is the attribute, N is the number of attributes, and Xji represents the various costs or benefits associated with each state There are some very signif-icant problems associated with determining the Xji, and these are well known in cost-benefit analysis literature (Layard, 1972; Dasgupta, 1993) Usually, the opti-mal decision is considered to be the maximization of the value of Equation (1.6), which then provides the possible decisions required Typically, this translates into desired (maximum) values for the probabilities, pi These values are obtainable through risk analysis, as are some of the values of Xji (where these are conse-quences) In practice, the optimization of Equation (1.6) can be constrained by regulatory requirements (Stewart and Melchers, 1997)

TABLE 1.5 (continued)

Summary of Performance Objectives Applicable to the Monticello Mill Tailings Repository

Media Standard

Point of Compliance

Period of

Compliance Regulation

Compacted soil layer in cover

Water percolationd <1 × 107 cm/s

Hydraulic conductivity of compacted soil layer in cover

Not defined 40 CFR 264.301

a If background is below this level.

bAn alternative concentration limit may be established under 40 CFR 192.02 (c)(ii)(A).

c Where secular equilibrium is obtained, this criterion will be satisfied by a concentration of 0.044 milligrams per liter For conditions of other than secular equilibrium, a corresponding value may be derived and applied, based on the measured site-specific ratio of the two isotopes of uranium dA unit gradient flow is assumed to equate percolation to hydraulic conductivity.

maxEVk pi X i

M

ji j

N

= 

  

   

= =

∑ ∑

1.3.2 REGULATORY CRITERIA

Typically, regulatory criteria are developed by public authorities acting broadly on behalf of the public and mandated by government Generally, regulatory authorities develop general safety goals and set specific safety standards, monitor system performance, and prosecute if specified safety standards are violated There are a number of possible measurement units available, including expected number of deaths, injuries or cost equivalents, short- and long-term expected health implications, and various environmental criteria None constitutes a com-pletely satisfactory accounting for the possible impact of a hazardous situation However, alternatives are difficult to find or suggest In a sense, all should be seen as convenient surrogates for much more complex accounting schemes; the associated issues are, ultimately, the same as those for economic decision criteria Table 1.6 shows an example of a typical set of safety targets for hazardous industries where neighboring land usage may be affected Regulatory standard enforcement can cause resentment and an adversarial situation that is not condu-cive to the operator or licensee committing to risk control and appropriate risk management Enforcement also requires periodic inspection of the facility by regulatory personnel

1.3.3 PRESCRIPTIVE DESIGN CRITERIA

The easiest approach, and perhaps the most used approach, is to show compliance with existing regulations and their prescriptive instructions for design This approach, strictly speaking, does not determine performance in the sense of per-formance measures noted above, but only shows that regulatory instructions for design have been met This is often adequate, because the prevailing knowledge base is often used to establish conservative limits for design and performance parameters However, the literature contains examples of barriers not meeting the objectives of the Resource Conservation and Recovery Act (RCRA) to protect human health, while being consistent with RCRA design guidance Thus, for future

TABLE 1.6

Typical Safety Targets for Land Use Analysis

Land Use

Individual Fatality Risk (××××10–6 per year)

Hospitals, schools, child-care facilities, old age housing 0.5

Residential, hotels, motels, holiday resorts

Commercial (including retail centers, offices and entertainment centers)

Sporting complexes and active open space 10

Industrial 50

barriers, it cannot be simply assumed that meeting RCRA design guidance will provide the protection that regulations and the public desire if the hazards, location, or design approach vary significantly from demonstrated RCRA design guidance

1.3.4 RISK CRITERIA

A more complete and systematic approach to developing long-term performance standards is to estimate risk to human health and the environment by considering possible exposure pathways, estimating the total exposure, and then (if needed) converting to risk This task can be performed with varying degrees of sophisti-cation depending on the situation The most common approach is deterministic Under the deterministic approach, a range of standard release and pathway sce-narios is constructed by the analyst, then doses are calculated under the assump-tion that the scenarios will occur without consideraassump-tion of the likelihood of the scenarios This approach is characteristically conservative, worst case, and tends to overestimate dose exposures to the receptors (Moore et al., 2001) An example is the use of a hydrological code, e.g., HELP, to estimate water movement through a barrier, then estimate exposure to the nearest population, convert to risk (if needed), and compare to exposure/risk requirements — all without including probabilities in any of the stages of analysis

For more complex, longer hazard, and/or higher hazard situations, probabi-listic approaches have been used “A probabiprobabi-listic approach to scenarios takes the likelihood of occurrence into account, allowing a mechanism to differentiate between site characteristics and accessibility” (Moore et al., 2001) Where uncer-tainties in key variables are significant, stochastic approaches are used to estimate uncertainties and sensitivities The uncertainties in key parameters are estimated and then propagated through the analysis of relevant scenarios to determine the resulting uncertainty in the relevant performance measures The term “probabi-listic” is sometimes used for such analyses because one is dealing with proba-bilities of a variable having a certain value However, the term can be used as a descriptor of the approach of analyzing the probabilities of different scenarios (as used in the nuclear industry and other industries) A stochastic approach can be combined with either deterministic sets of scenarios or probabilistic sets of scenarios For example, the United States Nuclear Regulatory Commission (USNRC) has established a set of regulatory guidelines for decommissioning sites (leaving residual hazards protected by barriers) that use stochastic parameter variations to estimate parameter uncertainties (Meyer and Gee, 1999; Meyer and Taira, 2001; Meyer and Orr, 2002)

A recent powerful method of stochastic barrier analyses was performed for the Mill Tailings Repository in Utah (Ho et al., 2001a,b, 2002a,b) The FRAMES shell with HELP was used to estimate the cumulative probability distribution for the following four performance measures:

• Peak Ra-226 dose

• Water transport through cover • Radio-gas transport through cover

The exposure scenarios were deterministic The stochastic analyses allowed for the estimation of the chance that a given performance measure would be exceeded and the relative contribution of different variables to the total uncertainty was covered Figure 1.7 provides information on the cumulative probability distribution for the peak Ra-226 dose that was determined in this analysis The figure shows that there is 100% probability that the dose will be below the 100 mrem/year (1 mSv/year) limit In this example, there is a 50% probability that the dose will be below 10–12 mrem/year (which is the value of a femto-Sv/year in international

dose units) The approaches described above can be applied in either a static or dynamic manner In static analyses, the probability of the occurrence of the scenarios; the boundary conditions; the mechanisms; the state of the barrier; and, hence, the failure rates are considered constant with time In dynamic analyses, one or more of the parameters is considered to vary with time Realistically, no barrier analysis is totally static because the boundary conditions of precipitation almost always vary over a year (or several years) Construction quality assurance/quality control (QA/QC) is often the key issue at the beginning of service life, but several

FIGURE 1.7 A typical risk-consequence matrix (From Stewart, M.G and Melchers R.E.,

1997 Probabilistic Risk Assessment of Engineering Systems, Chapman & Hall, London With permission.)

Likelihood

Consequences

Almost certain (5) Likely

(4) Moderate

(3) Unlikely

(2) Rare

(1)

Insignificant (1)

Minor (2)

Moderate (3)

Major (4)

Catastrophic (5)

High risk

Significant risk

Moderate risk

combinations of factors become the overriding issue during long service lives Indeed, the United States Environmental Protection Agency (USEPA) found QA/QC problems at several barriers studied in 1998 (USEPA, 1998) The challenge is to estimate the state of the system toward its end of life For illustration, consider the section ABCD through a conceptualized model of a cover system (Figure 1.8) The barrier AB is a cap composed of a number of layers of different permeabilities (and other properties) and may include a man-made membrane It may be assumed that the properties of this latter subsystem are reasonably well known On the other hand, it may be that CD represents a natural barrier with permeability and other properties that are known only with considerable uncertainties

1.3.5 DEMONSTRATING COMPLIANCE: THE SAFETY

CASE CONCEPT

A safety case consists of a document describing how the regulatory safety goals have been met Such a document is reviewed or audited by the regulatory authority to ensure the following:

• The study deals in sufficient depth with the facility under discussion (completeness requirement)

• Appropriate event probabilities and consequences have been considered • Compliance to the relevant regulations has been achieved and

documented

An important advantage is that the onus of proof is put on the licensee or operator Safety cases are used extensively in the off-shore, chemical and petro-chemical industries, particularly in Europe, and are similar to the demonstration

FIGURE 1.8 Schematic cap containment system showing potential fluxes and some

influences

Contaminant Differential

settlements

Flux Q1(t)

Environmental

influences E1

Barrier layers

resistance R2(t)

Barrier layers

resistance R1(t)

A B

C D

Flux Q2(t) Environmental

of compliance for operational activities or environmental impact assessments for new systems adopted internationally

1.3.6 MIXED CRITERIA

A mixed economic and regulatory framework for decisions is the so-called ALARP (as low as reasonably practical) or the ALARA (as low as reasonably attainable) approach Although terms such as “low,” “reasonably,” “possible,” and “attainable” are highly subjective and difficult to define, ALARP has been widely adopted (e.g., USNRC, off-shore industry members) The concept is sketched in Figure 1.9, and an overview is provided by Stewart and Melchers (1997)

1.3.7 QUALITATIVE AND INDEXING ANALYSES

This is the simplest level approach Numbers are not used in qualitative analysis, only subjective assessments, perhaps obtained from interaction between risk analysts and operators The results can then be put into a risk-consequence matrix This approach is easy to use and useful for nontechnical audiences However, it is difficult to use with quantitative approaches to risk assessment The rankings cannot be converted to numbers, as the outcome can be meaningless and incon-sistent An example of a qualitative risk-consequence matrix is shown in Figure 1.7 The indexing approach involves assigning ratings that usually are not ana-lytically derived, but represent performance assessments conducted on the basis

FIGURE 1.9 ALARP concept (From Stewart, M.G and Melchers R.E., 1997

Probabi-listic Risk Assessment of Engineering Systems, Chapman & Hall, London With permission.) Unacceptable region-risks cannot be justified (except in extraordinary circumstances)

ALARP region-risk reduction is impractical or costs are disproportionate to benefits gained

Acceptable region-ensure risks remain in this region

of experience with such systems or subjective estimates of performance proba-bilities This approach was used by Einarsson and Rausand (1998) in rating the survivability of an industrial system that was subjected to a number of stressors

1.4 QUANTIFICATION OF LONG-TERM DAMAGE SCENARIOS, EVENTS, AND MECHANISMS

Establishing the quantitative relationship between containment system or com-ponent reliability and design has been targeted by many researchers, among which are Gilbert and Tang (1995), Hartley (1988), Bogardi et al (1989), and Shackel-ford (1992) The limitations of current performance assessment models fall into one or more of the following categories:

• Description of only statistical reliability of sample test data for initial facility design

• Use of time-invariant barrier material characteristics and contaminant loading/stress levels to develop long-term performance estimates • Repetition of stress tests to a few cycles corresponding to short service

lives relative to the service lives of real facilities

1.4.1 CATEGORIESOF DEGRADATION MECHANISMS

The potential for damage of containment systems under applied stress is deter-mined by the interactions among three categories of factors that can change in magnitude with time: location factors, design and operational factors, and waste characteristics While some damage phenomena occur continuously, others are transient and can cause instantaneous damage if the resistance of the system or its components is exceeded by the applied stresses

1.4.1.1 Slow Physico-Chemical and Biological Processes

Among the physico-chemical processes that can damage barriers are barrier flocculation at low depth of burial; chemical attack and photo-aging of geotextiles, slurry settlement, dissolution; and freeze-thaw action (Chamberlain and Gow, 1978; Fernandez and Quigley, 1991; Fleming and Inyang, 1995; Elias et al., 1997; Daniels et al., 1999, 2001) Liu and Gilbert (2002) identified seepage-induced fluid pressure as a potential damage mechanism for landfill cover slopes Figure1.10

shows experimental results obtained by Fernandez and Quigley (1991) on the effects of permeant viscosity and effective stress on the hydraulic conductivity of compacted clays with mineralogies simulative of clay barriers

dielectric constant of the liquids was used as an index of physico-chemical aggressiveness The results indicate that the kaolinite compressive index decreased from 2.8 at a dielectric constant of 1.9 for heptane, to 0.75 at a dielectric constant of 24 for ethanol The compressive index was 2.06 for distilled water (which has a dielectric constant of about 80), but decreased slightly for formamide,

FIGURE 1.10 Hydraulic conductivity of water-compacted clay permeated with municipal

solid waste leachate–ethanol mixtures subsequent to prestressing water-wet clay at various levels of vertical effective stress (From Fernandez, F and Quigley, R.M., 1991 Canadian

Geotechnical Journal, 28, 338–398 With permission.)

Hy dra ulic c onduc tiv ity

, k (

cm/s)

10−5

10−6

10−7

10−8

10−9

Water compacted

σ′vo = kpa

Reference k (water) Final k for permeant

Water only Viscosity control Double layer control 50

% Ethanol in the permeant % Domestic waste leachate

100 50

100 Hy dra ulic c onduc tiv ity

, k (

cm/s)

10−5

10−6

10−7

10−8

10−9

Water compacted

σ′vo = 40 kpa

Reference k (water) Final k for permeant

Water only Viscosity

control

0 50

% Ethanol in the permeant % Domestic waste leachate

100 50

100 Hy dra ulic c onduc tiv ity

, k (

cm/s)

10−5

10−6

10−7

10−8

10−9

Water compacted

σ′vo = 80 kpa

Reference k (water) Final k for permeant

Water only Viscosity

control

0 50

% Ethanol in the permeant % Domestic waste leachate

100 50

100

% Domestic waste leachate

Hy dra ulic c onduc tiv ity

, k (

cm/s)

10−6

10−7

10−8

10−9

10−10

Water compacted

σ′vo = 160 kpa

Reference k (water) Final k for permeant

Water only

Viscosity control

0 50

% Ethanol in the permeant

100 50

100

H−C−C−OH H− H−

H

−

H

−

(Ethanol) є = 32

(a) (b)

which has a dielectric constant of 110 The authors computed the attractive forces for kaolinite in the various organic fluids using the Lifshitz theory and found that the attractive force variation agreed with the compressibility results qualitatively The reader should note that, as discussed in Chapter3, barrier components rarely comprise only a single mineral They are usually composite mixtures such that mineralogy, particle size distribution, and mix proportions determine their reac-tivity with permeants of a given chemistry under prevailing environmental con-ditions (Inyang et al., 1998a) As nontraditional barrier materials (e.g., paper mill sludges) are used more frequently, it is essential to consider physico-chemical interactions in barrier design and long-term performance assessment (Moo-Young and Zimmie, 1997)

one that will ultimately occur on the capping system While the forces of natural replacement can be managed, there are forces that require active intervention (i.e., expenditure of maintenance energy) to change or arrest the natural trajectory toward the successional equilibrium The bare soil cap, or cap planted with vegetation in most cases, represents a nonequilibrium condition As vegetation colonizes the cap and changes over time, it produces significant changes in the soil conditions relative to the initial soil conditions In addition to the changes noted above, root systems penetrate the soil structure, altering hydraulic properties by creating preferential flow paths through the soil and adding organic matter to the soil matrix The death and decay of above-ground plant material deposit organic matter on the soil surface, thereby altering the evaporative properties of the soil The increased organic content of the soil and development of preferential flow paths increases the moisture retention capacity of the soils of the cap system Similarly, the development of a mulch or litter layer on the soil surface retards run off, increases water infiltration rates, and decreases evaporation from the soil surface

The timing of successional sequences can vary dramatically, but the attain-ment of the local climax condition takes a long period of time Smith et al (1997) reported that in areas near Oak Ridge, Tennessee, the progression from old field grassland to a shrub vegetation stage is expected to take 10 to 25 years, while the continued development to a mature forest takes on the order of 65 to 150 years Processes in a Colorado subalpine forest can take 200 to 300 years to achieve the expected spruce-fir forest cover, and the rate of the processes is significantly influenced by such factors as soil nutrients, aspect (e.g., north or south facing slope), and rock cover in the terrain (Donnegan and Rebertus, 1998) Succession in nutrient-poor sand dune communities is similarly slow, with early successional species being lost within 100 years, while plant species were still being replaced after over 300 years Rates of change in the plant communities tend to be most rapid in the earlier stages of the successional sequence

One highly significant effect of vegetation is the alteration of the water balance Plants mine water and nutrients from the soil to support photosynthesis and growth Thus, the plant root systems pump water from the soil to the atmo-sphere throughout the growing season With this property in mind, much work has been performed in arid and semi-arid climates with water balance or evapo-transpiration caps (DOE, 2000) In this approach, the function of the vegetation is not only to hold the surface soil in place against wind and water erosion, but also to maintain the water balance of the cap Surface soil layers are designed by depth and texture to hold the annual input of moisture in the soil matrix, while the plants extract the moisture through the growing season, resulting in little or no deep penetration of moisture to the drainage layer Furthermore, by seeding these capping systems with the native climax vegetation, the successional sequence is jump-started, likely minimizing potential surprises as the capping systems mature

force Initial studies of plant root systems on capping systems were largely driven by concerns about root penetration into buried wastes Many cases are docu-mented of plant root systems penetrating into buried waste and transporting hazardous material to the surface (Arthur, 1982; Arthur and Markham, 1983) Recent capping system designs have included low permeability clay barriers between the vegetated topsoil and the buried waste Studies have shown that plant roots can penetrate into, if not through, clay barriers Particularly susceptible to penetration are barriers constructed of clays such as bentonite which have sig-nificant shrink-swell properties as moisture conditions change Drying of previ-ously hydrated clays of this type leads to cracking and permits points of entry for plant roots (Nyhan, 1989) One need only witness the ability of plants to establish in asphalt cracks and concrete pavements with the subsequent deterio-ration of those materials to understand that initial penetdeterio-ration into such clay barriers leads to further deterioration and potential localized movement of water through the barrier into the buried waste This poses the following two problems (i) Potential mobilization of contaminants to the surface through plant uptake — Because plant roots not absorb some elements, a significant factor in this consideration is the nature of the waste that is buried Root penetration into buried wastes at the Uranium Mill Tailings Remedial Action (UMTRA) site in Burrell, Pennsylvania, was deemed not to be a significant issue because the plants were not mobilizing the buried uranium mill tailing waste to an extent that resulted in significant human or ecological risks (UMTRA, 1992) Radioisotopes such as cesium and strontium, which are analogs for the biologically essential elements potassium and calcium, have significant remobilization potential if they are biologically available in the buried wastes Another type of slow process that affects the long-term performance of containment systems is global warming because of its impacts on regional hydrology and the response of vegetation, soils, and temperature conditions to expected patterns It should be noted that large-scale, long-duration events such as global warming may not directly affect containment system performance at initially discernable scales, but may cause significant changes in environmental conditions that, in turn, impact long-term performance Uncertainties in the estimates of the impact of large-scale, long-duration phe-nomena such as global warming translate to uncertainties in their impacts on future containment system performance Generally, a possible worse climate for the barrier should be estimated using climate change modeling and/or examining the geological and fossil record Then, the resulting effects can be estimated by modeling how the ecosystem would respond to the new climate and/or examining natural analogs The natural analogs can be from the site in question, or can be current ecology from a location that approximates the hypothetical new climate Ho et al (2002b) used this approach to compare estimates of cumulative probability distributions of a radon-226 dose from a shallow alluvial aquifer (Figure1.11)

of responses to climate shifts and secondary disturbances (e.g., fire) Pedogenic effects are inferred from measurements of key physical and hydraulic soil properties in natural and archaeological soil profiles that are considered analogous to future states of engineered soils Analogs of local responses to future global climate change exist as proxy ecological records of similar paleoclimates” (Waugh, 2001)

Indeed, the “present” and hypothetical “future” climate cases by Waugh (2002) were the basis for the two cases by Ho et al (2002b) in Figure 1.11 In that study, the future climate case was wetter, and the overall performance was calculated to be worse, but still acceptable

(ii) Induction of water movement into buried waste — The second and likely more significant issue associated with root penetration through the imper-meable layers is the induction of water movement into the buried waste Depend-ing on the magnitude of the water flow and the type of barrier systems below the waste, this type of failure can result in waste mobilization downward into the vadose zone and potentially to the water table, resulting in a contamination event that requires remedial action

1.4.1.2 Intrusive Events

Just as plants grow on the caps with the root systems seeking water, various animal species also invade capping systems with the primary objective of seeking food or shelter Both invertebrate and vertebrate species have been documented to invade waste isolation systems, resulting in the mobilization of buried mate-rials Bowerman and Redente (1998), Suter et al (1993), and Smith et al (1997) summarized experiences of animal intrusion into buried waste Studies at Hanford Washington and Idaho National Environmental and Engineering Laboratory

FIGURE 1.11 A schematic illustration of landfill deformation due to seismic activity (From

Inyang, H.I., 1992 Journal of Environmental Systems, 21(3), 223–235 With permission.) Deformed leachate

collection pipe

Drainage layer

(INEEL) in the U.S documented excavation by harvester ants to depths of to m below the ground surface through types of cover materials ranging from topsoil to gravel Studies at a variety of primarily arid and semi-arid sites in Washington, Idaho, New Mexico, and Colorado documented intrusion into cap-ping systems by pocket mice, deer mice, kangaroo rats, pocket gophers, prairie dogs, and ground squirrels These excavations represented invasions through a wide variety of cover systems with penetrations at least to m and probably significantly deeper; because in one case animals assimilated radionuclides buried under 2.4 m of soil cover These breaches of cover integrity can be complimented by slower migration of fines under overburden loads into the pore spaces of drainage layers as mathematically modeled by Bai et al (2000b)

Larger animals are also capable of invading burial sites, primarily in search of food, but sometimes in search of shelter Foxes, coyotes, badgers, and other predators excavate the tunnels of prey species such as mice and gophers In the southeastern region of the United States, the spread of armadillos represents the threat of disruption of surface cover systems as they search for insect prey As is the case of plants, these organisms are part of the surrounding landscape and significant (sometimes excessive) maintenance activities are required to eliminate them or minimize the effects of their activity once the capping system is con-structed Given the long expected requirements for cap integrity, high maintenance costs are not a desirable characteristic for a capping system, as they represent a substantial mortgage cost that can and should be avoided

1.4.1.3 Transient Events

FIGURE 1.12 Regions in the conterminous United States with greater than 90% proba-bility that the acceleration in bedrock will exceed 0.1 g in 250 years

FIGURE 1.13 Cumulative probability distribution for peak cumulative dose for Ra-226

and its progeny from the shallow alluvial aquifer for present and future conditions The “present” and “future” curves reflect the present and a hypothetical future climate (From Ho, C.K et al., 2002b Spectrum 2002, Reno, Nevada, August 2002 With permission.)

Cu

mu

la

tive pr

ob

ability

1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0

1.0E + 04

1.0E + 01

1.0E − 02

1.0E − 05

1.0E − 08

1.0E − 11

1.0E − 14

1.0E − 17

1.0E − 20

1.0E − 23

Present

Future

100 mr

em/y

r (D

OE or

der 5400.5 || 1.a)

are applied A comprehensive listing of stressing factors and their effects on the long-term performance of containment systems is presented in Table1.7

1.4.1.4 Cyclical Stressing Mechanisms

Several loading mechanisms of barrier systems are repetitive in nature (e.g., freeze-thaw and wet-dry cycling of cover systems) A barrier component or system that does not fail at constant loading or imposition of a single-load cycle can still be damaged by repetitive loads of a single magnitude The potential and magnitude of future containment system damage not depend exclusively on the level and number of repetitions of stress and intensity of physico-chemical and biological phenomena In general, three categories of interactive factors are recognized: location factors, design and operational factors, and waste factors A conservatively designed system is likely to resist the same level of stress better than a poorly designed system Furthermore, a well-constructed and maintained system should perform better under stresses that are imposed Various types of systems and barriers exhibit different levels of susceptibilities when exposed to different modes of physical stress and other damage processes Table1.8 shows a matrix of which stressors are relevant for which barrier systems Each model system has characteristic materials The matrix includes the natural subsurface for comparison; note that the natural subsurface is isolated from many stressors that have impacts on near-surface, engineered barriers Not shown in the table is an assessment of how important a stressor may be, or when a stressor may be important These aspects must be considered for individual barrier designs at specific locations (e.g., freeze-thaw cycles can be critical in northern climates and irrelevant in southern climates)

TABLE 1.7

How Stressors May Generate Effects on Near-Surface Barriers

Stressor Mechanical Effects

Physicochemical and Biochemical Effects Water (rainfall/snowmelt, surface water) Hydrostatic head

Erosion (run off, surface water, movement of materials within barriers, localized depressions pooling water)

Ice expansion/contraction

Wet-dry cycles Corrosion Leaching

Water influences plant, animal, microbial behavior

Water transports contaminants Surface water brings seeds → plant

ecology

Water brings microbes → microbial ecology

Temperature changes Differential thermal expansion

Freeze-thaw Desiccation

Ice expansion/contraction

Influences bio-chemical reaction rates Climate changes impact biota

Wind Mechanical loading of surface

Movement of objects Erosion of exposed surface Delayering (lifting layers)

Bring seeds → plant ecology Bring microbes → microbial ecology Add soil → change plant growing

conditions → change/hurt/help vegetation Mechanical loads (seismic, vibration, subsidence, impacting objects) Puncturing Mechanical loading

Settling of fines into coarse layers N/A

Plants Macro-porosity development

Evapo-transpiration

Uptake contaminated material and bring to surface

Impact animal ecology (food supply) Impact microbial ecology

(e.g., nutrient profiles) Evapo-transpiration

Animals Macro-porosity

Infiltration

Erosion (of excavated material) Digging and exposure of buried

material

Impact plant community/species Impact microbial ecology

(e.g., nutrient profiles)

Microbes Plugging of capillaries Bio-corrosion

Bio-leaching

Change surface tension, e.g., in pores and capillaries

Change PRB biochemistry Soil formation → change plant

to 10 freeze-thaw cycles occurred in the season and that the hydraulic conductivity of the samples increased by factors ranging from 50 to 300 More recently, effort has been made as exemplified by Daniels et al (2001) to develop quantitative schemes for interrelating field- and laboratory-based freeze-thaw permeability measurements and improving the resistance of barriers to freeze-thaw through soil stabilizer amendments In some cases, soil strength can improve at the expense of low hydraulic conductivity (Daniels et al., 1999)

TABLE 1.7 (continued)

How Stressors May Generate Effects on Near-Surface Barriers

Stressor Mechanical Effects

Physicochemical and Biochemical Effects

Radiation (UV, ionizing)

N/A Material property degradation

Waste Chemical attack Material property degradation

TABLE 1.8

Relevant Stressors for Various Barrier Systems

Relevant Stressors Earthen Cap Concr

ete Cap

Top of Gr

outed/

Entombed Structur

e

Ev

apor

ation

P

ond Liner Liner at Bottom of W

aste

Zones

Bottom of Gr

outed/

Entombed Structur

e

V

adose Zone Itself

(fate and tr

ansport,

not barrier per se)

Water (hydrostatic head, erosion) X X X

Water (wet-dry cycles) X X X X X

Corrosion/other chemical attack X X X X X X

Temperature changes, e.g., freeze-thaw cycles X X X X X X

Wind erosion X X

Imposed mechanical stress (subsidence, seismic, structural loads)

X X X X X X

Impacting objects, e.g., construction/operations activities, people and animals walking

X

Plant and/or animal intrusion X X X

Biocorrosion X X X

Other microbial impacts, e.g., plugging capillaries X X

Ultraviolet radiation X

Ionizing radiation X

Other cyclical degradation mechanisms of barrier systems include wet-dry cycling and desiccation in response to temperature and relative humidity profiles that can alternate but remain predominantly at high temperature and low humidity levels during most of the year At the fundamental level, the swelling potential of clays has been investigated by many researchers, among whom are Blackmore and Miller (1961) and Dasog et al (1988) From a practical standpoint, the inclusion of granular soils (sands and silts) in barrier mixtures and their compac-tion generally decrease shrink-swell potential

Several investigators have developed empirical methods for analyzing the relationships between the residual strength of structures and the number of load cycles Schaff and Davidson (1997) developed Equation (1.7) to describe the pattern of the Weibull scale parameter in terms of the residual strength of com-ponents of structures under fatigue loading:

(1.7)

where Rn is the residual strength scale parameter after n cycles of loading, Ro is the static strength scale parameter, Sp is the peak stress magnitude of the constant-amplitude loading, N is the scale parameter for the fatigue life distribution, and v is the strength degradation parameter Equation (1.7) describes the exponential line shown in Figure1.14 The reader should note that the strength measured for a particular cycle is observed as a distribution of magnitudes Thus, the line represents the connection of points defined by the specified statistical confidence level of strength data for each loading cycle Kachanov (1986) developed Equa-tion (1.8) for describing damage accumulaEqua-tion in materials and systems as a result of load repetition:

(1.8)

In Equation (1.8), D is the damage variable, S is the amplitude range of the repeated stress, n is the number of cycles, C > and m≥ are material constants In the case of containment systems, D could be fracture intensity or macroporos-ity In the field, load repetitions are not designed but observed, implying that observed data or their estimates need to be fitted to time functions for use in performance prediction equations

Quantitative techniques that are based on time-series analysis are useful in efforts to describe the loading pattern of containment systems in the field Khalil and Moraes (1997) developed a simple method of time-series analysis that is based on the linear least squares spectral analysis (LLSSA) For a given set of loading frequencies, the best-fit sinusoidal equation is found for observed data

R R R S n

N

n o o p

v

= −( − )

  

∂

∂ = −

 

 

D

n C

S D

m

The power of each frequency is taken as the square of the amplitude of the fit In this method, the function described by Equation (1.9) is fitted to a time series for each frequency By using the LLSSA, the parameters A and B can be found through Equations (1.10), (1.11), and (1.12)

(1.9)

(1.10)

(1.11)

(1.12) In the equations, Σ is the variance, γ is the parameter of interest, and t is the time The error on each of the two parameters A and B can be estimated using Equations (1.13) and (1.14), respectively Then, the total error of the power can be expressed as in Equation (1.15)

FIGURE 1.14 Strength distributions associated with a residual strength Weibull

relation-ship with number of loading cycles on materials (From Schaff, J.R and Davidson, B.D., 1997 Composite Materials: Fatigue and Fracture (Sixth Volume) ASTM STP 1285, pp 179–200 With permission.)

Re

sidual str

eng

th, R (n)

Peak

Static strength

Residual strength relation

Failed portion of distribution Stress

γi−γavg = •A cosωti+ •B sinωti

A= i ti∑ ti−∑ i ti∑ ti ti

1

∆∑∑γ cosω sin ω γ sinω cosω sinω 

B= i ti∑ ti−∑ i ti∑ ti ti

1

∆∑∑γ sinω cos ω γ cosω cosω sinω 

∆=∑cos2ωt∑sin2ωt −(∑cosωt sinωi)2

(1.13)

(1.14)

(1.15) The phase of the periodicity, Ø, and its error, σØ, can be estimated using

Equations (1.16) and (1.17)

(1.16)

(1.17)

With these parameters, the best-fit sinusoidal equation can be developed for data that fluctuate in cycles over time Khalil and Moraes (1997) applied this method to a different type of problem: determination of the concentration vs time function for methane in ice cores during the past 160,000 years (Figure1.15)

1.4.2 QUANTITATIVE LINKAGEOF CONTAMINANT RELEASE SOURCE

TERMSTO RISK ASSESSMENT AND COMPLIANCE LIMITS

Risk assessments of containment facilities require estimating the level of hazard posed by a containment system to human health and the environment The estimation of risks to human health and the environment is at the posterior end of the analyses At the anterior end, there is the step in which estimates of the probable quantities of contaminants that will be released from the facilities are made Such analysis is typical at the source term assessment stage illustrated in

Figure 1.16 by the USNRC (2000) as adapted from Kozak et al (1990) The overall framework for relating containment system performance (operationally defined in terms of source term concentrations) to human health and ecological risk assessment are illustrated in Figure 1.17 as developed by Nazarali et al (1998) System failure can be defined in terms of the exceedence of a given

σA γ γ ω

fit ti

N 2 2 =( − ) − ( ∑) sin ∆

σB γ γ ω

fit ti

N 2 2 =( − ) − ( ∑) cos ∆ σ2P=4(A2σ2A+B2σ2B)

Ø

/ / =

− >

− − < >

− − − −

tan ( )

tan ( ) ,

tan ( 1 0

B A A

B A π A B

B

B A/ )+ A< ,B<

   

 π 0 0

probability of release of specific quantities of target contaminants in the future Such a quantity can be specified on the basis of the known or assumed human health and environmental risks associated with exposures to the target contami-nants at contaminant levels above the specified values As shown in Figure1.18, releases of a specific contaminant from a containment system over a given interval occur as a distribution Several uncertainties plague efforts to make precise estimates of the probability of release of specific quantities of contaminants for selected future time frames Gallegos et al (1998) identified the following sources of uncertainty in long-term performance predictions of containment systems:

• Uncertainty in the likelihood of occurrence of future events

• Uncertainty in conceptual and analytical models that describe events and processes

• Uncertainty in parameter values

The establishment of the compliance of a containment system with selected performance criteria is characterized by the uncertainties stated above Indeed, one of the utilities of system dynamics approaches described by Siu (1994) is that possible interactions among various performance factors can be addressed, and uncertainties can be reduced by using the evolving knowledge base

FIGURE 1.15 Methane concentration vs time experimentally measured on ice cores

(From Khalil, M.A.K and Moraes, F.P., 1995 Journal of the Air and Waste Management Association, 45, 62–74 With permission.)

Conc

en

tra

tion (ppbv)

700

600

500

400

300

200

160 120

80

Kiloyears before present 40

Two common types of probabilistic analyses can be used in assessing con-tainment system performance Quantified risk analysis (QRA) gives all element performances and all probability estimates as numerical point estimates There is no estimate of the uncertainty in the result(s) QRA is widely employed in the

FIGURE 1.16 Conceptual model showing processes to be considered in an LLW

perfor-mance assessment (USNRC, 2000 A perforperfor-mance assessment methodology for low-level radioactive waste disposal facilities Recommendations of NRC’s Performance Assessment Working Group NUREG-1573 United States Nuclear Regulatory Commission, Office of Nuclear Material Safety and Safeguards, Washington, DC; modified from Kozak, M.W et al., 1990 U.S Nuclear Regulatory Commission, NUREG/CR-5532, Washington, DC.)

Infiltration

Container breach Engineered

barrier performance

Air transport

Vadose-zone transport Source

term

detail Waste from leach Facility release Unsaturated

Vadose-zone flow

Engineered barrier performance

Source term

Vadose-zone transport

Saturated-zone transport Saturated-zone

flow

Surface-water transport

Pathways and dosimetry

Dose to human

or Environmental Contaminant Containment & T

reatment

FIGURE 1.17 Risk calculation flow chart (From Nazarali, A.M et al., 1998 Proceedings of Topical Meeting on Risk-Based Performance

Assessment and Decision Making, Richland/Pasco, Washington, April 5–8, pp 143–150 With permission.)

