United States
            Environmental Protection
            Agency
           Office of Health and
           Environmental Assessment
           Washington DC 20460
EPA/60Q/8-88/075
May 1988
vvEPA
            Research and Development
Selection Criteria for
Mathematical Models
Used in Exposure
Assessments:
            Ground-Water Models

-------

-------
                                            EPA/600/8-88/075
                                            May  1988
     SELECTION CRITERIA FOR MATHEMATICAL

     MODELS USED  IN EXPOSURE  ASSESSMENTS:

             GROUND-WATER MODELS
          Exposure Assessment Group
Office of Health and Environmental Assessment
    U.S.  Environmental  Protection Agency
              Washington, D.C.

-------
                                   DISCLAIMER

     This document has been reviewed in accordance with U.S. Environmental
Protection Agency policy and approved for publication.  Mention of trade names
or commercial products does not constitute endorsement or recommendation for
use.
                                      ii

-------
                                    CONTENTS
Tables and Figures ..... . ......... . -. . . . : ,; ; ;  . .  . . .   v
Foreword	vi
Preface		.......... : .  . ... . vii
Abstract	viii
Authors, Contributors, and Reviewers ..	 . . .  . . .  .... .  ix

1.  EXECUTIVE SUMMARY	   1-1

     1.1.  Introduction	   1-1
     1.2.  Background Information	   1-1
     1.3.  General Guidelines and Principles of Model
           Selection Criteria	   1-2

          1.3,1.  Technical Criteria 	   1-3
          1.3.2.  Implementation Criteria	   1-4
          1.3.3.  Other Factors Affecting Model Selection	   1-5

     1.4.  Model Selection Decision Process	   1-5
     1.5.  Model Selection Example Problems	   1-8
     1.6.  Appendix A	   1-9

2.  INTRODUCTION	   2-1

3.  BACKGROUND INFORMATION 	 	   3-1

     3.1.  Primer on Ground-Water Flow	   3-1

          3.1.1.  Ground Water and the Hydro!ogic Cycle	   3-1
          3.1.2.  Porosity and Hydraulic Conductivity	   3-4
          3.1.3.  Flow in Ground-Water Systems 	   3-6

     3.2.  Primer on Contaminant Transport 	   3-8

          3.2.1.  Advective Dispersion Equation	   3-8
          3.2.2.  Attenuation and Degradation Mechanisms 	  3-11

               3.2.2.1.   Sorption	-	3-12
               3.2.2.2;   Degradation 	  3-13

          3.2.3.  Definition of Terms	  3-15

4.  GENERAL GUIDELINES AND PRINCIPLES
    OF MODEL SELECTION CRITERIA	   4-1

     4.1.  Overview:   Modeling Process 	   4-1
     4.2.  Overview:   Model Selection Criteria 	   4-4

          4.2.1.  Objectives Criteria	   4-4
          4.2.2.  Technical Criteria 	 '  4-6

                                      iii

-------
               4.2.2.1.  Transport and Transformation Processes	   4-6
               4.2.2.2.  Domain Configuration	   4-7
               4.2.2.3.  Fluid(s) and Media Properties 	   4-8

          4.2.3.  Implementation Criteria. .....  ...  ....;...,;-.  .  .  .   4-9
          4.2.4.  Other Factors Affecting Model Selection	   4-9

     4.3.  Model Selection vs Model Application	4-12
     4.4.  Familiarity with a Model.  ... . . . •.  .  . <.c.......  .  .  ...  4-13
     4.5.  Model Reliability 	  4-14

5.  MODEL SELECTION DECISION PROCESS  	   5-1

     5.1.  Technical Criteria used for Model Selection 	   5-1
     5.2.  Model Selection from the Reviewers  Point of View	5-29
     5.3.  Model Selection Worksheet  	  5-30
     5.4.  Waste Management Models .  .	5-40

          5.4.1.  Risk Assessment Methodology  for Regulatory
                  Sludge Disposal Through Land Application  .....  .  .  .  5-41
          5.4.2.  Risk Assessment Methodology  for Regulating
                  Landfill Disposal of Sludge	5-41
          5.4.3.  RCRA Risk/Cost Policy Model  (WET Model)   	  5-42
          5.4.4.  The Liner Location  Risk and  Cost Analysis Model   ....  5-43
          5.4.5.  Landfill Ban Model  	  5-45

6.  MODEL SELECTION EXAMPLE PROBLEMS  	   6-1

     6.1.  Screening Analysis Example Problem	   6-1

          6.1.1.  Objectives of the Study	   6-1
          6.1.2.  Conceptual Model of the Study Area  	   6-2
          6.1.3.  Model Selection Process	   6-2

     6.2.  Detailed Analysis Example  Problem	   6-4

          6.2.1.  Statement of the Problem 	   6-4
          6.2.2.  Objectives of the Study	   6-4
          6.2.3.  Conceptual Model of the Study.Area  	   6-5
          6.2.4.  Model Selection Process	   6-7

7.  REFERENCES	   7-1

APPENDIX A:  ANALYTICAL AND NUMERICAL MODEL SUMMARIES	   A-l
                                       IV

-------
                                     TABLES
5-1       Analytical Solutions,  Names,  and  References.	   5-32
5-2       Analytical Solutions Worksheet  	   5.34
5-3       Analytical and Numerical Models Worksheet.  .  .  .	5-37
                                    FIGURES
5-1
6-1
Code selection decision tree 	
Model region for the assessment-level example problem
5-2
6-6

-------
                                    FOREWORD

     When performing exposure assessments using predictive methods, assessors
frequently ask the following questions:  "How do I select the best fate model to
use in my assessment," "How can I tell if the model someone else used in their
assessment is appropriate," and "What are the strengths and weaknesses of these
models?"  This document is a first step in addressing these questions.
     One of the functions of the Exposure Assessment Group is to develop
guidelines for exposure assessments.  On September 24, 1986, the U.S.
Environmental Protection Agency published Guidelines for Estimating Exposures.
During the development of the guidelines and subsequent review and comment, four
areas were identified that required further research.  One of these areas was
selection criteria for mathematical models.  This document, which is the second
selection criteria document in the series, deals with ground-water models.  The
first dealt with surface water models.  Similar documents will follow dealing
with air models, and int he future, other types of models.
     This document is designed to help the exposure assessor evaluate the
appropriateness of models for various situations.  The report defines the terms
and discusses the general approaches that modelers take to a problem so that
exposure assessors may more readily evaluate the appropriateness of both new and
existing models.  In addition, step-by-step criteria are provided to enable the
assessor to answer the questions posed above.
                                                       Michael  A. Callahan
                                                       Director
                                                       Exposure Assessment Group

-------
                                     PREFACE

     The Exposure Assessment  Group  of the  Officeof Health and Environmental
Assessment  (OHEA) is preparing  several  documents  addressing selection criteria
for mathematical models used  in exposure assessments.   These,documents will
serve as technical support documents  for thetGuidelines for Estimating
Exposures, one of five risk assessment  guidelines published by the U.S.
Environmental Protection Agency in  1936.
     The purpose of this document is  to present criteria which provide a means
for selecting the most appropriate mathematical model(s) for conducting  an
exposure assessment related to  ground-water contamination.
             .'•     , • -      •"'.•''    •' • ' :   • ;'" ••..'•.-•;•- - '  , ' ' •••••>•• •(•;, , ,.;.-•',
     The literature search to support the models  discussed  in,this  report is
current to September 1986.
                                      vii

-------
                                   ABSTRACT

     Prior to the issuance of the Guidelines for Estimating Exposures in 1986,
the U.S. Environmental Protection Agency (EPA) published proposed guidelines in
the Federal Register for public review and comment.  The purpose of the
guidelines is to provide a general approach and framework for carrying out human
and nonhuman exposure assessments for specific pollutants.  As a result of the
review process, four areas were identified that required further research.  One
of these was the area of selection criteria for mathematical models used in
exposure assessments.
     The purpose of this document is to present criteria which provide a means
                           i
for selecting the most appropriate mathematical model(s) for conducting an
exposure assessment related to ground-water contamination.
     General guidelines and principles for model selection criteria are
presented  followed by a step-by-step approach to identifying the appropriate
model(s) for use in a specific application.  Several of the currently-available
models  are grouped into categories and a framework  is provided for selecting  the
appropriate model(s) based ,on the response to the technical criteria.  Brief
summaries  of all the currently available models discussed  in this report are
contained  in the appendix.
     Two site-specific example problems are provided to demonstrate the
procedure  for  selecting the appropriate mathematical model  for a particular
application.
                                      vm

-------
                      AUTHORS,  CONTRIBUTORS, AND REVIEWERS

     The Exposure Assessment Group within the U.S. Environmental Protection
Agency's Office of Health and Environmental Assessment was responsible for the
preparation of this document and provided overall direction and coordination
during the production effort.  The information  in this document has been funded,
wholly or in part, by the U.S. Environmental Protection Agency under Contract
No. 68-01-6939 to Camp Dresser & McKee Inc.  The work was performed by ICF
Technology Inc., a subcontractor (Seong Hwang,  Project Manager).

AUTHORS
                                                    f
     Frederick Bond
     ICF Northwest
     Richland, WA
     Seong Hwang
     Exposure Assessment Group
     Office of Health and Environmental Assessment
     U.S. Environmental Protection Agency
     Washington, DC
CONTRIBUTORS:
     Paul van der Heijde
     International Ground-Water Modeling Center
     Indianapolis, IN
     Scott Yates
     Robert S. Kerr Environmental Research Laboratory
     U.S. Environmental Protection Agency
     Ada, OK

REVIEWERS
     P.  B.  Bedient
     Rice University
     Houston, TX
                                       ix

-------
Carey Carpenter
Office of Ground-Water Protection
U.S. Environmental Protection Agency
Washington, DC                  •

Norbert Dee
Office of Ground-Water Protection
U.S. Environmental Protection Agency
Washington, DC

Ron Hoffer
Guidelines Implementation Staff
Office of Ground-Water Protection
U.S. Environmental Protection Agency
Washington, DC                             i

Joe Hughart
Water Management Division
U.S. Environmental Protection Agency, Region 4
Atlanta, GA

David Kyllonen
U.S. Environmental Protection Agency, Region 9
San Francisco, CA

James Mercer
Geo Trans, Inc.
Herndon, VA                        ;

Frank Mink
Environmental Criteria and Assessment Office
Office of Health and Environmental Assessment
U.S. Environmental Protection Agency
Cincinnati, OH

Charles Ris
Carcinogen Assessment Group
Office of Health and Environmental Assessment
U.S. Environmental Protection Agency
Washington, DC

John Segna
Exposure Assessment Group
Office of Health and Environmental Assessment
U.S. Environmental Protection Agency
Washington, DC

John Years!ey
Environmental Services Division
U.S. Environmental Protection Agency, Region 10
Seattle, WA

-------
     The following individuals, members of the EPA Ground-Water Model  Selection
Criteria Work Group, participated in meetings, provided advice on formulating
the criteria, and reviewed drafts of this document.  Their efforts were greatly

appreciated.
     Stuart Cohen                            >
     Office of Pesticide Programs

     Stephen Cordle
     Office of Research and Development

     Norbert Dee                         ,      •
     Office of Ground-Water Protection

     Annett Mold
     Office of Toxic Substances

     William Wood
     Office of Research and Development

     Saleem Zubair
     Office of Solid Waste

     Malcolm Field
     Exposure Assessment Group
     Office of Health and Environmental Assessment
     U. S. Environmental Protection Agency
     Washington, DC
                                       xi

-------

-------
                             1.  EXECUTIVE SUMMARY

1.1.  INTRODUCTION^
     This document presents a set of criteria which provide a means of selecting
the most appropriate mathematical model for conducting an exposure assessment
related to ground-water contamination.  These criteria were developed in
recognition of the growing use of exposure assessments across the U.S.
Environmental Protection Agency's regulatory programs.  Use of the criteria will
expedite the regulatory process by eliminating the use of unacceptable or
inappropriate models.  Their use will also improve the quality of data used in
the decision-making processes and promote consistency in exposure assessments.
     When performing a predictive exposure assessment, a major task is to
predict the transport of contaminants.  Since ground-water flow is an integral
part of contaminant transport, it is equally important, if not more so, to
accurately predict the ground-water flow.  Therefore, both ground-water flow and
contaminant transport mathematical models, and criteria for selecting these
models, are discussed in this document.
1.2.  BACKGROUND INFORMATION
     Some of the general background information necessary to understand the
selection of a ground-water flow and/or contaminant transport model is discussed
in this section.  This chapter is intended for the exposure assessor or the
non-modeler who is not completely familiar with hydrogeologic and modeling
terms.
     The first section provides a primer on ground-water flow.  The intent of
this section is to provide a brief summary of the background information
necessary to understand ground-water problems.  The chapter discusses the
                                      1-1

-------
general terms  used  to describe  and define ground-water flow and presents the
basic equation for  flow in  a ground-water system.
     The  second section provides  background  information on contaminant transport
in ground water.  The chapter presents the basic equation for
advective-dispersive transport  and discusses the important terms  in detail.
     The  last  section provides  definitions for terms used throughout the repor.t.
1.3.  GENERAL  GUIDELINES AND PRINCIPLES OF MODEL SELECTION CRITERIA
     In order  to enhance understanding and facilitate implementation of the
mathematical model  selection criteria, the following terms are defined:
mathematical model, process equation, analytical solution, analytical models,
numerical models, objectives criteria, technical criteria, and implementation
criteria.  The relationship between these terms may be thought of as follows.  A
mathematical model  consists of .two aspects:  a process equation and a solution
technique to solve  the  process  equation.  An analytical solution solves a very
simple process  equation  analytically by hand calculations.  An analytical model
solves a  more  complex,  but still  relatively  simple, process equation
analytically with a computer program.  A numerical model solves a simple or.
complex process equation numerically with a  computer program.  In the context of
this document, mathematical model refers to  all three solution techniques of a,
process equation.   The more detailed the specific application,, the more complex
the process equation.  The complexity of the process equation dictates the
solution  technique  required.
     There are three factors which dictate the level of complexity of the
mathematical model  chosen in the  selection process:
1.   objectives criteria;
2.   technical criteria; and
                                      1-2

-------
3.   Implementation criteria.
     The objectives criteria refer to the level of modeling detail required to
meet the objectives of the study.  There are many different objectives of
modeling studies, however, in the context of model selection, all objectives are
classified in two broad categories:  1) to perform a screening study or 2) to
perform a detailed study.
     A screening study is one where the purpose is to make a preliminary
screening of a site or to make a general comparison between several sites.  A
detailed study, on the other hand, is one where the objective is to make an
assessment of the environmental impact, performance, or safety of a specific
site.
     Based on the objectives of the study (screening or detailed levels), the
analyst or modeler will select either a screening or detailed model.  The
specific model to be used will be selected based on the technical selection
criteria discussed below.
1.3.1.  Technical Criteria
     The second level of consideration when selecting a mathematical model is
the technical criteria.  Technical criteria are those criteria related to the
mathematical model's ability to simulate the site-specific contaminant transport
and fate phenomena of importance.
     With regard to model selection, the technical criteria can  be divided into
three categories:
1.   transport and transformation processes;
2.   domain configuration; and
3.   fluid(s) and media properties.
                                       1-3

-------
 Transport and transformation process  criteria  relate  to  those  significant
 processes or phenomena known to occur on  site  that must  be modeled  in  order  to
                                                                    i
 properly represent the site.  Domain  configuration relates to  the ability  of the
 model  to accurately represent the  geohydrologic  system.  When  high  levels  of
 resolution are required to  predict contaminant concentrations  for comparison to
 health or design  standards,  it is  generally necessary to simulate site-specific
 geometry and dimensionality for which numerical  models are most appropriate.  If
 simplifying the site geometry can  be  defended  on a geologic and hydrologic
 basis, then the use of a simpler analytical model/solution may be justified.
 The third category of  technical  criteria  corresponds  to the ability of the
 mathematical  model  to  represent the spatial variability of fluid(s) atid media
 properties  of the  geohydrologic site.
     Once the level  of model  has been  decided, the technical criteria  will
 direct the  analyst  to  the specific type of model needed to properly simulate the
 transport and transformation  aspects  of the environmental setting.
 1.3.2.   Implementation  Criteria
     The  third  level of consideration when selecting  a mathematical  model is
 related to  the  implementation  criteria.   Implementation criteria are those
 criteria  dependent  on the ease  with which a model can be obtained and  its
 acceptability demonstrated.  Whereas the technical  criteria identify the models
 capable of  simulating-the relevant phenomena within the specified environmental
 setting,  the  implementation criteria identify documentation,  verification,
validation requirements, and ease of use so that the model  selected  provides
accurate, meaningful results.
                                      1-4

-------
1.3.3.  Other Factors Affecting Model Selection
     Other general factors related to model selection which are secondary to the
technical or implementation, criteria include data availability, schedule,
budget, staff and equipment resource, and level of complexity of system(s) under
study.  Schedule and budget constraints refer to the amount of
time and money available for the assessment.  If both analytical and numerical
models meet the selection criteria, time and cost may be considered factors for
electing to use an analytical approach.
1.4.  MODEL SELECTION DEC.ISION PROCESS
     The decisions to be made when selecting a ground-water flow model are
discussed in detail  in this section.  Some guidance  is provided for making the
decision and some discussion is provided regarding the errors  associated with
using the incorrect  model or feature(s) of a model.  The criteria for ground-
water flow are presented followed  by those for contaminant transport.
•    Are you simulating a water table  (i.e., unqonfined) or a  confined aquifer,
      or  a combination of both  (i.e., conditions  change spatially)?
•     Does the ground water  flow through porous media, fractures, or a
      combination  of  both?
•     Is  it necessary to simulate  three-dimensional flow  or can the
      dimensionality  be reduced without  losing  a  significant amount of  accuracy?
•     Are you  simulating a  single-phase  (i.e.,  water) or  a multi-phase  (i.e.,
      water  and  oil)  flow  system?
•     Can the  system  be  simulated  with  a uniform  value (homogeneous)  or spatially
      variable values (heterogeneous) of hydraulic conductivity,  porosity,
      recharge,  and/or specific storage?
                                       1-5

-------
 •    Is there a single or are there multiple hydrogeologic layers to be
      simulated?
 •    Is (are) the hydrogeologic layer(s) of constant or variable thickness
      spatially?
 •    Is the hydrologic system in a steady-state condition or do water levels
      fluctuate with time (transient condition)?
      After all these criteria has been satisfied,  in most cases there will  be
 several ground-water flow models which would be appropriate.  At this point the
 analyst can either select a ground-water flow model  and then  continue with  the
 selection  process to select a compatible (but separate)  contaminant  transport
 model,  or  the user can continue the process  to select a combined flow and
 transport  model.   It is quite common to develop a  fairly sophisticated flow
 model to predict  ground-water travel  paths and velocities  and link it with  a
 simpler transport model.
     The decisions to  be  made when  selecting  a contaminant transport  model  are
 discussed  in  detail  in this  section.   Some guidance  is  provided  to help in
 making  the decision  and some  discussion  is provided  regarding the errors
 associated with using  the  incorrect model or  feature(s)  of the model.
 t    Does  the  contaminant  enter the ground-water flow system at  a point or  is it
     distributed  along  a line  or over  an area  or a volume?
 •    Does  the  source consist of an initial slug of contaminant or is  it constant
     over  time?
•    Is  it necessary to simulate three-dimensional  transport or can the
     dimensionality be reduced without losing a significant amount of accuracy?
§    Does the model simulate dispersion?
                                      1-6

-------
t    Does the model simulate adsorption (i.e., distribution or partitioning
     coefficient) and, if so, does it simulate temporally and/or spatially
     variable adsorption?  Temporally or spatially variable adsorption is
     important where the soil conditions and/or concentrations change with time
     and space.                                                  ,
•    Does the model simulate first or second-order decay and/or radionuclide
     decay?
•    Does the model simulate density effects related to changes in temperature
     and concentration?  A truly coupled model is one where the ground-water
     flow is influenced by the density and viscosity of the water, which are
     influenced by the temperature of the water and the concentration of the
     solute.  In some cases  (i.e., large heat source or large fluctuations in
     solute concentration) it may be important to consider temperature and
     contaminant concentration effects on ground-water flow.
     After sequencing through the decision tree, there will, in most cases, be
several models which meet the desired criteria.  Since several models could meet
the desired criteria, it is  difficult to list a single model as a standard
model.  At this point the analyst can either select a transport model which is
compatible with the flow model selected above, or select a cpmbined ground-water
flow/contaminant transport model.
     Regardless of the approach selected, separate or combined flow and
transport model, it is likely that there will be several models which meet the
technical criteria.  The selection of the final model(s) should be based on the
implementation criteria, i.e., the model has been through a rigorous quality
assurance program so that it is thoroughly verified and the model is well
documented with user's manuals and test cases.
                                      1-7

-------
      If several models pass the quality assurance and documentation criteria,
 the final selection of a model should be based on familiarity with and,
 availability of the model, schedule, budget, and staff and equipment resources.
      A model selection worksheet is included in this section which facilitates
 the selection of the actual model or suite or models to be used based on the
 response to the technical criteria.  Separate worksheets are provided for both
 analytical  solutions and for analytical  and numerical  models (coded for the
 computer).   A summary of each of the models contained  in the worksheets is
 contained in Appendix A.
      A discussion of waste management models has been  included in  this section.
 Waste management models  are defined as models which  trace contaminant movement
 through the three primary environmental  pathways:  air,  surface water,  and/or
 ground water.   It is not  the objective of this document  to cover waste
 management  models in any  detail.   Rather, a few such models are described
 briefly to  make  the  reader aware  of them.   The models  discussed are:
 1.    risk assessment methodology  for regulatory sludge disposal  through  land
      application;
 2.    risk assessment methodology  for regulating landfill  disposal  of  sludge;
 3.    RCRA risk/cost  policy  model  (WET model);
 4.   the liner location risk and  cost analysis  model;  and
 5.   landfill ban model.
 1.5.  MODEL SELECTION EXAMPLE PROBLEMS
     Two site-specific example problems are provided in this section to
demonstrate the procedure for selecting the appropriate mathematical model for a
particular application.  The first example is an application where the objective
is to perform a screening study, while the objective of the second example is to
                                      1-8

-------
perform a detailed study.  The discussions of the example problems are presented
in the order that should be followed when conducting a ground-water flow and
contaminant transport model study, with model selecting being one element of the
process.
1.6.  APPENDIX A
     The appendix of the document contains a summary page for each of the
analytical and numerical mathematical models discussed in Table 5-3.  The models
are divided into seven categories:
1.   analytical flow models;
2.   analytical transport models;
3.   numerical flow models which can be applied to both saturated and
     unsaturated systems;
4.   numerical flow models which can only be applied to saturated systems;
5.   numerical contaminant transport models which can be applied to both
     saturated and unsaturated systems;
6.   numerical contaminant transport models which can only be applied to
     saturated systems;  and
7.   numerical contaminant and heat transport models which couple the solutions
     for pressure, temperature, and concentration (coupled models).
                                       1-9

-------

-------
                                2.   INTRODUCTION

     This document presents a set of criteria which provide a means of selecting
the most appropriate mathematical model for conducting an exposure assessment
related to ground-water contamination.  These criteria were developed in
recognition of the growing use of exposure assessments across the U.S.
Environmental Protection Agency's regulatory programs.  Use of the criteria will
expedite the regulatory process by eliminating the use of unacceptable or
inappropriate models*  Their use will also improve the quality of data used in
the decision-making processes and promote consistency in exposure assessments.
     These selection criteria are particularly directed toward exposure
assessments.  However, the same or very similar criteria would be applicable to
all aspects of managing ground-water contamination problems/sites.  To manage
contamination problems/sites, mathematical models are needed to perform initial
screening studies, assist in the design of disposal schemes, assess the probable
contaminant performance of specific sites, predict contaminant migration, and
aid in the design of monitoring programs and remedial action alternatives.
     In the context of this document, the term mathematical model refers to both
analytical and numerical solutions.  Analytical solutions refer to both those
which are coded for the computer as well as those which are suitable for hand
calculation.  Numerical solutions refer only to those which are coded for
solution on the computer since this is the only practical solution technique.
Throughout most of the report, the term mathematical model is shortened simply
to model.
     When performing an exposure assessment, the primary interest is in
predicting the transport of contaminants.  However, since ground-water flow is
                                      2-1

-------
an integral part of contaminant transport, it is equally important, if not more
so, to accurately predict the ground-water flow.  Therefore, both ground-water
flow and contaminant transport mathematical models, and criteria for selecting
these models, are discussed in this document.  Related models, such as
unsaturated flow, nonaqueous phase liquid, and geochemical models, are not
discussed in this document.         ='.-,.
     In order to develop a tractable, useful model, the importance of the
various processes controlling contaminant migration in ground water must be
identified.  Only the dominant processes are incorporated in the mathematical
models.  The explicit incorporation of every known or observed process is not
practical because the resulting model would require excessive computational time
and would contain too many internal coefficients that must be adjusted in the
                                  t                 ''
calibration or initial phase of a modeling study.  In addition, the effect of
the minor processes, in terms of the predicted concentration levels, is very
small in comparison with the dominant transpprt processes and attenuation
mechanisms.
     All of the contaminant transport models discussed in this report are
designed for situations where the contaminant is at trace concentration levels.
Trace concentrations, in this context, are defined as concentrations that have a
negligible impact on the density and viscosity of the fluid and, therefore, have
an insignificant effect on the ground-water movement.  The models are not
designed to be used in an emergency response framework,  such as an accidental
spill,  because in those situations 1) the contaminant is not likely to be at
trace concentrations initially and 2) the time required to set up the model, run
the model, and evaluate results is generally greater than the emergency response
times.
                                          2-2

-------
     This document is intended primarily to assist potential model users who are
not experts in water quality modeling.  A concerted effort has been made to
define specific terminology and to characterize the important assumptions and
limitations of the existing models.  The existing models are generally the best
available technology and are useful tools when applied properly.  However, model
accuracy is very sensitive to input parameters and calibration with field data
is essential.  In any modeling study the .assumptions and limitations of a
particular model along with the means by,which it is applied should be clearly
understood by the model user as well as by persons making decisions based, in
part, on the modeling results.  Although the guidance provided in this document
is primarily directed toward applications-oriented users of mathematical models,
the information presented is also  important to managers and other decision
makers who will have the ultimate  responsibility of assessing and controlling
contamination problems.
     In view of the diversity in typical modeling needs and objectives
associated with exposure assessments and related studies, these selection
criteria are formulated as a general guideline for selecting a relevant or
appropriate mathematical model.  They are not intended to serve as an absolute
set of standards for accepting or  rejecting models for possible use in exposure
assessments.
     These criteria deal only with the selection of existing mathematical models
for predicting ground-water flow and contaminant transport.  The  criteria do not
deal with the development of numerical algorithms for constructing new models.
Two other reports which address the issue of model selection criteria are
Simmons and Cole, 1985, and U.S. EPA, 1987.
                                       2-3

-------
     The  organization of this document is as fallows:
     Chapter 2 -- Background Information -- General  discussion of ground-water
flow and  contaminant.transport, description of ..equations .and processes, and
definition  of specific terms.            .                .   .
                        '   '•  '• M •'..•:.'  '•   ,.,',,•  j.i,: .  ••"'!, ./.- •>,.>.:. .  - ':' .'I-,.,''. ...... I'!?..
     Chapters -- General  Guidelines, and Principles  of  Model .Selection    ;i
Criteria  --  Overview,of the modeling p,roces,s, overview  of model  selection
criteria, and important issues .related to .model selection...
     Chapter 4 --Model Selection Decisipn Process — Step-by-step process  to,
identify the appropriate model(s)  for a specific application.         ,
     Chapter 5 -- Model Selection Example Problems, -.-Examples of how to use the
selection criteria.             ,   ,     .....:,
     Appendix A -- Analytical and Numerical Model, Summaries,   ; .,           ,  ;
                                      2-4

-------
                          3.  BACKGROUND INFORMATION

     Some of the general background information necessary to understand the
selection of a ground-water flow and/or contaminant transport model  is discussed
in this section.  This chapter is intended for the exposure assessor or the
non-modeler who is not completely familiar with hydrogeologic and modeling
terms.  It should be emphasized that the information presented here is a brief
overview.  For much more detailed discussions the reader is directed to the
following sources:  Freeze and Cherry, 1979; Chow, 1964; Davis and DeWest, 1966;
Javandel et al., 1984; Bachmat et al., 1980, and Mercer and Faust, 1981.
     The first section provides a primer on ground-water flow.  The second
section provides background information on contaminant transport in ground
water.  The last section provides definitions for terms used throughout this
report.
3.1.  PRIMER ON GROUND-WATER FLOW
     This section is  intended to provide a brief summary of the background
information necessary to understand a ground-water problem.  The chapter
discusses the general terms used to describe and define ground-water flow and
presents the basic  equation for flow  in a ground-water system.
3.1.1.  Ground Water  and the Hvdroloqic Cycle
     The hydrologic cycle can be defined as the endless circulation of water
between ocean,  atmosphere, and land.  The hydrologic cycle  is composed of
precipitation,  storage, runoff, and evaporation of the earth's water during the
cycle.  The total amount of water  is  essentially  fixed, but  its  form may change
 (i.e.,  solid,  liquid, gas).
                                       3-1