Risk

Risk plant

Great basin pocket mouse

Coyote

Red-tailed hawk

Residential exposure scenario

unit risk Industrial exposure scenario

unit risk Agricultural exposure scenario

unit risk Recreational exposure scenario

unit risk Ecological

Human health Transported unit

concentration t = 40 years

Transported unit concentration

t = 100 years

Transported unit concentration

t = 140 years

Transported unit concentration t = 10,000 years Source

concentraion t =

chemical and petrochemical industries Probabilistic risk analysis (PRA) is more advanced than QRA and has element performances treated as random variables Occurrence rates also can be given as random variables PRA is the method of choice in the nuclear industry Although more complex, more explicitness is required in the analysis and outcome uncertainty is recognized Both methods use essentially the same tools but more refined data and analysis is used in the

FIGURE 1.18 Probabilistic approach for treating model parameter uncertainty in an LLW

performance assessment (From USNRC, 2000 A performance assessment methodology for low-level radioactive waste disposal facilities Recommendations of NRC’s Perfor-mance Assessment Working Group NUREG-1573 United States Nuclear Regulatory Commission, Office of Nuclear Material Safety and Safeguards, Washington, DC.)

f(x) f(y) f(z)

x y

Estimate distributions of values for parameters x, y, and z

Input distributions into model

Produce distribution of model results

Compare with dose limits

Fr

equenc

y

Fr

equenc

y

Dose = g(x, y, z)

z

Dose

Dose limit

a

Dose

case of PRA Estimates of performance (i.e., failure or success) probabilities of components need to be made For a composite containment system that comprises many components, the following sequential steps can be taken:

1 Disassemble the system into components

2 Analyze the system (e.g., series, parallel, compound) Consider the technology and understand each component Predict future performance of each component (with uncertainty) Reassemble the components and their future behaviors

6 Assess critical components vs system behaviors

7 Develop future expected system performance vs time (or a surrogate) Estimate system uncertainty vs time

1.4.3 FRAMEWORKSFOR ASSESSMENTOF EVENT CONSEQUENCES AND CONNECTIVITIES AMONG CAUSES OF FAILURE

Two important assessment systems aid in the analysis of long-term performance characteristics of multi-component facilities and systems: fault trees and event trees

1.4.3.1 Fault Trees

The connectivity of failure causes is often represented by a fault tree (Figure 1.19) A fault tree describes the chain of events leading to system failure and is used extensively for estimating the reliability of mechanical systems such as rockets and aircraft Fault trees form the basis for quantitative estimation of failure probabilities in systems with simple pass-fail components, using point estimates of the probability of occurrence for each failure event They are less suited to reliability analysis for systems that are likely to fail mainly because of stochastic processes However, Cepic and Mavko (2002) discussed the use of the dynamic fault tree method to analyze the performance of multi-component systems

1.4.3.2 Event Trees

The consequences arising from component failure usually can be represented by an event tree (Figure 1.20) Again, usually point estimates of the probability of occurrence of each event in the event tree are used to predict outcome probabilities for each of the various outcomes

1.4.4 ESTIMATIONOF LONG-TERM FAILURE PROBABILITIES

must be unity Because the objective is to track the variability in barrier perfor-mance with time, both Rss and Fss can be considered to be time functions

(1.18)

1.4.4.1 System Failure Probability

The composite containment system illustrated in Figure1.8 can be subjected to a number of loads (i.e., structural and chemical), with a resulting condition in which contaminants have greater potential to travel through its barrier system Any of the component barriers can have subcomponents, but in order for the barrier to fail functionally, the contaminant must penetrate all layers Hence, in a simplified case, and as discussed by Stewart and Melchers (1997), the barrier can be considered as a system with n parallel components, with system failure Fss requiring failure Fi(i = 1, …, n) of each component barrier Both structurally and functionally, the configuration barrier system components may be such that some components are arranged in parallel mode, series mode, or a combination of both (i.e., compound mode) For a parallel system of components, Equation (1.19) represents the relationship between the failure of components and that of the system:

FIGURE 1.19 A typical fault tree for a waste containment system

Contaminant release volume/rate exceed

design value

Liner system damage

Loss of cover system effectiveness

Cover clay fails to contain

infiltration

Drainage layer clogs

Vegetative cover develops infiltration

channels

(1.19) When there are multiple failure paths perhaps through multiple barriers (Figure1.8), failure of the whole or a significant part of the whole system, Fs, may be brought about by failure of one or more of the subsystems, Fssj:

(1.20) where the component terms Fssj refer to any failure path and may be defined by Equation (1.19) in the case of a failure path having to cross through a barrier with multiple layers Note also that Equations (1.19) and (1.20) define the failure domain for the system (Stewart and Melchers, 1997)

1.4.4.2 Component Failure Probability

Theoretically, for a multi-component system, the system failure probability cannot be estimated precisely without computation of the failure probabilities of each of the system components For the containment system illustrated in Figure 1.8, if contaminant flux is selected as the performance factor, the probability of failure (exceedence of a specified flux within a specified time interval) can be modeled as a random process on a barrier parameter, Q(t), because the several factors

FIGURE 1.20 A typical event tree for a waste containment system

Liner system damage Drainage

layer clogs Vegetative layer

develops large infiltration channels Cover clay

fails to contains contaminant

Contaminant release volume and rate

exceed design value

Contaminant release

volume and rate do not

exceed design value

Contaminant release volume and rate

exceed design value

Contaminant release

volume and rate do not

exceed design value

Contaminant release

volume and rate do not

exceed design value Yes

No

Yes

No

Yes

No Yes

No

Consequences

Fss=F1∩ ∩ ∩F2 F3 etc

presented in Table1.8 have highly uncertain magnitudes When one layer in a barrier is subject to only one flux, the flux will be a function of time Because the flux is a complex process that is not well defined as a function of environ-mental influences (e.g., rainfall, temperature, water table fluctuations, ground movements, earthquakes), it is appropriate to model it as a random process, Q(t) (Figure 1.21) This is approach is not uncommon in other applications such as water resource projects

Also shown in Figure 1.21, a barrier parameter, R, must not have a value greater than r, where r is a deterministic quantity that represents the resistance offered by the containment system barrier layer(s) to the permeation of the flux Q(t) through it Failure is defined by the upcrossing event Q(t) >r

Of particular interest is the time, t1, to the first occurrence of a failure event (i.e., the first exceedence event) This time should be long for safe containment systems, and can be estimated readily if the exact trace of fluctuating loading is known However, because the loading is a stochastic process, the first exceedence event will be a random variable Its estimation is a central matter in reliability theory and is further discussed by Inyang (1994), Inyang et al (1995), and Melchers (1999) Following the analysis by Stewart and Melchers (1997), two probability density functions are shown to the left of Figure 1.21 The main one, instantaneous distribution, refers to all possible values of the load (flux) The point where the resistance, R = r, cuts across it has a small (shaded) part of the probability density function above it This is the probability that Q(t) > r Evi-dently, the shaded zone (i.e., the probability of failure) will be smaller for higher values of resistance, r Also, the time to the first exceedence event is expected to increase with r, indicating the importance of the resistance level

FIGURE 1.21 Realization of a continuous random process showing time to first

exceedence (From Melchers, R.E., 1999. Structural Reliability Analysis and Prediction, 2nd ed., Wiley, Chichester With permission.)

Extreme value distribution

Instantaneous distribution

Realization of load

First exceedence time t

Time t

Load Q

Barrier R = r

The other probability density function is an extreme value distribution that refers to the distribution of peaks (e.g., the maximum value recorded in any one year, month, or other time period) The reader is referred to standard textbooks on probability for deeper treatment of extreme values distributions More gener-ally, more than one flux can act on a given barrier or layer with the loading described by a vector process, Q(t) The details are not of concern here, except to note that the problem becomes an outcrossing of two processes: a continuous process and a pulse process (Figure 1.22) Also, the resistance becomes an envelope in the space of Q(t) The first exceedence time is now the time when either the load process or the combined action of the two processes outcrosses the envelope of capacity (i.e., the combined process or any individually crosses from the safe domain to the failure domain) Figure 1.9 shows a particular realization of the load process

The probability that the system fails in a given time period (0, tL) (e.g., the design life) can be stated as the probability that the system will fail when it is first loaded, denoted pf(0, tL), and the probability that it will fail subsequently given that it has not failed earlier This can be expressed as follows:

(1.21) where v is the outcrossing rate The expression is approximate because the second [ ] term is based on the assumption that failure events are rare and that such events therefore can be represented by the Poisson distribution, which leads to the expression shown

If the random load processes are assumed to continue indefinitely and have a stationary statistical nature (e.g., in the simplest case, the means and variances

FIGURE 1.22 Envelope of resistance showing a realization of the vector load process

and an outcrossing by one load component

Load

Load

Time t Envelope of resistance

Outcrossing event

do not change with time), then the rate at which they cross out of the safe domain (i.e., the outcrossing rate) can be estimated from the following:

(1.22)

where X = X(t) is a vector process and ( )+ denotes the positive component only.

The term E(XnX = x) = xn = n(t) · x(t) > represents the outward normal component of the vector process at the domain boundary Note that the integral extends over the safe domain, which is the complement of Equations (1.19) and (1.20)

The result (Equation (1.21) is valid only for rare outcrossings like those that might be associated with failure due to extremely rare, high-load events The result can be extended (i) to allow for gradual deterioration of structural strength with time with the result that pf(0, tL) and v become time dependent, and (ii) approximately for situations with outcrossings which are not rare

Approaches to evaluating expressions pf(0, tL) and v for the expected life and, hence, the reliability of the system are available through Monte Carlo simulation and further simplifications of the problem However, in both cases, Equations (1.21) and (1.22) must allow for the resistance R=R(t) to be a random variable as a function of time or a slow stochastic process (Stewart and Melchers, 1997)

1.4.4.3 Random Resistance

In practice, the actual resistance R=R(t) will not be known precisely Moreover, it will vary from point to point across the barrier, suggesting that actual resistance should be expressed as R=R(x,y,t) and modeled as a random field (Vanmarcke, 1983) However, this represents a difficult problem Consider first the case of a single point in (x,y) space Then R = R(x,y,t) becomes R= R(t) and could be modeled as a random variable, expressed through a probability density function (e.g., as in Figure 1.23) It follows that the line R =r shown in Figure 1.21 is just one realization of many possible outcomes

Usually, the resistance is made up of a number of components or is the outcome of a calculation procedure involving several variables These, too, can be uncertain and may be expressed as random variables or processes The mod-eling of R=R(t), therefore, can be complex The issues involved can be illustrated with a simple example Consider a random variable, S, that is a function of two others, M and A, given by:

(1.23) The probability density function fs(s) of S can be estimated from the corresponding probability density functions for M and A using Equation (1.23) In general, this

v= E Xn = f x dxX

 

 

• +

∫ X x

safe domain

( )

will require numerical integration and, in the case of complex functional rela-tionships, Monte Carlo simulation However, a simpler but approximate approach is to calculate only the first two moments of S using standard expressions, as follows:

(1.24) (1.25) where V = σ/µ is the coefficient of variation Here, µsis the mean (i.e., the expected value) and σ2

s is the variance The latter is a first estimate of the degree of uncertainty associated with S Extending this approach to a function that is more complex than Equation (1.23) leads directly to the so-called error propagation (or second moment) techniques already used in the environmental modeling arena, and earlier in the general system reliability evaluations (Shakshuki et al., 2002) Although these techniques are sometimes termed risk or reliability techniques, they are strictly techniques for estimating uncertainty involved in the algorithm For the case of a random field, R = R(x,y,t), two approaches are possible One is to use random field theory to represent R=R(x,y,t) and to apply Monte Carlo simulation in (x,y) space to estimate the outcrossing rate (Vanmarke, 1983) This is a major computational task for realistic problems A simpler approach is to estimate the probabilistic properties of the weakest failure path using extreme value theory The problem then reverts to the case described above, but now with minimum resistance for a given barrier area, Rmin =Rmin(t), with the associated probability density function (Melchers, 1999)

1.4.4.4 Simplifications of Theory

The theory sketched above can be simplified using a small number of assumptions The main outcome is that the formulation does not address time explicitly

FIGURE 1.23 Schematic probability density function of resistance

Probability

Mean resistance

Resistance

Although this implies significant limitations, the approach has been applied successfully to other problems The first assumption is that resistance remains constant with time, at least for reasonably long periods of time A step-wise approximation could allow for deterioration provided the stationarity remains valid The second assumption is that the load processes are stationary, that is, their statistical properties not change with time As a result, the outcrossing rate, v, is constant with time, meaning that the probability of failure for any given period of time is constant When only one load is primarily of interest, the above assumptions allow that load to be represented by only its probability density function This represents a significant computational simplification When mul-tiple loads act, usually one load dominates a particular failure scenario, allowing the loadings to be combined into load combinations that show the dominance of one load or high correlation between the dominant loads (such that they can be represented by the one random variable) Finally, under the above assumptions, the initial failure probability pf(0, tL) of Equation (1.21) can be subsumed into the random variable representation

Consider now the case with just one load Let this be the extreme value during the lifetime, with associated extreme value distribution fQ( ) (in this case for the maxima) Evidently, the maximum load is applied only once and the probability of failure is, thus, directly related to the probability distribution of the maximum load as follows:

Failure: (1.26)

in the event that the maximum load is applied or

(1.27) where Z is the safety margin It follows that the probability of failure is:

(1.28)

Allowing also for random strength with associated probability density, fR(r), Equation (1.28) becomes:

(1.29)

where FR( ) is the cumulative distribution function for R It is given by:

(1.30) r Q<

Z=r Q− <0

pf r Q Z f x dxQ

r

= < = =

∞ ∫ Prob( ) Prob( < 0) ( )

pf = R Q< = F x f x dxR Q −∞

∞ ∫

Prob( ) ( ) ( )

F rR R r f x dxR

r

( )= ( < =) ( )

−∞∫

Equation (1.29) is known as a convolution integral and can be interpreted loosely as follows Under the integral, the first term, given by Equation (1.30), denotes the probability of failure given that the actual load has the value Q=x The second term is the probability that the load takes the value Q= x This is then integrated over all possible values of x, a dummy variable In general, it is difficult to solve Equation (1.29) in closed form As seen below, an important exception is when R and Q are each represented by a normal distribution or, more generally, are completely described only by their means and variances Equation (1.29) could also have been written as follows:

(1.31) where G( ) is known as the limit state (or performance) function, and X = (R,Q) denotes the random vector of loads and resistances in general Expression G(X) < represents the condition that the bar will fail and was noted earlier

Then, Equation (1.30) can be extended to the case where several performance functions (e.g., Equations (1.19) and (1.20)) are met Equation (1.20), for exam-ple, becomes the following, with X collecting all of the random variables in the problem:

(1.32)

where fx( ) is the joint density function of the random variables, X The solution

of Equation (1.32) is not a simple matter One option is to use Monte Carlo simulation to perform the integration However, in its elementary form, this method is highly inefficient An alternative is to linearize the boundary∪

iGi(X) =

0 of the region of integration (i.e., a first-order approximation) and simplify the form of Equation (1.32) to each random variable being represented only by its first two moments These two simplifications allow the problem of integration to be bypassed altogether, because simple rules can be used for the addition of random variables represented by their first and second moments (i.e., mean and variance) For obvious reasons, this approach is called the first-order second moment (FOSM) method Because of its simplicity, it is widely used

In the FOSM method, the mean and standard deviation of the safety margin (Figure1.24) are, from probability theory rules:

(1.33)

(1.34) with, as before, failure denoted by Z < and survival by Z ≥ (Figure 1.24) Hence, the probability of failure becomes:

pf =Prob(R Q< )=Prob(Z<0)=Prob[ ( )G X <0]

pf f x dxx

G x i i

=

∪ ∫ <

∫… ( )

( )

µZ =µR−µQ

(1.35)

where Φ( ) is the standard normal distribution function with zero mean and unit standard deviation or variance It is extensively tabulated in statistics texts, at least for higher probability levels For low values of probability, more detailed tables are required (Melchers, 1999)

The parameter β is known as the safety or reliability index Evidently, β = µZ/σZ and measures, in the space of the safety margin (Figure 1.24), the distance from the mean of the safety margin to the failure condition in terms of the uncertainty, σZ, of the safety margin Evidently, a greater β implies a lower probability of failure and vice versa (Melchers, 1999)

1.4.4.5 The Multi-Dimensional Case

The above concepts carry over directly to problems involving multiple resistance parameters and multiple loads, but not load processes For load processes, time-dependent theory must be used It is conventional to transform all of the loads to the standard normal space Y (with zero mean, unit variance) The limit state function, which must be linear, is transformed also to g(y) = about the (as yet unknown) design point, y* Where there is dependence between the variables or where they are not normal, the Nataf, Rosenblatt, or some other transformation is required (Melchers, 1999)

Figure1.25 is a sketch of the problem in two-dimensional y space, showing contours of the hill described by the joint probability density function, fY(y), of all the transformed random variables, Y The probability of failure, pf, is represented

FIGURE 1.24 Probability of failure and safety index (From Melchers, R.E., 1999.

Struc-tural Reliability Analysis and Prediction, 2nd ed., Wiley, Chichester With permission.)

Z < Failure

Z > Safety fZ(z)

pf

βσz

z

µz σz

pf R Q Z Z

Z

= − < = < =  −



 = (− Prob( 0) Prob( 0) Φ µ Φ

by the volume under this hill in the failure region, i.e., the region for which g(y) < As before, rather than address pf directly, it is convenient to work with the safety index, β The central statement of the FOSM problem then becomes:

(1.36)

where yi represents the coordinates of any point on the limit state surface, g(y) = This is an optimization problem and can be solved using any appropriate minimization algorithms

Figure1.26 shows the same situation with m (i.e., multiple) limit states In the case of a series system, series bounds can be used to collect together the probabilities estimated separately using Equation (1.36) for individual limit states There are several such series bounds, the simplest (and least accurate) being:

FIGURE 1.25 Space of standard normal variables and linearized limit state function

(From Melchers, R.E., 1999. Structural Reliability Analysis and Prediction, 2nd ed., Wiley, Chichester With permission.)

Failure domain g(y) = Non-linear

g(y) = Linearized

Contours of fy (y)

y1

y2

y*

ν Safe domain

β

β= = 

 ∑=  min(y yT· ) yi

i n

1 2

(1.37)

with pfsystem denoting the probability of failure for the structural system with m limit state functions Bounds are available also for the intersections of two or more limit state functions (Melchers, 1999)

1.4.5 COMPONENTAND SYSTEM FAILURE IN CONTAINING CONTAMINANTS

The proper reliability estimation of a system such as in Figure1.8 is a significant task, recognizing the complex relationships governing the inputs and environ-mental effects Reliability estimates are heavily dependent on the information in the tails of probability distributions, implying that a good understanding of the input and output processes exists and that they can be modeled appropriately Although simplified models can be used, their use affects the quality of the outcomes from a reliability analysis Other alternatives are characterized by some deficiencies: simpler methods of safety or risk assessment, modeling and infor-mation about stochastic processes, and probabilistic variables that hide these

FIGURE 1.26 Series system representation in standard normal space (From Melchers,

R.E., 1999. Structural Reliability Analysis and Prediction, 2nd ed., Wiley, Chichester With permission.)

g1(y) = linearized

g1(y) =

g2(y) =

g3(y) = Contours of fy(y) = Safe domain

y1 y1*

y2

Failure domain Df

0

β1 β2 β3

max i

m

fi f system fi

i m

fi i

p p p p

=

= =

( )≤ ≤ −∏( − )≈

1

1

1

difficulties by ignoring the uncertainties Their application can lead to a false sense of security in future performance estimates

It follows that the simplifications introduced above in reliability theory also should be used with care As indicated, each requires significant assumptions about the system behavior Thus, when predicting the long-term reliability of containment systems, the time-variant approach is the most appropriate

The incompleteness of data and probabilistic models need not, however, be an insurmountable obstacle in the application of reliability theory As discussed briefly below, experience gained in-service through monitoring and observation can be used to refine understanding of the system and characteristics of the processes involved The tools to monitor and observe are Bayesian updating and system dynamics analysis, both of which are increasingly recognized as powerful tools in geoenvironmental engineering

A second and equally important aspect of the Bayesian method is the use of subjective probabilities in reliability analysis and elsewhere The probability distributions and the stochastic process representations in the exposition above were assumed to be completely known Classic statistical literature assumes that these can be inferred from observation of sufficient experiments, as in the use of monitoring data The issue of modeling vs monitoring with respect to contain-ment system performance assesscontain-ment has drawn attention from many researchers and practicing personnel Inyang (2003) provided an assessment of the general utility of monitoring and modeling in environmental assessments Following the issues discussed by Melchers (1999), it is not surprising that engineers and some applied scientists have taken a more pragmatic approach and assumed that even nonfrequent subjective information obtained from less formal observation and experience can be applied in probability theory and, hence, reliability theory Subjective information can, as Bayes implied, be used as prior information and refined as more data become available (see below) This is an important point with respect to applying probabilistic methods for the long-term performance analysis of waste containment systems In essence, there is convergence in utility between probabilistic analysis and the use of monitoring data, but monitoring data alone are insufficient as the bases for predictions and system management

1.4.6 RELATING PROBABLE CONTAMINANT CONCENTRATIONS TO RISKS

(1.38)

where Co is the source term concentration per unit width of the plume for contaminant fate and transport modeling (M/L2), q is the contaminant flow rate

per unit width in the vadose zone (L2/T), Ca is contaminant concentration in the

vadose zone (M/L3), d is the width of the saturated zone (L), Vh is the plume

velocity (L/T), and b is the plume thickness (L) Obviously, parameters Ca and q are determined mostly by containment system performance

Deterministic equations that express the flow rate and quantities of contam-inants through the components of a containment system (such as those presented in Chapter2) can be used to estimate the magnitude of these parameters With respect to the probabilistic analysis presented below, Ca and q can be computed as quantities with specified probabilities of occurrence for use in source term estimates in ecological and human health risk assessments

If the failure is specified in terms of the maximum allowable magnitudes of Ca,q, and even Co, then the probability of exceeding the specified magnitude of any of these three parameters can be used to determine failure In the latter case, if the probability of exceedence of a given magnitude is high enough (value to

FIGURE 1.27 An illustration of the influence of barrier damage on contaminant source

term for fate and transport modeling and risk assessments (From Reddi, L.N and Inyang, H.I., 2000 Geoenvironmental Engineering, Principles and Applications, 1st ed., Marcel Dekker, New York With permission.)

Surface impoundment

Liner

Unsaturated geomedia

Saturated

geomedia b

Bedrock

d

q, Ca

Leachate plume

Vh, Co

Water table Long term flow

channels α Dt

Ground surface

C C qd V b

o a

be specified), then the system would be considered to have failed Typically, contaminant migration models (i.e., fate and transport models) contain Co instead of Ca, thus eliminating the essential parameters that would enable a more complete appreciation of the decay, growth, or constancy of the source term with extended time In some cases where the source (e.g., the impoundment in Figure1.27) has been eliminated, focus on Co instead of Ca may be justified

The parameters Ca and Co can be quantitatively linked to exposure assessment equations that are usually incorporated into quantitative human health and eco-logical risk assessment frameworks An example of such a framework is the total risk integrated methodology (TRIM) described by USEPA (1999) and illustrated in Figure 1.28 The exposure-event function is an expression of the micro-environmental exposure of an individual or cohort to a contaminant in an exposure medium during a time step, t As shown in Equation (1.39), the exposure event function is the product of exposure concentration and exposure duration It should be noted that Equation (1.39) contains the parameter Cm, which relates to Ca and Co of Equation (1.38) Essentially, the critical utility of contaminant subsurface fate and transport models is to establish the quantitative linkage between Ca,Co, and Cm Usually, Ca and Co > Cm due to travel path attenuation factors, except for

FIGURE 1.28 An exposure-event simulation framework for the TRIM (From USEPA,

1999 Technical Support Document EPA-453/D-99-001 Research Triangle Park, North Carolina, pp 4.1–6.8.)

Ambient media concentrations Ci, air (t) Ci, water (t) Ci, soil (t) etc

Time step Averaging

time Time scale

matching

Uncertainty variability

Exposure-event function Inter media transfer

IFT (j, s –> m, k, t)

Contact medium Air Water

Food Soil

etc

Ambient zone

Microenvironment Activity

Time

Cm (i, k, l, t) ETz, m (i, k, l, t) E

unusual situations in which the contaminants accumulate at the sink where Cm is measured

The fate processes and transport rates of each contaminant that is released at initial but time-variable concentration, Ca, into the surrounding geomedia are affected by the homogeneity, isotropy, and continuity of the geomedium In some cases, as modeled by Inyang et al (2000a), the released contaminant can be transported away quickly because of the high permeability or high diffusion coefficient of the contaminant in the geomedium This transport would set up a high concentration gradient between the barrier and the bounding surface of the geomedium such that faster rates of contaminant release into the geomedium would result within the limits of the contaminant concentration for the contain-ment system

As discussed by Rowe and Fraser (1995), an impact assessment of the waste disposal facilities that comprise barrier systems is characterized by several uncer-tainties, some of which derive from hydrogeological factors Upon release from a containment system into the subsurface, the transport of a contaminant to an environmental sink can be retarded or accelerated by physico-chemical processes and travel path hydrogeology At adequate concentration and favorable pH-Eh condition, substances can precipitate out of pore solution, thereby blocking some pores and retarding contaminant transport to a sink Shu et al (2000) provided a phenomenological and quantitative description of pore plugging processes that can result from mineral substance precipitation Furthermore, Mazurek et al (1996) experimentally investigated and confirmed redox front entrapment in clay shales that resulted from contrasting solubilities of reduced and oxidized species Coupled with mineral sorption on pore walls and co-precipitation of secondary mineral phases that were produced, the contaminant was immobilized in the clay shale In fractured systems, contaminants can travel at high rates relative to values computed for an intact medium In some cases, the contaminants can travel as colloids at relatively fast rates from source to sink (Roy and Dzombak, 1997) Several investigators have developed and demonstrated techniques for character-izing textural characteristics of geomedia (Malone et al., 1986; Shi et al., 1999a,b) The characterization of fluid flow channels in geomedia is not always possible at the application scale of transport models such as those analyzed by Hathorn (1993) and Inyang et al (2000b)

(1.39) where Ez,m is the exposure experienced by person, z, from exposure medium, m, during time step, t, given that person z is in exposure district i in microenvironment k conducting activity i during that time step t. For example, the exposure in air might be measured in units of mg-h/m3 Note that the exposure time does not

need to be a whole time step; Cm is the concentration in exposure medium, m (e.g., air, water, soil), in exposure district, i, in microenvironment, k, associated with activity, l, during time step, t The units of measurement for air might be mg/m3, while the units of measurement for food might be mg/kg; ETz,m is the

exposure duration of individual or cohort, z, to exposure medium, m, in exposure district, i, in microenvironment, k, conducting activity, l, during time step, t; z is the individual or cohort; m is the exposure medium contacted (i.e., air, water, food); i is the exposure district; k is the microenvironment in which the exposure occurs (e.g., indoors at home, in a vehicle, indoors at work); and l is the activity

FIGURE 1.29 An example of an exposure or a potential dose profile and associated

measures, where B is the integrated exposure from time t = a to t = b; p is the time between peaks over x; and R is the respites between exceedences of x (From USEPA, 1999 TRIM Technical Support Document EPA-453/D-99-001 Research Triangle Park, North Carolina, pp 4.1–6.8; adapted from McCurdy, T.R., 1994 In McKee, D.J (Ed.), Tropospheric

Ozone: Human Health and Agricultural Impacts, Lewis, Ann Arbor, MI, pp 85–127 With permission.)

Conc

en

tra

tion

C(t) x

ci

P

B

R

Time t = a ti–1 ti t = b

∆ti

code that describes what the individual is doing at the time of exposure (e.g., resting, working, preparing food, cleaning, eating)

1.5 USE OF BARRIER DAMAGE AND PERFORMANCE MODELS FOR TEMPORAL SCALING OF

MONITORING AND MAINTENANCE NEEDS

Monitoring the state of the system provides some level of information about its operation For the most beneficial results, objectives should be set and the design of a monitoring program should be addressed prior to system construction System monitoring also requires an understanding of the manner in which results might be used in a system reliability context, as described in this section It should be noted, however, that the implementation of monitoring technologies and actual interpretation of the monitoring observations require appropriate expertise and ogies were also analyzed by Inyang et al (1995)

1.5.1 UPDATING

The facility risk analysis involves assumptions about the probability distributions for random variables, including means (expected values) and variations about the mean Monitoring can provide data to allow these initial assumptions to be modified The standard procedure in probability theory is to use the Bayes the-orem because it allows incorporation of experimental observations in an existing probability density function, so-called Bayesian updating This method can allow for imperfections in the data obtained from monitoring (e.g., known uncertainties introduced by instruments)

The concept can be illustrated quite simply (Figure 1.30) Consider some resistance property, R, which might, for example, represent some characteristic permeability Let represent the original (apriori) partial differential equa-tion of R New data collected from monitoring usually will not fit wholly in the original partial differential equation Let fv( ) represent the new data, here taken

for simplicity as a continuous partial differential equation Both are shown on Figure 1.30, together with the updated (posteriori) pdf

It should be clear that if the data have a lot of scatter [i.e., if fv( ) has a large

variance], the data not contain much useful information and little to help in refining the original partial differential equation Conversely, if the data have very little scatter, it is highly informative and will have a significant influence Simi-larly, if there is little understanding of the variable being considered, its apriori distribution, fR′( ), is highly uncertain and can be said to be noninformative The posteriori distribution will then be considerably influenced by the additional data Typically, the new evidence is monitoring data, which can be represented as a likelihood function, L (E/λ) This likelihood function is the conditional proba-bility of observing the set of observation outcomes, E, given that the value of the parameter about which there is uncertainty (e.g., the mean of a random variable)

′′

fR( )

′′

fR( )

technol-is λ According to Bayes theorem, the posterior probability distribution is then given by:

(1.40)

For example, if the observation data are the number of particular values above a given level during a given time interval or as a proportion of all readings, then the likelihood function is described by the Poisson distribution and is given by:

(1.41)

Typical prior, likelihood, and posterior distributions are available in standard texts The posterior distribution is usually highly dependent on the selection of the prior distribution Hence, care in selecting the original probability distributions is reflected in subsequent analysis, even after updating using observations

1.5.2 EFFECT OF UPDATING ON SYSTEM MANAGEMENT

Updating provides a better estimate of the random variables that influence the behavior of the system and, hence, the predicted probabilities of failure that should

FIGURE 1.30 Known (a priori) pdf for resistance fR′(r) as modified by new information

fV( ) and modified (posteriori) pdf fR″( ) (From Melchers, R.E., 1999. Structural Reliability

Analysis and Prediction, 2nd ed., Wiley, Chichester With permission.)

Posterior f ''R

Prio f 'R

Likelihood fV

f f L E

f L E d

"

'

'

( | ) ( | )

λ λ λ

λ λ λ

( )= ( )

( )

∞ ∫

0

L E T T

n

L L

n

( | ) exp !

enable the risk assessment to be updated as more appropriate monitoring infor-mation becomes available A simple example is sketched in Figure 1.31, which indicates the improved life expectancy of a system as a result of favorable observations leading to an improved estimate of system reliability, shown at the inspection time point

1.6 LIFE-CYCLE DECISION APPROACH AND MANAGEMENT

It is desirable to optimize the management of containment systems over time Examples might include minimizing the total expected costs, minimizing envi-ronmental impact, or reducing the likelihood of regulatory breaches and fines to a minimum With this approach, the possibility exists to optimize the times between discrete monitoring or the intensity of monitoring, if continuous, with respect to the desired objective (Figure1.32)

When contamination is left on-site in engineered containment systems, mea-sures must be taken to ensure that humans and the environment will continue to be protected from harmful exposures Monitoring, maintenance, institutional con-trols, and other stewardship activities may be needed for very long periods of time (hundreds to thousands of years), depending on the times over which the contaminants retain their hazardous characteristics A report by the National Academies (National Research Council, 2000) suggests that all engineered con-tainment systems, if left unattended, will eventually “fall” and recommends planning for fallibility

An ability to forecast system performance is needed for a number of reasons, not the least of which is the need to have a better understanding of the actual stewardship requirements associated with containment systems This knowledge

FIGURE 1.31 Schematic variation of reliability showing effect of an observation at

“inspection time point.”

Time Reliability

Design reliability

Acceptable reliability

can then be factored into remedial technology evaluation and remediation deci-sion-making Analytical forecasting tools that enable quantification of the likeli-hoods of events that could lead to failure, their potential consequences, and the associated response costs are needed so that more informed decisions can be made concerning the resources that will be needed and more effective stewardship planning can occur (Clarke et al., 2002; DOE, 2002) To achieve this requires the prediction of future environmental conditions and corresponding system responses Here, the use of natural analogue models is emerging as a valuable tool (Waugh et al., 1994)

Although there is a lack of performance data that can enable prediction verification at this point in time, the use of probabilistic approaches and scenario analyses can be helpful in determining the sensitivity of containment systems to specific events (Sanchez et al., 2002) With time, the ability to predict future performance to a reasonable period (a few decades) will increase as the knowledge base increases Improvement will occur through a combination of analytical models, natural analogs, and performance monitoring

REFERENCES

Arthur, W.J., III (1982) Radionuclide concentration sin vegetation at a solid radioactive waste disposal area in southeastern Idaho Journal of Environmental Quality, 11, 394–399

Arthur, W.J., III and Markham, O.D (1983) Small mammal soil burrowing as a radionu-clide transport vector at a radioactive waste disposal area in southeastern Idaho

Journal of Environmental Quality, 12, 117–122

Bai, M and Inyang, H.I (1999) Determination of contaminant migration in a barrier zone Proceedings of the International Symposium on Coupled Phenomena in Civil, Mining and Petroleum Engineering, Sanya, Hainan Island, P.R China, pp 1–15

FIGURE 1.32 Life-cycle reliability and assessment, showing effect of appropriate repairs

(schematic) (From Melchers, R.E., 1999. Structural Reliability Analysis and Prediction, 2nd ed., Wiley, Chichester With permission.)

Expected deterioration Acceptable risk level Risk of failure

fractured media ASCE Journal of Environmental Engineering, 122(5), 416–423 Bai, M., Elsworth, D Inyang, H.I and Roegiers, J.C (1997) Modeling contaminant migration with linear sorption in strongly heterogeneous media ASCE Journal of

Environmental Engineering, 123(11), 1116–1124

Bai, M., Inyang, H.I., Chien, C.C and Bruell, C (2000a) Factors for assessing flow and transport in fractured porous media.(Chapter 2) In Inyang, H.I and Bruell, C.J (Eds.), Remediation in Rock Masses, American Society of Civil Engineers, Wash-ington, DC, pp 12–27

Bai, M., Inyang, H.I., Meng, F., Abousleiman, Y and Roegiers, J.-C (2000b) Statistical modeling of particle migration in gravel packs Proceedings of the 4th Interna-tional Symposium on Environmental Geotechnology and Global Sustainable Development, Boston (Danvers), MA, pp 593–602

Benoit, G.R (1973) Effect of freeze-thaw cycles on aggregate stability and hydraulic conductivity of three soil aggregate sizes Soil Science Society of America Pro-ceedings, 37, 3–5

Benson, C.H., Abichou, T.H., Olson, M.A and Bosscher, P.J (1995) Winter effects on hydraulic conductivity of compacted clay Journal of Geotechnical Engineering,

ASCE, 121(1), 69–79

Bernabe, Y (1995) The transport properties of networks of cracks and pores Journal of

Geophysical Research, 100(B3), 4231–4241

Blackmore, A.V and Miller, R.D (1961) Tactoid size and osmotic swelling in calcium montmorillonite Proceedings of the Soil Science Society, 169–173

Bogardi, I., Kelly, W.E and Bardossy, A (1989) Reliability model for soil liners initial design Journal of Geotechnical Engineering, 115(5), 658–669

Bowerman, A.G and Redente, E.F (1998) Biointrusion of protective barriers at hazardous waste sites Journal of Environmental Quality, 27, 625–632

Cadwell, L.L., Eberhardt, L.E and Simmons, M.A (1989) Animal Intrusion Studies for Protective Barriers: Status Report for FY 1988 Pacific Northwest Laboratory, Richland, Washington

Cepic, M and Mavko, B (2002) A dynamic fault tree Reliability Engineering A: System Safety, 75(1), 83–91

Chamberlain, E.J and Ayorinde, O.A (1991) Freeze–thaw effects on clay covers and liners Proceedings of the 6th International Specialty Conference on Cold Regions Engineering, Hanover, NH, pp 136–151

Chamberlain, E.J and Gow, A.J (1978) Effect of freezing and thawing on the permeability and structure of soils International Symposium on Ground Freezing, Bochum, Germany, pp 31–43

Chen, J., Anandarajah, A and Inyang, H (2000) Compressibility of contaminated clay Proceedings of the 4th International Symposium on Environmental Geotechnology and Global Sustainable Development, Vol 2, Boston, MA, pp 1226–1235 Clarke, J.H., MacDonnell, M.M., Waugh, W.J., Smith, E D, Dunn, R J, Water, R.D and

Burns, D.E (2002) Engineered containment and control systems: nurturing nature Annual Meeting of the Society for Risk Analysis, December 2002 Daneshjoo, F and Hushmand, B (1999) Response properties of landfills during seismic

Daniels, J.L., Inyang, H.I., Howe, G.S and Ogunro, V.O (1999) Innovative use of triaxial cell permeameters for measuring the earthen barrier resistance to freeze-thaw degradation under field conditions Proceedings of the International Symposium on High Altitude & Sensitive Ecological Environmental Geotechnology, Nanjing, PR China, pp 326–333

Daniels, J.L., Inyang, H.I and Fang, H.Y (2001) Theoretical relationship for scaling freezing-induced damage in barrier systems under laboratory and field conditions

Proceedings of International Symposium on Application of Natural Materials for Environmental Geotechnology, Tokyo, Japan, pp 37–44

Dasgupta, P (1993) An Inquiry into Well-Being and Destitution, Clarendon Press, Oxford Dasog, G.S., Acton, D.F., Mermut, A.R and De Jong, E (1988) Shrink-swell potential and cracking in clay soils of Saskatchewan Canadian Journal of Soil Science, 68, 251–260

DOE (2000) Alternative Landfill Cover, Innovative Technology Summary Report, DOE/EM-0558, Office of Environmental Management, Office of Science and Technology

DOE (2002) Science and Technology Roadmap for Long-Term Stewardship, Idaho National Engineering and Environmental Laboratory

Donnegan, J.A and Rebertus, A.J (1998) Rates and mechanisms of subalpine forest succession along an environmental gradient Ecology, 80, 1370–1384

Einarsson, S and Rausand, M (1998) An approach to vulnerability analysis of complex industrial systems Risk and Analysis, 18, No

Elias, V., DiMaggio, J.A and DiMillio, A (1997) FWHA technical note on the degrada-tion-reduction factors for geosynthetics Geotechnical Fabrics Report, 24–26 Fernandez, F and Quigley, R.M (1991) Controlling the destructive effects of clay-organic

liquid interactions by application of effective stress Canadian Geotechnical

Jour-nal, 28, 338–398

Fleming, L.N and Inyang, H.I (1995) Permeability of clay-modified fly ash under thermal gradients Journal of Materials in Civil Engineering, 7(3), 178–182

Forrester, J.W (1961) Industrial Dynamics MIT Press, Cambridge, MA; currently avail-able from Pegasus Communications, Waltham, MA

Gallegos, D.P., Brown, T.J and Cochran, J.R (1998) A framework for treating uncertainty to facilitate waste disposal decision making — application of the approach to GCD performance assessment Proceedings from the Topical Meeting on

Risk-based Performance Assessment and Decision Making, Richland/Pasco, Washing-ton, pp 229–241

Gee, G.W and Ward, A.L (1997) “Still in Quest of the Perfect Cap,” Landfill Capping in the Semi-Arid West: Problems, Perspectives, and Solutions, Conference Pro-ceedings, Environmental Science and Research Foundation Report ESRF-019, May 1997, pp 145–164

Gilbert, R.B and Tang, W.H (1995) Reliability-based design for waste containment systems Proceedings of Special Conference Characterization, Containment,

Reme-diation, and Performance in Environmental Geotechnics, 2000, New Orleans, LA, pp 499–513

of Hazardous Waste: Engineering and Environmental Issues, Halsted Press, pp 245–258

Hathorn, W.E (1993) A review of fluid movement in compacted clay: the case of macropore flow Proceedings of the ASCE Hydraulics Division Conference on Engineering Hydrology, San Francisco, CA, pp 473–478

Ho, C.K., Arnold, B.W., Cochran, J.R., Tiara, R.Y and Webb, S.W (2001a) Development of a Risk-Based Performance-Assessment Method for Long-Term Cover Systems-Application to the Monticello Mill Tailings Repository, Sandia National Labora-tory report SAND2001-3032, October 2001

Ho, C.K., Webb, S.W and Arnold, B.W (2001b) Stochastic Simulations for Risk-Based Performance Assessments of Long-Term Cover Systems, Paper 144, 2001 Inter-national Conference & Remediation Technology Conference and Exhibition, Department of Energy Report DOE/EM-0620, Orlando Florida, 10–13 June 2001 Ho, C.K., Arnold, B.W., Cochran, J.R and Tiara, R.Y (2002a) Risk-Based Performance Assessment of Long-Term Cover Designs for Waste Isolation and Disposal at DOE Facilities, Project AL21SS22, DOE Environmental Management Subsurface Contaminants Focus Area (SCFA) Mid-Year Review, March 2002

Ho, C.K., Arnold, B.W., Cochran, J.R and Tiara, R.Y (2002b) Risk-Based Performance Assessment for Long-Term Cover Systems: Sensitivity and Uncertainty Analyses, Spectrum 2002, Reno, Nevada, August 2002

Hood, C.C., Jones, D.K.C., Pidgeon, N.F., Turner, B.A., Gibson, R., Bevan-Davies, C., Funtowicz, S.O., Horlick-Jones, T., McDermid, J.A., Penning-Rowsell, E.C., Ravetz, J.R., Sime, J.D and Wells, C (1992) Risk Management, Risk: Analysis,

Perception and Management, The Royal Society, London, pp 135–192 INEEL (2000) Subsurface Science Initiative Strategic Plan, Idaho National Engineering and

Environmental Laboratory report INEEL/EXT-2000–01094, August 2000, Rev Inyang, H.I (1992) Aspects of landfill design for stability in seismic zones Journal of

Environmental Systems, 21(3), 223–235

Inyang, H.I (1994) A Weibull-based reliability analysis of waste containment systems Proceedings of the First International Congress on Environmental Geotechnics, Calgary, Canada, 273–278

Inyang, H.I (1999) Balancing prescriptive and performance-based design specifications for waste containment systems Journal of Environmental Engineering, June, 491 Inyang, H.I (2003) On models, modeling and modelers in the environmental arena

Journal of Environmental Engineering, 129(6), 489–490

Inyang, H.I and Tomassoni, G (1992) Indexing of long-term effectiveness of waste containment systems for a regulatory impact analysis A Technical Guidance Document, Office of Solid Waste, U.S Environmental Protection Agency, Wash-ington, DC, p 29

Inyang, H.I., Betsill, J.D., Breeden, R., Chamberlain, G.H., Dutta, S., Everett, L., Fuentes, R., Hendrickson, J., Koutsandreas, J., Lesmes, D., Loomis, G., Mangion, S.M., Morgan, D., Pfeifer, C., Puls, R.W., Stamnes, R.L., Vandel, T.D and Williams, C (1995) Performance monitoring and evaluation Proceedings of the Assessment of Barrier Containment Technologies, International Containment Technology Workshop, Baltimore, MD, pp 355–400

Inyang, H.I., Parikh, J.M and Iskandar, A (1998a) Waste containment barrier In Encyclo-pedia of Environmental Analysis and Remediation, Wiley, New York, pp 5131–5141 Inyang, H.I., Daniels, J.L and Chien, C.C (2000a) Methods of analysis of contaminant migration in barrier materials In Remediation Engineering of Contaminated Soils, edited by D.L Wise, D.J Trantolo, E.J Cichon, H.I Inyang and U Stottmeister, Marcel Dekker, New York, pp 63–82

Inyang, H.I., Shu, Z., Bai, M and Roegiers, J.C (2000b) Prediction of pollutant migration in landfill sites Proceedings 4th International Symposium on Environmental

Geo-technology and Global Sustainable Development, Danvers, MA, pp 1487–1495 Jayawickrama, P.W and Lytton, R.L (1992) Conductivity through macropores in com-pacted clay Proceedings of the 7th International Conference on Expansive Soils, pp 99–104

Kachanov, L.M (1986) Introduction to Continuum Damage Mechanics, Martinus Nijhoff, Boston

Kavazanjian, E., Jr and Matasovic, N (2001) Seismic design of mixed and hazardous waste landfills, state of the art paper no 11, Proceedings of the Fourth International Conference on Recent Advances in Geotechnical Earthquake Engineering (on CD ROM), University of Missouri, Rolla, San Diego, California

Khalil, M.A.K and Moraes, F.P (1997) Linear least squares method for time series analysis with an application to a methane time series Journal of the Air and Waste

Management Association, 45, 62–74

Koerner, R.M and Daniel, D.E (1992) Better cover-ups Civil Engineering, 55–57 Kozak, M.W., Chu, M.S.Y and Mattingly, P.A (1990) A performance assessment

meth-odology for low-level waste facilities U.S Nuclear Regulatory Commission, NUREG/CR-5532, Washington, DC

Lake, C.B and Rowe, R.K (2000) Diffusion of sodium and chloride through geosynthetic clay liners Geotextiles and Geomembranes, 18, 103–131

Layard, P.R.G (1972) Cost-Benefit Analysis: Selected Readings Penguin, Harmondsworth Link, S.O., Cadwell, L.L., Petersen, K.L., Sackschewsky, M.R and Landeen, D.S (1995) The Role of Plants and Animals in Isolation Barriers at Hanford, Washington, PNL-10788, Pacific Northwest National Laboratory, Richland, Washington, 1995 Liu, C.N and Gilbert, R.B (2002) Distribution of seepage-induced fluid pressure in landfill covers slopes Proceedings of the Sixth International Symposium on Environmental Geotechnology and Global Sustainable Development, Seoul, Korea, pp 287–294 Lord, A.E and Koerner, R.M (1984) Fundamental aspects of chemical degradation of geomembranes Proceedings of the International Conference on Geomembranes, Denver, CO, pp 293–298

Malone, P.G., May, J.H., Brown, K.W and Thomas, J.C (1986) Use of x-ray radiographic methods in the study of clay liners Proceedings of the National Conference on Hazardous Wastes and Hazardous Materials, Atlanta, GA, pp 199–201 Matasovic, N., Kavazanjian, E., Jr., Augello, A.J., Bray, J.D and Seed, R.B (1995) Solid

waste landfill damage caused by 17 January 1994 Northridge earthquake The Northridge, California, Earthquake of 17 January 1994: California Department of Conservation, Division of Mines and Geology Special Publication 116, Sacra-mento, California, pp 221–229

Ozone: Human Health and Agricultural Impacts, Lewis, Ann Arbor, pp 85–127 Melchers, R.E (1999) Structural Reliability Analysis and Prediction (2nd ed.), Wiley,

Chichester

Melchers, R.E (2001) On the ALARP approach to risk management Reliability Engi-neering and System Safety, 71(2), 201–208

Merry, S.M and Bray, J.D (1997) Time-dependent mechanical response of HDPE geomembranes Journal of Geotechnical and Geoenvironmental Engineering, 123(1), 57–65

Meyer, P.D and Gee, G.W (1999) Information on Hydrologic Conceptual Models, Param-eters, Uncertainty Analyses, and Data Sources for Dose Assessments at Decom-missioning Sites, Nuclear Regulatory Commission report NUREG/CR-6656 or Pacific Northwest National Laboratory report PNNL-13091, November 1999 Meyer, P.D and Orr, S (2002) Evaluation of Hydrologic Uncertainty Assessments for

Decommissioning Sites Using Complex and Simplified Models, Nuclear Regula-tory Commission report NUREG/CR-6767 or Pacific Northwest National Labo-ratory report PNNL-13832, March 2002

Meyer, P.D and Taira, R.Y (2001) Hydrologic Uncertainty Assessment for Decommis-sioning Sites: Hypothetical Test Case Applications, Nuclear Regulatory Commis-sion report NUREG/CR-6695 or Pacific Northwest National Laboratory report PNNL-13375, January 2001

Miller, C.J and Lee, J.Y (1992) Effects of freeze-thaw on clay cover liner performance Proceedings of the Mediterranean Conference on Environmental Geotechnology, Cesme, Turkey, pp 205–211

Mohamed, A.O., Young, R.N., Caporuscio, F., Yanful, E.K and Bienvenu, L (1993) Chemical interaction and cyclic freeze-thaw effects on the integrity of the soil cover for waste amulet tailings Proceedings of the Joint CSCE-ASCE National Conference on Environmental Engineering, Montreal, Canada, pp 259–272 Moore, B., Ditmars, J., Minsker, B and Bachmaier, J (2001) Performance Evaluation

Using Expert Elicitation and Long Term Stewardship Monitoring Optimization for Long Term Stewardship, Paper 470, 2001 International Conference & Reme-diation Technology Conference and Exhibition, Department of Energy report DOE/EM-0620, Orlando, Florida, 10–13 June 2001

Moo-Young, H.K and Zimmie, J.F (1997) Waste minimization and reuse of paper mill sludges and landfill covers: a case of study Journal of Waste Management and Research, 15(6), 593–605

National Research Council (2000) Long-Term Institutional Management of U.S Depart-ment of Energy Legacy Waste Sites, National Academy Press, Washington, DC Nazarali, A.M., Rogers, P.M and Haass, C.C (1998) Long-term human health impact

assessment as a decision making tool in a complex system for environmental restoration and waste management Proceedings of Topical Meeting on Risk-Based Performance Assessment and Decision Making, Richland/Pasco, Washington, April 5–8, pp 143–150

Nyhan, J.W (1989) Development of technology for the long-term stabilization and closure of shallow land burial sites in semiarid environments LA-1128-MS Los Alamos National Laboratory, Los Alamos, NM

Piet, S.J., Dakins, M.E., Gibson, P.L., Joe, J.C., Kerr, T.A and Nitschke, R.L (2001) A Framework for Making Sustainable Cleanup Decisions Using the KONVER-GENCE Model, INEEL technical report INEEL/EXT-2001–01485

Reddi, L.N and Inyang, H.I (2000) Geoenvironmental Engineering, Principles and

Applications, 1st ed., Marcel Dekker, New York, 494 pp

Rowe, R.K and Fraser, M.J (1995) Effect of uncertainty in the assessment of the potential impact of waste disposal facilities Proceedings of Geoenvironment 2000, the ASCE Geotechnical Division Special Conference, New Orleans, LA, pp 270–284 Roy, S.B and Dzombak, D.A (1997) Chemical factors influencing colloid-facilitated transport of contaminants in porous media Environmental Science and

Technol-ogy, 31, 656–664

Sanchez, F., Clarke, J.H and Parker, F.L (2002) Evaluating requirements for stewardship of contaminated facilities Proceedings of the Waste Management 2002 Conference, Tucson, AZ

Schaff, J.R and Davidson, B.D (1997) A strength-based wearout model for predicting the life of composite structures Composite Materials: Fatigue and Fracture (Sixth Volume) ASTM STP 1285, pp 179–200

Shackelford, C.D (1992) Performance-based design of soil liners Proceedings of the Mediterranean Conference on Environmental Geotechnology, Cesme, Turkey, pp 145–153

Shakshuki, E.K., Ponnambalam, K and Wadaj, T (2002) Risk assessment in environ-mental systems Canadian Journal of Civil Engineering, 29(1), 1–7

Shi, B., Murakami, Y., Wu, Z., Chen, J and Inyang, H.I (1999a) Monitoring internal failure evolution in soils using computerized x-ray tomography (CT) Engineering

Geology, 54 (3–4), 321–328

Shi, B., Wu, Z., Inyang, H.I., Chen, J and Wang, B (1999b) Preparation of soil specimens for SEM analysis using freeze-cut-drying Bulletin of the International Association of Engineering Geology and the Environment, 58, 1–7

Shu, Z., Bai, M., Roegiers, J-C and Inyang, H.I (2000) Modeling selective plugging processes by induced precipitation for enhancing oil recovery Proceedings of the 4th International Symposium on Environmental Geotechnology and Global Sus-tainable Development, Boston (Danvers), MA, pp 309–317

Siu, N (1994) Risk assessment for dynamic systems: and overview Reliability Engineer-ing and System Safety, 43, pp 43–73

Smith, E.D., Luxmore, R.J andSuter, G.W (1997) Natural physical and biological pro-cesses compromise the long-term performance of compacted soil caps In: Barrier

Technologies for Environmental Management, Summary of a Workshop, National Research Council, Washington DC, pp D61–D70

STELLA (2001) Stella Research 7.0, developed by High Performance Systems, Inc., Stewart, M.G and Melchers R.E (1997) Probabilistic Risk Assessment of Engineering

Systems, Chapman & Hall, London

Suter, G.W., Luxmoore, R.J and Smith, E.D (1993) Compacted soil barriers at abandoned landfills will fail in the long term, Journal of Environmental Quality, 22, 217–226 UMTRA 1992 Vegetation growh patterns on six rock covered UMTRA project disposal

cells UMTRA Project Rep UMTRA-DOE/AL-400677.0000 Hanover, New Hampshire, 2001 (www.hps-inc.com)

EPA-453/D-99–001 Office of Air Quality Planning and Standards, United States Environmental Planning Agency, Research Triangle Park, North Carolina, pp 4.1–6.8

USNRC (2000) A performance assessment methodology for low-level radioactive waste disposal facilities Recommendations of NRC’s Performance Assessment Working Group NUREG-1573 United States Nuclear Regulatory Commission, Office of Nuclear Material safety and Safeguards, Washington, DC

Vanmarcke, E (1983) Random Fields: Analysis and Synthesis, MIT Press, p 382 Waugh, W.J (2001) Uranium Mill Tailings Covers: Evaluating Long-Term Performance,

Paper 244, 2001 International Conference & Remediation Technology Conference and Exhibition, Department of Energy report DOE/EM-0620, Orlando Florida, 10–13 June 2001

Waugh, W.J (2002) Characterization of the Environmental Envelope for the Design of Long-Term Covers, Project ID02SS21, DOE Environmental Management Sub-surface Contaminants Focus Area (SCFA) Mid-Year Review, March, 2002 Waugh, W.J., Peterson, K.L., Link S.O., Bjorstad, B.N and Gee, G.W (1994) Natural

analogs of the long-term performance of engineered covers In: Gee, G.W and Wing, N.R (Eds.), In-Situ Remediation: Scientific Basis for Current and Future Technologies, Battle Press

71

2 Modeling of Fluid

Transport through Barriers

Prepared by* Brent E Sleep

University of Toronto, Toronto, Canada Charles D Shackelford

Colorado State University, Fort Collins, Colorado Jack C Parker

Oak Ridge National Laboratory, Oak Ridge, Tennessee

2.1 OVERVIEW

As understanding of the mechanisms of contaminant transport through barrier tive approach to a performance design approach It is expected that reliance on models for predictive-based design will increase in the future, as the need for predicting long-term barrier system performance increases This chapter details the mechanisms and models for predicting the performance of components of passive barriers such as caps, permeable reactive barriers (PRBs), and walls and floors The relevant regulatory drivers and current state of practice are summa-modeling while this chapter focuses on the performance of components that constitute containment systems

* With contributions by Calvin C Chien, DuPont, Wilmington, Delaware; Thomas O Early, Oak Ridge National Laboratory, Oak Ridge, Tennessee; Clifford K Ho, Sandia National Laboratories, Albuquerque, New Mexico; Richard C Landis, DuPont, Wilmington, Delaware; Alyssa Lanier, University of Wisconsin, Madison, Wisconsin; Michael A Malusis, GeoTrans, Inc., Westminster, Colorado; Mario Manassero, Politecnico I, Torino, Italy; Greg P Newman, Geo-Slope International Ltd., Calgary, Canada; Robert W Puls, U.S Environmental Protection Agency, Ada, Oklahoma; Terrence M Sullivan, Brookhaven National Laboratory, Upton, New York

4040_book.fm Page 71 Wednesday, September 14, 2005 12:43 PM

prescrip-72 Barrier Systems for Environmental Contaminant Containment & Treatment

2.2 CAPS

2.2.1 FEATURES, EVENTS, AND PROCESSES AFFECTING

PERFORMANCEOF CAPS

Covers and caps are engineered structures that must perform within a larger dynamic natural system and, as such, must be designed with consideration of natural system influences Understanding these physical processes and applying appropriate numerical analyses to these processes can help the engineer to build an appropriate overall system that will perform with the desired objective The primary processes acting on a cap are described in the subsections below

2.2.1.1 Hydrologic Cycle

The purpose of a cap is usually to minimize water infiltration into underlying waste, and sometimes to minimize gas transport to the atmosphere As shown in the cap slope, cap soil properties, cap moisture conditions, and the duration and magnitude of precipitation, ponding and water run off can occur Water that does not run off of the cap is either stored in depressions in the cap surface, or infiltrated into the surface layer of the cap Water infiltrating into the surface layer of the cap is subject to evapo-transpiration Rates of evapo-transpiration depend on surface vegetation, soil properties, surface temperatures, soil and air relative humidities, and net solar radiation The remainder of the precipitation not trans-formed to run off or evapo-transpiration remains as storage in the cap, or, if the storage capacity of the cap is exceeded, the water percolates through the cap

Contaminant vapors can migrate through caps by advection or diffusion Advection rates depend on gas-phase permeabilities and pressure gradients across the cap Variations in barometric pressures can increase contaminant vapor advec-tion to the atmosphere Vapor diffusion is driven by the gas-phase concentraadvec-tion gradient existing across the cap Diffusion coefficients depend on soil porosity and water content, as well as contaminant molecular weight It is often assumed that diffusion at the ground surface occurs across a stagnant surface boundary conditions (Thibodeaux, 1981)

Water percolation and contaminant transport through the cap can also be

the migration of water or contaminant vapors through the system Natural events such as earthquakes, tornadoes, floods, and melting snow can also be disruptive

can be significant and should therefore be considered

Figure 2.1, water originates as precipitation that falls on the cap Depending on

Figure 2.2 Animal burrows or other passageways through the cap can accelerate altered by human or biointrusion into the cap and other natural events, leading to disparities between probable current and future percolation rates as shown in

Modeling of Fluid Transport through Barriers 73

FIGURE 2.1 Features, events, and processes associated with a long-term cap

FIGURE 2.2 Cumulative probability distribution of water percolation reaching the mill

tailings for present and future conditions (From Ho, C.K et al., 2001 Sandia National Laboratory Report SAND2001-3032; October.)

Climate

Transpiration

Precipitation

Run-on

Run-off Gas release

Evaporation

Storage Lateral drainage

Waste

Percolation/leaching Human intrusion/

bio-intrusion

10−13 10−12 10−11 10−10 10−9 10−8 10−7 10−6

0 20 40 60 80 100

Present

Future

Percolation Flux through Cover (cm/s)

Cu

mu

la

tive Pr

ob

ability

40 C

FR Par

74 Barrier Systems for Environmental Contaminant Containment & Treatment

2.2.1.2 Layers and Features

In very rare cases, a cap comprises a single soil layer over waste material Typically however, a cap is the unique combination of soils placed in layers on top of each other and in certain order that create the desired effect This section briefly outlines the general performance objective of each potential cap layer

• Ground surface layer — The top few inches of any surface soil may need to be treated as a unique soil region since, due to desiccation and drying effects, this zone generally has a much higher hydraulic con-ductivity than the soil a few inches below surface This zone is espe-cially important to include when simulating infiltration through cover systems using numerical models

• Vegetation layers — It is common to include a vegetation growth layer that may or may not be part of another cover layer In many cases, the vegetation can be a key to cap performance, but based on According to energy balance accounting, the sum of actual evaporation and transpiration are always less than the potential evaporation This means that for near-surface processes, the availability of water limits evapo-transpiration, and water that is not transpired through vegetation is removed through evaporation In other words, if vegetation were not present, actual evaporation would remove a similar amount of water The transpiration process becomes important when it is necessary to draw water from deeper beneath the surface, particularly when actual evaporation has significantly diminished at the surface due to drying of soils Vegetation is also critical for stability purposes on sloped covers, as well as erosion control

• Capillary break layers — These layers are generally created with coarse materials next to fine materials because, at a common negative water pressure, two different soils have different water contents Cap-illary breaks can be used in caps for various purposes When placed beneath a compacted layer, the capillary break limits percolation through the compacted material When placed above a compacted layer, the capillary break limits the evaporative drying of the compacted layer, because water cannot readily be drawn up in its liquid phase through the coarser capillary break layer when it is dry For this type of cover design, a model that includes coupled vapor flow should be used to assess the impact of vapor flux on barrier layer drying in the event that upward liquid phase flow has shut down

Modeling of Fluid Transport through Barriers 75 a barrier layer over a coarser layer to create a capillary break effect, and then place it beneath a vegetation growth layer It is not desired to have the root zone of the plant species extend into the barrier layer where damage can occur While long-term barrier layer performance is unknown and cannot be predicted with precision, the use of dense, well-graded materials for these layers has shown the best resistance to long-term performance deterioration (Wilson, 2002)

• Storage layers — These layers are generally made of loose well graded materials such that the hydraulic conductivity is sufficient to allow water to infiltrate and subsequently be drawn back out by evaporation and/or roots The thickness of a storage layer becomes a critical ques-tion in its funcques-tionality The cover must be thick enough to keep near-surface wetting and drying processes from interacting with the waste, and to withstand long-term erosion If the cover is to limit gas fluxes as well, there must be a zone of continual near-saturation within this layer over time and over prolonged dry periods; either that, or the storage layer must protect a deeper near-saturation barrier layer Long-term storage layer performance can be affected by coarse material breakdown, which can result in permeability loss

2.2.2 CURRENT STATE OF PRACTICE FOR MODELING

PERFORMANCEOF CAPS

Water movement through soils can be thought of as a three-component system consisting of the soil-atmosphere interface, the near-surface unsaturated zone, and the deeper saturated zone In the past, groundwater modeling has primarily focused on the saturated zone, which creates a discontinuity in the natural system because the unsaturated zone and the soil-atmosphere interface are not repre-sented Advances in unsaturated soil technology during the past decade have led to the development of routine modeling techniques for saturated and unsaturated soil systems However, modeling techniques for the third component, involving the detailed evaluation of the flux boundary condition imposed by the atmosphere, are not routinely available This section discusses some of the available codes that can be used for the predictive modeling of processes associated with cap performance A summary of the codes considered, and some of the key features different available software tools and their main solution processes, as well as feature overviews and source availability Table 2.2 lists the individual program’s solution options and features that are built into the various codes

2.2.2.1 Water Balance Method

or Environmental Contaminant Containment & T

reatment

Name Process Solved Parameters Technique Features/Limitations User Interface Availability

SoilCover 1D, Transient FEM

Pressure, temperature, vapor pressure with pseudo gas

Coupled, Simultaneous, nonlinear

Pre- and post-processor included; code unavailable Freeware

Text in Excel with dialogues; requires Excel 97 or 2000

www.vadose-science.com

Oxygen flux Subsequent VADOSE/W 2D, transient

and steady-state FEM

Pressure, temperature, vapor pressure; can be linked with slope stability software and contaminant transfer software

Coupled, simultaneous, nonlinear

Enhanced pre and post-processor included; climate and soils database included; user support included; commercially developed for cover/cap design

Full CAD data input and mesh generation; Microsoft certified for XP and lower OS

www.geo-slope.com

Oxygen or radon diffusion, dissolution, decay

Subsequent linear

Earthquake seismic analysis using VADOSE/W generated pore pressure data

Supplemental Integrated with program QUAKE/W

Slope stability analysis using VADOSE/W generated pore pressure data

Supplemental Integrated with program SLOPE/W

ranspor

t through Bar

riers

77

HELP 1D, quasi 2D, Analytical

Water balance Analytical Climate and soil database included; not physically based; limited design application; assumes unit gradient

Text in editor or Windows dialogues

www.wes.army.mil/el/ elmodels/helpinfo.html

UNSAT-H 1D, transient FEM

Pressure with vapor Nonlinear Pre- and post-processor available but excluded Code available

Text in editor or Windows dialogues

www.hydrology.pnl.gov/ unsath.asp

Temperature (optional) Subsequent linear HYDRUS-2D 2D, Transient

and steady-state FDM

Pressure, with vapor flow

Nonlinear Pre- and post-processor included; CAD mesh generation add-on

CAD and Windows dialogues

www.ussl.ars.usda.gov/ models/hydrus2d.HTM Temperature Subsequent linear

Contaminant transfer Subsequent nonlinear TOUGH 1D, 2D, 3D,

transient and steady-state IFDM

Pressure, temperature, vapor, gas in porous or fractured media

Coupled, Simultaneous, nonlinear

Limited pre- and post-processor available from independent suppliers Code available; users can customize

Limited CAD and text in editor

www-esd.lbl.gov/ TOUGH2

or Environmental Contaminant Containment & T

reatment

Available Software Overview

Software

Name Process Solved Parameters Technique Features/Limitations User Interface Availability

FEHM 1D, 2D, 3D, transient FEM/FVM

Multi-phase, multi-component heat, mass, gas, air including double porosity flow; can solve contaminant flow as advection/ dispersion or particle tracking

Coupled, simultaneous, nonlinear

Limited pre- and post-processor with 3D grid generator available from independent sources Unix or PC based; code included; user can customize; USA only

Limited CAD with text input

www-lanl.gov/EES5/ fehm.html

RAECOM 1D steady-state radon-gas diffusion

Radon-gas

concentration and flux through a multi-layer system

Linear Can automatically optimize layer thickness

Text entry RAECOM-cloned calculator available on the web:

wise/uranium/ctc.html Coupled, physical coupling between equations; simultaneous, more than one equation solved at same time (must be coupled); subsequent, more than one equation solved one after the other at each time step; supplemental, data from completed analysis used in separate analysis; linear, material properties not a function of variable http://www.antenna.nl/

being solved; nonlinear, material properties change with variable being solved, so iterations required; analytical, no partial differential equation, one pass solution

ranspor

t through Bar

riers

79

Available Software: Detailed Options

Software

Name Solved Parameters

Solution Complexity

Evapor -ation

Transpir-ation

Freez-ing

Run Off

Pond-ing

Soil Properties

SoilCover Pressure, temperature, vapor pressure with pseudo gas

RP RP RE SE RE — FF, CF

Oxygen flux A

VADOSE/W Pressure, temperature, vapor pressure; can be linked with slope stability software and contaminant transfer software

RP RP RE RE RE RE FF, CF

Oxygen or radon diffusion, dissolution, decay

RP Internally

calculated Earthquake seismic analysis using

VADOSE/W generated pore pressure data

RP FF

Slope stability analysis using VADOSE/W generated pore pressure data

RP FF

Contaminant transfer,

advection/dispersion, decay, particle tracking

RP FF

HELP Water balance A SE SE E E — CF

UNSAT-H Pressure with vapor RE SE RE — SE — CF Temperature (optional) RE

or Environmental Contaminant Containment & T

reatment

Software

Name Solved Parameters

Solution Complexity

Evapor -ation

Transpir-ation

Freez-ing

Run Off

Pond-ing

Soil Properties

HYDRUS-2D Pressure, with vapor flow RE SE RE — SE — CF

Temperature RE

Contaminant transfer RE TOUGH2 Pressure, temperature, vapor, gas in porous

or fractured media

RP — — — — — CF

FEHM Multi-phase, multi-component heat, mass, gas, air including double porosity flow; can solve contaminant flow as advection/dispersion or particle tracking

RP SE — — SE SE CF

RAECOM Radon-gas concentration and flux through a multi-layer system

A — — — — — CF

RP, rigorous physically based with assumptions limited to current understanding of real physical processes; RE, rigorous physically based but with empirical components or built-in limiting assumptions; SE, semi-empirical, equation based but user sets limits or there are limited built-in assumptions; E, empirically based, extreme limiting assumptions and little physical bases for generated data; A, analytically based — no partial differential equations; FF, free-form functions, user can customize; CF, closed-form functions, curve-fit parameters

Modeling of Fluid Transport through Barriers 81 and soil storage are considered The first method used for water balance calcu-lations was developed by Thornthwaite and Mather (1957) This method was used by Fenn et al (1975) to analyze leachate generation at municipal solid waste landfills

Typically, the water balance method is based on monthly climatic variables The monthly infiltration, I (cm), into a cover is given by:

I = P – R (2.1)

where P is precipitation (cm) and R is surface run off (cm) Surface storage was not considered by Fenn et al (1975) Run off is calculated from precipitation using a run off coefficient, C:

R = C P (2.2)

Fenn et al (1975) provided values of C for different soil types and slopes, with values ranging from 0.05 for sand with less than a 2% slope, to 0.35 for a steeply sloped (>7%) clay layer

Thornthwaite and Mather (1957) also provided tables for determining poten-tial evapo-transpiration (PET) as a function of mean temperature, heat index, and hours of sunlight When PET exceeds infiltration, moisture storage in the cap is expected to decrease unless the cap was already dry PET cannot exceed the water stored in the cap plus the infiltration for the month When infiltration exceeds PET, evapo-transpiration is equal to PET, and excess infiltration increases the moisture storage in the cap to field capacity Excess infiltration above the field capacity of the cap percolates through the cap

2.2.2.2 HELP

The hydrologic evaluation of landfill performance (HELP) model was developed by the United States Army Engineer Waterways Experimentation Station for the United States Environmental Protection Agency (USEPA) in 1984 The current version of the model, Version 3, was released in 1993

The HELP model is essentially a water balance model that includes subsur-face water routing It simulates both model cap and liner behavior in a landfill system The model is referred to as a quasi-two-dimensional model, as it simulates vertical flow in barrier and waste layers (assuming unit hydraulic gradient), and horizontal flow in drainage layers (using an analytical solution of the Boussinesq equation) Calculations are performed on a daily basis, and changes in soil moisture and surface storage are tracked (Peyton and Schroeder, 1993) The HELP model considers both rain and snow infiltration and accounts for interception by vegetation, surface evaporation, and surface storage

82 Barrier Systems for Environmental Contaminant Containment & Treatment

2.2.2.3 UNSAT-H

UNSAT-H (WinUNSAT-H) is a model for calculating water and heat flow in unsaturated media The model was developed at Pacific Northwest National Laboratory in Richland, Washington, to assess the water dynamics of near-surface, waste disposal sites The code is primarily used to predict deep drainage as a function of environmental conditions such as climate, soil type, and vegeta-tion UNSAT-H is a one-dimensional model that simulates the dynamics processes of infiltration, drainage, redistribution, surface evaporation, and uptake of water from soil by plants It uses a finite-difference approximation to solve the one-dimensional vertical form of Richards’ equation, which governs unsaturated moisture movement UNSAT-H was designed for use in water balance studies and has capabilities to estimate evaporation resulting from meteorological surface conditions and plant transpiration

The parameters required for each material type are saturated hydraulic con-ductivity, volumetric moisture content at saturation, irreducible moisture content, air entry head, and inverse pore size distribution index

2.2.2.4 SoilCover

SoilCover is a soil-atmosphere flux model that links the subsurface saturated/ unsaturated groundwater system and the atmosphere above the soil in an attempt to represent the soil-atmosphere continuum It is a one-dimensional finite element package that models transient conditions The model uses a physically-based method for predicting the exchange of water and energy between the atmosphere and a soil surface The theory is based on the well-known principles of Darcy’s and Fick’s Laws that describe the transport of liquid water and water vapor and Fourier’s Law that describes conductive heat flow in the soil profile below the soil-atmosphere boundary SoilCover predicts the evaporative flux from a saturated or an unsaturated soil surface on the basis of atmospheric conditions, vegetation cover, and soil properties and conditions The Penman–Wilson formulation is used to compute the actual rate of evaporation from the soil-atmosphere boundary, which is critical to modeling of evapo-transpirative caps (Wilson, 1990; Wilson et al., 1994)

The primary features and modeling capabilities of SoilCover are as follows: • Specification of detailed climate data, including minimum and

maxi-mum air temperature, net radiation, minimaxi-mum and maximaxi-mum relative humidity, and wind speed

• Specification of reduced climate data, including air temperature, rela-tive humidity, and potential evaporation (wind speed is optional) • Multi-layered soil profiles

• Optional specification of an internal liquid source/sink node

Modeling of Fluid Transport through Barriers 83 • User-defined or SoilCover-predicted thermal and hydraulic soil

prop-erty functions

• Internal adaptive time stepping scheme for daily simulations

• Relative convergence criteria for suction and temperature applied at every node

• Output data files providing daily profiles of volumetric and gravimetric water content, degree of saturation, matrix suction, total head, temper-ature, ice content, hydraulic conductivity, oxygen concentration, and vapor pressure

• Daily reporting of potential evaporation, surface flux, base flux, total evaporation, total run off, root flux, user-selected internal node flux and user selectable on-screen graphics during program execution show-ing continuous daily or cumulative fluxes in chart and table format plus daily updates of temperature and degree of saturation profiles The program user interface occurs in Microsoft Excel using dialogue boxes and custom menus, and the solver is a 32-bit Fortran executable file

2.2.2.5 HYDRUS-2D

HYDRUS-2D can be used to simulate two-dimensional water flow, heat transport, and the movement of solutes involved in consecutive first-order decay reactions in variably saturated soils HYDRUS-2D uses the Richards’ equation for simu-lating variably saturated flow and Fickian-based convection-dispersion equations for heat and solute transport The water flow equation incorporates a sink term to account for water uptake by plant roots The heat transport equations consider transport due to conduction and convection with flowing water The solute trans-port equations consider convective-dispersive transtrans-port in the liquid phase, as well as diffusion in the gaseous phase The transport equations also include provisions for nonlinear nonequilibrium reactions between the solid and liquid phases, linear equilibrium reactions between the liquid and gaseous phases, zero-order production, and two first-zero-order degradation reactions: one independent of other solutes and one that provides coupling between solutes involved in the sequential first-order decay reactions

84 Barrier Systems for Environmental Contaminant Containment & Treatment

2.2.2.6 VADOSE/W

VADOSE/W is a commercially developed two-dimensional finite element code that accounts for precipitation; evaporation; snow accumulation/melt/run off; groundwater seepage; freeze-thaw; ground vapor flow; actual transpiration from plants; and gas diffusion, dissociation, and decay It solves the same primary heat and mass differential equations as the SoilCover model except in two dimensions The gas diffusion equation is solved at the completion of each time step once water contents and temperatures are known throughout the domain

VADOSE/W uses the Penman–Wilson method (Wilson, 1990; Wilson et al., 1994) method for computing actual evaporation at the soil surface such that actual evaporation is computed as a varying function of potential evaporation dependent on soil pore water pressure and temperature conditions and independent of soil type and drying history The fully coupled heat and mass equations with vapor flow in VADOSE/W permit the necessary parameters at the soil surface to be available for use in the Penman–Wilson method VADOSE/W is currently the only numerical two-dimensional cap design model capable of calculating actual evaporation based on first-principle physical relationships, not empirical formu-lations that are developed for unique soil types, soil moisture conditions, or climate parameters

VADOSE/W can be used wherever accurate surface boundary conditions are required Typical applications include designing single or multi-layered soil cov-ers over mine waste and municipal landfill disposal sites; obtaining climate-controlled soil pore pressures on natural slopes or man-made covered slopes for use in stability analysis; and determining infiltration and evaporation as well as plant transpiration from agricultural irrigation projects

VADOSE/W comes with a built-in soil property database as well as full-year detailed climate data for over 40 sites worldwide Climate data can be easily scaled to suite specific conditions or the user can input specific climate data

2.2.2.7 TOUGH2

Modeling of Fluid Transport through Barriers 85 constitute the gas phase, are tracked and simulated separately Liquid- and gas-eous-phase flow can occur under pressure, viscous, and gravity forces according to Darcy’s Law, and interference between the phases is represented through relative permeability functions

A number of variations of the TOUGH2 code have been developed to include additional capabilities of modeling additional species, modeling fluctuating atmo-spheric boundary conditions, and inverse modeling The model parameters, initial conditions, and boundary conditions are typically entered into the code through text entry into a file that is read by the code Post-processing within TOUGH2 is limited and is typically performed by third-party software The source code for TOUGH2, written in standard FORTRAN77, is available from the United States Department of Energy (USDOE) Office of Scientific and Technical Information Energy Science and Technology Software Center in Oak Ridge, Tennessee

2.2.2.8 FEHM

Finite element heat and mass (FEHM) is a numerical simulation code for sub-surface transport processes (Zyvoloski et al., 1997) It models three-dimensional (3-D), time-dependent, multi-phase, multi-component, nonisothermal, reactive flow through porous and fractured media It can represent complex 3-D geologic media and structures and their effects on subsurface flow and transport FEHM uses a finite-element formulation to solve the governing equations of heat and mass transport Simulation of additional species (e.g., organics, radionuclides) can be performed simultaneously with the solution of heat, air, and water trans-port In addition, a particle-tracking module is also included that provides a more computationally efficient procedure to the solution of contaminant transport Millions of particles can be simulated that represent the effects of advection, diffusion, dispersion, and fracture-matrix interactions on transport

The entry of model parameters, boundary conditions, and initial conditions into FEHM is performed through the creation of text files that are read by the code FEHM does not perform any direct post-processing of the data for visual-ization, but the user has the option to output the data in formats that can be read by third-party software FEHM can be obtained free of charge in the United States