-------
      Our interest for the purposes of this report is in the land-based-portion
 of the cycle, particularly the portion that infiltrates the land surface and
 flows underground.  The subsurface distribution of water can be divided into
 five categories:
 1.   Soil Zone -- That area where evaporation and transpiration of water occurs
      (partially saturated).
 2-   Vadose (Unsaturated)  Zone — Partially-saturated zone consisting of
      sediments whose interconnected pore space (porosity)  is not completely
      filled with water.   Water flow in this region can be  both  horizontal  and
      vertical  as stratification of the sediments  will  cause significant
      conductivity contrasts.
 3'   The Capillary Fringe  —  A transition zone from the partially-saturated
      vadose zone to the  fully-saturated phreatic  (water table)  surface.
 4«   Phreatic  (Saturated)  Zone — Below the phreatic surface, the porous
      material  is fully saturated  with  water (with the  exception  of entrapped
      gas).   Water in this  zone is under hydrostatic  pressure.   The porous
      material  is called  the aquifer.
 5.    Dense  Rock  —  The phreatic zone merges at depth into  a  zone  of dense  rock
      with water  in  the pore spaces  but  few or  none of these  pore  spaces  are
      interconnected  and  flow  of water  is  severely  limited.
      That portion of the hydrologic cycle  beneath the land surface  can be called
the subsurface flow  system.   Inflow or recharge to the subsurface flow system
arrives as precipitation (in  the  form of rainfall  or snowmelt) or as
infiltration from surface water bodies.  Outflow or discharge occurs as
evapotranspiration or evaporation and as flow to surface water bodies.
Artificial recharge and discharge occur as injection or pumping, respectively.
                                          3-2

-------
     In this report, the selection of models developed to simulate ground-water
flow and contaminant transport in the saturated zone is discussed in detail.
Therefore, the rest of this background discussion is directed to flow in the
phreatic or saturated zone.
     Water that infiltrates the land s.urface generally moves vertically downward
to the water table and the phreatic zone.  Water in the phreatic zone generally
moves horizontally from areas of greater to areas of lesser hydrostatic head
(i.e., energy:  .see Section 3.1.3).  In both zones, unsaturated and saturated,
the prime moving force is gravity.
     The occurrence, movement, and storage of ground water are related to and
influenced by the porous media structure, lithology, thickness; hydraulic
conductivity,, hydraulic gradient, and porosity.  A formation, or group of
formations, that contains sufficient saturated permeable material to yield
significant quantities of water to wells and springs is defined as an aquifer.
Movement  and storage of water in an aquifer are chiefly controlled by the
aquifer hydraulic conductivity, permeability, and porosity.  Aquifers can be
subdivided  into three main types as follows:
1.    Unconfined (Water Table) Aquifer -- An aquifer in which the top of the
      saturated zone  (water table)  is in direct contact with the atmosphere
      through the open pores of the earth material above.
2.    Confined  (Artesian) Aquifer —. An aquifer which has an overlying layer
      which  does not  allow direct contact of the aquifer with the atmosphere.
      Water  in  a confined aquifer is under pressure and wells penetrating into
      the  aquifer will have a water level that reflects the pressure in the
      aquifer at the  point of .penetration.
                                           3-3

-------
 3.    Perched Aquifer — Beds of clay or silt or other materials  of limited  areal
      extent which present a restriction to flow of downward  moving water in the
      vadose zone may cause local  areas of saturation  above the regional  water
      table.  An unsaturated zone  is present between the  bottom of the  perching
      bed and the water table.
      Aquifers have a higher hydraulic conductivity than  adjacent units.   The
 lithologic units of low hydraulic conductivity  relative  to the aquifer are
 commonly called aquitards.   Appreciable quantities of water  can  move through an
 aquitard,  in most cases vertically upward and/or downward, from  aquifers  above
 and  below.   If very little  flow occurs,  the unit is termed an aquielude.
 3.1.2.   Porosity and Hydraulic  Conductivity
      The portion of an  aquifer's  volume which consists of openings  or  pores  and
 not  solid  material  is defined as  porosity.   Porosity  is  an index  of how much
 water can  be stored in  the  saturated  material and  is  usually expressed as a
 percentage  of the bulk  volume of  the  material.   Effective porosity  is defined as
 the  ratio  of the transmissive pore  volume  to the total unit volume, where the
 transmissive pore volume  is  that  portion which contributes to the net flow
 through  the  system.
      Hydraulic  conductivity  (K) is  a  measure of the capacity of a porous rock,
 soil, or sediment to  transmit water.  Aquifers having high hydraulic
 conductivity generally consist of clean coarse sands and mixtures of sand and
gravel and fine  sands, silts, or clays.  Aquifers having low hydraulic
conductivity generally consist of very fine sands,  silts, and clays, glacial
till, or stratified clays.
     Average values of hydraulic conductivity for different soil  classes  are as
follows:
                                      3-4

-------
                            Hydraulic Conductivity
               Soil  Class   (cm/sec)(apd/ft*)
               Clay
               Sand
               Gravel
10-6-KJ-3     10-2-10,
10-3-1.0      10, -10*
1.0 -102      104 -106
     If a hydraulic conductivity is independent of position in an aquifer, the
formation is homogeneous.  If the conductivity is dependent on position, the
formation is heterogeneous (i.e., K varies from point to point in the medium).
If the conductivity is independent of the direction of measurement at a point in
the aquifer, the formation is isotropic at that point.  If conductivity varies
with the direction of measurement at a point in an aquifer, the formation is
anisotropic at that point.  A formation can be isotropic and heterogeneous,
which is not at all uncommon.  Anisotropy is common when the conductivity varies
in the x (horizontal) and z (vertical) direction.  In many aquifers the vertical
conductivity can be estimated as one tenth the horizontal conductivity.
     The transmissivity of an aquifer is defined as the rate at which water of
the prevailing kinematic viscosity is transmitted through a unit width under a
unit hydraulic gradient.  Though spoken of as a property of the aquifer, it
embodies also the saturated thickness and the properties of the contained
liquid.  Transmissivity is also defined as the product of hydraulic conductivity
and aquifer thickness.
     The storage coefficient of an aquifer is defined as the volume of water
released from storage in a vertical column of 1.0 square feet when the water
table or other piezometric surface declines 1.0 feet.  In an unconfined aquifer,
it is approximately equal to the specific yield, which is defined as the amount
of water that drains from a soil due to the force of gravity.
                                      3-5

-------
3.1.3.  Flow In Ground-Water Systems
     In 1856 Henry Darcy reported that flow between two points in a soil column
is directly proportional to the difference in potential head (energy) between
the points and inversely proportional to the distance between the points.  These
two quantities together (difference in head divided by the distance between the
points) are known as the hydraulic gradient.  Darcy's Law can be written as
                               Q = -KA (h!-h2)/L                         (3-1)
where     Q «  flow rate (L3/t)
          K =  permeability or hydraulic conductivity of the media and fluid
               (L/t)
          A =  cross-sectional area of flow (L2)
      hi,h2 =  potential head at two points on a line parallel to flow (L)
          L »  length of flow path from hj to \\2 (L)
    hj-h2/L =  hydraulic gradient = head drop per unit distance
     The negative sign in Equation 3-1 indicates that ground-water flows from
high to low potential.  Headj potential, fluid potential, and potential  head are
all the same and can be defined as the mechanical energy per unit mass of a
fluid at any given point in space and time with respect to an arbitrary state
and datum (typically mean sea level).
     Darcy's Law is valid for steady flow with constant flux.  The law is only
valid for laminar flow, not when the flow becomes turbulent which is not common
in ground water.  Darcy's Law is a measure of the average or bulk velocity
through a given cross section of a porous medium.  The true ground-water
velocity between soil grains is defined as the Darcy velocity divided by the
porosity of the soil.
                                      3-6

-------
      The fundamental  equation  of ground-water flow can be derived from Darcy's
 Law
 plus  a  continuity  equation

 to yield
   q  -  Q/A =  -KVH

 S aH/at 3 - V. q

s aH/at =   .KVH
(3-la)

(3-2)

(3-3)
where  V - del operator, or
        = a2/ax2 + a2/ay2 + a2/az2
       S = storage coefficient
       H = hydraulic head
       K = hydraulic conductivity
Equation 3-3 can take many forms depending on whether the flow is steady state
or transient and whether the media is homogeneous or heterogeneous and isotropic
or anisotropic.  Refer to the references at the beginning of the chapter for the
various forms of the equation.  Steady-state flow occurs when the magnitude and
direction of the flow velocity are constant with time at any point in  the flow
field.  Transient flow (unsteady flow) occurs when the magnitude or direction of
the flow velocity changes with time at any point in the flow field.
     Each form of the ground-water flow equation has an infinite number  of
solutions.   To get a specific solution for a problem,  the initial  and  boundary
conditions  must be specified.   Initial conditions pertain to transient flow
cases only,  and specify the value of the dependent variable  (head)  initial  time
(t - 0).   Boundary conditions  are where the head or flux conditions  at the
boundaries  of the problem must be specified as  either  prescribed head
(Dirichlet)  or flux (Neumann).   For prescribed  head conditions,  the  head  is
                                      3-7

-------
specified at points along the boundary.  For prescribed flux conditions, the
flux of water (either in, out, or no flow) is specified at points along the
boundary.
3.2.  PRIMER ON CONTAMINANT TRANSPORT
     This section is intended to provide a brief summary of the background
information necessary to understand the transport of contaminants in ground
water.  The chapter begins with the basic equation for advective-dispersive
transport and discusses the important terms in detail.
3.2.1.  Advective Dispersion Equation
     Movement of contaminants in the soil can be described by the following
equation (van Genuchten and Alves, 1982)
                      ac/at = D* ac2/ax2 -  V* ac/ax  - kc                (3-4)
where C - solution concentration  (mg/1)
     D* = D/R
     V* - V/R
      R = 1 + (B/N) Kd = retardation factor (dimensionless)
      D « dispersion coefficient  (cm2/day)
      V » average interstitial pore-water velocity (cm/day)
      k - degradation rate  coefficient  (day'*)
      B » bulk density (g/cm3)
      N » effective porosity  (dimensionless)
     Kd <* partition coefficient  (ml/g)
Equation 3-4 states that the  change  in  contaminant concentration with  time  at
any distance, (X) is equal  to the algebraic sum of the dispersive transport (1st
term to right of equal sign), the convective  transport (2nd term),  and the
degradation or decay of  the compound (3rd term).  Van Genuchten  and Alves  (1982)
                                       3-8

-------
note that various modified forms of this same basic equation have been
used for a wide range of contaminant transport problems in soil science,
chemical and environmental engineering, and water resources.
     Equation 3-4 considers only one-dimensional transport of contaminants.
This equation considers dispersion, advection, equilibrium adsorption (linear
isotherm), and degradation/decay (first-order kinetics).
     A wide variety of physical processes occur in ground-water systems which
are important, to varying degrees, in the analysis of  contaminant fate and
transport.  A more detailed description of these processes is given by Fischer
et al. (1979) and Schnoor (1985).  Some of the important hydrologic transport
processes include:
e    Advection -- The process by  which solutes are transported by the bulk
     motion of the flowing ground water.  As a result of advection,  nonreactive
     contaminants are carried at an average rate equal to the average linear
     velocity of the water.
9    Molecular Diffusion -- The process whereby ionic or molecular constituents
     move under the influence of their kinetic activity in the direction of
     their concentration gradient.   Diffusion occurs in the absence  of any bulk
     hydraulic movement of the solution.  Diffusion is a dispersion  process
     which is important only at low velocities.
•    Hydrodvnamic Dispersion -- The tendency of a contaminant to spread  out from
     the path that it would be expected to follow according to the advective
     hydraulics of the flow system.  Dispersion  occurs because of mechanical
     mixing during fluid advection  and because of molecular diffusion due  to the
     thermal-kinetic energy of the  solute  particles.
                                      3-9

-------
•    Mechanical (Hydraulic) Dispersion -- Dispersion that is caused entirely by
     the motion of the fluid.
e    Longitudinal Dispersion -- Spreading of the contaminant in the direction of
     bulk flow.
t    Transverse Dispersion --' Spreading of the contaminant in the directions
     perpendicular to the bulk flow.
     Host contaminant transport models simulate hydrodynamic dispersion only and
disregard the molecular diffusion component because it is so small.  Some models
do simulate the molecular diffusion component for cases where the ground-water
velocity is small and the diffusive component can become significant.
     Host models available today simulate contaminant transport using a form of
the advective-dispersion equation.  Increasing evidence suggests that the
conventional advective-dispersive equation does not always adequately describe
contaminant transport in a natural geohydrologic system as a result of 1) random
variations in ground-water velocity that are induced by media heterogeneities,
and 2) the failure of Pick's diffusion law to properly describe hydrodynamic
dispersion at field scales (Gelhar et al., 1979; Gelhar and Axness, 1983; Smith
and Schwartz, 1980; Hatheron and DeHarsily, 1980; Dagan and Bresler, 1979; and
Dagan, 1982).  Field determinations of dispersivity have shown a scale
dependence, that is, the observed dispersivity was dependent on the size of the
experiment rather than being exclusively a media property.  A more detailed
discussion of recent findings regarding dispersivity can be found in Gelhar and
Axness (1983), Freyberg (1986), Gelhar et al. (1985), and Pickens and Grisak
(1981).
                                      3-10

-------
 3.2.2.   Attenuation and Degradation  Mechanisms
      The primary physical  processes  included  in  contaminant  transport  models  are
 sorption and  degradation.   The kinetic  formation and  rate  constants  used  to
 describe these  processes are  typically  based  on  laboratory measurements.   The
 results  of the  laboratory measurements  are  incorporated  in the models  as  source
 or  sink  terms in the general  advection  dispersion equation.
      The direct transfer of controlled  experimental results  to natural ground-
 water systems is not always straightforward.  Uncertainties  arise  in the
 definition of driving forces  such as whether  the system  is aerobic or  anaerobic,
 the pH of the natural  system,  and the chemical equilibria  of the natural  system
 (i.e., whether  other ions  are  present which may  catalyze or  retard various
 reactions and the organic  content of the soil).  One  of the  biggest problems
 with  simulating natural  ground-water conditions  in the laboratory is properly
 representing  the geologic  media  (i.e.,  layering, heterogeneity, hydraulic
 conductivity, porosity,  etc.).   In spite of the  uncertainties, these processes
 are incorporated in  many ground-water models, and some models have been
 calibrated  to field  conditions.  Careful calibration  has shown that they are
 useful for  representing the transport of various chemicals.
     Most of  the available models use some form of first-order reaction kinetics
 to represent  the different processes that will degrade or transform a specific
 chemical.   For  a simple first-order reaction, ignoring all  other mechanisms,  the
 concentration can be represented as a first-order differential  equation
                                  dc/dt - -kc                            (3-5)
where k is the rate constant (1/T).   In the simpler models, the rate constant
does not change, in the more complex models the rate constant(s)  may be variable
                                     3-11