2.2.2.9 RAECOM

Radiation attenuation effectiveness and cover optimization with moisture effects through a multi-layer cover (Rogers et al., 1984) Material properties, dimensions, and diffusion coefficients can vary among the different layers, and activity and emanation coefficients can be specified An online calculator that provides the for most applications via the web site http://ees-www.lanl.gov/EES5/fehm/

http://www.antenna.nl/wise/uranium/ctc.html

(RAECOM) is a code that simulates steady, one-dimensional radon gas diffusion

86 Barrier Systems for Environmental Contaminant Containment & Treatment

2.2.3 MODELING LIMITATIONSAND RESEARCH NEEDSFOR CAPS

There are many limitations to modeling the performance of caps, including data needs; lack of quality assurance and control of models and model usage; and lack of verification, validation, and calibration This section discusses these limitations and the associated research needs, as well as the role of modeling in designing caps

2.2.3.1 Role of Modeling

There is often a misperception of what a model can and cannot It is critical to get all stakeholders to understand and agree on the objectives of using the model Many believe that if the predictions arise from a sophisticated computer code that incorporates the fundamental physics as it is currently understood, the answer must be correct In fact, at best, the model output is a scientifically defensible, although not necessarily accurate, prediction of system behavior This belief in modeling leads to the development and use of more sophisticated models that advance the state of the science, but not necessarily provide more defen-sible predictions

In modeling cover system performance, the objective is to provide a measure of the ability of the cover to prevent water infiltration to the waste zone over long periods of time (i.e., tens of years to hundreds of years) It is not possible to precisely predict infiltration over long time periods due to the large number of uncontrolled variables (e.g., weather conditions, burrowing animals, root growth), heterogeneities in the physical properties of the system, and lack of precise understanding of the flow physics (e.g., hysteresis effects and soil characteristic curves are empirical relationships based on data) Therefore, the modeling approach should aim to demonstrate that the cover system limits infiltration to an acceptable level over a range of potential conditions This lends itself naturally, although not exclusively, to probabilistic modeling

2.2.3.2 Data Needs

The data required for modeling cap behavior depends on the model being used The simplest models such as the water balance method of Thornthwaite and Mather (1957) and Fenn et al (1975) require monthly climatic data such as precipitation, mean temperature, heat index, and hours of sunlight Soil types and cap slopes are also required to allow estimation of run off

Modeling of Fluid Transport through Barriers 87 required include porosity, field capacity, wilting point, hydraulic conductivity, and the United States Soil Conservation Society curve number for the surface layer The HELP model contains a list of default soil properties, and a database of climate data for a large number of North American cities (Peyton and Schroeder, 1993)

Other more rigorous models such as UNSAT-H, HYDRUS-2D, and VADOSE/W simulate unsaturated water flow by solving Richards’ equation Simulation of unsaturated water flow with Richards’ equation requires parameter specification of the soil characteristic curves for hydraulic conductivity and mois-ture content as a function of suction pressure, typically represented by empirical relationships such as those developed by van Genuchten (1980) or Fredlund and Xing (1994) These parameters are required for each unique soil layer in the cover system Saturated hydraulic conductivity and porosity are also required for each material Other parameters, such as the air entry pressure head, residual saturation value, vertical and horizontal saturated conductivity, and anisotropy parameters may be required depending on the model

Some of the models (i.e., TOUGH2, FEHM, VADOSE/W, HYDRUS-2D) also solve the heat transport equation to track evaporation and water vapor transport Therefore, these models require additional information regarding soil properties related to heat transport for the gas and liquid phase Parameters that are typically needed include thermal conductivity, specific heat capacity, latent heat of vapor-ization, surface tension, and parameters that describe the interactions between gas and liquids under flowing conditions (e.g., relative permeability)

2.2.3.3 Code Quality Assurance and Quality Control

88 Barrier Systems for Environmental Contaminant Containment & Treatment benchmark test documentation as would the required results If the new model cannot perform the basic benchmark tests, it is not acceptable for use in field design

A final point to consider is the establishment of a group of individuals who can assemble the benchmark tests and can review and update the tests as new and more advanced physics are introduced

2.2.3.4 Verification, Validation, and Calibration

The verification, validation, and calibration of numerical models are key compo-nents in the modeling process and are often the most poorly implemented and misunderstood

The key questions to ask when looking at models are what equations are being solved, what assumptions have been applied to the equations, and how are the equations being solved? For example, just because a model computes evap-oration does not mean that it does so based on sound physical relationships or that, if it is based on sound physics, the equations are solved properly After a model user has an understanding of the theory and physics incorporated into a numerical code, they should satisfy themselves that the numerical solution for that set of equations is correct This is the verification stage of the modeling process and is usually carried out by the model developer Verification has nothing to with site data and everything to with correct solution of the mathematics Verification and validation go together; where verification addresses solution techniques and validation is the process of obtaining confidence that the model applies to real situations represented by the theoretical formulations applied in the model Validation tests if the model theories actually apply to specific real observations — whether they are laboratory experiments or field studies

It is absolutely critical to validate a model based on known closed-form solutions, known physical observations, and laboratory tests where all parameters can be controlled and adjusted individually Models cannot be validated using field data alone because there is no direct control over or monitoring of all major model parameters For example, if a model is validated using site data where precipitation, run off, change in water storage, and bottom drain fluxes are mea-sured but actual surface evaporation and transpiration are not meamea-sured, then the source of discrepancies between measured and computed results cannot be deter-mined There could be error in the model estimate of evaporation, or there could be error in the field measurement of particular parameters The most appropriate use for field data in modeling is calibration of a previously validated model

Modeling of Fluid Transport through Barriers 89 physics truly represent the real physical processes in the ground If the model is rigorous enough and calibrated properly, then all physical processes measured in the ground and predicted by the model should match Calibration of nonrigorous models such as HELP must be interpreted with caution because, in many cases, the calibration can be achieved by adjusting only a single model parameter When this is done, the predicted and measured data only match for a single instance in time There is no guarantee that the adjustments made to the model to fit measured data represent the true physical properties in the field It may be possible to calibrate HELP to match measured percolation data, but it is very unlikely that parameters such as the temperature, water pressure, water stored in the soil, and the root depth match field conditions at the same instance or at some other instance in time

2.2.4 UNRESOLVED MODELING CHALLENGES

There are many challenges facing model developer users These challenges include the difficulties in modeling systems with time-varying properties and processes, the problems encountered in modeling infiltration at arid sites, and the role of heterogeneities in modeling

2.2.4.1 Time-Varying Material Properties and Processes

A major challenge facing modelers of cap performance is the time-varying nature of climate, vegetation, and soil properties All models of cap performance require extensive climatic data, including precipitation, temperature, and solar radiation to determine infiltration and evapo-transpiration Although historic data are avail-able for many locations, methods for estimating extreme values of these variavail-ables are not well developed

Physical deterioration of caps is commonplace, as they are easily impacted by surface and climate processes Changes in vegetation have an effect on run off generation and evapo-transpiration Establishment of shrubs and trees on caps can lead to cap penetration by roots, creating high conductivity pathways for water infiltration Similarly, burrowing animals can create high conductivity conduits through a cap Erosion and subsidence can seriously impact cap performance The cracking of clay layers in caps due to freeze-thaw cycles or desiccation (e.g., Albrecht and Benson, 2001) can significantly increase the effective hydrau-lic conductivity of caps, leading to greatly increased water infiltration or vapor escape Albrecht and Benson (2001) found that clay hydraulic conductivities increased by factors as high as 500 upon desiccation Subsequent resaturation did not lead to complete healing of dessication-induced cracks Although cap modeling can predict soil moisture levels in the cap, reliable models for changes in cap hydraulic properties due to dessication or freeze-thaw have not been developed

90 Barrier Systems for Environmental Contaminant Containment & Treatment over these long periods of time have not been conducted In addition, accurate predictions of long-term climate changes and the occurrence and impact of extreme events (e.g., earthquakes, floods, hurricanes, tornadoes) are not possible

2.2.4.2 Infiltration at Arid Sites

Arid sites are characterized by levels of precipitation that are almost balanced by loss mechanisms such as evaporation, transpiration, and run off For water balance models, the recharge is estimated by subtracting the losses from the predicted production Thus, small errors in either estimate can lead to large errors in recharge estimates

A second issue at arid sites is that evapo-transpiration models used in the water balance models for disposal cells that are sparsely vegetated are not accurate and tend to overpredict evapo-transpiration and underpredict recharge The use of physically based evapo-transpiration models (e.g., SoilCover, VADOSE/W) that are formulated to shut down actual evaporation as ground surfaces dry greatly improves infiltration estimates at arid sites

2.2.4.3 Role of Heterogeneities

The most commonly used models for estimating flow through cover systems assume uniform hydraulic and thermal properties for each layer of the cover system In practice, local heterogeneities are likely to be responsible for a large portion of the flow through cover systems The heterogeneities can arise naturally

and their impact on flow does not exist For example, desiccation cracking is known to occur in clay barriers and leads to increased flow However, the capa-bility to predict crack formation; the density of cracks; the changes in hydraulic conductivity that occur due to cracking and subsequently rewetting; and, more importantly, the change in flow through the layer does not exist

For field performance, localized failure will often control infiltration through the cover system This leads to the need to develop procedures to adequately represent these local failures using gross average properties for the layers

2.3 PRBS

In recent years, PRBs have evolved from the realm of an experimental method-ology to standard practice for containment and treatment of a variety of contam-inants in groundwater Like any remedial technology, the decision to use PRBs is conditioned by the characteristics of the natural system, target contaminants, and treatment objectives More than 60 sites have implemented this technology in the last few years to treat chlorinated solvents, fuel hydrocarbons, and various inorganic contaminants in groundwater As with any technology used to treat or Currently, the capability to predict the occurrence of local heterogeneities evolve in time (Section 2.3.5.1)

Modeling of Fluid Transport through Barriers 91 extract contaminants in the subsurface, successful implementation is contingent on effective site characterization, design, and construction Recent studies on long-term PRB performance at a number of sites emphasize the following key issues for successful use of PRBs:

• Performing adequate site characterization on the scale of the PRB — Site characterization approaches, typical of Resource Conser-vation and Recovery Act (RCRA) facility investigations (RFIs), are not adequate Performing additional localized characterization of the plume distribution in three spatial dimensions and with time, under-standing the local hydrogeology, and knowing the site geochemistry is required

• Understanding site hydrology to achieve successful implementation — PRBs must be located correctly to intercept the plume because once located in the subsurface, they cannot be moved It is therefore imper-ative that the PRB captures the plume at the present time and in the future allowing for variations in flow direction, velocity, and concen-trations of contaminants over time

• Developing contingency plans for failure to meet design objectives — It is surprising that site owners and regulators often fail to explicitly develop contingency plans Contingency plan development requires specification of design criteria and performance objectives and deter-mination of what constitutes a failure in order to clearly trigger con-tingency plan activation

2.3.1 FEATURES, EVENTS, AND PROCESSES AFFECTING

PERFORMANCEOF PRBS

Design of PRBs requires consideration of groundwater hydraulics, geochemical processes, and reaction kinetics and the interaction between these processes

2.3.1.1 Groundwater Hydraulics

As with any groundwater remediation technology, an understanding of the direc-tion and rate of groundwater flow spatially and temporally is critically important for successful design Groundwater hydraulics are particularly crucial for PRBs because the treatment system is immovable and passive yet must intercept the contaminant plume for effective treatment

characterize with precision and sufficient resolution given physical and budgetary constraints

To assess PRB system performance, information is needed on groundwater velocities through and near the planned PRB On the simplest level, these values can be estimated from observed hydraulic gradients and measured or estimated hydraulic conductivities Alternatively, groundwater velocities can be determined with a numerical groundwater flow model based on estimated hydraulic property distributions and hydrologic boundary conditions (i.e., water levels and/or fluxes on model boundaries and recharge and extraction rates), which can vary tempo-rally In many cases, it is important to consider the effects of temporal changes in flow direction and velocity due to variations in recharge, pumping of adjacent wells, or other disturbances It is not uncommon to observe changes in flow direction on the order of 30˚ or more over time due to transient boundary conditions Furthermore, the PRB permeability itself can change markedly over time in some situations (e.g., due to biological fouling or chemical precipitation in or near the PRB), which can substantially impact the hydraulic regime

Understanding site stratigraphy and lithology is crucial to understanding and predicting groundwater hydraulics If a low permeability layer exists at the site, the PRB can be keyed into this layer If one does not exist, then a hanging wall design can be employed, but uncertainty regarding plume capture may increase If the site has low permeability layers through which the PRB must be constructed, care must be taken during construction to avoid smearing of such layers, which could impact hydraulic contact between the formation and reactive media A thorough understanding of site stratigraphy is important when choosing a partic-ular construction method For example, the use of sheet piling to construct a reactive gate may not be a good choice where low permeability layers exist because of smearing potential

2.3.1.2 Geochemical Processes

The nature and extent of geochemical processes occurring within a PRB to a large degree determine the long-term treatment performance of the barrier The details of these processes are site specific and associated with chemical, physical, and biological factors such as the following:

• Reactive media type (e.g., zero-valent iron (ZVI), other metals, zeolite, organic materials)

• Influent groundwater chemistry (e.g., pH; amounts of cations, anions, and target contaminants)

• Microbiological environment within and around the PRB • Physical conditions (e.g., temperature)

summaries provided by the Air Force Research Laboratory (AFRL, 2000) and on the Remediation Technologies Development Forum (RTDF) web site that have been installed Of these, the vast majority (approximately 85%) use ZVI as the reactive medium Other types of reactive media that have been inves-tigated include other metallic materials (Gillham and O’Hannesin, 1992; Korte et al., 1995; Muftikian et al., 1995; Orth and McKenzie, 1995; Bostick et al., 1996; Hayes and Marcus, 1997), zeolite (Bowman et al., 2001; Rabideau and Van Benschoten, 2002), various organic materials (Benner et al., 1997), apatite (Conca et al., 2000; Fuller et al., 2002), and sodium dithionite injected as a solution (Fruchter et al., 1997) The AFRL (2000) summarizes different PRB media that have been investigated The AFRL (2000) and the RTDF web site also document the range of contaminants that are being treated by PRBs Chlorinated solvents such as trichloroethylene (TCE) and perchloroethylene (PCE) are the dominant target contaminants, but others include metals and radionuclides [e.g., Cr(VI), U(VI), Tc(VII)], other inorganics (e.g., NO3–, SO42–), and other

organics (e.g., pesticides, toluene)

Because of the dominance of ZVI as a reactive medium in PRBs, the following discussion focuses exclusively on geochemical processes occurring within it ZVI functions as a redox medium and treats contaminants by chemical reduction At the same time, the iron is sacrificially oxidized progressively from Fe(0) to Fe2+

and, finally, Fe3+ The oxidized species of iron potentially can react with other

components in the groundwater to precipitate a variety of amorphous and crys-have been formed by reactions occurring in ZVI PRBs

The reaction of groundwater with ZVI causes several major compositional changes that drive the formation of these reaction products ZVI begins to dissolve according to the following reactions:

2Fe0 + 2H

2O + O2 (aq) = 2Fe2+ + 4OH–

Fe0 + 2H

2O = Fe2+ + H2 (aq) + 2OH–

The first reaction involves the scavenging of dissolved oxygen by ZVI and is known to be a fast reaction because column and field studies show the complete absence of dissolved oxygen within a few centimeters of the influent face of a PRB The second reaction prevails once the oxygen is gone and is slower Both reactions result in a significant decrease in redox potential and a dramatic rise in pH, both of which are observed in typical ZVI PRBs The magnitude of change in pH depends on the detailed chemistry of the influent groundwater, its buffering capacity, and the rate of groundwater flow through the barrier For example, high alkalinity groundwater is more resistant to a change in pH However, the large available mass of ZVI in PRBs tends to overwhelm any redox buffering capacity of the groundwater

The oxidation of ZVI (and associated decrease in groundwater redox poten-tial) and the dramatic pH rise are the two principal factors that result in the formation of new solid phases, many of which are iron bearing (Table 2.3) Some of these phases that contain either Fe2+ (e.g., amorphous ferrous oxyhydroxides,

FeS, FeCO3) or mixed Fe2+ and Fe3+ (e.g., Fe3O4, green rust) also are effective

reducing agents for metals, radionuclides, and organics in groundwater Conse-quently, the formation of these reduced iron phases does not necessarily signifi-cantly diminish the reactivity of the barrier media However, not all phases formed in a PRB are iron-bearing For example, the increase in pH can also lead to precipitation of various carbonate minerals (e.g., calcite, aragonite) if the influent water has sufficient amounts of dissolved alkalinity and calcium The mix of solid phases formed and their order of precipitation depend on influent groundwater chemistry, the complex interplay of changing redox potential and pH in the system as ZVI dissolves, reaction rates, factors affecting the nucleation of phases, and groundwater flow rate The ability to predict these reactions and estimate their One concern associated with secondary mineral formation in PRBs is that these phases passivate the ZVI media, decreasing its reactivity and ability to treat contaminated groundwater Farrell et al (2000) reported an example of ZVI passivation with results of long-term column experiments in which they observed an over six-fold decrease in the reactivity of ZVI to TCE in the two-year experiment

TABLE 2.3

Examples of Precipitated Minerals Found in Fe(0) Field-Installed PRBs and Column Studies

Mineral Precipitate Group Minerals

Iron oxides and oxyhydroxides Goethite (α-FeOOH)

Akaganeite (β-FeOOH)

Lepidocrocite (γ-FeOOH) (Maghemite (Fe2O3)) Magnetite (Fe3O4)

Amorphous iron oxyhydroxides

Iron sulfides Mackinawite (Fe9S8)

Amorphous ferrous sulfide (FeS)

Carbonates Aragonite (CaCO3, orthorhombic)

Calcite (CaCO3, hexagonal) Siderite (FeCO3)

Green Rusts GR-I (CO32–) (Fe42+Fe23+(OH)12)(CO3⋅2H2O)

GR-I (Cl–) (Fe

32+Fe3+(OH)8Cl)

GR-II (SO42–) (Fe42+Fe23+(OH)12)(SO4⋅2H2O) Source: Liang, L., Sullivan, A.B., West, O.R., Kamolpornwijit, W and Moline, G.R., 2003 Predicting the precipitation of mineral phases in permeable reactive barriers Environ Eng Sciences. Vol 20(6): p 635

The authors found that the degree of passivation was related to the adhering ability of secondary minerals and not the overall mass of these phases formed

A number of PRBs have been cored and the media examined to understand the formation of secondary minerals (e.g., Puls et al., 1999a; Vogan et al., 1999; Phillips et al., 2000; Roh et al., 2000) Typically, cores are obtained by angle drilling through the vertical influent face of the barrier to provide a cross section extending into the PRB interior, capturing the precipitation that is expected to be the most significant at the sediment–ZVI interface Analytical methods such as X-ray diffraction (XRD) and scanning electron microscopy (SEM)typically are used to examine the solid phases that have formed Table 2.4 illustrates differences in groundwater chemistry and resultant secondary minerals observed in PRBs at the Canadian Forces Base Borden in Ontario, Canada (O’Hannesin and Gillham, 1998) and the USDOE Y-12 plant in Oak Ridge, Tennessee (Phillips et al., 2000) The low dissolved solids groundwater at the Borden site has resulted in little formation of new solid phases over a period of four years, and most precipitation

TABLE 2.4

Groundwater Chemistry of Two Different PRB Sites and the Secondary Phases Observed in Each

Chemical Constituent

Concentration (mg/L)

Canadian Forces Base Borden USDOE Y-12 Plant

Na 8.9

K 0.4 3.6

Ca 55 361

Mg 20.5

Total Fe <0.5 0.02

Cl 55

SO4 5–10 47

SiO2 Not available 3.8

NO3 Not available 904

Alkalinity (as CaCO3) 158 220

pH (unitless) 7.9 6.8

Eh (mV) 300 Not available

Dissolved oxygen 2.5-5 Not available

Secondary minerals observed:

Traces of iron oxides CaCO3 (aragonite)

CaCO3 Fe2(OH)2CO3

FeCO3 FeCO3

(After four years of operation; no cementation; mineralization confined to within several millimeters of influent face of the PRB)

Goethite Maghemite

Amorphous iron oxide Green rust

appears to be restricted to a very thin zone at the influent face of the PRB In contrast, the highly mineralized water from the Y-12 plant resulted in much more extensive formation of secondary phases, illustrated by the cementation of reac-tive media (Figure 2.3) The presence of NO3– at high concentrations in the Y-12

plant groundwater is an important factor in determining the degree of reaction occurring in this PRB because NO3– is readily reduced to NH

3 as iron is oxidized

and, therefore, is very corrosive to ZVI

In terms of groundwater treatment, the geochemical reactions between ZVI and the target contaminants are of primary importance within PRBs Currently, most PRBs are deployed to treat groundwater contaminated with chlorinated solvents such as TCE, PCE, and their daughter products A discussion of several possible degradation reaction pathways for TCE is provided in AFRL (2000) The following reaction illustrates the overall reductive dechlorination process for TCE:

3Fe0 + C

2HCl3 + 3H+ = 3Fe2+ + C2H4 + 3Cl–

As noted in AFRL (2000), there may be a number of reaction pathways resulting in a variety of potential intermediates, but experimental and field studies indicate that the net reaction is one of iron oxidation coupled with reductive dechlorination, leading to the production of dissolved ethene (and ethane) and chloride

FIGURE 2.3 Section of cemented core from a PRB at the USDOE Y-12 plant in Oak

Ridge, Tennessee

9 1 2 F

8

PRBs with ZVI also can be used to treat groundwater contaminated with some redox-sensitive toxic metals For example, dissolved species such as hexava-lent chromium, pertechnetate, and uranyl ions are known to react with ZVI Examples that conceptually illustrate these reactions are as follows:

Fe0 + 6H+ + CrO

42– = Fe3+ + Cr(OH)2+ + 2H2O

Fe0 +4H+ + TcO

4– = Fe3+ + TcO2 + 2H2O

2Fe0 + 3UO

22+ = 2Fe3+ + 3UO2

The reduction of these metals tends to make them less soluble and less mobile than the oxidized forms Because these contaminants generally are present in such low concentrations in groundwater, it has not been possible to identify specific solid phases where they are located within PRBs For example, Phillips et al (2000) did not observe uranium-bearing phases at the Y-12 PRB However, Fiedor et al (1998), and Gu et al (1998, 2002a) were able to confirm that U(VI) readily reduced to U(IV) in laboratory experiments with ZVI Gu et al (1998) also confirmed that precipitation, not sorption, was the overwhelmingly dominant process for immobilizing uranium For the PRB in Elizabeth City, North Carolina, Puls et al (1999b) report that the chromate contaminant in groundwater was reduced to the chromic (Cr3+) form in an insoluble mixed Cr–Fe hydroxide,

although that assumption is based on the sharp decrease in chromium concentra-tions within the PRB rather than characterization of specific chromium-bearing phases Based on their work at this PRB, Mayer et al (2001) also assert that chromic hydroxide is the likely Cr(III)-bearing phase formed There is no doubt that ZVI reduces these metals, but whether they are immobilized in the form of a separate reduced solid state, sorbed in phases such as iron oxyhydroxides, co-precipitated with other metals, or a combination of all of these processes has not been thoroughly studied

2.3.1.3 Reaction Kinetics

Reaction kinetics must be considered for chemical reactions occurring within PRBs From a design perspective, the target contaminants (and any toxic daughter products) must have adequate residence time within the barrier to react sufficiently so that effluent concentrations meet design expectations at the downgradient point of compliance (POC) Typical values of half-lives for common organic contam-inants with commercial iron tend to range from less than one to approximately 50 hours, depending on the contaminant and the source of iron used Tabulated summaries of measured half-lives for many compounds are given by Gillham (1996) and AFRL (2000); however, in the design phase of a PRB, site-specific half-lives of target contaminants usually are determined experimentally

Residence times are also partially dependent on the architecture of the ground-water flow system within the PRB Flow heterogeneities (e.g., preferential path-ways) generally exist and can permit target contaminants to migrate through the PRB more rapidly than designed, resulting in inadequate concentration reduction In addition, ZVI corrosion reactions and precipitation of secondary phases within a PRB can lead to progressive clogging of the media, resulting in localized flow diversion Tracer tests have been conducted at several PRBs and illustrate the heterogeneous nature of groundwater flow that exists (e.g., Battelle, 1998) Although the impacts of heterogeneities on groundwater flow and contaminant residence times in PRBs have received only limited attention, Benner et al (2001) modeled heterogeneous aquifer-barrier systems and provided some insights on how the effects on preferential pathways and residence time can be minimized

2.3.2 IMPACTS ON DOWNGRADIENT BIODEGRADATION PROCESSES

At many sites, the rate of natural biodegradation is not sufficient to meet remedial goals, and intervention is required in the form of additional treatment to accelerate or enhance the degradation rate ZVI treatment and natural biodegradation are compatible treatment processes for many chlorinated solvents Both are reductive processes that follow first-order reaction kinetics, and both involve the generation of partially dechlorinated daughter products with reaction rates that are typically slower than those of the parent compound Under appropriate circumstances, the two treatment processes may be synergistic in that the ZVI treatment can enhance or accelerate downgradient biodegradation rates by creating geochemical condi-tions more suitable to anaerobic bacterial metabolism A variety of mechanisms may be operative that stimulate biological processes

2.3.2.1 Enhancement of Geochemical Conditions Conducive to Anaerobic Biodegradation

for reductive biological processes In addition, many of the organisms involved in chlorinated solvent biodegradation are obligate anaerobes that cannot survive in the presence of oxygen

2.3.2.2 Overall Contaminant Concentration Reduction

In cases where a PRB does not achieve complete treatment of the parent com-pounds or reaction daughter products, partial treatment reduces the total loading of chlorinated contaminants in the downgradient aquifer Such incomplete treat-ment can be helpful to the downgradient biological processes in a number of ways In aquifers that are electron donor limited, the PRB can bring down the concen-tration of chlorinated solvents to a point where they can be fully dechlorinated by the available electron donor supply

Some chlorinated compounds are known to inhibit reductive dechlorination processes when present above a threshold concentration An example of this is the observed inhibitory effect of chloroform (CF) on the reductive dechlorination of chlorinated ethenes (Maymo-Gatell et al., 2001) This effect has been observed at CF concentrations above a few parts per million 1,1,1-trichloroethane (1,1,1-TCA) and carbon tetrachloride have also been observed to inhibit methanogenesis and the dechlorination of chloroethenes, although not a severely as CF (Adamson and Parkin, 2000) The PRB can create more favorable conditions for dechlorination by reducing the concentration of such compounds to below the level where they are inhibitory

A third beneficial effect of incomplete treatment is that the effects of competing electron acceptors can be reduced or eliminated When mixtures of chlorinated solvents are present in groundwater, the dechlorinating bacteria preferentially use the electron acceptors that yield the most energy for their metabolism The metabolic energy available from a given half-reaction is expressed as the Gibbs free energy of reaction, in units of kilojoules per mole The greater the Gibbs free energy available from dechlorinating a given compound, the more likely it is that the dechlorinating bacteria will preferentially use that compound as an electron acceptor This effect can be significant in plumes with mixtures of different chlorinated compounds

2.3.2.3 Production of Hydrogen

hydrogen-rich zone can have effects that extend further Dechlorinating organisms, notably Dehalococcoides ethenogenes, are known to be mobile in groundwater systems and can be carried downgradient with the groundwater flow (Ellis et al., 2000) As such, the hydrogen-rich zone immediately downgradient of the PRB can act as a robust source of dechlorinating organisms for the downgradient plume However, increased microbial activity can result in PRB biofouling (Gu et al., 2002b)

2.3.2.4 Electron Donor Production

The ZVI reaction products from chlorinated solvent treatment include fully and partially dechlorinated simple organic compounds that can serve as electron donors for downgradient biological dechlorination processes Examples include the production of formate from carbon tetrachloride and ethene from PCE and TCE Similarly, the treatment of 1,1,1-TCA with ZVI yields a significant amount of ethane, with lesser amounts of ethene, cis-2-butene, and 2-butyne (Fennelly and Roberts, 1998)

Typically, the daughter products of chlorinated solvents treated with ZVI are partially dechlorinated and therefore less highly oxidized than the parent com-pound Some of these partially reduced daughter products can be used as electron donors in downgradient biodegradation processes, particularly in aquifers that are not strongly reduced An example is the conversion of carbon tetrachloride to dichloromethane (DCM) in a PRB DCM is known to biodegrade rapidly under both aerobic and anaerobic conditions Under aerobic conditions, DCM can be biologically oxidized to carbon dioxide and hydrochloric acid Under anaerobic conditions, DCM can be converted to acetate by fermentation (Freedman and Gossett, 1991) The generated acetate can then serve as an electron donor Other examples of partially dechlorinated compounds that can serve as electron donors are cis-1,2-dichloroethylene (cis-1,2-DCE) and vinyl chloride, the daughter prod-ucts of PCE and TCE (Bradley and Chapelle, 2000), and 1,1-DCE, the daughter product of 1,1,1-TCA In aerobic aquifers, the biological transformation of fully chlorinated compounds such as PCE or carbon tetrachloride does not occur Under such conditions, a PRB can convert these compounds into lesser chlorinated species, which subsequently can be biologically oxidized

2.3.2.5 Direct Addition of Dissolved Organic Carbon

More commonly, the introduction of dissolved organic carbon can be an ancillary effect from ZVI emplacement, as in the case of biopolymer trenching or guar-based injection methods such as hydraulic fracturing or jetting When trenching is used as the construction method for a PRB, a high-density biode-gradable slurry is often used to hold the trench open during excavation and emplacement of the reactive media Typically, the slurry-filled trench is filled with the granular iron or sand/iron mixture from the bottom up using a tremie After the reactive media is emplaced, the biopolymer slurry is typically broken with an enzyme that converts it into simple soluble sugars Complete breaking of the slurry is essential to ensure that the completed PRB will have the desired permeability The resultant simple sugars are then dissolved in the groundwater and carried downgradient in the PRB effluent

Guar is used to suspend the granular iron during injection-based PRB emplacement methods such as jetting and hydraulic fracturing Guar is a highly soluble food-grade starch which, when chemically cross-linked, forms a highly viscous gel This viscous gel serves as a carrier fluid for the iron during the jetting or fracturing process As in the case of biopolymer slurry trenching, a breaker enzyme is added to the gel/iron slurry as it is injected into the subsurface, resulting in the transformation of the cross-linked slurry into simple soluble sugars

Dissolved organic carbon introduced as a by-product of PRB construction is transient and completely consumed after a period of time This period of time, however, may last as long as a few years depending on site-specific conditions such as groundwater velocity and the amount of biological activity During this period, the introduced organic carbon can significantly impact downgradient groundwater quality, particularly during the transition period from pre-PRB con-ditions to the new post-PRB steady-state In fact, the organic carbon can shorten the period of time necessary to reach steady-state by accelerating biodegradation and consequently accelerating the rate of desorption in the downgradient aquifer

2.3.3 PRB SYSTEM DYNAMICS

After a PRB is installed, treated groundwater begins to displace untreated ground-water in the downgradient aquifer Notwithstanding this influx of treated ground-water, contaminants continue to be present in the downgradient aquifer for some time after PRB installation, largely due to the slow desorption of contaminants from the aquifer solids This desorption occurs as the downgradient aquifer transitions into a new equilibrium with the treated PRB effluent Eventually, the reservoir of sorbed contaminants is depleted and downgradient contaminant concentrations are no longer replenished by the dissolution of sorbed contaminants This process can take several years, depending on site-specific factors such as aquifer grain size, fraction of organic carbon (FOC),groundwater velocity, and initial contam-inant concentrations

modeled using the Langmuir isotherm and is often represented as a graph of dissolved phase concentration vs time In groundwater systems, time is equivalent to the number of pore volumes of clean water that have passed through the aquifer material A typical desorption curve is shown in Figure 2.4

Sorbed phase contaminants are largely unavailable for biodegradation As contaminants desorb from aquifer solids and re-enter the dissolved phase, they become available to dechlorinating organisms in the downgradient aquifer Bio-degradation can accelerate the desorption process by suppressing contaminant concentrations in the dissolved phase Downgradient biodegradation processes can be accelerated by the presence of organic carbon from PRB construction and other processes discussed later

Both the desorption of parent compounds and the subsequent biodegradation of these compounds into fully or partially dechlorinated daughter compounds impacts downgradient groundwater quality until steady-state is reached Data from groundwater monitoring during this transient period can be confusing and can lead to erroneous conclusions about the performance of the PRB itself For example, the persistent presence of parent compounds such as TCE or PCE due to desorption in the downgradient aquifer can be interpreted as breakthrough or leakage of these compounds through the PRB because of faulty construction such as PRB holes or gaps Similarly, the presence of cis-1,2-DCE or vinyl chloride resulting from partial biodegradation of desorbed PCE or TCE can be incorrectly interpreted as daughter products in the PRB effluent resulting from insufficient residence time in the reactive treatment zone It is therefore useful to model these

FIGURE 2.4 Representative desorption curve

Theoretical Desorption Curve

Time

Nor

mali

ze

d Conc

en

tra

tion 0.8

1

0.2 0.6 0.4

processes to develop an estimate of the amount of time needed for the aquifer to reach steady-state at a given distance downgradient of the PRB

Figure 2.5 shows a conceptual model of a treatment train consisting of a PRB in combination with natural biodegradation processes at steady-state The con-centration of the target constituent (vertical axis) is shown vs distance in the downgradient direction (horizontal axis) C0 represents the initial concentration

in the upgradient area, which decreases in the downgradient direction at the intrinsic rate of natural biodegradation at the site This intrinsic rate can be slow or negligible at sites where additional treatment has been deemed warranted Ct is the target concentration that needs to be achieved prior to reaching the downgra-dient POC It is not necessary to design the PRB to achieve Ct at the downgradient edge of the PRB, if space is available downgradient of the PRB for biodegradation to further reduce the concentration prior to reaching the POC Cd is the PRB design concentration selected to achieve the target concentration at the POC, taking downgradient biodegradation into account

Consider the case of a treatment train consisting of ZVI treatment in a PRB followed by natural biological degradation for treatment of a carbon tetrachloride plume When treated by ZVI, carbon tetrachloride can be completely transformed into the fully dechlorinated end products carbon monoxide and formate, with a partial yield of trichloromethane (TCM) (approximately 40% on a molar basis) The generated TCM can then be completely transformed into a mixture of fully dechlorinated end products and DCM, which is not further treated by the ZVI The end result of this reaction series is the complete transformation of carbon tetrachloride and TCM, with production of DCM in the amount of approximately

FIGURE 2.5 Conceptual model of chlorinated solvent treatment using a PRB coupled

with natural biodegradation

Compliance point Permeable

barrier

Design basis Co

Cd

20% of the parent carbon tetrachloride on a molar basis Biodegradation can be an effective means of treating the generated DCM daughter product, thereby achieving complete dechlorination of the parent carbon tetrachloride to nonchlo-rinated end products prior to reaching the POC DCM is known to be rapidly biodegradable under a variety of environmental conditions, both aerobic and anaerobic (Cox et al., 1998)

In addition to carbon tetrachloride, the design approach shown conceptually chlorinated solvents where daughter products are generated that can be fully transformed in the presence of ZVI, but at a slower rate than the parent com-pounds The slower reaction rates of the daughter products necessitate a longer residence time in the reactive iron zone to achieve full dechlorination, resulting in a thicker PRB and correspondingly greater cost Examples of such compounds include PCE and TCE In the case of PCE and TCE, the daughter compounds cis-1,2-DCE and vinyl chloride are generated as part of the ZVI-driven degrada-tion reacdegrada-tions Both of these compounds have slower reacdegrada-tion rates than the parent PCE or TCE The generated vinyl chloride and cis-1,2-DCE can be fully dechlo-rinated in a ZVI PRB given a sufficient thickness of iron and corresponding residence time However, significant cost savings can be realized by using the available aquifer space downgradient of the PRB as a natural bioreactor to degrade residual cis-1,2-DCE and vinyl chloride rather than increasing PRB thickness This approach is only practicable at sites where the biodegradation rates of these daughter compounds is sufficiently rapid and where sufficient space and residence time is available downgradient of the PRB prior to reaching the POC

2.3.4 GEOCHEMICAL MODELING

groundwater compositions by the dissolution and/or precipitation of phases from a user-specified list of candidate phases

There are several general resources that discuss geochemical modeling in detail (Bethke, 1996; Paschke and van der Heijde, 1996; Zhu and Anderson, 2002) Those individuals interested in the chemical, mathematical, and numerical methods behind geochemical models and information about the limitations of the thermodynamic databases commonly used should consult these sources In addi-tion, these sources provide listings and references to many of the most common geochemical models in use today AFRL (2000), Yabusaki et al (2001), and Mayer et al (2001) are examples of studies that focus on geochemical modeling as applied to PRBs with ZVI as the reactive medium The specific examples in the following discussion are largely summarized from these and several additional sources

2.3.4.1 Speciation Modeling

Speciation models utilize the composition of groundwater (e.g., concentrations of dissolved species, pH, redox state) and temperature as input data to examine a large number of chemical reactions that potentially interrelate the chemical constituents of the water These models use the law of mass action to relate the various chemical species; identify the state of saturation of mineral phases; and determine the distribution of dissolved constituents among several different spe-cies (e.g., Ca2+, CaHCO

3+), including conversion of redox-sensitive species to

more stable forms consistent with the redox state of the system (e.g., NO3– to

NH3) At the heart of speciation models is a database containing the

thermo-dynamic properties for elements, ions in solution, solid phases, and gases from which the state of groundwater equilibrium is computed Speciation models also incorporate corrections for the effects of ionic activity and temperature

The saturation state of a specific phase in groundwater is determined by the saturation index (SI), which is defined by the relationship:

SI = log (IAP/K)

where K is the equilibrium constant of the reaction controlling formation of the phase and IAP is the ion activity product for the reaction (i.e., using the actual activities of the reaction species in groundwater) A state of equilibrium for the phase in the system is defined by SI = 0; SI > indicates oversaturation and SI < defines a condition of undersaturation

clay minerals nearly always have SI >> 0, even when similar clays are found as secondary minerals in the host rock

Phases that are known to form slowly, but are found to precipitate nonetheless complicate matters further For example, the abiotic reduction of SO42– to S2–

(with subsequent precipitation of sulfide phases) involves the transfer of eight electrons and is slow Typically, such reactions are considered unlikely during speciation modeling However, at the Portsmouth gaseous diffusion plant in Ohio, FeS was discovered in an ex situ, flow-through groundwater treatment canister filled with ZVI Investigation determined that the FeS resulted from microbial-catalyzed reduction of SO42– to S2– followed by precipitation of the phase

There-fore, the potential for microbial processes should not be ignored during modeling MINTEQA2 (Allison et al., 1991), EQ3/EQ6 (Wolery, 1992), and PHREEQC (Parkhurst and Appelo, 1999) are examples of the many geochemical models that can perform equilibrium speciation calculations for groundwaters and evaluate the saturation state for a large number of phases

AFRL (2000) contains an example of speciation modeling applied to a ZVI PRB at Naval Air Station (NAS) Moffett Field in Mountain View, California Groundwater samples were collected from monitoring wells placed in locations upgradient of, within, and downgradient of the PRB Using PHREEQC, it was possible to evaluate the changing saturation state of a variety of selected mineral phases for groundwater at different positions within the barrier system For example, upgradient groundwater at this site is observed to be close to equilibrium with respect to calcite (or aragonite) However, within the PRB near the upgra-dient side, calcite becomes oversaturated and then becomes sharply undersatu-rated within the PRB further in the downgradient direction A reasonable inter-pretation is that, as the water becomes oversaturated in calcite in the first few feet of the PRB, the phase precipitates Further downgradient of the PRB, chang-ing groundwater composition (e.g., decreaschang-ing concentrations of Ca2+ and

alka-linity) result in undersaturated conditions for the phase Analogous relationships are observed for other phases This modeling supports the general observation that most precipitation occurs near the sediment-PRB interface Speciation mod-eling is a qualitative and indirect method for evaluating the reactions that can occur within a PRB A more direct (and quantitative) approach is through reaction path modeling

2.3.4.2 Reaction Path Modeling

evolves and what phases (and their amounts) precipitate or redissolve at each step Computing the masses of different phases both dissolving and precipitating within PRBs as a function of reaction progress can yield estimates of the rate of pore space in-filling, an important consideration in predicting the lifetime of a barrier In an effort to match what is observed in the PRB, reaction path models permit the user to suppress the precipitation of phases that are not observed to form due to extremely slow kinetics Reaction kinetics can be incorporated into reaction path modeling if sufficient information about reaction rate equations is available (Gunter et al., 2000) Examples of geochemical models that have the ability to perform reaction path calculations include PHREEQC (Parkhurst and Appelo, 1999), EQ3/EQ6 (Wolery, 1992), and Geochemist’s Workbench (Bethke, 1994)

Reaction path modeling has been applied to the PRB at NAS Moffett Field and is reported in AFRL (2000) In contrast to results from speciation modeling at the barrier described above, reaction path modeling is able to provide a vivid conceptual picture of the complex interplay of the impacts of iron dissolution to groundwater chemistry (e.g., large, rapid increase in pH and decrease in redox potential) and associated onset of precipitation of new phases with increasing reaction progress (e.g., siderite, FeS2, aragonite and magnesite, followed by

Fe(OH)3 and green rust) In the case of aragonite, magnesite, and siderite,

pro-gressive decreases in Ca2+, Mg2+, Fe2+, and alkalinity concentrations in the later

stages of reaction progress lead to eventual dissolution of these early formed phases Sequential changes in the amounts of solid phases formed and dissolved constituents in groundwater can be graphically and quantitatively tracked In this example, a small number of plausible phases were selected by the modeler to participate in the reactions; all others were suppressed Clearly, the degree to which the model represents what happens in the PRB is dependent on this selection process, although other factors are important as well

Although reaction kinetics can be incorporated into reaction path modeling, it is not possible to develop a temporal and spatial picture of reactions occurring within a PRB without explicitly including flow and transport The following sub-section examines this more advanced approach to geochemical modeling

2.3.4.3 Reactive Transport Modeling

Reactive transport modeling is the most sophisticated form of geochemical mod-eling currently in use and incorporates groundwater flow, solute transport, and geochemical reactions in a fully coupled modeling system The primary advantage of coupled modeling is that there is an added degree of realism because the element of time is explicitly included in the simulations As a result, reaction kinetics can be incorporated into the modeling scheme and, in principle, the changes in groundwater chemistry and phases precipitating can be seen as a function of time (and space) as a packet of water passes through a PRB

not only are the thermodynamic data important, but kinetic relationships for the key phases, and sorption relationships for important dissolved species are needed In addition, hydraulic information from which the flow equations can be simulated is required The amount and variety of data needed to take full advantage of the capabilities of modern coupled models is significant and probably not something to be expected for normal PRB design activities However, when used as a research tool capable of capturing the important components of the site hydrology and geochemistry through sensitivity analysis, this type of modeling can be helpful in determining what level of sophistication of modeling is really necessary PHREEQC (Parkhurst and Appelo, 1999), MIN3P (Mayer, 1999), OS3D (Steefel and Yabusaki, 1996), and the Geochemist’s Workbench (Bethke, 1994) are examples of coupled flow, transport, and geochemical models

Continuing with the example provided by the PRB at NAS Moffett Field, Yabusaki et al (2001) presented results of a reactive transport investigation using the OS3D modeling code (Steefel and Yabusaki, 1996) For this study, one-dimensional transport was used As in the other modeling methods described above, a set of plausible phases was selected by determining if the phases were undersaturated in the background groundwater and oversaturated in the PRB Phases such as Fe(OH)3(am), Fe(OH)2(am), siderite, aragonite, and green rust were

included Reaction rate equations were selected for the key reactions to be modeled Hydraulic data was selected; dispersion and diffusion were ignored Although not without some problematic results, reactive transport modeling succeeded in providing improved understanding of the inter-relationship between transport and reaction rates occurring in the field

A second example of reactive transport modeling, applied to a ZVI PRB in Elizabeth City, North Carolina, was illustrated by Mayer et al (2001) In this example, Cr(VI) and TCE were the target contaminants The modeling results were able to closely match the observed behavior of these contaminants within the barrier, as well as that of other groundwater chemistry parameters such as pH and various dissolved inorganic species In addition, this modeling approach offered the possibility of hypothesizing important processes associated with more complex reactions

2.3.4.4 Inverse Modeling

nonunique and, being strictly a mass balance or mathematical operation, thermo-dynamics does not enter into the simulations One application of this methodology that might be used to help understand possible reactions in a future PRB is a flow-through column Because it is not thermodynamically based, issues such as reaction kinetics, inhibitions, and microbial catalysis not enter into the mod-eling If the list of plausible phases is inclusive, the combinations of phases and amounts of them can be determined mathematically to account for changes in groundwater chemistry It is up to the user to determine which, if any, of the acceptable solutions is realistic

NETPATH (Plummer et al., 1994) and PHREEQC (Parkhurst and Appelo, 1999) are two models that perform mass balance modeling

An example of an inverse modeling example based on data from the NAS Moffett Field PRB is presented in AFRL (2000) In this example, compositions of an upgradient groundwater and of a groundwater inside the PRB 0.5 feet from the upstream side of the PRB were selected for modeling Eight plausible phases were selected Only ZVI was allowed to dissolve; the other phases [Fe(OH)3,

siderite, marcasite, brucite, aragonite, magnesite, and CH4] were only allowed to

form Using the mass balance capability of PHREEQC, four acceptable solutions resulted that successfully related the two groundwater compositions Based on field observations at the PRB site, one of the models was rejected, but the remaining models were considered equally reasonable In general, one might imagine selecting the most probable model based on field observations of the identity and amounts of precipitated phases However, when the quantity of secondary phases is small, it is difficult to obtain an accurate estimate of their abundance with standard solids characterization techniques The relatively small number of solid samples that reasonably can be collected and analyzed from a PRB further compounds this difficulty Therefore, no selection among the three remaining models was possible

2.3.5 MODELING LIMITATIONSAND RESEARCH NEEDSOF PRBS

Geochemical modeling can be a powerful predictive tool when applied to PRB systems One major advantage of working with a ZVI barrier system in compar-ison to normal geologic systems is the inherent simplicity of the iron medium In effect, a single simple phase is present (Fe) that has well characterized physical, chemical, and thermodynamic properties However, as with any natural system (especially one with flowing groundwater), the simulation of multiple heteroge-neous reactions is not without complexities Attempts to make such models more realistic (temporally and spatially) have increased the burden on the modeler to obtain appropriate data For example, reaction rate equations, sorption relation-ships, hydraulic properties, and microbially catalyzed reactions are some of the complications that can be important

occurring in PRBs, especially as PRB materials age In spite of these limitations, experience gained investigating and modeling existing PRBs improves the under-standing of key processes, helps refine the selection of plausible phases, and helps focus on the most sensitive parameters to quantify

2.4 WALLS AND FLOORS

Both vertical and horizontal barriers can be used to mitigate the extent of sub-surface contaminant migration The most common type of vertical barrier is the slurry cutoff wall, whereas horizontal barriers can include constructed engineered barriers (floors) and natural geologic formations such as aquitards and aquicludes

2.4.1 VERTICAL BARRIERS

The three main types of cutoff walls used as vertical barriers for subsurface containment of contaminated groundwater are soil–bentonite (SB) walls, cement–bentonite (CB) walls, and composite slurry walls (CSWs) SB cutoff walls are constructed by displacing the bentonite slurry in an excavated trench by backfilling with a mixture of the bentonite slurry and the excavated trench spoils CB cutoff walls are constructed using a mixture of cement and bentonite slurry to maintain the stability of the excavated trench and then allowing the mixture to set to form the cutoff wall CSWs are constructed by inserting a geomembrane into the slurry along the centerline of the trench during construction In most applications involving the containment of contaminated groundwater, vertical cutoff walls are keyed into naturally occurring horizontal barriers formed by low-permeability geologic formations, such as aquitards or aquicludes, to impede contaminant migration beneath the zone of contamination

2.4.2 HORIZONTAL BARRIERS

In the case where a suitable aquitard or aquiclude is too deep, the construction of a horizontal barrier beneath the zone of contamination may be required The more common options for horizontal barrier construction based on existing tech-nologies are permeation grouting and jet grouting Permeation grouting involves the injection of a low viscosity slurry material into the ground through a series of overlapping injection wells Construction of a basal barrier by permeation injection requires the grout material to penetrate the soil completely and remain in place until solidification is complete For jet grouting, a single fluid (grout), two fluids (grout/air), or three fluids (grout/air/water) are injected into the soil under high pressure (over 300 atm) through a small orifice (1 to mm diameter) to erode or cut the soil and simultaneously place and mix the grout, resulting in a homogeneous columnar mass (e.g., Kauschinger et al., 1992; Tausch, 1992) Jet grouting is feasible in virtually all soil conditions ranging from clays to gravels (Kauschinger et al., 1992)

inter-penetrating grout discs In this application, a borehole is driven to the required depth, and a jet grout monitor is inserted The monitor is fitted with a high-pressure jet-cutting nozzle and can be rotated to cut a disc-shaped hole at the required depth The monitor is raised during cutting to form a disc of the required thickness Jet grouting can also be used to form inclined barriers (Dwyer et al., 1997) The barrier in this case consists of two honeycombed rows of interconnected vertical and inclined Portland-based grout columns forming a V-shaped trough in which the contamination is located The inside of the cement V-trough can be lined with a low viscosity, chemically resistant polymer to form a secondary barrier to contaminant movement This approach offers the advantage of not having to drill through the contamination

Another possible approach is to use horizontal directional drilling to form a barrier under the contamination (Sass et al., 1997) Although not yet practical, this technique potentially overcomes the shortcomings of vertical or inclined drilling and has been successful in contamination detection (Katzman, 1996; Anon, 2000)

2.4.3 CURRENT STATE OF PRACTICE FOR MODELING

PERFORMANCEOF WALLSAND FLOORS

Given the limitations of the prescriptive design approach, a performance-based design approach may be more appropriate to ensure successful containment In a performance-based design, individual properties or elements of a containment system are not prescribed Instead, the design is based on demonstration that the containment system will meet the overall objective For example, the overall objective may be to maintain the concentration of a target pollutant at a level below risk-based standards [e.g., drinking water maximum contaminant levels (MCLs)] at a downgradient POC (e.g., fence line, monitoring well) or point of exposure (POE) (e.g., drinking water well) Predictive contaminant transport modeling will be a critical component for demonstrating successful performance in this regard

This chapter provides a comprehensive description of the current state of the art for prediction of the performance of vertical cutoff walls for waste contain-ment The description is focused on applications in the saturated zone, with emphasis on performance of cutoff walls for containment of aqueous phase miscible contaminants However, given the recent interest in vadose zone walls, liquid- and gas-phase transport processes that govern the performance of vadose zone walls merit some consideration and are described in this chapter to provide a foundation for future development of design criteria and/or transport models for prediction of wall performance

the system approach, multiple components of the flow domain, such as a vertical cutoff wall located within an aquifer, are modeled as a system resulting in the need to consider the effect of heterogeneous media This approach generally requires the use of more complex semi-analytical or numerical models (e.g., finite difference, finite element)

These two approaches differ primarily with respect to the point at which the contaminant migration is evaluated, such as the POC or POE The component or analytical modeling approach is relatively simple to use, but is potentially con-servative in that the POC typically must be assumed to be located at the outer boundary of the barrier rather than at some other location downgradient of the containment location As a result, the design based on a component analysis may be too conservative and, therefore, ineffective in terms of cost

In most cases, environmental regulations allow for a POC at some location downgradient of the containment facility (e.g., the interface between a confined site and the upper confining layer, a property boundary, some other location) such that the impact of contamination reaching the outer boundary of the barrier is not the primary concern In such cases, the system approach using either semi-analytical or numerical models may be more appropriate (i.e., less conservative) and, therefore, may result in more cost-effective designs However, the systems approach generally requires more input data than the component approach, tend-ing to offset the difference in cost between the two approaches

2.4.4 CONTAMINANT TRANSPORT PROCESSES

Most barriers employed in geoenvironmental applications are designed to provide containment of dissolved contaminant plumes in groundwater Thus, the discus-sion herein is limited to processes that govern the liquid-phase migration of solutes The reader is referred to Bear (1972), Corey (1994), Pankow and Cherry (1996), and Charbeneau (2000) for information regarding the migration of immis-cible fluids

2.4.4.1 Aqueous-Phase Transport

Aqueous-phase contaminant transport in porous media is controlled by a variety of physical, chemical, and biological processes The primary physical processes governing miscible contaminant transport are advection, diffusion, and dispersion Diffusion tends to be the dominant transport process in relatively low flow rate situations, such as those that occur through clay barriers (Rowe, 1987; Shackel-ford 1988, 1989) However, advection and dispersion dominate in relatively high flow rate situations, such as contaminant migration through coarse-grained aquifer materials

the contaminants For example, the radioactive decay of the initial contaminant results in by-products that also can represent a potential adverse environmental impact Similarly, subsequent desorption of a previously adsorbed contaminant or dissolution of a previously precipitated contaminant can result in negative environmental impacts

Typically, only the physical processes of advection, diffusion, and dispersion, and the chemical processes of sorption and radioactive decay are included in practical modeling applications Although prototype models that include the more complicated chemical processes (e.g., oxidation/reduction, precipitation, hydrol-ysis, complexation, biodegradation) have been formulated, these models are con-sidered not yet suitable for routine use in practice (National Research Council, 1990) Thus, the development contained herein pertains primarily to modeling advection, dispersion, and/or diffusion with sorption and radioactive decay

For applications involving subsurface containment barriers, aqueous-miscible solute transport traditionally is described using the one-dimensional form of the advective-dispersive transport equation, in which the total flux of a solute, J, is represented as the sum of advective, diffusive, and dispersive fluxes, or:

(2.3)

where Ja is advective flux, Jd is diffusive flux, Jm is mechanical dispersive flux, qh [= khih, where kh = hydraulic gradient] is hydraulic liquid flux, C is the solute concentration, θ is the volumetric water content, which is equal to the porosity (n) for saturated media, D* is the effective diffusion coefficient, Dm is the coef-ficient of mechanical dispersion, vS (= qh/θ) is seepage velocity, D is the hydro-dynamic dispersion coefficient, and x is the direction of transport For one-dimensional transport, the hydrodynamic dispersion coefficient, D, can be expressed as follows (Freeze and Cherry 1979, Shackelford 1993):

(2.4) where αL is the longitudinal dispersivity of the porous medium in the direction of transport For transport through low permeability cutoff walls, the seepage velocity, vs, is often sufficiently low that mechanical dispersion is negligible

The governing equation for transient solute transport through porous media based on conservation of mass within a representative elementary volume (REV) is as follows:

(2.5)

J J J J q C D C

x v C D

C x

a d m h S

= + + = − ∂

∂ = −

∂ ∂

θ * θ θ

D D= *+Dm=D*+αL Sv

∂

∂ =−∇ ⋅ +

(θR C)

t J S

where Rd is the dimensionless retardation factor that accounts for instantaneous, linear, reversible adsorption of the solute to the solid phase, and S is a general source (>0)/sink (<0) term for other chemical and/or biological reactions For example, first-order decay of a chemical species can be included through a sink term as follows (van Genuchten and Alves, 1982):

(2.6) where Λ is a lumped decay constant [T–1] given as follows (Rabideau and

Khan-delwal, 1998a):

(2.7) where λw is the decay constant for first-order decay of contaminant in aqueous solution, and λs is the decay constant for first-order decay of contaminant on the solid phase The value of Rd is equal to unity for a nonreactive solute (i.e., Rd = 1) and greater than unity for a reactive solute (i.e., Rd > 1) For linear sorption, the retardation factor is written as:

(2.8) where ρd is the dry bulk density of the soil, and Kd is the distribution coefficient that relates the change in adsorbed concentration of a solute to a change in the liquid-phase solute concentration at equilibrium For organic contaminants, the distribution coefficient commonly is related to the organic carbon partition coef-ficient, Koc, as follows:

(2.9) where foc is the mass fraction of organic carbon in the soil The lumped decay constant assumes several reduced forms depending on special conditions For example, for nonadsorbing solutes (i.e., Rd = 1) or solutes that undergo decay only in the aqueous phase (i.e., λs = 0), Λ = λw, whereas for equal rates of decay in both the aqueous and solid phases (i.e., λw = λs = λ), Λ = λRd

Based on the assumptions that the porous medium is homogeneous, isotropic, and rigid (nondeformable); the water is incompressible; and the liquid flux is steady (i.e., vs = constant), the combination of Equations (2.3) through (2.8) results in the following form of the advection-dispersion reaction equation (ADRE) governing one-dimensional aqueous miscible transport:

(2.10) S=−θΛC

Λ=λw +λs(Rd−1)

Rd = +1 ρd Kd θ

Kd=Koc⋅foc

R C

t D

C

x v

C

x C

d∂ S

∂ =

∂

∂ −

∂

∂ −

2

Equation (2.10) is often presented in dimensionless form:

(2.11)

where:

(2.12)

where L is the barrier thickness

Solution and application of Equation (2.11) requires specification of initial and boundary conditions Consider conditions as illustrated in Figure 2.6, which can be viewed as a vertical cross section for a horizontal barrier or a plan view of a vertical wall A complete solution of transport in this system requires a two-dimensional (or 3-D) description However, a very common approach is to only consider one-dimensional contaminant transport through the barrier and to choose boundary conditions that are an approximation of reality These boundary con-ditions consist of entrance boundary concon-ditions on the inside of the barrier and exit boundary conditions on the outside of the barrier

The most common entrance boundary condition used is the first type, con-sisting of specification of a temporally constant concentration at the inlet (x = 0):

C (0,t) = C0 (2.13)

FIGURE 2.6 Boundary conditions for horizontal (vertical cross-section) or vertical barrier

(plan view)

Entrance: C(0, t)

L Horizontal or vertical barrier

Groundwater flow Exit: C(L, t)

Contaminated material

∂

∂ =

∂

∂ −

∂

∂ −

C t

C

x Pe

C

x C

* *

* *

*

* * *

2

2 Λ

C C

C t

tD R L x

x L Pe

v L D

L D d

s

* * * *

; ; ; ;

= = = = =

0

2

van Genuchten and Alves (1982) also give boundary conditions that represent an exponentially decaying entrance boundary concentration Third-type boundary conditions consist of specifying that the sum of the advective and dispersive fluxes away from the boundary is constant, given by the product of temporally constant fluid velocity and contaminant concentration:

(2.14)

These boundary conditions are appropriate for advection-dominated transport typical of laboratory columns, but can give inaccurate flux predictions for the low flow conditions typical of low permeability barriers (Rabideau and Khandel-wal, 1998a) Rowe and Booker (1985a) gave a modified third-type boundary condition that assumed finite initial mass in a completely mixed source zone and decreasing source concentration due to transport into the barrier This boundary condition was also used by Rabideau and Khandelwal (1998b)

The most common outlet boundary conditions used for general modeling of contaminant transport are semi-infinite boundary conditions of first type:

(2.15) or second type:

(2.16)

The use of a semi-infinite boundary condition implies no change in material properties and flow perpendicular to the barrier on the aquifer side of the barrier However, in many cases such as groundwater flow beneath a liner or parallel to a vertical barrier, contaminants leaving the barrier are diluted, thus reducing the concentration at the exit boundary and increasing the concentration gradient and diffusive fluxes In these cases, a semi-infinite exit boundary condition (of any type) is not appropriate because contaminant fluxes through the barrier would be underestimated

If the flow rate in the aquifer is rapid relative to the contaminant flux across the barrier, the contaminant concentration at the barrier-aquifer boundary can approach zero and a first type boundary condition may be appropriate:

C(L,t) = (2.17)

Rowe and Booker (1985a) gave boundary conditions that accounted for the rate of contaminant removal at the barrier-aquifer interface as a function of groundwater velocity, diffusion coefficient, barrier width, and aquifer depth Equation (2.17) represents a more conservative assumption as it maximizes flux

vC vC t D C

t 0= 0( , )0 − ∂∂

C( , )∞t =0 ∂ ∞

∂ =

C t

x ( , )

When transport is advection dominated, a second-type boundary condition is sometimes applied at the barrier-aquifer interface This application is not appro-priate for systems where diffusion is significant because it assumes that diffusion across the boundary is negligible (Rabideau and Khandelwal, 1998a)

2.4.4.2 Coupled Solute Transport

Although solute transport analyses for engineered soil barriers of low hydraulic conductivity typically are based on advective-dispersive theory as described above, advective-dispersive transport theory represents a limiting case of the more general coupled flux transport theory in that coupling terms (e.g., chemico-osmosis) are assumed to be negligible (e.g., Yeung, 1990; Shackelford, 1997) While advective-dispersive theory is considered acceptable for coarse-grained soils (e.g., aquifers), use of advective-dispersive transport theory for clay-rich soil barriers may not be appropriate For example, results of several laboratory studies have shown that some clay soils have the ability to act as membranes that restrict the transport of charged solutes (i.e., ions) (e.g., Kemper and Rollins, 1966; Olsen, 1969; Kemper and Quirk, 1972; Fritz and Marine, 1983; Malusis et al., 2001; Malusis and Shackelford, 2002a) This solute restriction also results in chemico-osmosis, or the movement of liquid in response to a solute concen-tration gradient, but opposite to the direction of solute diffusion (Olsen, 1969; Mitchell et al., 1973; Olsen, 1985; Barbour, 1986; Barbour and Fredlund, 1989; Neuzil, 2000)

Several solute transport models that account for the presence of soil mem-brane behavior have been developed (Bresler, 1973; Greenberg et al., 1973; Barbour and Fredlund, 1989; Yeung, 1990; Yeung and Mitchell, 1993; Malusis and Shackelford, 2002b) In all of these models, membrane behavior is repre-sented by a chemico-osmotic efficiency coefficient, ω, or reflection coefficient, σ, that ranges from zero (ω = σ = 0) for nonmembranes to unity (ω = σ = 1) for ideal membranes that completely restrict the passage of solutes (Staverman, 1952; Kemper and Rollins, 1966; Olsen et al., 1990; Keijzer et al., 1997) Clay soils that exhibit membrane behavior typically only partially restrict the passage of solutes (i.e., < ω, σ < 1) and, therefore, are considered nonideal membranes

In the absence of electrical current, the general expression for total coupled flux of a chemical species, j (neglecting mechanical dispersion), can be written as follows (Shackelford et al., 2001):

(2.18)

where Jha,j is the hyper-filtrated advective flux of solute j, Jπ,j is the chemico-osmotic counter-advective flux of solute j, Jd,j is the diffusive flux of solute j, and qπ is chemico-osmotic liquid flux The hyper-filtrated advective mass flux repre-sents the traditional advective transport term that is reduced by a factor of (1 – ω)

J J J J q C q C nD C

x

j ha j j d j h j j

j

= + + = − + − ∂

∂

due to the membrane behavior of the soil [i.e., Jha,j = (1 – ω)Ja,j] In physical terms, the factor (1 – ω) represents the process of hyper-filtration whereby solutes are filtered out of solution as the solvent passes through the membrane under an applied hydraulic gradient The chemico-osmotic counter-advective flux represents the transport of solutes due to chemico-osmotic liquid flux opposite to the direc-tion of diffusion (i.e., from low solute concentradirec-tion to high solute concentradirec-tion) The chemico-osmotic liquid flux, qπ, can be written as follows (Malusis and Shackelford, 2002c):

(2.19) where γw is the unit weight of water and π is chemico-osmotic pressure For dilute

chemical solutions, chemico-osmotic pressure, π, is related to solute concentra-tion in accordance with the van’t Hoff expression or (Tinoco et al., 1995):

(2.20)

where R is the universal gas constant [8.143 J/mol⋅K], T is the absolute temper-ature [K], N is the total number of solutes Thus, the summation term accounts for the concentrations of all chemical species in solution, including species j For example, if a solution contains the cation and anion of a binary, fully dissociating salt (e.g., sodium chloride), the chemico-osmotic pressure can be expressed as follows:

(2.21) where Ca and Cc are the concentrations of the salt anion and the salt cation, respectively Fritz (1986) notes that the van’t Hoff equation is valid for concen-trations up to 1.0 M for monovalent salts (e.g., sodium chloride, potassium chloride)

For transient transport, the total coupled solute flux equation (Equation (2.18)) is substituted into the continuity equation (Equation (2.3)) to yield the following:

(2.22)

where vπ is the osmotic seepage velocity that is related to the chemico-osmotic liquid flux, qπ, or

q K x h w π = ωγ ∂∂π π= = ∑ RT Ci

i N

1

π=RT C( a+Cc)

R C t v C x v C x C v x D dj j S j j j ∂ ∂ = − ∂ ∂ − ∂ ∂ − ∂ ∂ + ∂ ( ) *

ω π π

(2.23)

Equation (2.22) must be written separately for each chemical species in solution The equations are nonlinear and require a numerical method (e.g., finite difference, finite element) to solve for the concentration distribution of a solute within a cutoff wall at any point in time In addition, because the transport of a solute is dependent on the presence of other solutes in the system based on Equation (2.20), an iterative solution method must be utilized in conjunction with the following electro-neutrality constraint that must be satisfied at all points within the system:

(2.24)

where Ci is expressed in molar concentration and Zi is the ionic charge of solute i Further information with regard to solution of the coupled solute transport equations is given by Malusis and Shackelford (2002c)

2.4.4.3 Modeling Water Flow through Barriers

The simplest approach to modeling water flow through barriers is to apply Darcy’s Law, assuming one-dimensional, steady-state saturated flow If the hydraulic head be applied to single- or multiple-layer barriers In cases where head differences may not be known, a barrier is usually modeled as part of a larger system such as a landfill or a subsurface system For landfills, the most commonly used model is the HELP model (Schroeder et al., 1994a,b) The HELP model is an integrated

geomembrane The model predicts water buildup above the barrier layer to predict leakage rates through the barrier layers

To predict leakage rates through vertical barriers installed in complex hydro-geologic settings, a variety of numerical models can be employed, with MODFLOW (McDonald and Harbaugh, 1988) being the most widely used Particular attention must be given to grid discretization issues because barriers are often small in size in comparison to the total size of the system being simulated, and sharp hydraulic head gradients and changes in groundwater flow directions can occur in the vicinity of the barrier

When geomembranes are employed in horizontal and vertical barriers, fluid flow through defects may be the dominant mode of water flow and contaminant

v q K

g RT

C x h

w

i

i N

π = θπ =θ ρω  ∂∂

=

∑

1

C Zi i i

N =

=

∑

1

difference across the barrier is known, this approach is straightforward and can

transport Foose et al (2001) used MODFLOW to examine several analytical models (e.g., Giroud, 1992; Rowe, 1998) for leakage rates through composite liners They concluded that the existing analytical solutions had shortcomings and provided a series of recommendations for modification of these solutions for a variety of different conditions

2.4.4.4 Analytical Models

Shackelford (1988, 1989) identified three possible scenarios (Figure 2.7) for the design of low permeability walls based on a one-dimensional conceptualization of the system: (a) pure diffusion; (b) diffusion with advection and (c) diffusion against advection The pure diffusion scenario represents the limiting case Dif-fusive transport generally is significant through relatively thin (≤ m) barriers with K ≤ 10–7 centimeters per second (cm/s) and dominant through thin barriers

with K ≤ × 10–8 cm/s (Shackelford, 1988, 1989) The diffusion with advection

scenario occurs when water levels on the inside of the barrier exceed those on the outside The diffusion against advection scenario occurs when the water level is drawn down on the containment side of the wall to induce inward flow and reduce the outward flux of contaminants

A variety of analytical models can be applied to predict one-dimensional Selection of the appropriate model involves a choice of steady-state or transient conditions and the choice of the appropriate boundary conditions Analyses can be performed to evaluate contaminant concentrations on the aquifer side of the

FIGURE 2.7 Design scenarios for low permeability walls: (a) pure diffusion, (b) diffusion

with advection, (c) diffusion against advection (From Shackelford, C.D., 1989. Geotech-nical Engineering 1989, TRB, NRC, National Academy Press, Washington, DC, pp 169–182; Manassero, M and Shackelford, C.D., 1994. Rivista Italiana di Geotecnica, AGI, 28(1) With permission.)

Diffusion

(b) Advection

Diffusion

(c) Advection

+x +x +x

Diffusion

(a)

co c < co co c < co co c < co

barrier, or flux rates of contaminants through the barrier as a function of wall thickness; drawdown across the wall; and the associated pore water velocity, effective diffusion coefficient (influenced by soil type), and adsorption capacity of the wall

Rubin and Rabideau (2000) illustrated the impact of the Peclet number on steady-state fluxes through one-dimensional barriers with constant concentration entrance boundary condition, fixed-zero concentration exit boundary condition, and no decay For zero concentration at the barrier-aquifer interface, the flux rate, F, through the barrier is:

(2.25)

In dimensionless form this becomes:

(2.26)

The dimensional flux, F*, is thus the ratio of steady-state advective-dispersive flux to steady-state dispersive flux alone

When decay occurs as the contaminant is transported through the barrier, the dimensionless flux is given by (Rabideau and Khandelwal, 1998a):

(2.27)

For the case of no decay (Rubin and Rabideau, 2000), this simplifies to:

(2.28)

Consider a m thick barrier with a hydraulic conductivity of 10–9 m/s, a

porosity of 0.4, and an effective diffusion coefficient of 10–10 m2/s Assuming

negligible mechanical dispersion gives a Pe number of 25∆h, where ∆h is the hydraulic head difference across the barrier The impact of ∆h on flux is plotted results for zero hydraulic head gradient correspond to diffusive transport with varying rates of decay The figure illustrates the significance of advection and decay on contaminant flux rates

F v C D C

x s = − ∂ ∂     θ F FL DC PeC C x * * * * = = − ∂ ∂ θ F Pe Pe Pe * * * exp sinh =     + +

(2 )

4

2

2 Λ Λ F Pe Pe * exp( ) = − −

In the case of a decaying contaminant source (i.e., a finite source), design should be based on a transient analysis In many cases, the rate of increase of flux with time or the time to reach some critical concentration at the barrier-aquifer interface can be estimated, necessitating the use of a transient solution A variety of solutions have been used for transient analysis of barriers Many of these (Shackelford, 1989, 1990; Acar and Haider, 1990) are based on the Ogata Banks solutions for transient, one-dimensional contaminant transport with type-one entrance boundary conditions, and type-type-one semi-infinite boundary conditions Rabideau and Khandelwal (1998a) give transient solutions for the combination of type-one entrance boundary condition and perfect flushing exit boundary condition, as well as a variety of other boundary conditions The combination of a first-type entrance condition and flushing exit condition again resulted in the most conservative mass flux estimates at the exit end of the barrier In comparison, estimated flux rates from a model based on a third-type entrance boundary condition and a zero-gradient exit boundary condition were about a factor of 20 lower

to a vertical barrier is important, and the perfect flushing boundary condition is deemed to be too conservative, models that simulate the aquifer adjacent to the applied Equation (2.10) with a finite mass entrance condition to one-dimensional transport across the barrier layer They assumed a completely mixed aquifer with known depth and seepage velocity below the barrier and developed a Laplace transform solution that was inverted using a numerical Laplace inversion routine Manassero and Shackelford (1994) presented a steady-state solution for a similar conceptual model, but with a constant concentration entrance condition Rabideau and Khandelwal (1998a) developed a transient model (with numerical Laplace

FIGURE 2.8 Steady-state flux across a barrier as a function of hydraulic head gradient

0.0 0.5 1.0 1.5 2.0 2.5 3.0

0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.10

Hydraulic Head Gradient

Dimensionle

ss Flu

x

, F*

No decay Decay constant = Decay constant = 10

transform inversion) similar to that of Rowe and Booker (1985a), and showed that a range of mixing zone flow rates spanned behavior from the perfect flushing condition to that of a semi-infinite boundary condition Because transverse vertical dispersion coefficients can be quite small (on the order of molecular diffusion), definition of an appropriate mixing zone for use of one-dimensional solutions is an open question

In the case where greater accuracy in predicting contaminant transport through barriers is desired, multi-dimensional models should be used In this case, Rowe and Booker (1985b, 1986) developed Laplace transform solutions for transport through a variety of barrier–aquifer configurations, including a system with four layers in descending order consisting of a clay layer, a sand layer, a second clay layer, and a lower sand layer However, as with the one-dimensional analytical and quasi-analytical models, transport in the sand layers was assumed to be one-dimensional and fully mixed vertically

In many cases, horizontal and vertical barriers can include geomembranes Inorganic contaminants have negligible diffusion rates through intact geomem-branes, while organic compounds diffuse readily through geomembranes (Foose et al., 2001) Foose et al (2001) developed simplified analytical solutions for solute transport through composite liners incorporating a geomembrane and a soil layer Modeling the transport of contaminants through CSWs that contain geomembranes is, however, complicated by the possibility of defects in the geomembranes, particularly at joints Estimating contaminant flux rates through poorly characterized defects in geomembranes is difficult, but necessary for a realistic prediction of contaminant flux rates into the environment Foose et al (2002) evaluated the impact of geomembrane defects on contaminant transport using a finite difference numerical model and demonstrated the importance of defective joints on transport through composite geosynthetic clay liners

2.4.5 MODELING LIMITATIONSAND RESEARCH NEEDSOF WALLS AND FLOORS

Limitations to modeling barrier performance include difficulties in determining input parameters, dealing with measurement accuracy and uncertainty, and time-varying properties Models are also limited with respect to their ability to simulate coupled solute transport and membrane effects in clay soils

2.4.5.1 Input Parameters and Measurement Accuracy

As in the case of compacted clay soils, the basic parameters of CB and SB mixtures for slurry cutoff walls (SCOWs) refer to (1) hydraulic conductivity, (2) sorption and diffusion, and (3) compatibility The main factors influencing the hydraulic conductivity of CB mixtures are the solids content, curing time, confinement stress, and stress–strain behavior For the common basic composi-tions of CB mixtures, the hydraulic conductivity can vary from 10–8 to 10–9 m/s

(one order of magnitude change) The use of additives can lower these values by another order of magnitude (De Paoli et al., 1991)

The confinement stress history is an important factor that influences the hydraulic conductivity of CB mixtures (Manassero et al., 1995) The confinement stress in reducing hydraulic conductivity is more effective if applied at short curing times, although after 50 days it is also possible to observe hydraulic conductivity reduction due to an increase of confinement effective pressure For confining stresses ranging from 0.1 to 10 MPa applied on 28 days cured samples, the hydraulic conductivity varies between × 10–8m/s to 5.× 10–10 m/s (Manassero

et al., 1995)

Stress–strain behavior can influence the in situ hydraulic conductivity of SCOWs due to the possible stress and strain induced by surrounding ground movements The level of confinement stress can determine different kinds of mixture behavior under deviatoric stress as observed by Manassero et al (1995) The different kinds of mixture behavior are brittle-softening, brittle-hardening, ductile-softening, and ductile-hardening The latter is the most favorable to keep hydraulic conductivity low On the basis of preliminary experimental results (Manassero et al., 1995) and to obtain satisfying stress–strain behavior, the minimum effective confinement stress provided to the CSW CB backfilling mix-ture should be higher than 0.8 qu with qu being the unconfined compressive strength of the CB mixture