-------
and calculated as a function of changing environmental  conditions.   The
analytical solution to this equation when k is a constant is
                                C(t) =  C0 e'kt
                                                                         (3-6) ,
where Co is the initial concentration.  From this equation an estimate of the
time required for the process to reduce the contaminant concentration below a
fixed "action level" can be determined.
                                 t  = Ln Cm/C0                           (3-7)
                                         -k
     Often the reaction rate of various chemicals subject to different kinetic
processes are characterized in terms of their half life, tj/2-  This is a
measure of the time required for some  kinetic process to degrade or transform
the specific chemical to one-half of the initial concentration.  The half-life
is calculated from Equation 3-7 with the C(t)/C0 set to 1/2.
     A brief description of the sorption and degradation processes is included
for potential users unfamiliar with the terminology.  Much more detailed
descriptions including assumptions, limitations, kinetic formulations, and
methods for estimating rate constants  are given by Bohn et al . (1979), Bolt and
Bruggenwert (1978), Peterson et al.  (1986), and Freeze and Cherry (1979).
3.2.2.1.  Sorption -- Sorption is a transfer process whereby dissolved chemicals
in the ground water become attached to sedimentary materials and/or organic
matter.  The process is commonly described using a partition coefficient.  The
definition of the partition coefficient is the ratio of the mass of chemical
sorbed on the solid phase divided by the mass of chemical left in solution at
equilibrium, as shown in Equation 3-4.  The important assumptions in using this
                            1   '      t                                       • 0*
formulation are:  1) the chemical is at trace concentrations, hence the sorption
isotherm may be assumed to be linear;  and 2) the system is at equilibrium.
                                      3-12

-------
     Some problems associated with field application of this concept include:
1) many chemicals exhibit nonreversible sorption characteristics, hence,
desorption from sediments to the water column may not be correctly represented;
2) sorption characteristics are dependent on particle size (sand, silt, clay)
and particle size is variable in a natural system; 3) sorption characteristics
are dependent on pH and ionic strength which are often variable in a natural
system; and 4) the presence of other compounds in a natural system results in
competition for sorption sites on the soil matrix.  Some of the compounds that
may be strongly affected by sorption include heavy metals and many organic
compounds.  For organ compounds, partition coefficients are frequently
normalized to the organic carbon content of the soil (Karickhoff, 1980, 1981,
and 1984; and Gschwend and Wu, 1985).  As noted previously, the relationship
between dissolved and adsorbed forms of a contaminant is usually represented in
the form of an equilibrium partition coefficient (Kd).  The partition
coefficient is defined as the ratio of the mass of the substance adsorbed to the
particulates (per unit mass of particulates) over the dissolved concentration of
the solute.  The retardation factor is calculated from the partition
coefficient, bulk density, and the effective porosity, and is a number which
describes how many times slower than water a contaminant travels through a
porous media system.
3.2.2.2.  Degradation -- Degradation in ground water systems may result from one
or more of three mechanisms:  biological transformations, hydrolysis, or
chemical reactions.  Biological transformations are reactions due to the
metabolic activity of aquatic microbes, primarily bacteria.  Depending on the
specific chemical, the transformations may be very fast due to the presence of
enzymes and for other compounds the process may be very slow.  The rate and
                                      3-13

-------
nature of by-products will also be dependent on the availability of oxygen.  In
unsaturated aerated zones, aerobic degradation will predominate, while anaerobic
mechanisms will be controlling in anoxic zones.  For chemicals where
the transformation is fast, degradation is often the most important
transformation process in the aquatic environment.  Various kinetic formulations
have been proposed including first- and second-order forms.  The rate
coefficients are known to be a function of temperature, pH, and available
nutrients.  The second-order kinetic formulations describe the degradation rate
as a function of the concentration of the compound and the bacterial population
which is changing as the compound is degraded.  A variety of organic compounds
may be subject to degradation.  A discussion of various compounds is provided by
Klecka (1985).
     Hydrolysis is the reaction of chemicals with water.  Typically a compound
is altered in a hydrolysis reaction by the replacement of some chemical group of
the compound with a hydroxyl group.  The hydrolysis reactions are commonly
catalyzed by the presence of hydrogen or hydroxide ions and hence the reaction
rate is strongly dependent on the pH of the system.  Hydrolysis reactions alter
the structure of the reacting compound and may change its properties.  The new
compound is usually less toxic than the original, compound, but this is not
always the case depending on the specific reaction.  Neely (1985) lists several
functional groups that are susceptible to hydrolysis reactions including alkyl
ha!ides, amines, carbamates, carboxylic acid esters, epoxides, lactones,
phosphoric acid esters, and sulfonic acid esters.  For many functional groups,
and therefore a considerable number of compounds, hydrolysis will not occur.
     Chemical reactions refer to the interaction of contaminants with other
chemicals in the ground water besides the water itself.  Many chemical reactions
                                      3-14

-------
are of the oxidation-reduction type (redox).  Redox reactions involve the
transfer of electrons from one chemical (the reducing agent) to the other (the
oxidizing agent).  In the process, toxicity and solubility properties are often
changed.  Possible couples for redox reactions can be determined from oxidation
potential tables.
3.2.3.  Definition of Terms
     Throughout this report a number of terms or phrases are used which may be
interpreted as having somewhat different meanings by different readers with
various backgrounds, experience, and general inclinations.  In an attempt to
avoid any misinterpretations, this section defines the specific meaning intended
for a few key terms.
 1.  Calibration --  In this document we will use the term calibration to
     describe the  initial phase of a modeling study where the input coefficients
     of a model  are  adjusted  in an attempt to match measured field data (e.g.,
     velocity, concentration).  The types of coefficients that are commonly
     adjusted in a ground-water flow model  are recharge and discharge, hydraulic
     conductivity  or transmissivity, and porosity.  The types of coefficients
     that are commonly adjusted in a transport model include dispersion
     coefficients, degradation rate constants, sorption properties,  and possibly
     source  and  sink terms.
 2.  Validation  -- The term  validation will be used to describe a  separate  step
     of a modeling study where the calibrated model  (i.e.,  fixed coefficients,
     no more adjustments) is  applied to  a different  set of  conditions  and the
      results are compared with  a  separate set of  field data.  The  validation
      phase  is an attempt to  see  if the model  can  reproduce  field data  under
      conditions  different than  those used  in  the  calibration  phase.  This
                                      3-15

-------
    distinction between calibration and validation is easy to define and
    appropriate from an idealistic point of view.  Real applications may find
    it impossible to obtain a separate data set or the validation phase may
    indicate poor model performance and the validation data set might be used
    for additional calibration.  Under these circumstances, larger uncertainty
    in the model results, and this needs to be incorporated into any
    decision-making process.
3.  Verification -- The term verification will be used to define the process of
    comparing the results of one model against those of another model.  For
    example, it is common to compare the results of a tested and accepted
    analytical solution/model against a more sophisticated numerical model in
    order to verify the numerical solution technique.
4.  Mathematical Model -- The term mathematical model will be used to describe
    the mathematical representation of the physical system.  The model may
    represent an analytical solution to these equations and in other cases the
    model may be an approximate numerical representation of these equations.
    In some cases, the models based on analytical solutions are simple enough
    that the calculations can be performed using a hand calculator.  In other
    cases the analytical models are more complex and often are implemented as a
    program to be run on a computer.  All of the numerical models are
    implemented as programs to be run on a computer.  Computer models are often
    referred to as codes or computer codes.  We have used the term model, as
    opposed to code, wherever appropriate in this document.  Mathematical model
    in this definition is equivalent to mathematical systems model  involving
    multi-processes.
                                     3-16

-------
 5.   Process Equation -- The mathematical  representation of a physical
     phenomenon or process for a system.   For example,  solute transport in a
     saturated system may be described by the solution  of the advective
     dispersion equation.  A process equation in this definition is equivalent
     to a multi-process equation involving several  different phenomena.
 6.   Analytical Models -- A computer program written to solve a particular
     process equation.  For example, the AT123D model (Yeh, 1981) is a computer
     program written to analytically solve the advective dispersion transport
     equation for a variety of simple initial and boundary conditions.
 7.   Numerical Model -- A computer program written to solve a particular process
     equation for which no general solution exists.  For example, the CFEST
     model (Gupta et a!., 1986) is a computer program,  part of which is written
     to numerically solve the advective dispersion transport equation for a
     variety of simple or complex initial and boundary conditions.
     The relationship between the above terms may be thought of as follows.  A
mathematical model consists of two aspects:  a process  equation and a solution
technique to solve the process equation.  An analytical solution solves a very
simple process equation analytically by hand calculations.  An analytical model
solves a more complex, but still relatively simple, process equation
analytically with a computer program.  A numerical model solves a simple or
complex process equation numerically with a computer program.  In the context of
this document, mathematical model refers to all three solution techniques
(analytical solution, analytical model, numerical model) of a process equation.
The more detailed the specific application, the more complex the process
equation.  The complexity of the process equation dictates the solution
technique required.
                                      3-17

-------
     The term analytical or numerical code typically refers to the computer
program (the set of computer instructions written in a programming language and
acted on by a computer), whereas an analytical or numerical model is the
implementation of the code with a specific data set (either site specific or
generic) to test the simulated representation of the system against observed or
measured behavior.  In this document the analytical solutions referenced in
Table 5-1 and the analytical and numerical codes summarized in Appendix A are
all included in the general category of mathematical models.
 8.  Objectives Criteria -- Criteria related to the level of modeling detail
     required to meet the objectives of the study.  In the context of model
     selection, objectives are classified into two categories:  perform a
     screening or generic study where simple analytical solutions/models would
     most likely be used, or perform a detailed study using numerical models.
 9.  Technical Criteria -- Criteria related to a code's ability to simulate the
     transport and fate phenomena of importance.  These criteria are based on
     the physical, chemical, and biological characteristics of the site and the
     contaminant of interest.
10.  Implementation Criteria -- Criteria related to the ease with which a code
     can be obtained and its acceptability demonstrated.   Relevant factors
     include sources of the code,  documentation, verification, and validation.
11.  Screening and Detailed Assessment -- The types of modeling analysis that
     are described in this document may be very broadly categorized as screening
     or detailed (site specific) exposure assessment studies.   Obviously, these
     categories cannot be distinguished by definitive criteria but rather there
     is more of a "gray" area between the two.  We have chosen to use the term
     screening analysis to represent studies where limited calibration and
                                     3-18

-------
validation data are available and the uncertainty associated with the
predicted results Is comparatively large, somewhere In the nature of an
order of magnitude.  The term detailed or site-specific analysis is used to
represent studies where a smaller uncertainty in the predicted results is
necessary, on the order of a factor of two to ten.  Calibration and
validation data are necessary in order to reduce the uncertainty inherent
in the results and also attempt to quantify the bounds associated with the
uncertainty through the validation phase and sensitivity studies.  The
models used for a screening analysis are generally easier to use but make
certain restrictive assumptions.  The more complex site-specific assessment
models are more difficult to use and generally do not make as many
restrictive assumptions; the input data requirements, however, may be
substantially greater.
                                3-19

-------
.)•« f

-------
                     4.  GENERAL GUIDELINES AND PRINCIPLES
                          OF MODEL SELECTION CRITERIA
     This chapter discusses some of the general guidelines and principles
related to model selection criteria.  The first section discusses the steps that
should be taken in the overall modeling process.  The second section provides a
general discussion of the model selection process and discusses the three
principle criteria used to select a model:  objectives criteria, technical
criteria, and implementation criteria.  The last three sections discuss some
important issues related to model selection:  model selection vs model
application familiarity with a model, and model reliability.
4.1.  OVERVIEW:  MODELING PROCESS
     The selection of models for the analysis of exposure to contaminants
involves factors addressing a number of issues, not all of which are amenable to
expression in specific criteria.  Certain judgmental factors are better suited
to statement in the form of general guidelines and principles.  Many of these
arise from the nature of the overall modeling process of which model selection
is but a single step.  Five general steps may be identified in the modeling
process.  Although model selection is meant to be the primary emphasis of this
report, the different steps influence each other and need to be described.  The
five general steps are:
1.   Problem Characterization -- The analyst clearly identifies the exposure
     assessment study objectives and constraints.
2.   Site Characterization -- The analyst reviews available data on the site,
     develops a conceptual model identifying processes of interest, performs a
     screening analysis; if a modeling study is necessary, the analyst then
     identifies data needs and fill those needs.  The results of the site
                                      4-1

-------
     characterization will determine technical specifications for model
     selection by identifying the single processes at the site.
3.   Model Selection Criteria -- The analyst matches the objective, technical,
     and implementation criteria to available models and selects the most
     appropriate model(s).
4.   Code Installation -- If the model selected is a computer code, the code is
     installed on the computer system and tested to document proper installation
     and ability to reproduce accepted solutions to standard problems.
5.   Model Application -- The verified model uses site characterization data as
     input for the exposure assessment simulation.
     These five general steps are not the model selection criteria but rather
the overall process by which a problem is identified and a model selected to
perform an exposure assessment study.  Model selection criteria is listed as the
third step in this process.  The two previous steps, problem characterization
and site characterization, are crucial in the selection of an appropriate
model(s).  While the steps can be considered sequential in nature, it is
important to recognize interactions and feedback mechanisms between them.  For
instance, knowledge of the model selection criteria is important to assure that
site characterization is adequate and properly formatted.  An understanding of
code installation procedures is required for proper scheduling and resource
allocation.  Familiarity of candidate models is needed to assure that site
characterization provides necessary input data.
     Problem characterization is important because a wide variety of models and
modeling approaches are available.  Different modeling techniques are suitable
for different objectives and physical problems.  The exposure assessment
objective must define what the goal of the analysis should be and must also be
                                      4-2

-------
defined in a manner consistent with known project constraints such as schedule,
budget, and other resources.
     Site characterization is an important step that needs more description
because the conceptualization of the physical system, whether it is a specific
site or a generic problem, will obviously influence any additional steps.  If
the objectives of the exposure assessment are to evaluate an existing
contamination problem, then this step should include the availability of field
measurements in the specific study area.  Depending on the problem, the field
measurements may include geologic structure data, pump tests, recharge/discharge
data, elevations of surface water bodies, contaminant sampling, and source
characterization.  Field measurements may identify the extent of the
contamination problem and whether or not the concentration levels are above some
regulatory or dangerous level.  In addition, if these initial studies identify a
contamination problem and a modeling study is to be performed, then the field
measurements will be used in the selection of an appropriate model and for model
calibration.
     Model selection criteria  (the primary goal of this report) is entirely
dependent on the first two  steps.  This  step is covered in more detail in the
rest of this chapter and  in Chapter 4.0.
     Code installation only applies when the model chosen in the third step is a
computer code.  When a code is  first obtained and installed on a specific
computer system, it  is essential that the model be tested to verify that it is
working correctly and can reproduce suitable example problems.  Various computer
systems and the necessary model software may have a  variety of differences, some
distinct and others  more  subtle.  These  differences may require modifications  to
an acquired code  (e.g., double  precision arithmetic  or changing output formats)
                                       4-3

-------
 on different computer systems.  Verification assures that modifications have not
 changed model results significantly.  This step should be performed by the
 person doing the actual model analysis.
      Model application relates the use of a model in an attempt to answer the
 questions to the use of a model in an attempt to answer the questions defined in
 the objectives.  Depending on the objectives of the analysis,  this step may
 consist of several  parts including calibration, validation,  and application to
 evaluate different  conditions or scenarios.
 4.2.   OVERVIEW:  MODEL SELECTION CRITERIA
      There are three factors which dictate the  level  of complexity of the
 mathematical  model  chosen in the selection process:   1)  objectives criteria,  2)
 technical  criteria,  and  3)  implementation criteria.
 4.2.1.   Objectives  Criteria
      The first  level  of  consideration when selecting  a mathematical model  is
 related  to the  objectives of the study.   Based  on the objectives,  the  analyst
 can limit  the choice  to  either  simple analytical  solutions/models  or to more
 complex  numerical models.
      The objectives criteria  refer to the  level of modeling detail required to
meet  the objectives of the  study.  There are many different objectives of
modeling studies, however,  in the context of model selection, all objectives can
be classified in two broad categories:  1) to perform a screening study or 2) to
perform a detailed study.
     A screening study is one where the purpose is to make a preliminary
screening of a site or to make a general comparison  between several sites.   A
                                      4-4

-------
screening assessment can also be generic (i.e., not site specific) where the
objective is to compare several hypothetical cases or designs.
     A screening study may be appropriate when sufficient data is not available
to properly characterize the site.  Although lack of data is no excuse for not
modeling a site correctly, often there are times when a screening comparison of
cases/sites can be quite helpful in analyzing a problem.  A screening study can
help direct the data collection effort at a site.  For generic studies, it is
most often not possible to select a detailed data set which is representative of
all the hypothetical sites/cases being simulated.  By virtue of the fact that it
is a generic study, the data base is typically quite simple such that it is
representative of several sites/cases.
     A detailed study, on the other hand, is used when the objective is to make
a detailed assessment of the environmental  impact, performance, or safety of a
specific site.  This type of study requires detailed data for a specific site or
for a number of sites, and the  results of the study are typically used to make
specific decisions regarding the site or sites.  For example, a detailed study
might involve predicting concentration of a particular contaminant at a specific
aquifer location.  The results  of the study would be used by decision makers to
determine whether remedial action is needed at the site.
     Screening and detailed studies usually require the use of screening and
detailed models, respectively.  A screening model would typically be ,an
analytical solution or model with minimal data requirements.  Usually these
analytical solutions/models are used for special, simplified physical conditions
                                                *
that represent the behavior of  particular physical processes when isolated from
other effects.  Compared to numerical models,  screening models require less
data, are easier to implement,  and are less expensive to run.
                                       4-5

-------
      Detailed models are typically numerical models which have more extensive
 data requirements, are more difficult to implement, and cost more to operate
 (take longer to set up, calibrate, and evaluate results and require more
 computer time) than analytical models/solutions.  Detailed models usually have
 more realistic initial and boundary conditions and possibly more time-dependent
 inputs and outputs than screening models.
      Based on the objectives of the study  (screening or detailed),  the analyst
 or modeler will  select either a screening  or a detailed model.   The specific
 model  to be used will  be selected based on the technical  selection  criteria
 discussed below.
 4.2.2.   Technical  Criteria
     The second  level  of consideration when selecting a mathematical  model  is
 the technical  criteria.   Technical  criteria are those criteria  related to the
 mathematical  model's ability to simulate the site-specific contaminant transport
 and fate phenomena of  importance.   These criteria  are based on  the  physical,
 chemical,  and  biological  characteristics of the site  and  the contaminant  of
 interest.   The characteristics  of the  site  and  the  processes that need to be
 simulated  are  determined  from the  hydrogeologic and contaminant  data  and  the
 conceptual  model of the site.
     With  regard to model  selection, the technical  criteria can  be  divided  into
 three categories:   1) transport  and transformation  processes, 2) domain
 configuration, and  3) fluid(s)  and media properties.
 4.2.2.1.  Transport and Transformation Processes -- Transport and transformation
 process criteria relate to those significant processes or phenomena known to
occur on site that must be modeled in order to  properly represent the  site.  The
transport process is the physical migration process controlled by adsorption,
                                      4-6

-------
attenuation, diffusion, dispersion, volatilization, and density effects related
to temperature and/or concentration.  The transformation processes that effect
contaminant migration can be divided into chemical and biological processes.
Chemical processes include complexation, hydrolysis, chemical degradation, and
oxidation-reduction.  Biological processes include biodegradation, biological
transformation, metabolism, and respiration.  The mathematical model selected
must be able to simulate all the relevant physical processes occurring within
the specified environmental setting.
4.2.2.2.  Domain Configuration -- The second category of the technical criteria
relates to the ability of the model to represent the domain configuration of the
geohydrologic system.  The relevant parameters related to geometry include:
t    Water table or confined flow system
t    Porous media or fracture flow
t    Steady-state or transient flow
t    Single- or multi-phase flow
•    Constant, flux, or no-flow boundary conditions
•    Single- or multi-layer system
•    Constant or variable thickness layers
•    One-, two-, or three-dimensional system
•    Source configuration
          Constituents
          Point, line, or area source
          Initial value, constant, or variable source
     When high levels of resolution are required to predict contaminant
concentrations for comparison to health or design standards, it is generally
necessary to simulate site-specific geometry and dimensionality for which
                                      4-7

-------
numerical models are most appropriate.  If simplifying the site geometry can be
defended on a geotechnical basis, then the use of a simpler analytical
model/solution may be justified.
     Analogously, the natural dimension of the system can be reduced if at least
one dimension can be integrated to a single value.  For example, if a
contaminant is distributed evenly in a single aquifer in the vertical direction,
the level of contamination can be expressed as a single value in a
two-dimensional transport model of the horizontal plane (x-y).  In effect, the
contaminant mass in the vertical (z) dimension is integrated into a single
representative value.
4.2.2.3.  Fluidfs) and Media Properties -- The third category of technical
criteria corresponds to the ability of the mathematical model to represent the
spatial variability of fluid(s) and media properties of the geohydrologic site.
The relevant issue in this category is whether the site can be considered
homogeneous or heterogeneous with regard to hydraulic conductivity, recharge,
porosity, and specific storage.  If homogeneous conditions can be assumed, it is
often possible to simulate the site with an analytical solution/model.
Heterogeneous conditions almost always require a numerical model.  For example,
an aquifer may consist of several geologic material types all having different
hydraulic conductivities.  Proper simulation of the spatially variable hydraulic
conductivity would require the use of a numerical model.
     As stated above, the objectives criteria will direct the analyst to select
either a screening or a detailed model. Once the level of model has been
decided, the technical criteria will direct the analyst to the specific type of
model needed to properly simulate the transport and transformation aspects of
the environmental setting.
                                      4-8

-------
4.2.3.  Implementation Criteria
     The third level of consideration when selecting a mathematical model is
related to the implementation criteria.,  Implementation criteria are those
criteria dependent on the ease with which a model.can be obtained and its
acceptability demonstrated.  Whereas the technical  criteria identify models
capable of simulating the relevant phenomena within the specified environmental
setting, the implementation criteria identify documentation, verification,
validation requirements, and ease of use so that the model selected provides
accurate, meaningful results.
     Relevant questions to be considered concerning the implementation of a
particular model include:  1) what is the source of the model and how easy is it
to obtain (i.e., is it a proprietary model); 2) are documentation and user's
manuals available for the model and, if so, are they well written and easy to
use; 3) has the model been verified: against analytical solutions and other
models and, if so, are the test cases available so the analyst can test the
model on his computer system; and.5) has the model  been validated against field
data?
     The technical criteria can be used to narrow the model selection to a few
codes in the same general category. The implementation criteria can then be used
to further narrow the decision to one or several of the technically acceptable
models.                                    >
4.2.4.  Other Factors Affecting Model Selection
     Other general factors related to model selection which should never
override the technical or implementation criteria include data availability,
schedule, budget, staff and equipment resources, and level of complexity of
system(s) under study.  Schedule and budget constraints refer to the amount of
                                      4-9

-------
 time and money available for the assessment.  If both analytical  and numerical
 models meet the selection criteria, .time and cost may be considered factors for
 electing to use an analytical  approach.   Analytical  solutions/models are easier
 to install  on a computer system, are more quickly mastered,  and are dependent on
 more easily-obtainable input data than numerical  models.  The time factor will
 be of increased importance if staff are  not familiar with any of the appropriate
 models.   In this case, schedule considerations  may dictate selection of a
 different modeling team; one experienced with the appropriate models.
      Staff  resources are also  a major consideration  in  modeling.  Regardless of
 the quality of the model selected,  the expertise  of  the analyst has a major
 impact on model  results.  This can  also  impact  model  selection when the analyst
 is familiar with one or more of the appropriate models.   In  the simplest case,
 if the analyst has direct experience with an acceptable model,  then that model
 is preferred.   Similarly,  if the analyst has experience with  a specific type  of
 model  (e.g.,  finite element  vs finite difference), one  of that type should  be
 selected.   In  certain  cases, familiarity with a model more complex  than required
 may dictate use  of that model  rather than a  simpler  one  since  there
 will  be no  loss  of resolution  and the added  staff experience would  compensate
 for time  and cost  differences.  In no case, however,  should familiarity  with a
 model  dictate  its  selection when  it  does  not satisfy the  technical  and
 implementation criteria.   In practice, many models are rightfully applied to
 situations  which are not fully  compatible with  the model's design
 characteristics.   However, justification  of the choice should  indicate  the
 correctness of the model's use.
     Hardware requirements are  similar to staff requirements.  The more complex
mathematical models require more powerful computers with larger mass storage
                                      4-10

-------
devices and extra peripheral equipment.  If both analytical and numerical models
meet the selection criteria, available hardware may dictate the use of the
simpler analytical models.  If a sophisticated model is required and adequate
equipment is not available, alternate means of conducting the modeling must be
found.  Equipment constraints cannot be used to select a model other than those
meeting the selection criteria.
     As noted in the preceding paragraphs, subjective factors such as
objectives, schedule, budget, and staff and equipment may be used to select one
model from a group of models all found to meet selection criteria.  Alternately,
subjective factors may dictate use of a different modeling team to improve the
quality of results obtained from a given model.  Under no circumstances,
however, shall subjective  factors be used to select a model which otherwise
fails the selection criteria.
      In summary, the first step of the model selection process is to define the
objectives of the study.   If the objective is to perform a screening-level or
generic type study, a simple analytical solution/model or  a simple numerical
model should be  selected.   If the objective is to perform  a detail-level  study,
a more complex numerical model should be selected unless the  study area  is such
that  it can be simulated on a technically sound basis with a  simpler model.
      The decision as to which category of models or type of model can  accurately
simulate a site  or perform an assessment is based on the
technical criteria.  The technical criteria are based on a model's ability to
correctly simulate the transport and transformation processes, domain
configuration, and fluids  and media properties of a site.  Depending on  the
objective of the study, the technical  criteria can  be used to narrow the
selection of either  screening or detailed models.
                                      4-11

-------
      Once the appropriate models have been selected on the basis of technical
 criteria, the list can be further narrowed by use of the implementation
 criteria.  The implementation criteria relate to factors such as availability  of
 models, documentation, verification, and validation.
      If a list of models still remain after satisfying the objectives,
 technical, and implementation criteria,  the final  decision should be based on
 such subjective factors as schedule, budget,  and staff and equipment resources.
 Under no circumstances, however,  should  subjective factors be used to select a
 model which otherwise fails the selection criteria.
 4.3.  MODEL SELECTION VS MODEL APPLICATION
      While this document is concerned with the issue of model  selection,  it is
 important to realize  that in most cases  the biggest  differences  in model
 predictions are a result of how the  model  is  applied,  not  which  model was
 selected.   For most modeling studies,  there are  typically  a number of models
 which satisfy the objectives,  technical,  and  implementation criteria.   If  all  of
 these models are  applied to the same data  set, they  will all obtain  virtually
 the  same  results.  Differences  may arise  based on  how  the  initial  and
 boundary  conditions and grid (numerical models)  are  specified  in the model, but
 in all  cases these differences  should  be  insignificant.
      The big difference in  modeling  assessments  results from how the model  is
 applied, not from which model is  selected.  The  difference in how the model is
 applied stems from the  fact  that  different modelers  interpret the same data set
differently.  For example,  if you gave ten different modelers the same raw data
set and the  same model  code, they would probably develop ten different models  of
the site based on ten different interpretations of the data.  The differences  in
the models might be small in many cases,  but it is not unreasonable.that large
                                     4-12

-------
differences (i.e., completely different ground-water flow directions or
contaminant concentrations) could occur; especially for large, complex data sets
requiring a detailed modeling approach.  If, on the other hand, you gave ten
different modelers the same model input data set (i.e., the inputs are defined
for them, not a raw data set) and ten different models, with which they were
very familiar, they should all arrive at about the same results.
     Where the major differences occur in any modeling study is in the
interpretation of the raw data and in the conceptualization of the study area.
Since it is impossible to know everything about a ground-water flow/contaminant
transport system, raw data are always subject to different interpretation.  As a
result of this basic fact, it is important to remember that the major
differences in modeling studies will probably result from differences in the
application of the model(s), not from the selection of the specific model to
use.
4.4.  FAMILIARITY WITH A MODEL
     It  is important to emphasize the importance of familiarity with a model,
especially with regard to the more complex models.  Detailed models, especially
numerical models, can be quite complex with a large number of  input variables,
switches, outputs, and simulation/computer related requirements.  Often  it
requires months or even years of experience and several studies to fully
comprehend all the aspects of a model.  Because of this, it is strongly  advised
that an  analyst select a model with which he/she is familiar if it possesses all
the selection criteria.
     Many people  feel that a more complex model should not be  selected if a
simpler  model can do the job.  They argue that the complex model requires too
much data, will take longer to implement, and will cost more to run on the
                                      4-13

-------
computer than  the  simpler model.  This may not be the case at all.  Most complex
codes can be run in  a less complex mode which requires less data and
implementation time.  Even if computer costs are a little higher, the costs are
almost always  far  offset by the savings in labor costs because the analyst is
familiar with  the  model and can use it far more efficiently.
     When dealing  with the more sophisticated numerical models, only analysts
familiar with  the  selected model should perform the assessment. If experienced
persons are not available within the group, expert help from outside the group
should be obtained.  Simpler models can be used by analysts with little or no
specific experience with the model.  However, the work should be reviewed by a
more experienced user to ensure that the model was set up and the results were
interpreted correctly.
4.5.  MODEL RELIABILITY
     Because a natural ground-water system is very complex and heterogeneous, a
model will never form an exact replica of the system.  Because every model will
inevitably be  a simplification of the actual system,  mathematical models should
not be used to make "exact" predictions of ground-water flow or contaminant
concentrations.  Ideally, models should only be used to make comparisons between
cases, whether site specific or generic.
     When developing a model  of a study area, simplifications must be made at
every step in the process.   The simplifications are a consequence of limitations
in the acquisition of field data,  limitations in developing a conceptual model
of the systems, limitations in properly representing the data and physical
processes in the model,  and limitations in predictive mathematical  theory for
some physical  processes.   Because of all the limitations and simplifications,  it
is often difficult to place a great deal of confidence in predictive model
                                     4-14