Some information on physico-chemical interactions between CB mixtures and chemical compounds to be contained by a SCOW can be found in Ziegler et al (1993), Gouvenot and Bouchelaghem (1993), Finsterwalder and Spirres (1990), Muller-Kirchenbauer et al (1991), Jessberger (1994), and Mitchell et al (1996) The diffusion parameters are fully comparable with the same parameters from compacted clay liners (CCLs) (on the order of 10–10 m2/s) even though the total

porosity of typical CB mixtures can be to 10 times greater From preliminary experimental results, the sorption capacity of CB mixtures seems to be rather effective for some organic pollutants and for anions in solution (Fratalocchi et al., 1996) This is probably due to the alkaline environment in the pore space However, further validation of these results is necessary

in the short term Jefferis (1992) and Tedd et al (1993) showed that only strong acids and/or sulfates could cause problems for the CB mixtures in the long term As far as the laboratory tests for compatibility assessment problems are concerned, further information is provided in Manassero et al (1995) In terms of further development, more research is needed to evaluate the temperature effects and desiccation problems, which can be particularly serious for SCOW mixtures

The hydraulic conductivity of SB slurry trench cutoff walls can be less than 10–9 m/s The stress state in the SB backfill can have a strong influence on

in-service hydraulic conductivity Also, contaminated permeants can increase the hydraulic conductivity of barrier soil, but the effect is less significant when the soil is under high confining pressure

To measure the hydraulic conductivity, it is possible to perform laboratory or in situ tests The main difficulty in performing the laboratory tests is to obtain undisturbed samples In fact, to bore into the wall after construction is difficult due to the soft consistency of the backfill

Large-scale pumping tests can be difficult to interpret due to the impact of gravity drainage from porous soils as the groundwater level is lowered within the contained area and the possibility of leakage through an underlying aquitard

2.4.5.2 Time-Varying Properties and Processes

Properties of barrier materials often change in time as barrier materials age A comprehensive experimental study on the curing time effect and solids content of CB mixtures was carried out by Fratalocchi (1996) The decrease in the hydraulic conductivity vs time was fitted by an exponential equation of the type:

(2.29)

where K and Kr are the hydraulic conductivities at time, t and tr, respectively, and the exponent, α is the coefficient of reduction of the hydraulic conductivity in time The α parameter for the best fitting functions has also been related to the cement to water ratio For other types of cement and/or bentonite, an inde-pendent determination of α is recommended (Manassero, 1996)

The reduction of SB SCOW hydraulic conductivity vs time is generally quicker than for CB mixtures Due to the short drainage paths, most of the consolidation process of an SB mixture occurs in a few months and, after this period, the hydraulic conductivity reduction is negligible On the other hand, an increase of hydraulic conductivity in the long term can occur using compatibility problems with the pollutants to be contained

2.4.5.3 Influence of Coupled Solute Transport

Vertical cutoff walls typically are designed to prevent or minimize the spread of a miscible contaminant plume in a groundwater aquifer as shown schematically

K K t

t r

r

= 

  

across the wall such that advective contaminant flux, Ja, typically occurs in the same direction as the diffusive flux, Jd(i.e., positive advection) (Figure 2.7a) As a result, breakthrough of the contaminant plume through the wall cannot be prevented without active pumping within the contaminant source area to reverse the hydraulic gradient such that the advective flux occurs opposite to the direction of diffusion (i.e., negative advection) (Figure 2.7b) In this case, contaminant breakthrough can be prevented if the diffusive flux is greater than the opposing advective flux

If the soil within the wall exhibits membrane behavior (i.e., ω > 0), chemico-osmotic flux of liquid from low concentration to high concentration (i.e., from the receptor side to the source side of the wall) results in a counter-advective contaminant flux, Jπ, that opposes Jd regardless of the direction of the hydraulic gradient The advective contaminant flux in response to the hydraulic gradient is reduced by the factor (1 – ω) due to hyper-filtration, as explained previously Thus, membrane behavior within the cutoff wall provides additional protection against contaminant breakthrough and can reduce the need for pump and treat to establish a counter-advective hydraulic gradient The coupled solute flux equation (Equation (2.22)) indicates that the significance of the potential benefit of mem-brane behavior depends on the magnitude of the chemico-osmotic efficiency coefficient, ω, for the soil

2.4.5.4 Membrane Behavior in Clay Soils

Membrane effects (e.g., hyper-filtration, chemico-osmosis) are attributed to elec-trostatic repulsion of charged solutes (ions) by the diffuse double layers of adjacent clay particles that extend into the pore space (e.g., Hanshaw and Coplen, 1973; Marine and Fritz, 1981; Fritz and Marine, 1983; Fritz, 1986; Keijzer et al., 1997) As stated previously, a soil membrane that completely restricts the trans-port of ions and, thus, exhibits a chemico-osmotic efficiency coefficient equal to unity (i.e., ω = 1) is considered an ideal membrane In this case, the diffusive double layers of adjacent particles overlap in the pore space, leaving no free space for solute transport However, values of ω for clay membranes typically fall within the range < ω < because the pores vary over a range of sizes relative to the thickness of the diffuse double layers such that not all of the pores are restrictive (Kemper and Rollins, 1966; Olsen, 1969; Bresler, 1973; Barbour, 1986; Barbour and Fredlund, 1989; Mitchell, 1993; Keijzer et al., 1997) Thus, the degree of solute restriction and the resulting value of ω is affected by a combination of physical and chemical factors, including the state of stress on the soil, the types and amounts of clay minerals in the soil, and the types and concentrations of the solutes (Kemper and Rollins, 1966; Bresler, 1973; Olsen et al., 1990; Mitchell, 1993) In general, ω increases with an increase in stress (lower porosity), an increase in the amount of high activity clay minerals, and a decrease in the valence and concentration of the solute

For example, consider the effects of salt concentration and ion valence illus-trated in Figure 2.9 (also see Kemper and Rollins, 1966) The results in Figure 2.9 pertain to recent tests performed at Colorado State University on bentonite-based geosynthetic clay liner (GCL) specimens using potassium chloride solu-tions, as described by Malusis et al (2001) and Malusis and Shackelford (2002a,c) The results illustrate that ω can vary over almost the entire range of < ω < For a given porosity (n), values of ω increase with an increase in the average salt concentration across the soil and a decrease in valence (Ca2+ vs Na+

or K+) Both of these trends are consistent with expected behavior, in that the

thickness of the diffuse double layers of adjacent clay particles within the soil pores decreases as the ion concentration and cation valence in the pore water increases (e.g., Mitchell, 1993)

The results in Figure 2.9 suggest that membrane behavior can be significant in clay soils containing an appreciable amount of sodium bentonite Sodium bentonite often is the principal clay mineral in soil-based vertical cutoff walls for waste containment due to the low hydraulic conductivity (e.g., ≤ 10–7 cm/s)

typically required in these applications Thus, consideration of membrane effects

FIGURE 2.9 Chemico-osmotic efficiency (From Malusis, M.A and Shackelford, C.D.,

2002b. Journal of Geotechnical and Geoenvironmental Engineering, ASCE, 128(2), 97–106 With permission.)

0.0 0.1 0.2 0.2 0.3 0.4 0.5 0.5 0.6 0.7 0.8 0.8 0.9 1.0

0.0001 0.001 0.01 0.1

n = 0.74 n = 0.78–0.80 n = 0.86 n = 0.80 n = 0.84 n = 0.91 n = 0.84 n = 0.84

Average Salt Concentration (M) CsCl

NaCl

Closed Symbols: Malusis and Shackelford (2002 b) Open Symbols: Kemper and Rollins (1966) Crossed Symbols: Kemper and Quirk (1972)

KCl NaCl

Chemic

o-Osmotic Efficienc

y Co

efficien

t,

may be warranted in predictive assessments of the waste containment perfor-mance of cutoff walls

2.5 COMPLICATING FACTORS

Several factors can complicate the rather simplified modeling approach described thus far A brief discussion of some of these factors is provided here to indicate a sense of the potential problems associated with contaminant transport modeling in environmental geotechnics

2.5.1 CONSTANT SEEPAGE VELOCITY ASSUMPTION

Shackelford (1997) has observed that a potentially significant limitation in the analytical modeling approach for modeling contaminant migration using models based on the ADRE is the assumption of a constant seepage velocity The assump-tion of steady-state flow is unrealistic, particularly in relatively short-term situaassump-tions A simplified analysis of the accuracy of the constant seepage velocity assump-tion for flow through a compacted clay liner was presented by Shackelford (1997) While admittedly limited in application, this analysis indicates that the constant seepage velocity assumption can lead to potentially unconservative results in that the depth of penetration of the contaminant front is underestimated in the short term Thus, care should be exercised when using models based on solution to the ADRE for modeling scenarios that not maintain a constant seepage velocity

2.5.2 CONSTANT VOLUMETRIC WATER CONTENT ASSUMPTION

Thus far, conditions in which the volumetric water content, θ, is constant (i.e., θ≠ f(t)) have been assumed However, complications arise in the formulation when the conditions initially are unsaturated and θ is not maintained constant [i.e., θ = f(t)], as in the case of transient infiltration processes In these cases, the time-dependent change in θ affects the magnitude of the seepage velocity [i.e., v = f(θ)] and the dispersion coefficient [i.e., D = f(θ)]

A number of investigators have studied diffusive transport through unsatur-ated soils, including sandy clay loam (Rowell et al., 1967), clay and loam (Porter et al., 1960), silt (Rowe and Badv, 1996a), sand (Lim et al., 1994; Rowe and Badv, 1996b), and gravel (Conca and Wright, 1990; Rowe and Badv, 1996b; and Badv and Rowe, 1996) The diffusion coefficient for chloride has been shown to decrease with decreasing θ In addition, Rowe and Badv (1996b) found that the following relationship could be used to estimate the effective diffusion coefficient in unsaturated soil, Dθ*.

(2.30) D

n D

θ θ θθ

*

min *

= −

−  

where θmin = minimum volumetric water content at which there is no

intercon-nected water through which diffusion can occur For some soils, Rowe and Badv (1996b) found that θmin≈ such that Equation (2.30) reduces to:

(2.31)

Rowe and Badv (1996b) also experimentally and theoretically examined advective-diffusive transport through layered systems consisting of a near-satu-rated lower permanent liner over an unsatunear-satu-rated layer (including silt, fine sand, fine gravel, and 38-millimeter stone) and demonstrated that conventional theory (e.g., Rowe and Booker, 1987, 1994) could be used to obtain good predictions of contaminant transport provided that allowance was made for the effect of θ on D* through Equation (2.30) For low degrees of saturation, such as for stone, dispersion was found to be significant in the unsaturated zone for q≥ 0.12 m/year, whereas diffusion was important for q = 0.017 m/year

The modeling formulation presented thus far includes only liquid-solid par-titioning The transport of volatile organic compounds (VOCs) also can lead to gas-liquid partitioning Gas-phase diffusion coefficients of VOCs typically are 104

to 105 times higher than the corresponding liquid-phase diffusion coefficients A

detailed description of the influence of variably saturated conditions is beyond the scope of this chapter; a comprehensive presentation on the subject is provided by Charbeneau and Daniel (1993)

2.5.3 ANION EXCLUSIONAND EFFECTIVE POROSITY

In some clay soils, anion exclusion or negative adsorption results from the repul-sion of anions from the negatively charged surfaces of clay particles(e.g., Bohn et al., 1985) The exclusion of anions from the pores of clay soils during transport contributes to an effective porosity effect in terms of anion migration That is, not all of the pore space is available for contaminant migration In such cases, the effective porosity effect is indicated by values of Rd < for nonreactive tracers, typically anions, where the actual value for Rd equals the effective porosity ratio, ne/n, where ne is the effective porosity and n is the total porosity (Shackelford, 1995a,b; Shackelford et al., 1997a,b) For example, Wierenga and van Genuchten (1989) attributed Rd values for Cl– of 0.78 to an effective porosity effect due to anion exclusion

2.5.4 NONLINEAR SORPTION

In formulating the adsorption process, linear sorption was assumed However, in many practical applications, the contaminant concentrations are sufficiently high such that nonlinear adsorption isotherms result In such cases, the partitioning between liquid and solid phases is a function of the pore water concentration In

D nD

θ θ

the case of nonlinear adsorption, the standard form of the ADRE (Equation (2.10)) is not strictly applicable, and more complex numerical procedures are required Approximate methods for utilizing analytical models with nonlinear adsorp-tion isotherms are given by Shackelford (1993) for column tests and by Manassero et al (1997) for diffusion tests These approximate methods simplify the analyses considerably and tend to provide reasonably good indications of the extent of contaminant migration, particularly for cases where the advective flow rates are low However, the approximate methods can result in significant errors in esti-mating the distribution of contaminants

2.5.5 RATE-DEPENDENT SORPTION

The local equilibrium assumption (LEA) for adsorption in the formulation of the retardation factor generally is valid when the reaction time between the adsorbate (contaminant) and the adsorbent (soil) is fast relative to the flow rate of the pore water Khandelwal et al (1998) examined the transport of organic solutes through SB barrier materials and concluded that deviations from local equilibrium were not likely significant However, for relatively high flow rates that typically occur in coarse-grained systems such as aquifers, the LEA may not be strictly valid In such cases, kinetic or rate-dependent (nonequilibrium) adsorption reactions may be required In general, incorporation of kinetic or rate-dependent reactions in the governing transport equations requires a numerical solution, although the semi-analytical finite-layer technique recently developed by Rabideau and Khan-delwal (1998b) may be used in some cases

The existence of a rate-dependent adsorption process tends to result in less adsorption than would be predicted based on the LEA Thus, failure to recognize the existence of rate-dependent adsorption can result in an underestimation of the actual extent of contaminant migration

2.5.6 ANION EXCHANGE

Some soils possess the ability to adsorb anions as well as cations In fact, anion

– –

nonreactive (i.e., Rd = 1) during transport may not be appropriate in all soils, particularly soils that contain appreciable amounts of the clay minerals with AEC > as shown in Table 2.5 Measured adsorption of anions, principally Cl–

and Br–, has been reported recently on the basis of both laboratory studies

(Shackelford and Redmond, 1995; Alshawabkeh and Acar, 1996) and field studies (Seaman et al., 1995)

Anion adsorption generally is attributed to a pH-dependent surface charge associated with exposed hydroxyls on the edges of clay minerals (layered sili-cates) and/or metal hydroxides (e.g., Fe2O3) In general, the exposed hydroxyls

carry either a partial positive charge (–OHδ+) at low pH or a partial negative

charge (–OHδ–) at high pH that results in the ability to adsorb either anions or

cations, respectively The partial charges on the exposed hydroxyls result from incomplete charge balance with the interior structure of the mineral The pH at which the net anion adsorption capacity equals the net cation adsorption capacity is referred to as the zero point of charge (ZPC) or the point of zero charge (PZC) Thus, anion adsorption is favored for pH < ZPC, whereas cation adsorption is favored for pH > ZPC However, as indicated by Shackelford and Redmond (1995), simultaneous adsorption of both anions and cations probably occurs during solute transport in soils that exhibit a pH-dependent surface charge

2.5.7 COMPLEXATION

Analyses performed using models based on the ADRE typically neglect the potential effects of complexation For example, when a metal, M2+, is dissolved

in water, the metal may exist in several different forms or complexes, such as free metal species, M2+, or as metal hydroxides such as MOH+ and M(OH)

2

These three different complexes migrate at different rates due to the difference in charges and sizes associated with each species For example, adsorption of these three species based on consideration of charge is expected to be favored in the following order: M2+ > MOH+ > M(OH)

2 However, in most simplified

modeling applications, the potential transport of complexed species is simply ignored and the transport is assumed to be associated with the principal free ionic form of the species (i.e., M2+) Thus, it is important to note that an evaluation of

the contaminant transport of a given species may actually encompass the transport of several different complexed species that not migrate at the same rate

2.5.8 ORGANIC CONTAMINANT BIODEGRADATION

Certain organic contaminants can diffuse readily through both geomembranes (e.g., Britton et al., 1988; Saleem et al., 1989; Park and Nibras, 1993; Park et al., 1995; Rowe et al., 1995, 1996a) and clay (e.g., Rowe et al., 1995, 1997) Con-sideration of biodegradation is important for modeling the potential impact of these contaminants Studies of biodegradation of VOCs in leachate are limited

TABLE 2.5

The Cation and Anion Exchange Capacities of Common Soil Minerals and Organic Matter

Soil Component CEC (cmol/kg) AEC (cmol/kg)

Illite (2:1) 10–40

Smectite (2:1) 90–120

Vermiculite (2:1) 100–150

Kaolinite (1:1) 3–15 3–5

Oxides/Hydroxides Below 5

However, Rowe (1995) and Rowe et al (1996b, 1997) showed that there can be substantial degradation both within the landfill and in the soil The rate of deg-radation can have a profound effect on the suitability of different liner systems (Rowe et al., 1996a)

2.5.9 TEMPERATURE EFFECTS

Although most of the transport parameters are functions of temperature, and a significant amount of study has been devoted to freeze-thaw effects on the hydraulic conductivity of fine-grained soils, temperature effects generally are ignored in modeling simulations that cover long-term effects However, in some applications, temperature effects may not be negligible For example, the large temperature gradients typically found between landfills (approximately 60°C) and the sur-rounding soil (approximately 20°C) may require incorporation of heat transport to provide accurate predictions of mass transport Therefore, the potential effects of temperature should be recognized when extrapolating results over extended periods that involve significant temperature fluctuations

REFERENCES

Acar, Y.B and Haider, L (1990) Transport of low-concentration contaminants in saturated earthen barriers Journal of Geotechnical Engineering, 116(7), 1031–1052 Adamson, D.T and Parkin, G.F (2000) Impact of mixtures of chlorinated aliphatic

hydrocarbons on a high-rate, tetrachloroethene dechlorinating enrichment culture

Environmental Science and Technology, 34(10), 1959–1965

AFRL (2000) Design guidance for application of permeable reactive barriers for ground-water remediation Prepared for the Air Force Research Laboratory, Tyndall AFB, FL, March 31

Albrecht, B.A and Benson, C.H (2001) Effect of dessication on compacted natural clays

Journal of Geotechnical and Geoenvironmental Engineering, 127(1), 67–74 Allison, J.D, Brown, D.S and Novo-Gradac, K.J (1991) MINTEQA2/PRODEFA2, a

geochemical assessment model for environmental systems USEPA Report EPA/600/3-91/021

Alshawabkeh, A.N and Acar, Y.B (1996) Electrokinetic remediation II: Theoretical model Journal of Geotechnical Engineering, ASCE, 122(3):186–196

Anderson, M.P and Woessner, W.W (1992) Applied Groundwater Modeling — Simulation

of Flow and Advective Transport, Academic Press, San Diego

Anon (2000) Real-time soil testing during horizontal drilling Civil Engineering, 70(3), 28 Badv, K and Rowe, R.K (1996) Contaminant transport through a soil liner underlain by an unsaturated stone collection layer, Canadian Geotechnical Journal, 33(2), 416–430 Barbour, S.L (1986) Osmotic Flow and Volume Change in Clay Soils Ph.D Dissertation,

University of Saskatchewan, Saskatoon 255 pp

Battelle (1998) Performance evaluation of a pilot-scale permeable reactive barrier at former Naval Air Station Moffett Field, Mountain View, California Prepared for Naval Facilities Engineering Service Center, Port Hueneme, CA, Nov 20 Bear, J (1972) Dynamics of Fluids in Porous Media, American Elsevier, New York Bear, J and Verruijt, A (1987) Modeling Groundwater Flow and Pollution, Reidel Publ

Co., Dordrecht

Benner, S.G., Blowes, D.W and Ptacek, C.J (1997) A full-scale porous reactive wall for prevention of acid mine drainage Ground Water Monitor Remediation, 17(4), 99–108

Benner, S.G., Blowes, D.W and Molson, J.W.H (2001) Modeling preferential flow in reactive barriers: Implications for performance and design Ground Water, 39(3), 371–379

Bethke, C.M (1994) The Geochemist’s WorkbenchTM, version 2.0, A User’s Guide to Rxn, Act2, Tact, React, and Gtplot Hydrogeology Program, University of Illinois Bethke, C.M (1996) Geochemical Reaction Modeling: Concepts and Applications

Oxford University Press, New York/Oxford

Bohn, H., McNeal, B and O’Connor, G (1985) Soil Chemistry, 2nd ed., Wiley, New York, 341 pp

Bostick, W.D., Jarabek, R.J., Slover, W.A., Fiedor, J.N., Farrell, J and Heferich, R (1996) Zero-valent iron and metal oxides for the removal of soluble regulated metals in contaminated groundwater at a DOE site K/TSO-35-Lockheed, P Martin Energy Systems, Inc., Oak Ridge, TN

Bowman, R.S., Li, Z., Roy, S.J., Burt, T., Johnson, T.L and Johnson, R.L (2001) Pilot test of a surfactant-modified zeolite permeable barrier for groundwater remedia-tion In J.A Smith and S Burns (Eds.), Physical and Chemical Remediation of Contaminated Aquifers Kluwer Academic/Plenum, New York, pp 161–185 Bradley, P.M and Chapelle, F.H (2000) Aerobic microbial mineralization of

dichloroet-hene as sole carbon substrate Environmental Science and Technology, 34(1), 221–223

Bresler, E (1973) Simultaneous transport of solutes and water under transient unsaturated flow conditions Water Resources Research, 9(4), 975–986

Britton, L.N., Ashman, R.B., Aminabhavi, T.M and Cassidy, P.E (1988) Prediction of transport properties of permeants through polymer films A simple gravimetric experiment Journal of Chemical Education, 64(4), 368–370

Charbeneau, R.J (2000) Groundwater Hydraulics and Pollutant Transport, Prentice-Hall, Upper Saddle River, NJ

Charbeneau, R.J and Daniel, D.E (1993) Contaminant transport in unsaturated flow (Chapter 15) In Maidment, D.R (Ed.), Handbook of Hydrology, McGraw-Hill, New York, pp 1–54

Conca, J.L and Wright, J (1990) Diffusion coefficients in gravel under unsaturated conditions Water Resources Research, 20(5): 1055–1066

Conca, J.L., Lu, N., Parker, G., Moore, B., Adams, A., Wright, J and Heller, P (2000) PIMS — Remediation of metal contaminated water and soils Proceedings of the Second International Conference on Remediation of Chlorinated and Recalcitrant Compounds, May 2000

ethane and chloroform in groundwater at a superfund site In Wickramanayake, G.B and Hinchee, R.E (Eds.), Remediation of Chlorinated and Recalcitrant Compounds, Vol 1, Battelle Press, Columbus, OH

De Paoli V., Marcellino P., Mascardi C and Paviani A (1991) Le Applicazioni dei Diaframmi nella Difesa Ambientale Proceedings of 15th Conference of

Geotech-nics of Torino, Engineers and Architects Society of Torino, Italy (in Italian) Dwyer, B.P., Heiser, J., Stewart,W and Phillips, S (1997) In situ containment of buried

waste at Brookhaven national laboratory Proceedings of the International Con-ference on Containment Technology, St Petersburg (Florida), pp 474–480 Ellis, D.E., Lutz, E.J., Odom, J.M., Buchanan, R.J., Bartlett, C.J., Lee, M.D., Harkness,

M.R and Deweerd, K.A (2000) Bioaugmentation for accelerated in situ anaer-obic bioremediation Environmental Science and Technology, 34(11), 2554–2260 Farrell, J., Kason, M., Melitas, N and Li, T (2000) Investigation of the long-term perfor-mance of zero-valent iron for reductive dechlorination of trichloroethylene. Envi-ronmental Science and Technology, 34, 514–521

Fenn, D.G., Hanley, K.J and DeGeare, T.V (1975) Use of the water balance method for predicting leachate generation from solid waste disposal sites, EPA/530/Sw-168, USEPA, Cincinnati, OH

Fennelly, J.P and Roberts, A.L (1998) Reaction of 1,1,1-trichloroethane with zero-valent metals and bimetallic reductants Environmental Science and Technology, 32(13), 1980–1988

Fiedor, J.N, Bostick, W.D., Jarabek, R.J and J Farrell (1998) Understanding the mech-anism of uranium removal from groundwater by zero-valent iron using x-ray photoelectron spectroscopy Environmental Science and Technology, 32, 1466–1473 Finsterwalder K and Spirres J (1990) Applications of a new mineral liner material for the encapsulation of a contaminated ground Proceedings of the International Conference Contaminated Soil ’90, Karlsruhe, Kluwer Academic, Dordrecht Foose, G.J., Benson, C.H and Edil, T.B (2001) Predicting leakage through composite

landfill liners Journal of Geotechnical and Geoenvironmental Engineering, ASCE, 127(6), 510–520

Foose, G.J., Benson, C.H and Edil, T.B (2002) Comparison of solute transport in three composite liners Journal of Geotechnical and Geoenvironmental Engineering, ASCE, 128(5), 391–403

Fratalocchi, E (1996) Comportamento idraulico delle miscele ternarie acqua-cemento-bentonite, Ph.D Thesis, Politecnico di Torino (in Italian)

Fratalocchi E., Manassero M., Pasqualini E., Roccato M (1996) Basic parameters for the design of cement-bentonite slurry cutoff walls Proceedings of The Impact of Industry on Groundwater Resources, GEAM, Torino

Fredlund, D.G and Xing, A (1994) Equations for the soil water characteristic curve

Canadian Geotechnical Journal, 31, 521–532

Freedman, D.L and Gossett, J.M (1991) Biodegradation of dichloromethane and its utilization as a growth substrate under methanogenic conditions Applied Envi-ronmental Microbiology, 57(10), 2847–2857

Freeze, R.A and Cherry, J.A (1979) Groundwater, Prentice Hall, Upper Saddle River, NJ Fritz, S.J (1986) Ideality of clay membranes in osmotic processes: a review Clays and

Fritz, S.J and Marine, I.W (1983) Experimental support for a predictive osmotic model of clay membranes Geochimica et Cosmochimica Acta, 47, 1515–1522 Fruchter, J.S., Cole, C.R., Williams, M.D., Vermeul, V.R., Teel, S.S., Amonette, J.E.,

Szec-sody, J.E and Yabusaki, S.B (1997) Creation of a subsurface permeable reactive barrier using in situ redox manipulation Proceedings of International Containment Technology Conference and Exhibition, St Petersburg, FL, Feb 9–12, pp 704–710 Fuller, C.C., Bargar, J.R., Davis, J.A and Piana, M.J (2002) Mechanisms of uranium interactions with hydroxyapatite: implications for groundwater remediation

Envi-ronmental Science and Technology, 36, 158–165

Gillham, R.W (1996) In situ treatment of groundwater: Metal-enhanced degradation of chlorinated organic contaminants In Aral, M.M (Ed.), Advances in Groundwater Pollution Control and Remediation, Kluwer Academic, New York, pp 249–274 Gillham, R.R and O’Hannesin, S.F (1992) Metal-catalyzed abiotic degradation of halo-genated organic compounds IAH Conference: Modern Trends in Hydrogeology, Hamilton, Ontario, May 10–13 pp 94–103

Giroud, J.P (1992) Rate of leakage through a composite liner due to geomembrane defects

Geotextiles and Geomembranes, 11(1), 1–28

Gouvenot, D and Bouchelaghem, A (1993) Barrieres antipollution souterraines Proceed-ings of the International Symposium Geoconfine ’93, Montpellier, A.A Balkema, Rotterdam, pp 207–212 (in French)

Greenberg, J.A., Mitchell, J.K and Witherspoon, P.A (1973) Coupled salt and water flows in a groundwater basin Journal of Geophysical Research, 78(27), 6341–6353 Gu, B., Liang, L., Dickey, M.J., Lin, X and Dai, S (1998) Reductive precipitation of uranium (VI) by zero-valent iron Environmental Science and Technology, 32, 3366–3373 Gu, B., Watson, D.B., Phillips, D.H and Liang, L (2002a) Biogeochemical, mineralogi-cal,and hydrological characteristics of an iron reactive barrier used for treatment of uranium and nitrate In Naftz, D.L., Morrison, S.J., Davis, J.A and Fuller, C.C (Eds.), Groundwater Remediation in Trace Metals, Radionuclides, and Nutrients,

with Permeable Reactive Barriers, Academic Press, New York

Gu, B., Watson, D.B., Wu, L., Phillips, D.H., White, D.C and Zhou, J (2002b) Microbi-ological characteristics in a zero-valent iron reactive barrier Environmental Mon-itoring and Assessment, 77, 293–309

Gunter, W.D., Perkins, E.H and Hutcheon, I (2000) Aquifer disposal of acid gases: modeling of rock-water reactions for trapping of acid wastes Applied

Geochem-istry, 15, 1085–1095

Hanshaw, B.B and Coplen, T.B (1973) Ultrafiltration by a compacted clay membrane — II Sodium ion exclusion at various ionic strengths Geochimica et Cosmochimica Acta, 37(10), 2311–2327

Hayes, J.J and Marcus, D.L (1997) Design of a permeable reactive barrier in situ

remediation system, Vermont site Proceedings of the In Situ Remediation of the Geoenvironment Conference 12, ASCE, Minneapolis, MN, Oct 5–8, pp 56–67 Ho, C.K and Webb, S.W (1998) Capillary barrier performance in heterogeneous porous

media Water Resources Research, 34(4), 603–609

Grouting, Soil Improvement and Geosynthetics, ASCE, New York

Jessberger, H.L (1994) Emerging problems and practices in environmental geotechnics

Proceedings of the 13th ICSMFE, New Delhi, Oxford and IBH Publishing, New Delhi

Katzman, M.M (1996) Directional drilling in environmental geotechnology Proceedings of the 3rd International Symposium on Environmental Geotechnology, San Diego, pp 659–668

Kauschinger, L.J., Perry, E.B and Hankour, R (1992) Jet grouting: State of the practice

Grouting, Soil Improvement and Geosynthetics, GSP No 30, ASCE, Vol 1, pp 169–181

Keijzer, Th.J.S., Kleingeld, P.J and Loch, J.P.G (1997) Chemical osmosis in compacted clayey material and the prediction of water transport In Yong, R.N and Thomas, H.R (Eds.), Geoenvironmental Engineering, Contaminated Ground: Fate of

Pol-lutants and Remediation, Thomas Telford, London, pp 199–204

Kemper, W.D and Quirk, J.P (1972) Ion mobilities and electric charge of external clay surfaces inferred from potential differences and osmotic flow Soil Science Society of America Proceedings, 36, 426–433

Kemper, W.D and Rollins, J.B (1966) Osmotic efficiency coefficients across compacted clays Soil Science Society of America Proceedings, 30, 529–534

Khandelwal, A., Rabideau, A.J and Shen, P (1998) Analysis of diffusion and sorption of volatile organic contaminants in soil/bentonite barrier materials Environmental

Science and Technology, 32, 1333–1339

Korte, N.E., Liang, L and Clausen, J (1995) The use of palladized iron as a means of treating chlorinated contaminants Emerging Technologies in Hazardous Waste Management VII, Extended abstracts for the special symposium, Atlanta, GA, pp 42–45

Lim, P.C., Barbour, S.L and Fredlund, D.G (1994) Laboratory determination of diffusion and adsorption coefficients of inorganic chemicals for unsaturated soil First International Congress on Environmental Geotechnics, Edmonton, Alberta, BiTech, Richmond, BC, Canada, pp 319–324

Malusis, M.A and Shackelford, C.D (2002a) Coupling effects during steady-state solute diffusion through a semipermeable clay membrane Environmental Science and Technology, ACS, 36(6), 1312–1319

Malusis, M.A and Shackelford, C.D (2002b) Chemico-osmotic efficiency of a geosyn-thetic clay liner Journal of Geotechnical and Geoenvironmental Engineering, ASCE, 128(2), 97–106

Malusis, M.A and Shackelford, C.D (2002c) Theory for reactive solute transport through clay membrane barriers Journal of Contaminant Hydrology, 59, 291–316 Malusis, M.A., Shackelford, C.D and Olsen, H.W (2001) A laboratory apparatus to

measure chemico-osmotic efficiency coefficients for clay soils Geotechnical Test-ing Journal, ASTM, 24(3), 229–242

Manassero, M and Shackelford, C.D (1994) The role of diffusion in contaminant trans-port through soil barriers Rivista Italiana di Geotecnica, AGI, 28(1)

Manassero, M., Fratalocchi, E., Pasqualini, E., Spanna, C and Verga F (1995) Contain-ment with vertical cutoff walls Key-Note Lecture, Proceedings of International Conference The Geoenvironment 2000, ASCE GSP no 46, New York

Manassero, M., Di Molfetta, A., Pacitti, M and Veggi, S (1997) The case history of a landfill for hazardous industrial wastes Key-Note Lecture, Proceedings of the 1st ANZ Conference, Geoenvironment ’97, Balkema, Rotterdam

Marine, I.W and Fritz, S.J (1981) Osmotic model to explain anomalous hydraulic heads

Water Resources Research, 17(1), 73–82

Mayer, K.U (1999) A numerical model for multicomponent reactive transport in variably saturated porous media Ph.D thesis, Dept of Earth Sciences, University of Waterloo, Waterloo, Ont., Canada

Mayer, K.U., Blowes, D.W and Frind, E.O (2001) Reactive transport modeling of an

in situ reactive barrier for the treatment of hexavalent chromium and trichloro-ethylene in groundwater Water Resources Research, 37(12), 3091–3103 Maymo-Gatell, X., Nijenhuis, I and Zinder, S.H (2001) Reductive dechlorination of

cis-1,2-dichloroethene and vinyl chloride by Dehalococcoides ethenogenes

Environ-mental Science and Technology, 35(3), 516–521

McDonald, M.G and Harbaugh, A.W (1988) A modular 3-dimensional finite-difference ground-water flow model: U.S Geological Survey Techniques of Water-Resources Investigations, Book 6, 586 pp

Mitchell, J.K (1993) Fundamentals of Soil Behavior, 2nd ed., Wiley, New York Mitchell, J.K., Greenberg, J.A and Witherspoon, P.A (1973) Chemico-osmotic effects

in fine-grained soils Journal of the Soil Mechanics and Foundations Division, ASCE, 99(SM4), 307–322

Mitchell, J.K with contributions from Filz, G (session co-leader), Anderson, L., Betsill, D., Carter, E., Davidson, D., Day, S., Esnault, A., Evans, J., MacMullen, S., Manassero, M., Martinenghi, L., Stamnes, R., Tamaro, G., van Meurs, G.A.M and Yang, D (1996) Design, construction, performance of soil- and cement-based barriers International Containment Technology Workshop Sponsored by U.S Department of Energy, U.S Environmental Protection Agency, DuPont, Baltimore Muftikian, R., Fernando, Q and Korte, N (1995) A method for the rapid dechlorination of low molecular weight chlorinated hydrocarbons in water Water Research, 29, 2434

Muller-Kirchenbauer, H., Rogner, J and Friedrich, W (1991) Isolamento delle fonti di inquinamento mediante diaframmi impermeabili e misure idrauliche Proceedings of International Conference Il risanamento di siti inquinati, Istituto per l’Ambi-ente, Milano (in Italian)

National Research Council (1990) Ground Water Models: Scientific and Regulatory

Appli-cations, National Academy Press, Washington, DC

Neuzil, C.E (2000) Osmotic generation of “anomalous” fluid pressures in geologic environments Nature, 403(6766), 182–184

O’Hannesin, S.F and Gillham, R.W (1998) Long-term performance of an in situ iron wall for remediation of VOCs Ground Water 36(1), 164–170

Olsen, H.W (1969) Simultaneous fluxes of liquid and charge in saturated kaolinite Soil

Geotechnical Journal, 22, 238–241

Olsen, H.W., Yearsley, E.N and Nelson, K.R (1990) Chemico-osmosis vs diffusion-osmosis Transportation Research Record No 1288, Transportation Research Board, Washington DC, pp 15–22

Orth, R.G and McKenzie, D.E (1995) Reductive dechlorination of chlorinate alkanes and alkenes by iron metal and metal mixtures Emerging Technologies in Hazard-ous Waste Management VII, Extended abstracts for the special symposium, Atlanta, GA, p 50

Pankow, J.F and Cherry, J.A (1996) Dense Chlorinated Solvents and Other DNAPLs in

Groundwater, Waterloo Press, Portland, OR

Park, J.K and Nibras, M (1993) Mass flux of organic chemicals through polyethylene geomembrane Water Environment Research, 65(3): 227–237

Park, J.K., Lakti, J.P and Hooper, J.A (1995) Effectiveness of geomembranes as barriers for organic compounds Proceedings, Geosynthetics ’95, 3, IFAI, February, pp 879–892 Parkhurst, D.L and Appelo, A.A.J (1999) User’s guide to PHREEQC (version 2)-A computer program for speciation, batch reaction, one-dimensional transport, and inverse geochemical modeling USGS Water-Resource Investigations 99-4259 Paschke, S.S and van der Heijde, P (1996) Overview of chemical modeling in

ground-water and listing of available geochemical models International Groundground-water Modeling Center, Report GWMI-96–01, Colorado School of Mines, Golden Peyton, R.L and Schroeder, P.R., Water balance for landfills, in Daniel, D.E (Ed.),

Geotechnical Practice for Waste Disposal, Chapman & Hall, New York, 1993 Phillips, D.H., Gu, B.,Watson, D.B., Roh, Y., Liang, L and Lee, S.Y (2000) Performance

evaluation of a zerovalent iron reactive barrier: Mineralogical characteristics

Environmental Science and Technology, 34(1), 4169–4176

Plummer, L.N., Prestemon, E.C and Parkhurst, D.L (1994) An Interactive Code (NETPATH) for Modeling NET Geochemical Reactions along a flow path USGS Water-Resources Investation 91-4078

Porter, L.K., Kemper, W.D., Jackson, R.D and Stewart, B.A (1960) Chloride diffusion in soils as influenced by moisture content Soil Science Society of America Pro-ceedings, 24(6), 460–463

Pruess, K (1991) TOUGH2 — A general-purpose numerical simulator for multiphase fluid and heat flow, LBL-29400, Lawrence Berkeley Laboratory, Berkeley, CA Puls, R.W., Paul, C.J and Powell, R.M (1999a) The application of in situ permeable

reactive (zero-valent iron) barrier technology for the remediation of chromate-contaminated groundwater: a field test Applied Geochemistry, 14:989–1000 Puls, R.W., Blowes, D.W and Gillham, R.W (1999b) Long-term performance monitoring

for a permeable reactive barrier at the U.S Coast Guard Support Center, Elizabeth City, North Carolina Journal of Hazardous Materials, 68, 109–124

Rabideau, A and Khandelwal, A (1998a) Boundary conditions for modeling transport in vertical barriers Journal of Environmental Engineering, ASCE, 124(11), 1135–1139 Rabideau, A and Khandelwal, A (1998b) Nonequilibrium sorption in soil/bentonite

barriers Journal of Environmental Engineering, ASCE, 124(4), 329–335

Roh, Y., Lee, S.Y and Elless, M.P (2000) Characterization of corrosion products in the permeable reactive barriers Environmental Geology, 40, 184–194

Rogers, V.C., Nielson, K.K and Kalkwarf, D.R (1984) Radon Attenuation Handbook for Uranium Mill Tailings Cover Design, NUREG/CR-3533, U.S Nuclear Regulatory Commission, Washington, DC

Rowe, R.K (1987) Pollutant transport through barriers Proceedings of ASCE Specialty Conference, Geotechnical Practice for Waste Disposal ’87, Ann Arbor, June, pp 159–181

Rowe, R.K (1995) Leachate characterization for MSW landfills Proceedings of the 5th International Landfill Symposium, Vol 2, Sardinia, Italy, pp 327–344

Rowe, R.K (1998) Geosynthetics and the minimization of contaminant migration through barrier systems beneath solid waste Keynote Lecture, 6th International Confer-ence on Geosynthetics, Atlanta, IFAI

Rowe, R.K and Badv, K (1996a) Chloride migration through clayey silt underlain by fine sand or silt ASCE Journal of Geotechnical Engineering, 122(1), 60–68 Rowe, R.K and Badv, K (1996b) Advective-diffusive contaminant migration in

unsatur-ated sand and gravel. ASCE Journal of Geotechnical Engineering, 122(12), 965–975.

Rowe, R.K and Booker, J.R (1985a) 1-D Pollutant migration in soils of finite depth

Journal of Geotechnical Engineering, ASCE, 111(GT4), 479–499

Rowe, R.K and Booker, J.R (1985b) 2D pollutant migration in soils of finite depth

Canadian Geotechnical Journal, 22(4), 429–436

Rowe, R.K and Booker, J.R (1986) A finite layer technique for calculating 3-dimensional pollutant migration in soil Geotechnique, 36(2), 205–214

Rowe, R.K and Booker, J.R (1987) An efficient analysis of pollutant migration through soil (Chapter 2) In Lewis, R.W., Hinton, E., Bettess, P and Schrefler, B (Eds).,

Numerical Methods in Transient and Coupled Systems, Wiley, New York, pp 13–42 Rowe, R.K and Booker, J.R (1994) POLLUTE v — 1D Pollutant migration through a non-homogeneous soil Geotechnical Research Centre, University of Western Ontario, London, Canada ©1983 (1990).

Rowe, R.K., Quigley, R.M and Booker, J.R (1995) Clayey Barrier Systems for Waste

Disposal Facilities E & F.N Spon, Eds., Chapman and Hall, London, 390 pp Rowe, R.K., Hrapovic, L and Armstrong, M.D (1996a) Diffusion of organic pollutants

through HDPE geomembrane and composite liners and its influence on ground-water quality Proceedings of the 1st European Geosynthetics Conference, Maas-tricht, October, pp 737–742

Rowe, R.K., Hrapovic, L., Kosaric, N and Cullimore, D.R (1996b) Biodegradation of dichloromethane in leachate North American Water and Environment Congress ’96, Anaheim, CA, June, paper 356 (CD Rom) p, Session GW-9, Groundwater Contaminant Fate and Transport

Rowe, R.K., Hrapovic, L., Kosaric, N., and Cullimore, D.R (1997) Anaerobic degradation of dichloromethane diffusing through clay Journal of Geotechnical Engineering,

ASCE, 123(12), 1085–1095

Rowell, D.L., Martin, M.W and Nye, P.H (1967) The measurement and mechanism of diffusion in soil III The effect of moisture content and soil solution concentration on the self diffusion of ions in soils Journal of Soil Science, 18, 204–222 Rubin, H and Rabideau, A.J (2000) Approximate evaluation of contaminant transport

through vertical barriers Journal of Contaminant Hydrology, 40, 311–333 Saleem, M., AsfourAfa, Dekee, D and Harrison, B (1989) Diffusion of organic penetrants

ronmental problems Proceedings of the International Conference on Containment

Technology St Petersburg (Florida), pp 489–498

Schroeder, P.R., Aziz, N.M., Lloyd, C.M and Zappi, P.A (1994a) The hydrologic eval-uation of landfill performance (HELP) model: User’s Guide for Version 3, EPA/600/R-94/168a, September 1994a, U.S Environmental Protection Agency Office of Research and Development, Washington, DC

Schroeder, P.R., Dozier, T.S., Zappi, P.A., McEnroe, B.M., Sjostrom, J.W and Peyton, R.L (1994b) The hydrologic evaluation of landfill performance (HELP) model: engineering documentation for version 3, EPA/600/R-94/168b, September 1994 U.S Environmental Protection Agency Office of Research and Development, Washington, DC

Seaman, J.C., Bertsch, P.M and Miller, W.P (1995) Ionic tracer movement through highly weathered sediments Journal of Contaminant Hydrology, 20, 127–143 Shackelford, C.D (1988) Diffusion as a transport process in fine-grained barrier materials

Geotechnical News, 6(2), 24–27

Shackelford, C.D (1989) Diffusion of contaminants through waste containment barriers, Transportation Research Record No 1219, Geotechnical Engineering 1989, TRB, NRC, National Academy Press, Washington, DC, pp 169–182

Shackelford, C.D (1990) Transit-time design of soil liners Engineering Geology, 29, 79–94

Shackelford, C.D (1993) Contaminant transport (Chapter 3) In Daniel, D.E (Ed.),

Geotechnical Practice for Waste Disposal, Chapman & Hall, London, pp 33–65 Shackelford, C.D (1995a) Analytical models for cumulative mass column testing In Acar, Y.B and Daniel, D.E (Eds.), Geoenvironment 2000, ASCE Geotechnical Specialty Publication No 46, ASCE, New York, 355–372

Shackelford, C.D (1995b) Cumulative mass approach for column testing Journal of Geotechnical Engineering, ASCE, 121(10), 696–703

Shackelford, C.D (1997) Modeling and analysis in environmental geotechnics: An over-view of practical applications In Kamon, M., (Ed.), 2nd International Congress on Environmental Geotechnics, IS-Osaka ’96, Vol 3, Osaka, Japan, Nov 5–8, Balkema, Rotterdam, pp 1375–1404

Shackelford, C.D and Redmond, P (1995) Solute breakthrough curves for processed kaolin at low flow rates Journal of Geotechnical Engineering, ASCE, 121(1), 17–32

Shackelford, C.D., Cotten, T.E., Rohal, K and Strauss, S (1997a) Acid buffering a high pH soil for zinc diffusion Journal of Geotechnical and Geoenvironmental Engi-neering, ASCE, 123(3), 260–271

Shackelford, C.D., Cotten, T.E., Davis, M.M., Strauss, S.H and Rohal, K.M (1997b) Characterizing zinc migration through a high pH sand-clay mixture XIV

Interna-tional Conference on Soil Mechanics and Foundation Engineering, Vol 3, Hamburg, Germany, Sept 6–12, Balkema, Rotterdam, pp 1935–1938

Shackelford, C.D., Malusis, M.A and Olsen, H.W (2001) Clay membrane barriers for waste containment Geotechnical News, 19(2), 39–43

Staverman, A.J (1952) Non-equilibrium thermodynamics of membrane processes Trans-actions of the Faraday Society, 48(2), 176–185

Tausch, N (1992) Recent European developments in constructing grouted slabs Grouting, Soil Improvement and Geosynthetics, GSP No 30, ASCE, Vol 1, pp 301–312 Tedd P., Paul V, Lomax C (1993) Investigation of an eight year old slurry trench cut-off

wall Proceedings of International Conference Green ’93, Bolton University, Bolton, Balkema, Rotterdam

Tinoco, I., Sauer, K and Wang, J.C (1995) Physical Chemistry, Prentice-Hall, Upper Saddle River, NJ

Thibodeaux, L.J (1981) Estimating the air emissions of chemicals from hazardous waste landfills Journal of Hazardous Materials, 4, 235–244

Thornthwaite, C.W and Mather, J.R (1957) Instruction and tables for computing potential evapo-transpiration and the water balance, Publications in Climatology, 10(3), Drexel Institute of Technology, Centerton, NJ

van Genuchten, M.T (1980) A Closed-form equation for predicting the hydraulic con-ductivity of unsaturated soils Soil Science Society of America Journal, 44(5), 892–898

van Genuchten, M.Th and Alves, W.J (1982) Analytical solutions for the one-dimensional convective-dispersive solute transport equation Technical Bulletin No 1661, USDA, Washington, DC

Vogan, J.L., Focht, R.M., Clark, D.K and Graham, S.L (1999) Performance evaluation of a permeable reactive barrier for remediation of dissolved chlorinated solvents in groundwater Journal of Hazardous Materials, 68, 97–108

Wierenga, P.J and van Genuchten, M.Th (1989) Solute transport through small and large unsaturated soil columns Ground Water, 27(1), 35–42

Wilson, G.W (1990) Soil evaporative fluxes for geotechnical engineering problems Ph.D Thesis, University of Saskatchewan, Saskatoon, Canada

Wilson, G.W (2002) Can apples be compared to cover systems Waste Geotechnics, March Wilson, G.W., Fredlund, D.G and Barbour, S.L (1994) Coupled soil-atmosphere

mod-elling for soil evaporation Canadian Geotechnical Journal, 31(2), 151–161 Wolery, T.J (1992) EQ3/EQ6, a software package for geochemical modeling of aqueous

systems, package overview and installation guide (version 7.0) Lawrence Liver-more National Laboratory, Technical Report UCRL-MA-110662(1)

Yabusaki, S., Cantrell, K., Sass, B and Steefel, C (2001) Multicomponent reactive trans-port in an in situ zero-valent iron cell Environmental Science and Technology, 35, 1493–1503

Yeung, A.T (1990) Coupled flow equations for water, electricity and ionic contaminants through clayey soils under hydraulic, electrical, and chemical gradients Journal

of Non-Equilibrium Thermodynamics, 15, 247–267

Yeung, A.T and Mitchell, J.K (1993) Coupled fluid, electrical, and chemical flows in soil Geotechnique, 43(1), 121–134

Zhu, C and Anderson, G (2002) Environmental Applications of Geochemical Modeling, Cambridge University Press, Cambridge

Ziegler, C., Pregl, O and Lechner, P (1993) Pollutant transport through sealing medium in contaminated land and sanitary landfills Proceedings of the 4th International Landfill Symposium, Sardinia ’93, Vol 2,S Margherita di Pula, Environmental Sanitary Engineering Centre (CISA), Cagliari, Italy, pp 1599–1612

143

3 Material Stability

and Applications

Prepared by* Craig H Benson

University of Wisconsin at Madison, Madison, Wisconsin

Stephan F Dwyer

Sandia National Laboratories, Albuquerque, New Mexico

3.1 OVERVIEW

This chapter focuses on material properties and behavior for caps, cutoff walls, and permeable reactive barriers (PRBs), with an emphasis on understanding the mechanisms and factors that affect their durability in full-scale systems Infor-mation obtained from laboratory tests are analyzed in this context The reader is referred to the preceding book in the containment series, Assessment ofBarrier ContainmentTechnologies (Rumer and Mitchell, 1995), as well as Daniel (1993), Gavaskar et al (1998), LaGrega et al (2000), Blowes et al (2000), Naftz et al (2002), and Reddi and Inyang (2000) for detailed information on the general characteristics of barrier materials mix design approaches and performance issues In this chapter, the emphasis is on fundamental factors and laboratory and field observations that relate to the long-term performance of materials used in con-structing various types of containment systems The overall performance of these systems has been analyzed holistically using the systems approach in Chapter

Chapter dealt with models of water and contaminant fate and transport through components of containment systems It is herein recognized that material properties

144 Barrier Systems for Environmental Contaminant Containment & Treatment play a significant role in overall system performance This chapter is divided into three primary subsections, each of which addresses materials performance for a specific type of containment structure

3.1.1 THE ROLE OF BARRIER MATERIAL MINERALOGY AND MIX

COMPOSITION ON PERFORMANCE

Earthen materials or geomaterials are the most frequently used materials in containment system barrier construction Generally, barrier mixes are composites of particles of various sizes and minerologies For barriers that are designed to minimize flow rates and retard contaminant solute transport through physico-chemical interactions, clays are commonly used in mixes with silts; sands; and amendments such as resins, activated carbon, slags, polymers, and ash The clays are usually alumino-silicates native to the barrier material, or they may be added to the barrier mix in cases where the natural clay content of the barrier material is insufficient to provide the required mix characteristics In other cases, barrier materials are fabricated and used to provide specific functions An example is a geomembrane that can be incorporated as a component into a containment structure for fluid retention, separation of clay to minimize the chance of attack by aggres-sive permeants, and diversion of gas flow to desirable control points Table 3.1 provides a general listing of various characteristics of barriers that affect classes of phenomena that relate to the most significant barrier design objectives Some of

TABLE 3.1

Containment System Design Considerations and Material Characteristics that are Usually Evaluated in Bench-Scale Tests

Physico-Chemical

Design Consideration Phenomena of Concern

Significant Barrier Material Properties

Reduction of contaminant release and transport

Advection Hydraulic conductivity

Density Moisture content Gradation Porosity Crack density

Diffusion Porosity

Dispersion Tortuosity

Leachability Crack density

Chemical compatibility Inadequate retardation Density

Physical durability Chemical attack Mineralogy relative to

contaminant chemistry

Radiation transport Density, mass attenuation

Material Stability and Applications 145 the barrier parameters such as hydraulic conductivity, porosity, and crack density apply to compacted, cemented, and fabricated materials

For granular barrier materials that may be compacted or cemented into barrier layers, the component material mineralogy and specific surface area are key material factors that, in combination with the emplacement density, control the initial and long-term barrier material textures when exposed to physical stresses and chemical contact Mineralogy controls the physico-chemical interactions (including the reactivity) of a barrier component with permeating fluids under a given environmental condition Under the most frequently encountered temper-ature, pressure, and pH–Eh conditions in the field, clays (comprising mostly aluminosilicates) react with permeants much more aggressively than sands (com-prising mostly silica) Because of their mineralogy, the charged clay surfaces present opportunities for the chemisorption of charged contaminants such as heavy metals as summarized by Inyang (1996) in Table 3.2

For a barrier material that has favorable mineralogy (i.e., a mineralogy that favors its interaction with permeating fluids in reactions that remove solutes without degrading the barrier), the opportunity for its interaction with the per-meant is enhanced if its specific surface is high The specific surface is the ratio of surface area to weight of a material, and it is inversely proportional to the grain size of the material For surface reactions like cation exchange and adsorp-tion that are prevalent in barriers, their role in increasing the contaminant distri-bution coefficients (i.e., cleaning the permeating fluid in terms of its entry vs exit chemistries) increases as the specific surface of the component material increases, as reflected in results plotted by Milne-Home and Schwartz (1989) presented in Figure 3.1 Often, even when a specific barrier component exhibits a desirable material characteristic, it may not be adequate with respect to another characteristic For example, a clay mineral such as sodium montmorillonite may be sorptive enough for heavy metals but inadequate in terms of providing strength against desiccation Yet still, cost considerations usually preclude the use of single-component barrier systems in waste containment Essentially, most barrier materials are composites, the proportions of which are designed to optimize performance characteristics at minimal cost In the case illustrated in Figure 3.2, D’Appolonia (1980) evaluated the effects of fines (% minus #200 sieve) on the permeability of soil-bentonite (SB) backfill candidate materials and found that for both plastic fines and nonplastic/low-plasticity fines, the permeability decreased as the fines content increased Permeability values for the plastic fines were generally lower than those of the nonplastic/low-plasticity fines Presumably, the plastic fines comprise more moisture-sensitive or expansive minerals than the nonplastic/low-plasticity fines Figure 3.3 shows the effects of bentonite (mont-morillonite) content on the permeability of the SB backfill candidate material mixes A bentonite content of 3% (by dry weight) was adequate to reduce the permeability values from × 10–5 to × 10–3 centimeters per second (cm/s) to

about 10–7 cm/s for well-graded coarse materials.

146 Barrier Systems for Environmental Contaminant Containment & Treatment

TABLE 3.2

Sorption Characteristics of Soil Minerals and Chemical Additives for Hazardous and Radioactive Metals

Single Component Material General Properties Metals Tested Test Type Test Conditions Results Montmorillonite (Garcia-Miragaya and Page, 1976) CEC = 94.8 meql/L; particle size <2 µm

Cd2+ Batch Initial pH

range = 4.6–7.3

95%, 95%, and 90% of Cd2+ sorbed by Na-, Ca-, and K-montmorillonite, respectively Montmorillonite from Texas (Puls and Bohn, 1988) Ca — saturated

Cd2+, Zn2+, Ni2+

Batch Initial pH =

5.5, 6.5, 7.5

50% of metals were adsorbed at pH range of 4–5.81 Vermiculite

(Ziper et al., 1988)

K — fixed, 500–1000 µm particle size, SSA = 22.5 m2/g

Cd2+ Batch Initial pH =

5.0, 10–9–10–5 M

0.9 moles of Cd2+ adsorbed per kg

Kaolinite (Puls and Bohn, 1988)

Fine particles Cd2+, Zn2+, Ni2+

Batch Initial pH =

5.5, 6.5, 7.5

Adsorption followed the order: Cd > Zn > Ni 50% of metals were adsorbed within pH range 4.49–5.80 Kaolinite (Yong

and Galvez-Cloutier, 1993)

LI = 61%, SSA = 24 m2/g; 84% below µm

Pb2+ Batch Initial pH =

3.0 g of Kaolinite in 40 mL of lead solutions

Maximum Pb2+

adsorption decreased at high pH due to precipitation Goethite (iron oxide) (Coughlin and Stone, 1995) SSA = 47.5 m2/g

Mn2+, Co2+, Ni2+, Cu2+, Pb2+

Batch Initial pH =

3–8 NaNO3

used to maintain selected ionic strength

Coordination chemistry of oxides affects adsorption 50% of Cu2+, Pb2+, Co2+, Ni2+ removed at pH 4.5, 4.8, 6.3, 6.8, respectively Goethite

(iron oxide) (Kuo, 1996)

Zn2+, Cd2+, Ca2+

Batch Initial pH =

5.3–8.3 NaNO3 used to maintain selected ionic strength

Material Stability and Applications 147

permeability ranges of three mix compositions for a fly ash cement-slurry wall, the results of which are presented in Figure 3.4 Test results developed (Fleming and Inyang, 1995) for fly ash amended materials, which may, in some cases, exhibit cementation if the ash mineralogy is favorable or some cementing agents are added, show that initial and longer term permeabilities of cemented barrier materials can be significantly influenced by reactions among the mix components

Figure 3.5 shows the conceptual textural patterns proposed by Fleming and Inyang (1995) in a comparative study of the effects of class F (nonreactive) fly ash and class C (reactive) fly ash amendment of barrier clay on changes in permeability under freeze-thaw action The patterns are similar, but the reactive fly ash exhibits initial and final permeabilities that are lower than those of the nonreactive ash

3.1.2 APPROACHES TO MATERIAL EVALUATION AND SELECTION

Bench-scale tests provide the best opportunity to evaluate the fundamental char-acteristics of barrier materials However, holistic assessments of a barrier system performance are most meaningfully performed through a combination of bench-scale testing and field quality assurance and monitoring tests The bench-bench-scale approach has been widely used to evaluate barrier material parameters in batch

TABLE 3.2 (continued)

Sorption Characteristics of Soil Minerals and Chemical Additives for Hazardous and Radioactive Metals

Single Component Material General Properties Metals Tested Test Type Test Conditions Results

Fly ash (Singer and Berkgaut, 1995) Hydrothermal ly treated, CEC = 2.5–3 meq/g

Pb2+, Sr2+, Cu2+, Zn2+, Cd2+, Cs2+

Batch Initial pH =

5.0 Total concentration of competing ions = 0.1 N

Selectivity order: Pb2+ > Sr2+ > Cu2+ > Cd2+ > Zn2+ > Cs2+ at 25 mg/L lead concentration, absorbed Pb = 35 µg/g Pyrolusite

(MnO2) (Ajmal et al., 1995)

Crushed samples

Pb2+, Cd2+, Zn2+, Mg2+

Batch Washed and

dried at 40°C; pH range of about 2–8

At pH = 6.5, 100% of initial 22.7 mg/L of Pb2+ was sorbed; other results show high sorption for Zn2+ and Cd2+ but low sorption for Mg2+

148 Barrier Systems for Environmental Contaminant Containment & Treatment

systems, monoliths of scaled down dimensions, or columns of media The latter can be densely compacted, as in the case of earthen materials considered for fluid/contaminant transport barriers or loosely emplaced as in reactive columns Most of the granular barrier material characteristics that are usually targeted are summarized in Table 3.3 Not all of these tests need to be performed for all barrier materials Some tests, exemplified by porosimetry, are not usually performed because the influence of the pore size distribution measured is represented along with barrier material density and reactivity with specific contaminants in data obtained from column tests for contaminant retardation coefficient estimation The tests listed in Table 3.3 have designations that vary from one country to another, although they are most standardized under the American Society for

FIGURE 3.1 Specific surface vs bulk cation exchange capacity for various sediments

and minerals (From Milne-Home, W.A and Schwartz, F.W., 1989.Proceedings of the

Conference on New Field Techniques for Quantifying the Physical and Chemical Proper-ties of Heterogeneous Aquifers, Dallas, Texas, pp 77–98 With permission.)

Sp

ec

ific sur

fa

ce

(m

2/g

.)

1000

100

10

1

0.1

0.1

Bulk C.E.C (meq/100 g)

10 100 Montmorillonite

Illite

Kaolinite

Explanation

American Petroleum Institute Reference clays (Patchett, 1975) Shales (Patchett, 1975) Milk River formation Mome L’Enfer, Erin formations Belly River formation (GENPAR 2) Sandstones

Discrete particle clays Pore ilning clays Pore bridging clays

1

1 1

2

2

Material Stability and Applications 149

rials such as geomembranes are tested under protocols that are different from those of granular barrier materials Fundamental tests are important because they can provide data that are helpful in performing a general durability evaluation of barrier materials and understanding mechanisms that are determinants of their durability

3.1.3 GEOSYNTHETICS AND THEIR DURABILITY IN BARRIER SYSTEMS

In general, the ability of barrier materials to retard fluid transport, resist chemical and biological attack, and maintain structural integrity under externally imposed stresses depends on their composition, emplaced thickness, and the quality assur-ance practices implemented during construction Early in the development of containment system design configurations, earthen and cementitious barrier mate-rials were used almost exclusively A more recent development, particularly within the past two decades, is an increase in the use of geosynthetic materials to enhance containment system barrier layer performance Both earthen and geosynthetic barrier materials have advantages and disadvantages Earthen bar-riers are most commonly clayey soils that are either compacted into layers as in landfills and surface impoundments or emplaced as slurry backfill as in slurry cutoff walls While they can retard contaminant transit through a variety of processes (e.g., sorption, induced precipitation of dissolved substances within inter-particle pore spaces), significant variability and uncertainty can exist in the

FIGURE 3.2 Effects of fines content on the permeability of soil-bentonite backfill (From

D’Appolonia, 1982 Proceedings of the 13th Annual Geotechnical Lecture Series,

Phila-delphia Section, American Society of Civil Engineers, Philadelphia, PA With permission.)

% Min

u

s #200 sie

ve

80 70 60 50 40 30 20 10

0 10−9 10−8 10−7 10−6 10−5 10−4

Plastic fines

SB Backfill permeability, cm/sec

Nonplastic or low plasticity fines

mate-150 Barrier Systems for Environmental Contaminant Containment & Treatment

spatial distribution of barrier transport parameters such as hydraulic conductivity and diffusion coefficient Furthermore, under aggressive chemical environments and sustained desiccation processes, earthen barriers can develop enlarged flow channels that allow contaminants in both the gaseous and liquid phases to travel through the barrier easily Geosynthetic materials such as geomembranes have less

FIGURE 3.3 Effects of bentonite content on the permeability of SB backfill (From

D’Appolonia, 1982 Proceedings of the 13th Annual Geotechnical Lecture Series,

Phila-delphia Section, American Society of Civil Engineers, Philadelphia, PA With permission.)

FIGURE 3.4 The effects of cement/water ratio and fly ash/cement ratio on the

perme-abilities of slurry wall mixtures (From Ryan, C.R and Day, S.R., 1986.Proceedings of the 7thNational Conference on Management of Uncontrolled Hazardous Waste Sites,

Washington, DC With permission.)

P

er

me

ability of SB b

ack

fill, cm/s

ec

10–2

10–3

10–4 10–5

10–6

10–7

10–8 10–9

0

% Bentonite by dry weight of SB backfill

3 Well-graded coarse

gradations (30–70% + 20 sieve) w/10 to 25% nonplastic fines

Poorly graded silty sand w/30 to 50% nonplastic fines

Clayey silty sand w/30 to 50% fines

Mix

Mix

Mix

10–7 10–6

K, (cm/sec.)

C/W FA/C Mix 0.20 0.00 Mix 0.20 0.24 Mix 0.25 0.60

10–5

Material Stability and Applications 151

variability in the spatial distribution of transport parameter magnitudes because they are manufactured in tightly controlled processes Furthermore, they are less permeable to fluids and offer the opportunity to minimize the overall design thickness of a barrier layer On the other hand, punctures, poor joints, and internal degradation can diminish their effectiveness as barrier layers Giroud et al (1992, 1997) have developed quantitative methods for estimating liquid transport through geomembrane defects

Geosynthetic barrier materials have been used as barrier layers that comple-ment the functions of earthen barrier layers Many composite cover designs such

FIGURE 3.5 Effects of reactions among barrier constituents on the permeability of

ash-modified clayey barrier soil subjected to freeze-thaw cycling (From Fleming, L.N and Inyang, H.I., 1995.ASCE Journal of Materials in Civil Engineering, 7(3), 178–182 With permission.)

Before freezing

After freeze - thaw cycling

a Class F fly ash-modified clay soil

c Class F fly ash-modified clay soil

P

er

me

ability

b Class C fly ash-modified clay soil

d Class C fly ash-modified clay soil

Longitudinal fracture Reactive ash particle Clay platelet Reacted rim

Nonreactive ash particle

PCA

POA

PCB

0 tCB

No of freeze-thaw cycles or time

tCA

152 Barrier Systems for Environmental Contaminant Containment & Treatment

as those consistent with the minimum design standards developed for the Resource Conservation and Recovery Act (RCRA), comprise both soil barrier layers and geosynthetic materials Othman et al (1997) have performed studies of the performance of such barrier configurations in the field The results indicate that with adequate quality control, such systems can perform effectively, at least within the few decades that they have been in service Another composite barrier system that typically produces desirably low hydraulic conductivities in barrier systems is the geosynthetic clay liner (GCL) that has been studied by many researchers (Estornell and Daniel, 1992; Rad et al., 1994; and Petrov et al., 1997) The GCL is gaining wider acceptance in the containment industry because of its cost effectiveness, relatively easy installation, and low barrier thickness Instal-Although test protocols, design methods, and quality assurance methods have been developed [Koerner and Daniel, 1997; Haxo, 1987; United States Environ-mental Protection Agency (USEPA), 1985], concerns about the long-term dura-bility of geosynthetic materials in barrier systems remain This concern is driven by the knowledge that all materials that are exposed to stressors degrade with time Such degradation in the long term is not limited to geosynthetic materials, but extends to emplaced earthen barrier materials as well For geosynthetic

TABLE 3.3

General Testing Approaches and Methods for Significant Characteristics of Batch and Compacted Barrier Materials

Dependent Property Test Method(s)

Soil Texture

Densitya Direct measurement

Dispersivity Indeterminate; evaluate experimentally

Gradation Sieve, hydrometer tests

Hydraulic conductivitya,b Permeameter tests

Moisture content Drying tests

Path length/tortuositya Indeterminate; evaluate experimentally

Plasticityb Atterberg limits

Pore size distribution Porosimetry

Porosity (effective)a Empirical methods, porosimetry

Soil Composition

Chemical (elemental) composition Chemical tests (e.g., x-ray fluorescence)

Mineralogy (crystallinity) Mineralogy tests (e.g., x-ray diffraction)

a Denotes a property dependent on compaction.

bDenotes a property dependent on mineralogy.

Source: Adapted from Inyang, H.I et al (1998) Physico-Chemical Interactions in Waste Containment Barriers, Encyclopedia of Environmental Analysis and Remediation, Vol 2, Wiley, New York, pp 1158–1165

Material Stability and Applications 153 materials that have been effectively installed, degradation mechanisms include aging, chemical attack, and photo-oxidation To assess the potential effectiveness of geosynthetic barriers in containment systems over 500 service time frames, Badu-Tweneboah et al (1999) analyzed prospective effects of various degradation processes of a 1.5-millimeter (mm)-thick high-density polyethylene (HDPE) geomembrane that was installed within a landfill cover They used data from studies performed on geomembranes and other polymeric materials to evaluate the damage potential under sustained contact with aging agents such as oxygen, microorganisms, heat, ultraviolet radiation, and radioactivity, as well as flaw development due to abrasion, thermal stresses, animal burrowing, and plant root penetration The analysis led to the conclusion that up to 5% reduction in yield strain can occur per 25 years of service, resulting in an estimated yield strain of zero if a liner deterioration pattern is assumed or 36% of the original yield strain in 500 years if a logarithmic deterioration pattern is assumed

On the basis of their analysis, Badu-Tweneboah et al (1999) estimated that the progressive stiffening of the geomembrane due to molecular rearrangement under induced stresses in common containment system configurations would likely result in stress cracking after 300 years of service The challenge is to relate the damage potential to flaw sizes and numbers — a necessary step for estimating potential fluid transport rates through geosynthetic materials

3.2 MATERIAL PERFORMANCE FACTORS IN CAPS

Caps or surface barriers in general are used to isolate buried wastes or contam-inated soils from the atmosphere and biota on the earth’s surface To design an effective cap, it is necessary to consider multiple objectives, including biota intrusion (i.e., intrusion of plants, animals, and humans into the underlying waste or contaminated soils), wind and water erosion, gas control, and percolation of water into underlying waste The material performance criteria established for each of these objectives depend on the type of waste to be contained and the risks imposed by the waste on the nearby environment For example, stringent mix design criteria may need to be used for facilities containing long-lived and toxic radioactive wastes, whereas less stringent criteria can be applied to facilities containing largely inert construction and demolition wastes The life span over which the cap must function is generally associated with the type of waste as well (e.g., 1,000 years for radioactive wastes or 30 years for solid wastes) In most containment applications, however, there is no intent of ever exhuming the waste Thus, a cap must meet the performance criteria as long as the material being contained poses a risk to the surrounding environment In most cases, this means that caps need to be designed for perpetuity and that a plan be in place to monitor and maintain the cap as needed

(e.g., erosion, biota intrusion, gas control) Two general cap designs are used: resistive designs and water balance designs Examples of resistive designs are shown in Figure 3.6; examples of water balance designs are shown in Figure 3.7 Resistive designs employ a barrier system with high hydraulic impedance to limit percolation (Benson, 2001) The barrier system can consist of geomembranes, fine-grained earthen materials, asphalt layers, or combinations of these or similar materials A drainage system is often used to limit the driving head on the barrier and ensure physical stability The water balance approach employs the store and release principle to limit percolation to an acceptable amount (Benson, 2001) Materials are selected that have adequate capacity to store infiltrating water during wet periods without appreciable percolation Vegetation is used to remove the stored water and return it to the atmosphere so that the cover has the capacity to store water during subsequent infiltration events

The resistive and water balance design approaches are fundamentally different The resistive design approach is predicated on constructing and maintaining a system that blocks natural water movement In contrast, the water balance approach

FIGURE 3.6 Profiles of caps relying on a resistive barrier: (a) Compacted clay barrier

and (b) composite barrier

Vegetated surface layer (150 mm)

Clay liner (>600 mm)

Clay liner (>600 mm) (a)

(b)

Geomembrane Geocomposite drain

uses natural processes to limit natural water movement The natural approach used for water balance covers is considered by some to be superior The logic is that a system that works with nature (i.e., water balance cap) is believed to be less likely to fail over the long term than a system that works against nature (i.e., resistive cap) However, currently there is no direct evidence demonstrating that one approach is superior, provided that the cap is designed and constructed properly

3.2.1 MATERIAL PERFORMANCE FACTORS IN COMPOSITE BARRIERS

Resistive designs generally employ engineered materials to provide the hydraulic impedance needed to meet a percolation criterion These materials include com-pacted natural clays, bentonites used alone in layers (e.g., as in a geosynthetics clay liner) or mixed with other earthen materials (e.g., a compacted mixture of sand and bentonite), polymeric sheets known as geomembranes, and asphalt and asphalt concrete layers (Koerner and Daniel, 1997) During the last decade, a wealth of experience has accrued regarding the characteristics of these materials and the elements that are required to reduce percolation to small amounts Expe-rience has shown that systems that rely solely on an earthen barrier (i.e., compacted

FIGURE 3.7 Schematic water balance caps: (a) Monolithic cap design and (b) two-layer

capillary barrier Finer-grained

soil

Coarser-grained soil Capillary

break

(b) Thick layer of

finer-grained soil

Waste

clay barrier or GCL) are prone to failure, even after short service lives, whereas composite designs that combine a geomembrane underlain by an earthen barrier appear to function extremely well, at least for the relatively short experience record (<10 years) that currently exists (Benson, 2001, 2002) The performance of caps that rely solely on a geomembrane or asphalt layer is largely unknown

The following two examples illustrate how resistive designs that rely solely on an earthen barrier can fail soon into their service lives One is a cap employing a compacted clay barrier consisting of 460 mm of compacted clay placed on compacted subgrade and overlain with 150 mm of topsoil vegetated with Bermuda and rye grasses This type of cap is often the presumptive remedy (i.e., the default design) for sites in the United States Superfund program, as was the case for the cap described here The other is a similar design, except a GCL was used instead of a compacted clay barrier, and 600 mm of “protective cover soil” was placed between the GCL and the topsoil layer The topsoil layer was vegetated with crown vetch to minimize erosion

The clay barrier was compacted in a manner that yielded a field hydraulic conductivity of x 10–8 cm/s at the time of construction (the design criterion was

10–7 cm/s) The cap was intended to transmit less than 30 mm/year of percolation.

Concerns about long-term cap performance led to installation of a system for monitoring all components of the water balance (Benson, 2002; Roesler et al., 2002) and, most importantly, the percolation rate Water balance data collected from the cap since the time of construction are shown in Figure 3.8

FIGURE 3.8 Water balance data for the clay cap

0 500 1000 1500 2000 50 100 150 200 250 8/1/02 Surface runoff Soil water storage No rain Drying soil W at er a pplie d, e va p otranspira tion, sur fac e r

unoff, and p

er cola tion (mm) S oil-w at er st orage (mm) Evapo-transpiration Percolation Applied water

Approximately 10 months after construction (September/October 2000), a period with little precipitation persisted for approximately six weeks During this period, the cap desiccated as evidenced by the monotonic decrease in soil-water storage during this period Prior to this period, the cap transmitted percolation at rate of approximately 30 mm/year, which is consistent with the design criterion Afterward, the percolation rate was approximately 500 mm/year (approximately one half of annual precipitation) Inspection of the clay barrier after it desiccated showed that the barrier contained desiccation cracks (Albright and Benson, 2002; Roesler et al., 2002) that served as preferential flow paths, causing the large percolation rate increases that were measured and the stair-step character of the cumulative percolation record

Concerns about the field performance of a cap that relies solely on a GCL also led to percolation rate monitoring using two 10 m by 10 m lysimeters (Thorstad, 2002) The cumulative percolation recorded by the lysimeters is shown in Figure 3.9 Excessive percolation was first noticed during the spring thaws of 1997 The GCL was exhumed in June 1997 and inspected to determine the cause of the excessive leakage rates GCL thinning due to pressure applied by gravel in the lysimeter was the suspected cause of the high percolation rate, but no quantitative assessment of the failure mechanisms was made A layer of sand was added to the lysimeter above the gravel as a cushion, a new GCL was installed, and the over-lying soil layers were replaced

Percolation monitoring continued after the lysimeters were rebuilt in 1997 Approximately 15 months after reconstruction, the percolation rate became exces-sive again Monitoring continued until October 1999, when one of the lysimeters (BL2) was exhumed to inspect the GCL Monitoring of the other lysimeter (BL1) continued Percolation recorded by lysimeter BL1 continued relatively steadily and averaged 211 mm/year

Inspection of the GCL exhumed from directly over lysimeter BL2 revealed that the bentonite was dry and cracked No thinning due to uneven pressure applied by the underlying soil was observed Hydraulic conductivity tests on samples of the GCL exhumed from inside and outside the lysimeter showed a saturated hydraulic conductivity ranging between 1.4 × 10–6 cm/s and 1.0 × 10–4 cm/s or

as much as 50,000 times the as-built hydraulic conductivity (2 × 10–9 cm/s).