-------
results (I.e., prediction of a contaminant concentration at a well).  When
comparing various cases or scenarios with a model, however, all the cases are
subject to the same limitations and simplifications.  Therefore, model results
are more reliable in a situation where one case or alternative is compared to
another.
     Another method by which model results can be made more useful, is to perform
a sensitivity analysis.  A sensitivity analysis is an analysis that quantifies
the change in a specific performance assessment measure (i.e., ground-water
velocity or concentration) resulting from a change in a specific input parameter
or set of parameters to the model.  For example, the model could be used to
predict a range of ground-water velocities based on a range of hydraulic
conductivities and gradients thought to be representative of the aquifer.  A
sensitivity analysis is a means of dealing with the inherent uncertainty which
exists in the measurement of many hydrogeologic and contaminant transport
parameters.
     Although the many limitations and simplifications make it difficult to
place a lot of confidence in model predictions, the use of models to make such
predictions can often be justified in that, in most cases, a model  is the only
means or the best means available to make the prediction.   For sites with
complex hydrogeologic conditions, the use of a model is often the only means of
integrating all the data into a meaningful package.  In all cases,  no matter how
simple the site or how extensive the input data, model output or predictions
always need careful  evaluation because they are generally only as accurate as
the model  input data and the knowledge of the system upon  which the model is
based.  Model results,  whether "exact" predictions, a comparative analysis or a
sensitivity run should always be considered as a guide to  the probable system
                                     4-15

-------
response.  Despite their limitations, when properly applied,  models are often
the best tools available for assistance in making decisions on ground-water flow
and contaminant transport problems.
                                      4-16

-------
                      5.  MODEL SELECTION DECISION PROCESS
                                                           • . „ •  --     •,:..     i
      This chapter discusses the technical  criteria used to select the
 appropriate mathematical  model  for a specific application.   The first section
 provides  a detailed discussion  of the technical  criteria and the errors
 associated with  selecting one option over  another.   The second section  briefly
 discusses the  model  selection process from the point of view of a reviewer from
 a  regulatory agency.  The last  section groups the  available  models into
 categories and provides the framework for  selecting  the appropriate model  or
 category  of models  based  on the response to the  technical  criteria.
 5.1.  TECHNICAL  CRITERIA  USED FOR MODEL SELECTION
      The  technical  criteria used  to  select  the appropriate performance
 assessment mathematical model(s)  for ground-water  applications  are outlined in
 Figure 5-1.
      The  first task  in any  contaminant  transport analysis  is  to  simulate the
 ground-water flow.  The ground-water flow can  be simulated separately from the
 contaminant  transport, or both can be  simulated with one model.   However, each
 approach  requires sequential  simulations; first the ground-water  flow followed
 by the contaminant transport.  For this reason, Figure  5-1 is divided into two
 sections.  The first section  addresses the technical decisions that need to be
made to select the appropriate ground-water flow model  (or flow portion of a
combined ground-water flow and contaminant transport model), while the second
section addresses the technical  decisions that need to be made to select the
appropriate contaminant transport model.
     The decisions that need to be made when selecting a ground-water
flow/contaminant transport model are discussed in detail below.  Guidance is
                                      5-1

-------
                     Ground-Water Flow
              Hater Table or Confined Aquifer? I
               Porous Media or Fracture Flow?
                 I  1.  2,  or 3  Dimensional? I
               |  Single  Phase  or  Multi-Phase
                Homogeneous  or  Heterogeneous?
Hydraulic Conductivity. Recharge. Porositv. Specific Storage
               I Single Laver or Multi-Layer? I
          I Constant or Variable Thickness Layers? I
                 I Steady-State or Transient 1
            Select the Appropriate Analytical or
              Numerical Ground-Water Flow Code
                             or
    Continue with the Decision Tree and Select a Combined
      Ground-Water Flow and Contaminant Transport Model
          Figure 5-1.   Code selection  decision  tree.
                             5-2

-------
       I Contaminant Transport
    Point. Line, or Area! Source?")
 I Initial Value or Constant Source?
      I 1, 2. or 3 Dimensional? I
             Dispersion?
             Adsorption?
       « Temporal Variability
        • Spatial  \
ariability
            Degradation?
        • 1st  Order/2nd Order
        • Radioactive Decay
          Density Effects?
   t Thermal and/or Concentration
Select the Appropriate Analytical or
Numerical Contaminant Transport Code
      Figure 5-1.  (continued)
                 5-3

-------
provided for making each decision  and some discussion is provided regarding the
errors associated with making  an incorrect choice  (i.e., selecting a model that
does not contain the desired feature); for example, the use of a porous media
model to simulate fracture flow.
     In all cases, the impact  of an incorrect decision or the use of an
                    L               "      '=-..-      •'     -
incorrect feature of a model is difficult to quantify.  For many cases, the
errors are difficult to quantify because they are  site specific; they depend on
the actual site data.  For example, the error associated with using a
homogeneous hydraulic conductivity distribution where a heterogeneous
distribution should have been  used can only be quantified on a site-by-site
basis.  In other cases, the errors are difficult to quantify even for a
site-specific application because  the necessary parameters are unknown or cannot
be measured.  An example where it  is difficult to  quantify the associated error
would be using a model that uses a uniform value of recharge over a study area
versus one that uses a recharge distribution because recharge is difficult to
measure in the field.  '
     Many of the decisions that need to be made when selecting an appropriate
model can be made on the basis of  past experience.  Although this document
provides insight regarding decisions based on experience, there is really only
one way to gain the knowledge that comes from experience and that is to get
involved in modeling studies and in the selection  of the appropriate models for
those studies.
     Several questions need to be  answered when selecting an appropriate ground-
water flow model.  Such questions  deal primarily with the flow media, but some
questions address such aspects as  steady-state or transient flow and whether the
fluid is single or multi-phase.  The following paragraphs list the questions
                                      5-4

-------
needing to be answered when selecting an appropriate flow model with a brief
explanation of each.

1.   Are you simulating a water table (i.e., unconfined) or a confined aquifer,
     or a combination of both (i.e., conditions change spatially)?
                                                  t
     The analyst needs to select a model which can simulate the type of aquifer
     conditions which exist at the site, either water table,  confined, or a
     spatial combination of both.  Most ground-water flow models simulate
     confined aquifer conditions, whereas not all  models can  simulate water
     table conditions or a combination of both.  The reason for this is that the
     solution for the water table case is more complicated and requires more
     time to solve.   As a result, it is not uncommon that a confined model  will
     be used to solve a problem with water table or mixed water table and
     confined conditions.

     The problem associated with using a confined  model  for unconfined aquifer
     conditions is that the aquifer thickness is not allowed  to vary.   As a
     result, the transmissivity (defined as the aquifer  thickness  times the
     hydraulic conductivity) remains constant when it should  be adjusting with
     the fluctuations in the water table.   The error associated with using  a
     confined model  for unconfined aquifer conditions is small  as  long as the
     fluctuations in the water table are small  compared  to the  total  thickness
     of the aquifer.   Small  fluctuations are usually defined  as where  the water
     table elevation changes by less than  10% of the total  aquifer thickness.
     This criterion  is valid under most  circumstances.   Typically, the criterion
                                     5-5

-------
     is violated when there are large stresses on the system such  as  pumping  or
     recharge,  or when the aquifer is very thin (in general,  less  than  25-ft
     thick).

2.   Is the ground-water flow through porous media, fractures,  or  a combination
     of both?
     The analyst needs to select a model which can simulate the types of media
     which exist at the site, either porous media, fractured media, or a
     combination of both.  Most ground-water flow models simulate  porous media.
     Although few models are designed to simulate fracture flow, most numerical
     models can handle soils with different porosity (i.e., dual porosity).  As
     a result,  to date most fracture flow systems are approximated by porous
     media models.  The validity of this assumption is dependent on two factors:
     1) how highly fractured are the media, and 2) what is the scale (size) of
     the fracture system.  In this approximation, it is generally assumed that a
     highly fractured system can be thought of as a rock that is so highly
     fractured that it resembles a continuum porous medium.  A quantitative
     definition of a highly fractured system does not exist.  However, for this
     approximation, highly fractured could possibly be considered as fractured
     rock with an effective porosity resembling the effective porosity of a
     porous media which  is typically on the order of 10 to 25%.

     The ability to simulate a fractured system with a porous media model is
     also scale dependent.  The larger  the scale of a fractured system, the more
     the flow pattern through fractures represents flow in porous media.  As the
                                       5-6

-------
     scale gets smaller and smaller, It eventually gets down to simulating the
     flow in a single fracture which cannot be simulated with a porous media
     model.                                          •

     In some instances, a finite-element model can be used to simulate flow in
     faults, where a fault is defined as a large fracture.  The fault can have a
     higher or lower permeability than the surrounding rock depending on the
     nature of the rubble in the fault zone.  In this case, the fault is
     simulated as a series of long, narrow elements in the model and the
     hydrologic properties (hydraulic conductivity, porosity, etc.) of those
     elements are set accordingly.

3.   Is it necessary to simulate three-dimensional flow or can the
     dimensionality be reduced without losing a significant amount of accuracy?
     The analyst needs to select a model which can simulate the dimensionality
     needed to properly represent the site, either one-,  two-, or
     three-dimensional.

     In general,  models should be selected in three dimensions unless it can be
     shown that the degree of media homogeneity and spatial symmetry of the
     aquifer are such that they justify the selection of a lower dimensional
     model.   If,  for example,  the problem consists of a single aquifer with
     uniform hydraulic properties and concentrations in the vertical  (z)
     dimension,  then a two-dimensional  x-y simulation is  justified.   If the
     hydraulic properties and  concentrations are also uniform in one of the two
                                      5-7

-------
horizontal dimensions  (x or y in addition to z), then one is justified in
making a one-dimensional simulation.

Lower dimensional simulations are not always made because they are
justified based on the homogeneity or spatial symmetry of the data.  In
many instances, a lower dimensional simulation is made because the data are
not available to perform a three- or two-dimensional simulation, or even to
know if the higher dimensional simulation is required.  The use of a
three-dimensional model without adequately detailed site characterization
data can easily lead to incorrect results.

Even though a lower dimensional simulation is performed where a higher
level model was warranted, the results can still be quite useful in
assessing a particular problem.  The lower dimensional model can be used as
a screening-level model to make numerous, low-cost runs which, in many
cases, can provide valuable insight into a problem.

It is important to remember that performing a complete three-dimensional
analysis does not eliminate uncertainty in the results.  It is virtually
impossible to completely characterize a ground-water flow/contaminant
transport system, and no matter how well the system is characterized, the
model, whether one-, two-, or three-dimensional, is always a simplified
representation of the real system.  Three-dimensional model results, as
well as one- and two-dimensional model results, need to be analyzed very
carefully to determine if they make sense.
                                 5-8

-------
     The dimensions of the ground-water flow model  can be dictated by the needs
     of the contaminant transport model.  For example, if the flow is occurring
     in a single homogeneous aquifer of constant thickness,  a two-dimensional
     x-y model  should be selected.  However, if the contaminant plume is not
     uniformly distributed over the thickness of the aquifer or if a remedial
     action requires pumping from a screened interval  less than the total
     aquifer thickness, then a three-dimensional transport model is required.
     If a three-dimensional transport model  is required,  then the flow should
     be simulated in three dimensions.  It is possible to simulate the ground-
     water flow with a one-dimensional model and then simulate the contaminant
     transport with a three-dimensional model, but in this case the transport
     simulation will not truly be a three-dimensional  simulation since the flow
     vector is in only one direction.

4.   Are you simulating a single-phase (i.e., water) or a multi-phase (i..e.,
     water and an insoluble contaminant) flow system?
     The analyst needs to select a model which can simulate the type of flow
     system which exists at the site, either single-phase or a multi-phase
     system.

     Very few mathematical models are available today which can be practically
     applied to a multi-phase ground-water flow problem.   Much of the original
     research in this area was performed in the petroleum industry for the
     analysis of three-phase flow of oil, gas, and water.  The petroleum
     industry work has been adopted in the ground-water industry to study
                                      5-9

-------
     multi-phase flow between water and contaminants,  but the  ground-water
     efforts are in the infant stages.

     Host contaminant transport analyses assume the contaminant is  soluble in
     water.  This single-phase approach is accurate enough for almost all
     practical purposes.  For the case  where the contaminant is immiscible in
     water, the single-phase approach does not apply.   If a single-phase model
     with proper boundary conditions such as an area source of continuous
     release is applied to a multi-phase problem, it can provide some insight
     into the extent of plume migration for a relatively large source of the
     immiscible contaminant phase which would float on the aquifer's surface or
     sink to the bottom of the aquifer.  Significant inaccuracies are likely to
     arise if the contaminant travels with ground water along  the flow line.
5.   Can the system be simulated with uniform (homogeneous)  or spatially
     variable (heterogeneous) values of hydraulic conductivity,  porosity,
     recharge, and/or specific storage?
     Homogeneous in this sense refers to spatially uniform values in the x
     direction or the x-y plane.  Some mathematical models,  particularly
     analytical solutions or simple numerical codes, only simulate a single
     value of a hydraulic property spatially.  For example,  the hydraulic
     conductivity or recharge may have to be uniform over the entire model
     region.  In a heterogeneous model, on the other hand, many of the
     parameters (particularly the hydraulic conductivity and the recharge) can
     be specified on a node-by-node (finite difference) or an element-by-element
     (finite element) basis.  Thus, it is possible to represent a spatial
     distribution of certain parameters over the study area.
                                      5-10

-------
Quantifying the errors associated with using a single average value versus
using the actual distribution is difficult and very site specific.  In a
general sense, many of the criteria discussed above apply here.  With
regard to hydraulic conductivity, if all the spatial values are within a
factor of 10, using a single average value is probably justified.  If the
spatial values of recharge, porosity, and storage do not vary by more than
about 25%, using a single average value is probably justified.  In most
modeling studies, the recharge, porosity, and storage are not well enough
known to define a spatial distribution.  For site-specific, detailed
studies, if spatially-variable values are required but are not available,
additional data should be obtained.  However, model results using
spatially-averaged values can still be valid and quite helpful for
comparing alternatives, for conducting screening-level studies, for
identifying data deficiencies, and/or for gaining some general insight into
the nature of a problem.

As stated above, the error associated with using an average value versus a
spatial distribution is very site specific.  For example, the average
ground-water velocity over a model region could be the same if the region
is simulated ,with one average value of hydraulic conductivity or if a
spatial distribution is used with a range of high and low values.  In
another similar case, the low values of hydraulic conductivity could be
distributed such that they control the flow, and the velocities for the two
cases could be very different.
                                 5-11

-------
6.   How many hydrogeologic layers are to be simulated?
     The analyst needs to select a model which can simulate either single or
     multiple hydrogeologic layers/aquifers depending on the conditions which
     exist at the site.

     Some mathematical models, particularly analytical solutions, can only
     simulate a single aquifer.  The single-aquifer approach is valid for cases
                                           .      ' '   " - -  ,=  ' ' » J '  • L •'  . '  ' '  • S ,
     where the major portion of the ground-water flow and contaminant transport
                ,                         :        '       *" '   !" "''  5 '    .>>' '• '•  ' '' "*
     occurs in a single layer of a multi-layered system, or where the hydraulic
     properties of all the layers of a multi-layered  system are very similar.
     If the hydraulic properties of the various layers are significantly
     different and contaminants are being transported in more than one layer,
     then a multi-layered model should definitely be  applied.
               '.       '•         -          . . , - i  '       "'  '.-'',•-•' ^'- "i T '  '..' '• '' ' •!
     Significant differences between geologic layers  are difficult to define.
     For hydraulic conductivities, a significant difference might be anything
     greater than a factor of 10.  Hydraulic gradients and porosity  should at a
     minimum be within a factor of 25%, and flow directions should be
     essentially the same (both horizontally and vertically).   For contaminant
     transport considerations, the layers should have similar compositions so
     their sorption properties can accurately be simulated with  one  value.
                                      5-12

-------
7.   Do the hydrogeologlc layers vary in thickness spatially?
     The analyst needs to select a model which can simulate constant or variable
     thickness hydrogeologic layers depending on the conditions which exist at
     the site.

     Some mathematical models,  particularly analytical  solutions,  can only
     simulate constant or uniform thickness within a layer or layers.  If the
     thickness of a layer(s) does not change significantly spatially, the
     uniform assumption is valid.  If the thickness changes by more than about
     10% of the average thickness, a model  should be used that can simulate
     spatially-variable thickness.

     The principle problem associated with not simulating variable thickness is
     that the transmissivity distribution in the model  will be incorrect
     (transmissivity = hydraulic conductivity x aquifer thickness).  Using the
     wrong transmissivity distribution could have a significant impact on the
     model results depending on the specifics of the problem being simulated.
     One of the few ways to determine the magnitude of  the impact  is to run a
     sensitivity analysis with  site-specific data.  In  general, changing
     thicknesses are analogous  to changing diameters in pipe flow.    Velocities
     will increase or decrease  accordingly, thus affecting travel-time
     predictions.
                                     5-13

-------
8.   Is the hydrologlc system in a steady-state condition or do water levels
     fluctuate with time (transient condition)?
     The analyst needs to select a model which can simulate the type of
     hydrologic condition which exists at the site, either steady state or
     transient.

     Steady-state flow is defined as the condition when the magnitude and
     direction of the flow velocity (as defined by the hydraulic conductivity
     and hydraulic gradient) are constant with time at any point in the flow
     field.  Transient flow (unsteady flow) is when the magnitude or direction
     of the flow velocity changes with time at any point in a flow field (Freeze
     and Cherry, 1979, p.49).  The application of a steady-state model or
     approach .is technically valid only in the somewhat unrealistic case where
     the water table maintains the same position for some extended period of
     time (throughout the entire simulation period).  In most actual cases,
     variations in recharge and discharge introduce transient effects on the
     flow system.  Therefore, technically speaking, a transient model or
     approach should always be applied.  However, when applying models, the
     general rule that is followed is that if the fluctuations in the water
     table are small in comparison with the total vertical thickness of the
     aquifer (or hydrologic flow system), and if the relative configuration of
     the water table remains the same throughout the cycle of the fluctuations
     (i.e., the high and low points remain highest and lowest, respectively),
     then the transient system can be simulated as a steady-state system with
     the water table fixed at its mean position.  As was the case for confined
                                      5-14

-------
versus unconflned flow, small fluctuations 1n the water table are generally
defined as less than about 10% of the total thickness of the aquifer (or
flow system).

Transient simulations require much more data and are more difficult to
implement than steady-state simulations.  Typically, the additional data
(time series of water-level measurements, recharge, discharge, etc.) are
not available and one  is forced to make a steady-state approximation.  In
many  instances the steady-state approach can yield valuable  insight and
information provided that the assumptions used are conservative  and that
the limitations of the approach are  fully understood.

An approach  that can be used to approximate  a fully  transient  simulation
but that requires less data  and is easier  to implement  is  a  series of
steady-state solutions.  In  this  approach  a  new  potential  or head
distribution is  solved for whenever  a significant stress  is  imposed  on  the
system.  A significant stress can be a pumping well(s)  being turned  on  for
a pump-and-treat remedial  action, or a cyclic recharge  event where it  is
known,  for example,  that the majority of the recharge occurs in  the  winter
months.

The errors associated with using  a steady-state  approximation  of a
 transient system are difficult  to quantify because they depend on each
 specific case.  For example, if a water table exhibits very consistent and
 small fluctuations over a yearly cycle due to a consistent pumping and
 recharge pattern (i.e., pumping occurs between April and September for
                                  5-15

-------
      irrigation  and  natural  recharge  1s large  1n the winter and small  in the
      summer),  a  steady-state approximation may be valid.   If on the other hand,
      the pumping schedule  is sporadic and it significantly alters hydraulic
      gradients and directions of flow, and the recharge is variable from season
      to season and year to year, a transient simulation should be performed.

      After these questions (criteria) have been addressed, in most cases the
analyst will find that there are several ground-water flow models which meet the
desired criteria.  At this point the  analyst can either select a ground-water
flow  model and then  continue the selection process to choose a compatible (but
separate) contaminant transport model, or the  user can continue the process to
select a combined flow and transport  model.  It is quite common to find a fairly
sophisticated  flow model linked with  a simpler transport model since the
transport model  parameters are generally less well known.
      The decisions to be made when selecting a contaminant transport model  are
discussed in more detail below.  Some guidance is provided to help in making the
decision and some discussion is provided regarding the errors associated with
incorrectly using the model  or feature(s) of the model.
     The errors  associated with incorrectly selecting a transport model or
feature of a model are difficult to quantify.  Some guidance is given below, but
in many cases these errors are site specific and,  therefore,  cannot be
quantified in general terms.   Following are the questions to be answered when
selecting a contaminant transport model.
                                     5-16

-------
1.    Does the contaminant enter the ground-water flow system at a point or is it
     distributed along a line or over an area or a volume?
     The analyst needs to select a model which can simulate the type(s) of
     source(s) that exists at the site, either point, line, or area source.

     A point source is characterized by contaminants entering the ground water
     at a single location such as a pipe outflow or injection well.  A line
     source is characterized by contaminants entering the ground water along a
     line as in the case of leachate emanating from the bottom of a trench.  An
     area! source or nonpoint source is characterized by contaminants entering
     the ground water over an area as  in the case of leachate emanating from a
     waste lagoon or an agricultural field.  A volume source releases
     contaminants in a form of volume  in ground water.

     A point  source can be simulated with either  a one-, two-, or
     three-dimensional mathematical model whereas a  line or  area!  source  should
     be  simulated with a  higher  dimensional  model.   Most two-  or
     three-dimensional models  can  simulate  point, line, and  area!  sources.
     Three-dimensional models  are  appropriate  for simulating a volume  source.
     All  one-dime;nsional  models,  particularly  analytical  solutions,  simulate
     point  sources.   Line or area sources can  be  simulated with  one-dimensional
     models by assuming  that the contaminant concentration is uniform  except in
     the dimension  simulated.   Because the  analyst  typically lacks the necessary
     data to perform conceptually correct  transport simulations, it is quite
     common to use a one-dimensional  transport model with averaged values of
                                      5-17

-------
      concentration and velocity to simulate  line  or  areal  sources.   This
      approach  is  discussed in  more detail  below in the  section  on
      dimensionality.

      Typically, when  using a one-dimensional transport model to represent a line
      or areal  source,  the  plume width  is underestimated and its concentration is
      overestimated.   This  effect diminishes  as the contaminant migrates farther
      and farther  from the  source.  Also, typically when performing
      one-dimensional  transport  simulations,  the analyst uses the peak measured
      concentration in  the  interest of being  conservative.  The analyst must keep
      in mind these conservative aspects of the approach when analyzing the
      results.

2.   Does the source consist of an initial  slug of contaminant or is it constant
     over time?

     The analyst needs to select a model which  can simulate the type(s)  of
     source release that occurs at the site,  either pulse,  constant or a
     combination of both.

     Contaminants  can  enter the ground water  either as  an  instantaneous  pulse  or
     as a continuous release over time.   A  continuous release may be either
     constant or variable.   Variable releases may  be  due to source  decay,
     variable precipitation on  the  source,  or intermittent  source application
     (e.g.,  dumping).   Many of  the  simpler  mathematical models can  only  simulate
     slug or constant  input releases.   Some of the one-dimensional models and
                                     5-18

-------
    most  of  the  higher  dimensional  models  can  simulate  all  the  various means  by
    which contaminants  can  enter the  ground water.

     In  many  transport modeling studies,  a  one-dimensional  model  which cannot
     properly represent  the  distribution  of the source in  time will  be used at
     the screening-level stage to make a  conservative prediction of  the
     concentration.   The errors associated  with simplifying the  contaminant
     input pulse  are site specific and, therefore,  difficult to  quantify.   In
     general, use of a continuous input model  will  over-predict  concentrations
     if the peak  value is set as the input.  If an  average value is  set,  the
     model will under-predict discrete peaks.   The  effects of using  an  average
     flux will diminish with distance from the source.

3.   Is it necessary to simulate three-dimensional  transport or can  the
     dimensionality be reduced without losing a significant amount of accuracy?

     The analyst  needs to select a model  which can simulate the dimensionality
     needed to properly represent the site, either one-, two-, or
     three-dimensional.

     All transport models should be selected in three dimensions unless it can
     be shown that the contaminant migration or remediation scheme can be
     accurately represented  in a lower dimension.  Cases where a contaminant
     source  is distributed both areally  (x-y) and vertically  (z) would, in most
     instances, require a fully three-dimensional model.  Similarly, remedial
     action  simulations employing pumping wells screened at specified depths
     and/or  cutoff walls not fully  penetrating an aquifer would most likely
                                      5-19

-------
 require a three-dimensional model.  Cases  in which the aquifer is uniformly
 contaminated  in  the  vertical direction and where geologic layering is not
 important can be simulated  in two dimensions, generally,either in x-y or
 x-z.  Very simple cases  consisting of a large plane source, a simple
 contaminant distribution, and a uniform velocity distribution can be
 simulated with a one-dimensional model.

 Because  the analyst  typically lacks the necessary data to develop a two- or
 three-dimensional  transport model, it is quite common to simulate
 transport in  one dimension,.  Only in recent years have documented and
 verified three-dimensional models become available,and responsible parties
 have been willing  to spend the time and money required to develop and run
 site-specific models.  The lack of necessary data is no excuse for running
 a lower  dimensional model if it is not suited for the specific application,
 although  this is the approach which apparently is often taken.  A
 one-dimensional model, however, can still  provide valuable results for a
 screening  study where conservative predictions or alternatives are
 compared.                .

The limitations of the one-dimensional  approach are:

•    it cannot simulate multiple sources;
§    it can only simulate a large plane source or an  average concentration
     distribution over a large x-y,  x-z,  or y-z plane for an areal  source;
     and
                                5-20

-------
o    it cannot simulate transverse dispersion (perpendicular to the flow
     path).

Although errors associated with these limitations are difficult to
quantify, their general effect is the prediction of very conservative
(higher) concentrations.  A conservative prediction can be useful for
screening studies, but often the results are so conservative that they have
little use.

The  dimensionality of the transport model  should be equal to or  less than
the  dimensionality of the groundrwater  flow  model.  Although it  is  possible
to simulate the  ground-water  flow with  a one-dimensional model  and  then
 simulate contaminant transport with  a three-dimensional model,  the
 transport simulation would  not truly be a  three-dimensional  simulation
 since the flow vector is in only one direction.

 It is important to remember that performing a complete three-dimensional
 analysis does not by any means eliminate uncertainty in the results.   It is
 virtually impossible to completely characterize a ground-water
 flow/contaminant transport system, and no matter how well  the system is
 characterized, the model, whether one-, two-, or three-dimensional, is
 always  a simplified  representation of the real system.  Three-dimensional
 model results, as well as one- and two-dimensional model results,  need  to
 be  analyzed  very carefully to make sure they are reasonable.
                                  5-21

-------
4.   Does the model simulate dispersion?

     Dispersion is the spreading of a solute in porous medium caused by
     mechanical mixing of water through the media.  Dispersion cannot be
     expressed in terms of the mean ground-water velocity alone.   Local
     variations in flow velocity direction and magnitude caused by the natural
     heterogeneity in media properties causes dispersion.   The spreading process
     is usually associated with unknown and uncharacterized geologic variation
     of hydraulic properties.   The dominant property causing the  spreading is
     usually the hydraulic conductivity variation in space(Simmons and Cole,
     1985).

     Host  all  contaminant transport models simulate  some form of  dispersion.
     There is  much concern,  however,  as to whether dispersion can be adequately
     described in  a model  because  it  is related  to spatial  scale  and variations
     in  hydraulic  properties which  are  difficult to  simulate in a model.   There
     seems to  be a clear  consensus  among ground-water  transport researchers that
     the conventional  convective-dispersive equation may be  an inadequate  and
     inappropriate description  of field-scale dispersion (Pickens  and  Grisak
     1981a and  b;  Jury 1982; Smith  and  Schwartz  1980;  Gelhar  et al.  1979;
     Matheron  and  DeMarsily 1980; Dagan  and Bresler 1979; Simmons  1982a and b).
     Some recognized inadequacies of the conventional  approach are prediction of
     non-physical  upstream migration, failure to predict increased dispersion
     caused by larger  scales of heterogeneity, inappropriate dependence on
    diffusion-like boundary conditions, and incorrect representation of
    non-Fickian, asymmetric solute distributions (Simmons and Cole, 1985).  For
                                    5-22

-------
5.
the purposes of this report, it will be assumed that the currently
available models are adequate.

It is difficult to quantify the errors associated with simulating
dispersion because they are site specific.  Turning off dispersion in a
model only considers the movement of contaminants by the advection process
and hence yields a "worst case" prediction of peak contaminant
concentration.  However, this causes the leading edge of the plume to take
longer to arrive at a point downgradient.  The larger the value of
dispersion, the greater the spread of the plume and the lower the peak
concentration.  Since dispersion coefficients are very difficult to measure
in the field, the best way to choose a value is in the model calibration
process, often requires a large amount of data for field calibration.

Does the model simulate adsorption (i.e., distribution or partitioning
coefficient) and, if so, does it simulate temporally and/or spatially
variable adsorption?  Temporally or spatially variable adsorption is
important when the soil conditions and/or concentrations change with time
and space.

There are a number of chemical  and biological processes that affect the
rate and manner in which contaminants travel  in the ground water
(adsorption, ion exchange,  degradation,, biotransformation, etc.).  Since
adsorption and degradation  (decay) are the only processes that are
typically simulated in contaminant transport models,  they are the only
processes discussed in this report.   Adsorption is discussed in this
section and degradation in  the next  section.
                                     5-23

-------
Adsorption  is  usually represented  by  a distribution or partitioning
coefficient which  relates  contaminant concentration in solution to
contaminant concentration  associated  with  (adsorbed to) the soil.
Distribution coefficients  are  used in transport models to approximate
contaminant retardation  in the subsurface.  Contaminant retardation relates
ground-water velocity to contaminant  velocity.