Chemical analysis showed that the exchange complex of the bentonite was dominated by calcium and magnesium, whereas sodium was originally the pre-dominant cation (Thorstad, 2002) The exchange of calcium and magnesium for sodium reduced the swell potential of the bentonite sufficiently so that cracks that formed during drier periods could not swell shut during wetter periods As a result, the hydraulic conductivity of the GCL became unacceptably high

installation (2.4 mm/year on average), suggesting that the composite barrier is far superior to the GCL alone

Positive field performance of caps that employ a resistive design with a composite barrier has been reported by others as well (Melchior, 1997; Dwyer,

FIGURE 3.9 Profile (a) and cumulative percolation record (b) for GCL cap

Vegetated surface layer (150 mm)

Silty base layer (600 mm) Protective layer (600 mm) GCL

(a)

0 100 200 300 400 500 600 700 800

2000 Original BL1

Original BL2 Rebuild BL1 1st rebuild BL2 2nd rebuild BL2

Elapsed time (days) 1996

1997 1998

1st rebuild

2nd rebuild

BL2

2001 2000

(b)

C

u

m

u

la

tive p

er

cola

tion (mm)

1999

2001; Albright and Benson, 2002) Melchior (1997) reported percolation rates between 0.8 and 3.0 mm/year for a cap in Germany employing composite barrier design The barrier consisted of 600 mm of clay (saturated hydraulic conductivity less than 10–7 cm/s) overlain by a 1.5-mm-thick HDPE geomembrane, a sand

drainage layer 250 mm thick, and a vegetated topsoil layer 750 mm thick Dwyer (2001) reported an annual percolation rate of 0.1 mm/year for a cap in semi-arid Albuquerque, New Mexico, having a design similar to Melchior’s cap Dwyer (2001) also reported a percolation rate of 1.8 mm/year for a similar cap in Albuquerque employing a composite barrier with a GCL as the earthen component The USEPA’s Alternative Cover Assessment Program (ACAP) is also moni-toring the percolation rate from seven caps employing composite barrier layers consisting of a geomembrane underlain by a GCL or compacted clay barrier (Albright and Benson, 2002; Roesler et al., 2002) Percolation rates from these caps are summarized in Table 3.4 The percolation rates generally are near zero in semi-arid and arid climates, and less than mm/year in humid climates Thus, the composite barrier generally seems to be effective, largely because the geomembrane is nearly impervious and the fine-grained soil beneath the geomem-brane provides impedance to flow at points where the geomemgeomem-brane may contain defects

The exception is the cap located in Monterey, California This cap is located in a semi-arid environment, but is transmitting 18 mm/year of percolation (Table 3.4) The cover soil placed on the geomembrane for this cap consisted of soil from

TABLE 3.4

Summary of Precipitation and Percolation Rates from Caps with Composite Barriers Monitored by ACAP

Site

Duration

(Days) Climate

Total Precipitation

(mm)

Percolation (mm/year)

Altamont, CA 517 Arid 487 0.0 (0.0%)

Apple Valley, CA 156 Arid 115 0.0 (0.0%)

Marina, CA 684 Semi-arid 466 18.1 (3.9%)

Boardman, OR 485 Semi-arid and seasonal 181 0.0 (0.0%)

Polson, MT 847 Semi-arid and seasonal 744 0.2 (0.1%)

Cedar Rapids, IA 381 Humid and seasonal 772 0.9 (0.1%)

Omaha, NE 552 Humid and seasonal 719 3.7 (0.5%)

Percentage of precipitation in parentheses

demolition projects and contained a variety of debris, including reinforcing bars and angular chunks of concrete These materials may have caused puncturing of the geomembrane, which may be responsible for the higher percolation rates (Roesler et al., 2002) This example illustrates an important point: caps con-structed with suitable barrier materials can function poorly if other aspects of the design are not properly implemented

Although the performance record for caps with composite resistive barriers is good, the record is short relative to the life span over which the caps are intended to function Melchior’s study has the longest record (eight years) Dwyer’s record is four years, and the monitoring is continuing at ACAP sites In general, composite barriers that have been exhumed appear to be in excellent condition even after several years of service, including those barriers located in the arid desert in southern California (Corser and Cranston, 1991; Melchior, 1997) Additionally, several studies suggest that geomembranes should perform adequately for hundreds of years, if not longer (Hsuan and Koerner, 1998; Clarke, 2002; Rowe and Sargam, 2002) However, these predictions are primarily heu-ristic or based on ancillary measurements (e.g., depletion rate of anti-oxidants) The reality is that little hard data exist that can be used to make reliable predictions regarding the life span of geomembranes in composite covers Given the dearth of information on life expectancy, this is an area in need of research given that caps employing composite barriers are ubiquitous

3.2.2 MATERIAL PERFORMANCE FACTORS IN WATER

BALANCE DESIGNS

Water balance designs generally employ broadly graded finer-textured soils because of their capacity to store significant amounts of water with little drainage and their ability to deform without cracking Coarse-grained materials are also used to form capillary breaks that enhance storage in the finer layer or divert water under unsaturated conditions The coarse material can also be used to remove water from the barrier through advective drying (Albrecht and Benson, 2002; Stormont et al., 1994) Caps that employ a single layer of fine-textured soil are generally referred to as monolithic barriers, whereas those with two or more layers with contrasting particle size are referred to as capillary barriers (Figure 3.7)

has been monitored under natural conditions and conditions that are extremely wet for the region

Because a 1000-year life without maintenance was required, natural construc-tion materials that are known to have existed in place for thousands of years were selected The top-to-bottom profile consists of a 2-m-thick layer of vegetated silt-loam overlying layers of sand, gravel, basalt rock (riprap), and asphalt (Figure 3.10) Each layer serves a distinct purpose The silt-loam is for storing infiltration (600 mm of water can be stored in the silt loam before it will drain) and provides the medium for establishing plants that are necessary for transpiration The coarser materials placed directly below the fine soil layer create a capillary break that enhances the storage capacity of the silt-loam Placement of the silt-loam directly over coarser materials also creates an environment that encourages plants and animals to limit their natural biological activities to the near surface, thereby reducing biointrusion into the lower layers The coarser materials also help deter inadvertent human intruders The asphalt layer (asphalt concrete overlain by layer of fluid-applied asphalt) acts as a secondary barrier that employs a resistive approach to impede and divert water passing through the capillary break A shrub and grass cap was established on the cap in November 1994 Two sideslope configurations, a clean fill gravel on a 10:1 slope and a basalt riprap on a 2:1 slope, were also part of the overall design and testing

FIGURE 3.10 Hanford cross section of Hanford cap showing (a) interactive water balance

processes, (b) gravel sideslope, and (c) basalt riprap sideslope

Lateral drainge Upper neutron probe

access tube resistant gravel admix Runoff Existing grade (a) (b)

1 10

1 50

2 50

Clean fill side slope (pit run gravel)

(c)

Basalt side slope Vertical drainage Waste crib Precipitation (P) Evapo-transpiration Neutron probe access tube Upper silt w/admix 1.0 m Lower silt 1.0 m Sand filter 0.15 m Gravel filter 0.3 m

Basalt rock Riprap 1.5 m Drainage gravel

0.3 m Composite asphalt (asphaltic concrete coated w/fluid applied asphalt 0.15 m min.)

Top course 0.1 m Sandy soil (structural) fill

From November 1994 through October 1997, sections of the cap were sub-jected to an irrigation regime of three times the long-term average annual pre-cipitation, which included a simulated 1,000-year storm event (70 mm of water) during the last week of March for three years (1995 through 1997) Percolation did not occur from the cap until the third year, and then only a small amount (less than 0.2 mm) was transmitted from one section subjected to the enhanced irrigation treatment No drainage has occurred since then from this section or from any other portion of the cap In fact, the percolation that was recorded has been attributed to lateral flow from water diverted off an adjacent roadway rather than flow through the cap (USDOE, 1999)

Despite the large amount of water that was applied, all available stored soil water was removed from the entire soil profile by late summer each year by evapo-transpiration (Figure 3.11), which maintained the water storage in the silt-loam layer well below the estimated drainage limit of 600 mm If the silt-silt-loam thickness was reduced from m to 1.5 m, the storage data indicate that little or no percolation would be expected However, if the silt-loam thickness was

FIGURE 3.11 Temporal variation in mean soil water storage in the silt-loam in the

Hanford cap Monitoring was interrupted 1998–2000 Horizontal dashed lines represent estimated storage limits for caps with silt-loam layers m, 1.5 m, and 1.0 m thick (From USDOE, 1999 200-BP-1 Prototype Barrier Treatability Test Report DOE/RL-99-11, U.S Department of Energy, Richland, WA; Ward, A and Gee, G., 2000 In Looney, B and Falta, R (Eds.), Vadose Zone Science and Technology Solutions, Battelle Press, Columbus, OH, pp 1415–1423 With permission.)

9/30/1994

Date

100 200 300 400 500 600 700

Nonirrigated average Irrigated average

2.0 m silt loam

1.5 m silt loam

1.0 m silt loam Drainage

under natural conditions

W

at

er st

orage (mm w

at

er)

9/30/2002 9/29/2000

reduced to m, it appears that the cap would not perform well under extremely wet conditions

The cap tested at Hanford represents perhaps the most sophisticated and redundant type of water balance design ever considered The level of complexity associated with the cap is needed for the radioactive wastes that it is designed to isolate For many sites (e.g., municipal solid wastes, demolition debris, contam-inated soils), however, less sophisticated water balance caps are needed An assessment of more typical water balance caps is being conducted by ACAP under natural climatic conditions (Bolen et al., 2001; Albright and Benson, 2002) The caps tested by ACAP are intended to meet a target percolation rate that ranges between and 30 mm/year depending on the type of waste, the regulations in place at each site, and the climate (semi-arid or arid vs humid) Laboratory measurements of unsaturated and saturated soil properties were used in conjunc-tion with common methods accepted in practice to design each cap (Bolen et al., 2001) Typically, an unusually wet condition was used for the design calculations Percolation rates measured for the ACAP water balance caps as of April 2002 are summarized in Table 3.5, along with the design percolation rates Nine monolithic barriers and five capillary barriers are being evaluated The design criterion is being achieved at eight of the 10 semi-arid sites, but at none of the humid sites The factors contributing to the higher than anticipated percolation rates are currently under evaluation, but the data illustrate that water balance caps not necessarily perform as intended

One key factor contributing to the higher than anticipated percolation rates appears to be the influence of pedogenesis on hydraulic properties near the surface Samples are currently being collected from the surface of each test section as large undisturbed blocks to characterize the hydraulic property changes that have occurred A summary of the saturated hydraulic conductivity measurements obtained to date is provided in Table 3.6 The saturated hydraulic conductivity has increased due to factors such as desiccation and root penetration at three of the four sites for which tests have been conducted At the fourth site, the hydraulic conductivity has remained about the same Understanding how the hydraulic properties change over time is critical to predicting how water balance caps will perform over the long term Long-term performance prediction is an issue in need of research before water balance caps can be considered a long-term solution for containment Another important issue probably contributing to higher than antici-pated percolation rates is scaling between hydraulic properties measured in the laboratory and those operative in the field Additional study of scaling issues and how they can be incorporated in design is needed to understand long-term cap performance

3.2.3 COUPLING OF VEGETATION AND MATERIAL

PERFORMANCE FACTORS

or Environmental Contaminant Containment & T

reatment

Design and Measured Percolation Rates and Precipitation Summary for Water Balance Caps Monitored by ACAP

Site

Design Criterion (mm/year)

Duration

(Days) Climate Cover Type

Total Precipitation

(mm)

Percolation (mm/year)

Altamont, CA 517 Arid Monolithic barrier 487 1.0 (0.3%) Apple Valley, CA 156 Arid Monolithic barrier 115 0.0 (0.0%) Marina, CA 684 Semi-arid Capillary barrier 466 61.8 (13.3%) Sacramento, CA 847 Semi-arid and seasonal Monolithic barrier 1080 mm thick 744 48.4 (11.1%) Monolithic barrier 2450 mm thick 3.1 (0.7%) Polson, MT 847 Semi-arid and seasonal Capillary barrier 744 0.2 (0.1%) Helena, MT 905 Semi-arid and seasonal Monolithic barrier 385 0.0 (0.0%) Boardman, OR 485 Semi-arid and seasonal Monolithic barrier 1220 mm thick 181 0.0 (0.0%) 485 Semi-arid and seasonal Monolithic barrier 1840 mm thick 181 0.0 (0.0%) Monticello, UT 607 Semi-arid Capillary barrier 514 0.0 (0.0%) Albany, GA 30 722 Humid Monolithic barrier with trees 1983 91.3 (7.2%) Cedar Rapids, IA 381 Humid and seasonal Monolithic barrier with trees 772 143.1 (15.6%) Omaha, NE 552 Humid and seasonal Capillary barrier, 760 mm storage

layer

719 3.7 (0.5%) 552 Humid and seasonal Capillary barrier, 1060 mm storage

layer

719 3.7 (0.5%)

Percentage of precipitation in parentheses

Vegetation reduces erosion and, for water balance caps, is mostly responsible for removing water stored in the cap There are three important factors that affect the success associated with establishing vegetation: proper preparation of the cap surface (e.g., not over-compacted), provision of nutrients, and selection of veg-etation that is consistent with the surrounding environment (e.g., a heavy grass cover should not be used for a water balance cap in the desert of Las Vegas, Nevada)

When these issues are considered during design and construction, vegetation has largely been successful For example, at the Hanford site, the survival rate of transplanted shrubs has been remarkably high (97% for sagebrush and 57% for rabbitbrush) Heavy invasions of tumbleweed have occurred (e.g., in 1995), but have not persisted Grass cover consisting of 12 varieties of annuals and perennials, including cheatgrass, several bluegrasses, and bunch grasses, currently dominates the surface Approximately 75% of the surface remains covered by vegetation requiring no maintenance, which is a value typical of shrub-steppe plant communities (Gee et al., 1996) A similar example is shown in Figure 3.12

for the water balance caps at the ACAP site in Sacramento, California Within one year of construction, a healthy cover of grasses and forbs was established with a leaf area index on the order of 1.4 (Roesler et al., 2002)

Characterizing the transpiration that can be expected from vegetation is a more challenging issue (Figure 3.13) Figure 3.13 shows water balance quantities for the thinner (1,080 mm) monolithic water balance cap in Sacramento being monitored by ACAP (test section on right-hand side of photographs shown in Figure 3.12) During the first growing season after construction (2000), the vegetation was able to extract the water and deplete the soil-water storage to the wilting point (approximately 180 mm), thereby providing an adequate soil res-ervoir for storing water during the subsequent winter However, the vegetation was far less effective in extracting the water in Spring 2001, even though the precipitation record was similar in both years, the water stored at the end of both wet seasons was comparable (approximately 400 mm), and the vegetation

TABLE 3.6

Summary of Saturated Hydraulic Conductivities of Samples Retrieved from the Surface of Covers being Monitored by ACAP

Site

Geometric Mean Hydraulic Conductivity (cm/s) End of Construction Summer 2002

Albany, GA 1.9 × 10–7 2.8 × 10–5

Cedar Rapids, IA 1.5 × 10–5 4.6 × 10–4

Helena, MT 5.0 × 10–7 1.6 × 10–7

appeared no different during either growing season Despite these similar condi-tions, the vegetation removed approximately 140 mm less water during the 2001 growing season Inadequate water removal resulted in inadequate storage capacity the following wet season As a result, the storage capacity (approximately 430 mm) was quickly exceeded during the wet period, and most of the water that infiltrated the cap surface became percolation

The inadequate transpiration observed during the 2000 growing season did not persist During the 2001 growing season, the vegetation removed all of the available stored water However, the reason for these differences remains a mys-tery, and efforts are currently underway to better understand why transpiration was greatly lower in 2001 This example illustrates, however, that characterizing and understanding the characteristics of vegetation is as important as understand-ing other materials used for caps, particularly for water balance caps that rely on transpiration as a critical barrier system process

FIGURE 3.12 ACAP test sections in Sacramento, CA, at the end of construction (a) and

one year after construction (b)

(a)

3.3 MATERIAL PERFORMANCE FACTORS IN PRBS

In contrast to most containment systems, which are usually designed to impede the flow of water, PRBs provide containment by treating contaminated water that passes through them PRBs rely on a reactive material placed in the subsurface (or manipulation of the physico-chemical properties of the subsurface environ-ment) to treat contaminated groundwater (Figure 3.14) As contaminated water passes though the PRB, reactions occur between the contaminants and the reactive medium, resulting in effluent that meets a target concentration, such as a maximum contaminant level (MCL) (depicted as “remediated water” in Figure 3.14)

A variety of reactive media are used for PRBs, including granular iron metal, granular activated carbon, zeolitic minerals, compost, limestone, and other “solid” materials placed in the subsurface to promote the physical, chemical, and bio-logical conditions necessary for contaminated groundwater treatment A summary of many of the materials being used is provided in Table 3.7 A photograph of granular iron and clinoptilolite is shown in Figure 3.15

The most commonly used treatment material is granular iron metal, which is effective for treating groundwater affected by both organic and inorganic constit-uents (Gillham and O’Hannesin, 1994) Although the proportion of all PRB applications using granular iron has not been computed, a reliable estimate is that 70% to 90% of PRBs installed as tests or full-scale applications have used

FIGURE 3.13 Water balance quantities for thin cover (1080 mm thick) monolithic water

balance covers being monitored by ACAP in Sacramento, CA (Data from Roesler et al., 2002 Field Hydrology and Model Predictions for Final Covers in the Alternative Cover Assessment Program — 2002, Geo-Engineering Report No 02-08, University of Wisconsin, Madison, WI; Albright, W and Benson, C., 2002 Alternative Cover Assessment Program 2002 Annual Report, Publication No 41182, Desert Research Institute, Reno, NV.)

0 200 400 600 800 1000 1200 1400 50 100 150 200 7/1/99 Soil-water storage Precipitation Percolation Surface runoff

1080-mm monolithic cover

Evapo-transpiration C um ula tive pr ec ipit ation, e va p o-transpira tion, and s oil-w at er st orage (mm) P er cola

tion and sur

granular iron as the reactive medium Other materials, such as granular activated carbon (GAC), compost, crushed limestone, alumino-silicates such as zeolitic minerals, and other materials are less used thus far, but are being tested in a variety of diverse applications

3.3.1 APPROACH TO SELECTION OF PRB MATERIALS

The criteria for selecting a reactive material are described by Blowes et al (2000) and include an assessment of the range of materials that can be used to remove contaminants and an assessment of the duration of material reactivity These criteria, coupled with an assessment of the potential for the release of hazardous

FIGURE 3.14 Schematic of a PRB used to intercept and treat a plume of contaminated

groundwater

TABLE 3.7

List of Reactive Materials that have been Used in PRBs

Treatment Materials Contaminants Treated

Zero-valent metals (including iron) (may or may not include metal couples)

Methanes, ethanes, ethenes, propanes, chlorinated pesticides, freons, nitrobenzene, certain metals (Cr, U, As, Tc, Pb, Cd)

Ferric oxides Mo, U, Hg, As, P, Se

Zeolites Sr, Pb, Al, Ba, Cd, Mn, Ni, Hg, certain organics

Activated carbon Mo, U, Tc, chlorinated VOCs, BTEX

Limestone Cr, Mo, U, acidic water

Compost Metals, acidic water

Alumina As

Peat, humate Mo, U, Cr, As, Pb

Sawdust, compost Nitrate

Oxygen Aromatic hydrocarbons, MTBE, vinyl chloride

Phosphates Mo, U, Tc, Pb, Cd, Zn

(a)

Waste area

Groundwater

Plume

Remediated water ARTZ

materials or contaminant by-products (e.g., release of vinyl chloride due to the reductive dechlorination of dichloroethylene), can be used to assess the potential of the barrier material to provide adequate groundwater treatment Interactions between natural groundwater constituents can result in extensive formation of secondary mineral precipitates within the barrier These precipitates can hinder barrier performance by clogging the pore space and reducing barrier permeability, or by obscuring reactive particle surfaces The assessment can be combined with an understanding of contaminant concentrations, groundwater geochemistry, and site hydrogeology to determine whether a practical remedial system can be constructed Then, a preliminary cost estimate can be developed and compared to remedial alternative estimates

Implementation of a remedial system employing a PRB can proceed through a series of steps, with accompanying decision points leading to the installation of an optimized system These steps start with a theoretical assessment of the potential for treatment using existing PRB materials State and federal guidance manuals have documented the PRB materials that were employed at existing PRB installations, the contaminants that were treated, and the contaminant removal that was attained [e.g., Interstate Technology Regulatory Council (ITRC), 1999a,b] This information can be used in conjunction with theoretical calculations, such as the use of geochemical speciation/mass transfer computer codes or the use of

FIGURE 3.15 Examples of reactive media used in PRB applications: granular iron metal

pH–Eh diagrams to assess the potential for contaminant removal If contaminant removal is possible, then laboratory treatability testing is considered

Laboratory treatability tests can be used to assess the potential for contami-nant removal and develop reaction parameters to assist in barrier design Batch experiments can be conducted to determine contaminant reactivity and measure reaction rates under static conditions Column experiments can be used to measure rates of contaminant removal under dynamic flow conditions and assess the poten-tial for the precipitation of secondary minerals and barrier clogging Where pos-sible, mineralogical examination of column materials following the testing program can be used to verify the presence and structure of secondary precipitates to assess the stability of these precipitates within the barrier and evaluate the potential for barrier clogging Complementary geochemical modeling, including reactive transport modeling, can be used to develop design parameters at this stage The geochemical modeling, coupled with groundwater flow and transport modeling, can be used to provide preliminary estimates of barrier performance and longevity and to design parameters for pilot- or full-scale installations

3.3.2 EVALUATION OF FIELD PERFORMANCE USING

PILOT TESTING

The decision whether to conduct a pilot-scale test or move directly to full-scale implementation depends on the history of the technology and the confidence of the client and regulators Many PRB technologies have been demonstrated suf-ficiently to satisfy regulators that the treatment processes are well understood and the installation success depends on site-specific processes Pilot-scale installations vary in scale and degree of monitoring, from small-scale column experiments conducted ex situ at a field site to large-scale installations that ultimately form a portion of a full-scale PRB

The key objective of the pilot-scale installation is to simulate conditions in a full-scale system as closely as possible Using the candidate reactive materials and natural aquifer materials in contact with site groundwater and typical con-taminant concentrations provides a close approximation to the characteristics of full-scale systems The small size of pilot installations provides an opportunity for monitoring at a level of detail that is sufficient to provide design parameters for the full-scale installation Pilot-scale installations should be sufficiently ver-satile so that variability in treatment media and groundwater flow rates can be assessed The results of the pilot-scale installation can be used to confirm con-taminant reactivity and assess the potential for negative secondary reactions such as scaling or clogging The pilot-scale system should also have well-defined dimensions and performance characteristics to simplify scaling up to the final remedial system

are placed along the central axis of the borehole for groundwater sampling Several well casings with slots at different depths can be used to obtain multiple samples at different depths A peristaltic pump is used to collect low-flow samples from the slotted section of each casing for analysis

RTWs were first used to test the efficacy of different reactive media for removing arsenic from groundwater at a DuPont site in East Chicago, Indiana Data collected from the RTWs over a nine-month period were used to select PRB material Basic oxygen furnace (BOF) slag was selected for use in the PRB based on data collected from RTWs, whereas laboratory studies indicated that another material was more appropriate

The coaxial configuration of the RTW ensures that groundwater passed through approximately 75 mm of reactive material before sampling regardless of the local groundwater flow direction For accelerated tests, groundwater can be continually extracted through the casing Because RTWs are simpler and less costly than a full-scale pilot wall, multiple RTWs can be installed at a given site to test different materials or act as controls RTWs also have several advantages over ex situ field demonstrations (Table 3.8)

Installation quality is important in a RTW providing reliable data The drilling process should not create a smear zone at the well interface that might impede flow Centrality of the well casing is also important so that the flow path through the reactive medium is the same at all points in the well A centralizer consisting of a plastic disk with threads that match the well string is generally placed at the bottom of a RTW, along with conventional stainless-steel centralizers along

FIGURE 3.16 RTW using passive groundwater flow

Groundwater samples

Bentonite seal Reactive material in

300-mm borehole

Slotted well casing

the length of the well casing (Figure 3.17) The stainless-steel centralizers are installed above the slotted section so that the water being sampled is not exposed to any extraneous reactivity

Data from a RTW can be interpreted at several levels, from strict demonstra-tion of contaminant removal to development of break-through and capacity cor-relations and projections of service life By manipulating flow rates, kinetic expressions can also be developed A passive RTW (i.e., operating under natural groundwater flow conditions) can provide an assessment of effectiveness in nearly real time, i.e., one month of field data is equivalent to one month of ultimate PRB exposure To project the ultimate life of a PRB, an extractive RTW can be employed In this technique, groundwater is pumped out of the central casing at an accelerated rate, analogous to using higher throughputs in a laboratory column Avoiding kinetic limitations (i.e., from an extraction rate that is too high) with an extractive RTW is important unless a kinetic study is intended An appropriate rate should be determined in the laboratory and then translated to the field test

3.3.3 EFFECTS OF HYDRAULIC CONSIDERATIONS ON REACTIVE

MATERIAL PERFORMANCE

To date, PRB research has focused mostly on the reaction mechanisms, kinetics, and conversion efficiency associated with the reactive materials (Tratnyek et al., 2003) Much less effort has focused on factors that affect PRB hydraulics, even

TABLE 3.8

Comparison of Reactive Test Wells vs Ex Situ Field Tests

Parameter In SituReactive Test Well Ex SituPacked Columns Key Technical Parameters

Groundwater chemistry Actual Can be different

Contaminant losses in system Essentially none Can be significant

Potential data quality High Varies significantly

Flow rate through bed Natural (uncontrolled, cross-flow); or

enhanced (pumped), radial flow

Controlled, precise, axial

Logistical Parameters

System complexity Low Medium to high

Effluent disposal None Problematic

Multiple location tests Concurrent Sequential

Duration limit Unlimited, at low cost Limited by cost, etc

Weather protection required None Can be significant

Overall Assessment

Final wall approximation Very close; a mini-wall Approximation

Technical certainty High Varies significantly

though hydraulic factors can have as large an impact on PRB effectiveness (Eykholt et al., 1999; Elder et al., 2002)

As more PRB systems are implemented and monitored, performance data suggest that hydraulic characteristics of PRB materials need to receive greater attention during design Recent reviews of PRB applications have suggested that most cases of unintended performance are due largely to inadequate hydraulic performance Few cases are related to inadequacies in the chemical treatment methodology (Warner and Sorel, 2001; Battelle, 2002) These findings indicate that designers need to consider hydraulics as a critical factor affecting successful PRB deployment, and approach hydraulic design with the same level of care as reaction effectiveness Hydraulic aspects that can have a large impact on PRB effectiveness are aquifer material heterogeneity and spatial variability of the groundwater flow field The importance of geological heterogeneities and the need for characterization was illustrated in a recent case study of a PRB con-structed near Kansas City, Missouri (Laase et al., 2000) The PRB was installed in an alluvial aquifer to intercept a plume containing trichloroethylene (TCE) Data from a hydrogeological study were used as input to a groundwater model used to select PRB orientation and breadth The breadth was to be sufficiently

FIGURE 3.17 Centralizers for maintaining casing position in a RTW: Base centralizer

large to capture the entire width of the plume An extensive set of monitoring wells (12 upgradient, 16 downgradient, and 10 adjacent to the ends of the PRB) was installed to monitor influent and effluent conditions and check for bypassing Data from the monitoring program showed that the wall was not functioning as intended While the reaction mechanisms appeared to have been accounted for properly, a sandy gravel region toward the southern end of wall was not detected during hydrogeological characterization and caused a portion of the plume to bypass the PRB, as shown in Figure 3.18 In addition, reversals in the hydraulic gradient during recharge events caused the southerly extent of the plume to curl northward and, at times, flow backward through the PRB Bypassing was occur-ring along the northern end of the PRB as well

Few PRBs are monitored as closely as the PRB in Kansas City Thus, the frequency of problems caused by heterogeneity is unknown However, a recent modeling study by Elder et al (2001, 2002) suggests that geological heterogeneity may be having a much larger impact on PRB effectiveness than previously thought Elder et al (2001, 2002) constructed a series of heterogeneous aquifers containing PRBs and simulated flow and transport through the aquifer and PRB Because a model was used, effluent concentrations were characterized in far greater detail than is possible in the field, even with a dense network of monitoring wells

Typical results reported by Elder et al (2001, 2002) are shown in Figure 3.19 The simulation consisted of a TCE source with a uniform concentration of 1000

FIGURE 3.18 Schematic of plume bypassing southern end of PRB installed near Kansas

City (Adapted from Laase, A et al., 2000 In Wickramanayake, G et al (Eds.), Chemical

Oxidation and Reactive Barriers, Remediation of Chlorinated and Recalcitrant Com-pounds, Battelle Press, Columbus, OH, pp 417–424)

N Plume

Flow

micrograms per liter (µg/L) located 20 m upgradient of the PRB By the time the plume reached the PRB, dispersion induced by aquifer heterogeneities caused the TCE concentration to range from 0.1 to 1000 µg/L As groundwater flowed through the wall, the TCE concentration decreased, but not always below the target level (5 µg/L) In fact, the effluent concentration was as high as 500 µg/L

FIGURE 3.19 Concentrations at source (a), influent face of PRB (b), and effluent face of

PRB (c) in a heterogeneous aquifer (Adapted from Elder, C et al., 2002.Water Resources Research, 38(8), 27-1 to 27-2)

Ele va tion (m) 45 40 35 30 25 20 15 10 10

Lateral distance (m)

Ele va tion (m) 45 40 35 30 25 20 15 10 10

Lateral distance (m)

Ele va tion (m) 45 40 35 30 25 20 15 10 10

Lateral distance (m)

TCE Concentration (mg/L)

1000 500 100 10 0.1 0.01

Monitoring well screens

Boundary of PRB

Monitoring well screens

Boundary of PRB (a) Source concentrations

(b) Influent concentrations

in some locations These high concentrations were due to preferential flow through the wall as a result of heterogeneity in the adjacent aquifer materials

Elder et al (2001) assessed whether the range in effluent concentrations, as well as the peak effluent concentrations, could be detected using typical PRB monitoring schemes Most monitoring schemes were found to be too sparse to capture most of the variability, and effluent concentrations detected by typical systems were found to underestimate peak effluent concentrations by an order of magnitude or more

The findings reported in Elder et al (2001, 2002) illustrate the need for better characterization of PRB flow rates and flow paths A variety of methods can be used for characterization Tracer tests have been used to establish groundwater flux and flow paths through PRBs Dissolved tracers are well suited to determining the direction of groundwater flow, but are limited by the number of injection wells that can be used without overlapping tracer plumes In addition, flow velocities obtained from tracer studies are sensitive to the number and distribution of monitoring points and to chemical dispersion Accurate determinations of dispersion are often lacking, particularly within a PRB where settling or other construction-related effects can be significant Another method is the use of downhole flow sensors These instruments rely on dispersion of a heat pulse or measurement of colloidal particle velocities to determine groundwater flow veloc-ities Velocities measured with this technique often vary considerably within an individual well and between wells and may differ from those in the aquifer due to the effects of an open borehole In situ sensors embedded in the aquifer can eliminate the effects of an open borehole, but the higher cost associated with dedicated sensors generally limits their application to a few points within or near a PRB Even so, the extent of representation of velocities from flow sensors is unclear In a comparison of three different downhole flow sensors, Wilson et al (2001) concluded that the three methods “rarely measured the same velocities and flow directions at the same measurement stations” and that repeat measure-ments “failed to consistently reproduce either flow direction, flow magnitude, or both.”

A promising method that can be used to evaluate the average velocity in an operating PRB using granular iron is reaction path monitoring Geochemical reactions within a granular iron PRB cause the precipitation of solids containing the major constituents in the groundwater (e.g., calcium, magnesium, manganese, carbonate), as well as contaminants (e.g., arsenic, molybdenum, selenium, uranium, Reaction path monitoring involves estimating a time-averaged groundwater veloc-ity from constituent concentrations in the aqueous and solid phases

iron (based on volume) at the upgradient face, (2) 1.2 m wide central portion containing 100% granular iron, and (3) 0.5-m-wide downgradient panel that contains only pea gravel

The PRB was placed within a groundwater plume emanating from the former mill site Groundwater entering the PRB has a uranium concentration of about 0.4 mg/L The plume also contained arsenic (0.01 mg/L), molybdenum (0.06 mg/L), nitrate (61 mg/L), selenium (0.02 mg/L), and vanadium (0.4 mg/L) as it entered the PRB The PRB was very effective at treating each of these contaminants Concentrations of all contaminants decreased to low levels in the PRB during 2.7 years of its operation (Morrison et al., 2001)

FIGURE 3.20 Schematic of groundwater cutoff walls and PRB installed at Monticello,

Utah, site (ZVI, zero-valent iron.)

0.5 m of gravel/ZVI 1.2 m of 100% ZVI 0.5 m of gravel with

air-sparging pipe

North slurry wall (29.6 m) Groundwater flow

PRB (31.4 m)

Solid phases in the gravel-iron panel and the iron-only panel were sampled using direct-push coring in February 2002 Cores were collected at 70 random locations within 10 evenly spaced PRB segments (Figure 3.21) The cores were cut into 4.1-cm lengths; 614 samples were collected, of which 279 were digested and analyzed for calcium, uranium, and vanadium Constituent concentrations dissolved in the groundwater were also measured at 10 evenly spaced time intervals from six wells upgradient and six wells downgradient of the gravel-iron panel

Core analysis showed that nearly all the uranium (Figure 3.22) and vanadium were deposited in the gravel-iron panel In contrast, calcium was deposited in both iron-containing panels (Figure 3.23) The distribution of calcium is more pervasive and indicates a slower rate of transfer to the solid phase Calcium, uranium, and vanadium are distributed along the entire length of the PRB, indi-cating that the entire mass of iron is being used to treat the contaminated ground-water, and that groundwater has not flowed preferentially through specific portions of the PRB

Mean concentration differences ()Cw) were calculated from the aqueous phase concentrations (Table 3.9) The mean groundwater flux (Qw) was then computed for each solid phase species using:

(3.1)

where Cs is the mean solid-phase concentration, Mg is the mass of solid material initially in the zone (70.2 mg), and )t is the deposition period (2.7 years) The mean groundwater fluxes computed following this approach using uranium and calcium data are summarized in Table 3.9 The mean groundwater flux (24 L/min) through the gravel-iron panel zone calculated using the calcium data was identical to that calculated using uranium and was considerably less than the design value of 189 L/min

3.3.4 STRUCTURAL STABILITY FACTORS IN PERFORMANCE

Little post-construction assessment occurs regarding the integrity of the in-place reactive medium from the perspective of sustainability (e.g., reactivity, conduc-tivity) or structural stability (e.g., settlement, movement, strength) Properties such as density or specific gravity, shape, and water content can affect the in-place density, porosity, and hydraulic conductivity In addition, when two or more materials are used in a PRB or the PRB contains multiple sections of materials, mixing uniformity or complete separation must be ensured For example, the specific gravity of the reactive materials must be considered so that a construction procedure can be developed that will promote uniform mixing

Q C M

t C w

s g

w =

FIGURE 3.21 Locations of cores for solid-phase samples and monitoring wells for aque-ous phase analyses in PRB installed at Monticello, Utah, site

Me

ters 15

10

5

0

0 0.5 1.0 1.5 20

25 30

Meters

Core Well Flow

FIGURE 3.22 Distribution of uranium in the solid phase (mg/kg) in PRB installed at Monticello, Utah, site Contour interval is 50 mg/kg

Me ters 30 25 20 15 10

0 0.5 1.0 1.5 Meters

Flow

Gravel/ZVI ZVI 70.0 319.3 0.1 0.10.1

0.1 0.1 0.9 384.4 206.5 190.3 437.0 1.0 0.1 1.1 0.0 0.2 156.6 92.7 249.8 0.0 0.0 0.0 0.0 0.0 0.0 0.9 172.4 226.0 288.8 807.5 330.2 200

10.5 0.0 0.0 0.0 0.1

0.4 0.0 0.1

0.1 0.1 0.1 1.0 0.5 0.1 251.8 396.8 558.4 334.5 343.5 350.0 539.5 200 371.0 269.0 147.1 6.3

(http://www.copyright.com/ w Hampshire, 2001 (www.hps-inc.com). August 1998 (www.clu-in.com) for most applications via the web site http://ees-www.lanl.gov/EES5/fehm/ http://www.antenna.nl/wise/uranium/ctc.html. (http://www.rtdf.org/public/permbarr/prbsumms/ an Benschoten, J.E (2002) Online at http://www.eng.buffalo.edu/ees/research/Remediation_Radioactive_GW.htm.