In screening studies,  transport models not simulating adsorption are often
used to make conservative  estimates of contaminant concentrations.  These
conservative estimates can be  useful  as long as the predicted
concentrations are not so  high  that they completely misrepresent the
problem.  A general rule for determining when to simulate adsorption might
be that if  the contaminant is  only mildly retarded, say less than a factor
of 5 times  slower than the ground  water, then the conservative approach of
not simulating adsorption  may  be informative in a screening effort.  If the
contaminant  has a retardation  factor  larger than 5, a model should
be selected  that simulates retardation.  In any detailed modeling study
adsorption  should always be simulated for all retarded contaminants. The
partitioning coefficients  for all  contaminants are dependent on the
equilibrium  solution concentration.  However, in dilute solutions,  the
partitioning coefficient is constant for all  practical purposes (linear
isotherm).  Stated in a different way, a contaminant's absorption
properties remain the same at low concentrations (linear),  whereas at
higher concentrations adsorption decreases as the contaminant concentration
in the ground water increases (nonlinear).  This nonlinear property can
become very important when predicting the migration of the center of mass
                                5-24

-------
versus the leading edge of a contaminant plume, or when predicting required
pumping time for a pump-and-treat remedial action.  At high concentrations,
the linear assumption will over-predict adsorption and travel times and
under-predict ground-water contaminant concentrations.  For hydrophobic
contaminants, the threshold below which linearity can be assumed is about
one half the solubility (Neely and Blau,  1985).  For hydrophilic
substances, the threshold must be determined on a case-by-case basis
because of interactions between the water and the contaminant.

Very  few currently available transport models can simulate the nonlinear
adsorption case,  so  adsorption is nearly  always simulated as  a linear
isotherm.  For very  low concentrations, this approximation can produce mass
balance errors which typically overestimate the contaminant mass  in the
solution phase and underestimate the mass in the  solid  phase.  Therefore,
the  approximation with the  linear isotherm  is  more  conservative
 (i.e.,  it would  predict a higher concentration  at a monitoring well at  an
earlier time than it should actually arrive).

Certainly,  it would  be more accurate to use a  nonlinear isotherm model  for
 all  cases  in which  it applies.   The currently  accepted practice  in
 transport  modeling,  however,  is  to  lump all chemical  and biological
 processes  into  one term,  the adsorption or  retardation term,  and simulate
 it as a linear  isotherm.   Geochemical  models  are  employing a more
 state-of-the-art approach with regard  to simulating chemical  and biological
 processes.  Until the science is better understood, however, the accepted
 practice will be to lump them all  into one  term.
                                  5-25

-------
6.   Does the model simulate first or second-order decay and/or radionuclide
     decay?

     Degradation or radlonucllde decay are factors which can significantly
     affect predicted concentrations and total  mass of contaminants  In  a ground-
     water system.   As with adsorption,  conservative estimates  of concentrations
     can be made by neglecting degradation 1n a screening level  or "worst-case"
     analysis.   For a detailed analysis,  however,  a degradation  model should be
     used.

     Host transport models  simulate  first-order decay.   Only  a few models  can
     simulate higher-order  decay.  No  transport models  are currently available
     which  can simulate the transformation of one chemical constituent  into  a
     breakdown product and  then simulate the transport  of the transformed
     constituent(s).  One approach that can be  used to  simulate the transport of
     a parent species and all  its transformation products is to simulate the
    transport of a parent  as  if there is no degradation and then
    apportion the resulting concentration on the basis of an assumed
    transformation efficiency (often 10%) and the ratio of molecular weight.
    This only works well  if the adsorption coefficients are similar  between the
    parent and by-product contaminants.   It also assumes that the by-product
                                                               i'.
    does not further transform.

    Several transport models,  both analytical and numerical,  can simulate
    radioactive  decay.  Many of the  models can  account for the  generation and
    transport of daughter products in  both straight and branched decay  chains.
                                    5-26

-------
7.   Does the model simulate density effects due to changes in temperature and
     concentration?  A truly coupled model  is one in which the ground-water flow
     is influenced by the density and viscosity of the water,  which,  in turn,
     are influenced by the temperature of the water and the concentration of the
     solute.  In some cases (i.e., involving a large heat source or large
     fluctuations in solute concentration)  it may be important to consider
     temperature and contaminant concentration effects on ground-water flow.

     A basic assumption of most ground-water flow and contaminant transport
     models is that gradients of fluid density, viscosity, temperature, and
     concentration do not affect the velocity distribution.  In most  cases these
     gradients are small and this is a safe assumption.  In some cases, however,
     these gradients can become large and can significantly affect flow
     patterns.  In these cases a coupled model solving for pressure,
     temperature, and/or solute concentration as functions of  fluid density and
     viscosity should be used.

     An example of such a case is the effect of heat generated by the burial of
     high-level radioactive waste on the ground-water flow in  the vicinity of
     the repository.  The heat from the repository will have a buoyant effect on
     the surrounding water causing it to rise.  The extent of  the rise and the
     magnitude of the impact can be predicted with coupled models.

     Under most naturally-occurring situations, the assumption that flow is not
     affected by temperature,  density,  and  concentration gradients is valid.  In
     cases where extremely deep systems are being simulated, however, the
     naturally-occurring geothermal  gradients may influence ground-water flow
     patterns and should therefore be simulated.   The only other cases where one
                                     5-27

-------
     typically  needs  to  be  concerned with thermal flow patterns are
     anthropogenic phenomenon  such as geologic repositories or injection of
     heated or  cooled fluids.

     Similarly, under most  naturally occurring circumstances, density and
     concentration gradients are small enough they can be neglected.  The
     classic example  of  a naturally-occurring density problem that should be
     simulated with a coupled  model is that of saltwater intrusion.  Most
     man-made contamination problems result in low enough ground-water
     concentrations that they  do not affect the ground-water flow.  In a few
     instances, concentrations become large enough that the density of the water
     is affected and  a coupled model is required for an accurate prediction of
     the flow and transport.  An important example is leachate from landfills.
     Concentrated leachate plumes often move deep into the aquifer before they
     migrate laterally.  If the effect of concentration is neglected, the
     analysis may under-predict concentration at deeply screened wells.

     After sequencing  through the decision tree, there will, in most cases, be
several models which meet the desired criteria.  Since several models could meet
the desired criteria,  it is difficult to list a single model as a standard
model.   At this point  the analyst can either select a transport model which is
compatible with the flow model selected above, or select a combined ground-water
flow/contaminant transport model.
     Regardless of the approach selected,  whether separate or combined flow and
transport models, it is likely that there will be several  models which meet the
technical criteria.  The selection of the final model(s)  should be based on the
                                      5-28

-------
implementation criteria; i.e., the model meets the criteria that it has been
through a rigorous quality assurance program so that it is thoroughly verified,
and the model is well documented with user's manuals and test cases.
     If several models pass the quality assurance and documentation criteria,
the final selection of a model should be based on familiarity with and
availability of the model, the modeling schedule, available budget, and staff
and equipment resources.
5.2.  MODEL SELECTION FROM THE REVIEWERS POINT OF VIEW
     The selection criteria are described primarily from the point of view of an
analyst performing an exposure assessment study.  The basic process begins with
defining project objectives, assessing a physical situation, and then selecting
a model to represent the important processes relevant to the objectives and
physical conditions.  Another use of the selection criteria is from the point of
view of a regulatory agency that is reviewing an exposure assessment study.
Under these circumstances the reviewer needs to evaluate the choice of the model
used in performing the study.  The selection criteria must be fundamentally the
same for both applications, however, some differences exist in how the criteria
is used or evaluated.  This section is included to describe those differences.
The proper selection of a model is one step in performing an exposure assessment
study.   All models are sensitive to the choice of input parameters.  The choice
of these parameters will be, in most cases, at least as important as the choice
of a particular model and must be considered in the review process.
     The first step for the person reviewing the exposure assessment study is to
identify the characteristics and capabilities of the specific model.  This
information can best be obtained from a model user's manual.
                                     5-29

-------
     The  single  most  important  step  in choosing a model is to define project
objectives.   Similarly,  the most  important step for the reviewer is to have a
clear  understanding of the study  objectives.  The objectives of the analysis
should clearly state  whether  it is performed as a generic analysis or if a more
detailed  site-specific analysis with appropriate data for calibration and
validation is intended.   In cases where the objectives are not clearly defined,
then the  only option  is  to interpret how the results and conclusions of the
study  are presented (e.g., associated uncertainty, potential impacts, and
importance of decisions  based on  the results).  This is difficult and will
obviously be subjective  in some cases.  After this initial step, the rest of the
selection criteria are essentially the same for both the selection of a model
and review of a  model  selection by another person.
     It is important  that the reviewer examine the presentation of results and
conclusions to see that  they are consistent with the study objectives, model
choice, and model application.  Any decisions based on model results must
incorporate the  uncertainty inherent in the predictions.  For a generic
analysis, the level of uncertainty will be about an order of magnitude at best.
For more  detailed analysis, the validation phase of a modeling study may provide
some guidance in defining the uncertainty associated with the model predictions.
5.3.  MODEL SELECTION  WORKSHEET
     Section 5.1 discussed the  technical criteria used to select a mathematical
model for a specific  application.  This section provides a model selection
worksheet which  facilitates the selection of an actual  model or suite of models
based on  the response  to the technical criteria.
                                      5-30

-------
     The model selection worksheets are shown  in Tables 5-1, 5-2, and  5-3.
Tables 5-1 and 5-2 are the analytical solutions worksheets and Table 5-3  is the
analytical and numerical model  (coded for the  computer) worksheets.
     Table 5-1 lists a total of 63 analytical  solutions by name and reference,
tells which calculator or personnel computer they have been programmed for (if
any), and lists any pertinent comments.  The first seven solutions are for
ground-water mounding problems  and the remainder are for contaminant transport
problems.
     Table 5-2 lists the analytical solutions  (by number) in the same  order they
are listed in Table 5-1.  The technical selection criteria (listed across the
top) are in the same order as they are discussed in Section 5.1 (Figure 5-1).
     Table 5-3 presents a list  of some of the  currently available, documented,
mathematical models.  The models are divided into seven categories:
1.   Analytical flow models
2.   Analytical transport models
3.   Numerical flow models which can be applied to both saturated and
     unsaturated systems
4.   Numerical flow models which can only be applied to saturated systems
5.   Numerical contaminant transport models which can be applied to both
     saturated and unsaturated  systems
6.   Numerical contaminant transport models which can only be applied  to
     saturated systems
7.   Numerical contaminant and  transport models which couple the solutions for
     pressure, temperature, and concentration  (coupled models).
Within these major categories,  the models are listed in alphabetical  order.  The
technical criteria or specifications for the models (listed across the top) are
                                      5-31

-------
      TABLE  5-1.     ANALYTICAL  SOLUTIONS,   NAMES,   AND  REFERENCES

1>
H
11
4)
1)
()
71
1)
»
10)
11)
11)
11)
14)
15)
10
17)
U)
H)
20)
ID
»)
23)
24)
Analytical Solution names
Analysts »f Creun*at>r Mounting leneatli
Tallin* Pen*
Circular Recharge Area
Ctralir lulu Htcaarje Mound
Hwnd Decay
Circular Recharge Are*
HanxHng
River's Solution
junction ind Dispersion — Regional Flew
MAP Flue
H.UC
rtosve
H.0t CROSS-SECTION
RAIDCM UAU
RAW* WALK
U03M9
Htn
Advectlon and Dispersion fro* a Stream
Advectton and Dispersion from a Single
ruling Veil
Advec$1ve Mass Transport Thels Particle Hover
Streamlines and Travel Times for Regional
Flew Affect* by Sources and Sinks
Advectlve Transport Model
S-PaUu
Ground Vater Dispersion
Plume Management Model
Computer Type
TI-SJ
TI-S9
HP-41

TRS-80, Osborne,
Kaypro. IBM
Apple. Victor,
Kaypro II. Vector
Apple
(3)
Apple. Victor,
Kaypro II, Vector
Osborne, Super-
brain. Kaypro, IBH
Osborne
Apple, 'Victor,
Kaypro II, Vector
(6)
(7)
Osborne
Osborne
TltS-80. (3)
TRS-80, (3)
Tl-59
HP-41
HP-41
HP-41
TI-SB/59
TI-59
Comment!
Predicts Hydraulic Head
Predicts Hydraulic Head
Predicts Hydraulic Head
Predicts Hydraulic Head
Preclcts Hydraulic Head
Predicts Hydraulic Head
Predicts Hydraulic Head and
Discharge
Predicts Contaminant Concentration
Predicts Contaminant Concentration
Predicts Contaminant Concentration
Predicts Contaminant Concentration
Predicts Contaminant Concentration
Predicts Contaminant Concentration
Predicts Contaminant Concentration
Predicts Contaminant Concentration
Predicts Contaminant Concentration
Predicts Contaminant Concentration
Predicts Contaminant Concentration
Predicts Contaminant Concentration
Particle Location Nlth Time
Particle Location ulth Time
Predicts Mass Loading
Predicts Contaminant Concentration
Predicts Contaminant Concentration
Deference
Sandberg et •!., INI
Prlckttt an* Vorhees,
Walton, 1983
Ulrlck (1)
Bear, 1972
Wilton. 1983
(2)
Holden, Sunada, and
Varner, 1984
Ballon, 1983
(2)
(2) (4)
(5)
(2)
(2)
(3)
(5)
(5)
Walton. 1983
Walton, 1983
Sandberg et al . , 1981
Prtckett and Vorhees,
Olsthoorn, 1984 (4)
Ulrlck, 1983 (1)
Oberlander and Nelson
Kelly. 1982
Sandberg et al., 1981
Prlckett and Vorhees,

INI

















.1981


, 1984

1981
 25)  Calculator Code for Evaluation Landfill        TI-59
     Leaehate Plumes  *

 tfl)  Dissipation of a Concentrated Slug of          TI-59
     Contaminant

 27)  Advectlon and Dispersion from a Single Solute   KP-41
     Injection Veil

 a)  COAST — Analytical  Solution for One-
     Otmenslonal Contaminant Transport

 21)  TOAST — Analytical  Solution for Twq-
     Dlmenslonal Contaminant Transport

 X)  LT1KD — Semi-Analytical Solution to Radial
     Dispersion In Porous Media            '

 31)  KSSq — Semi-Analytical Contaminant
     Transport

 32)  RT — Mapping Concentration Distribution In
     an Aquifer Based on  a Time Series Data
     Collection Concept

 33)  Material Release on  the Surface with One-
     Dimensional Vertical Dovnward Transport

 34)  Tho-Dlmenslonal  Horizontal  Flow ulth a Slug
     Source

 3S1  Tn-Otmonslonal  Horizontal  Flo* ulth a
     Continuous Source

 36)  Instantaneous Source, Infinite Aquifer Depth

 37)  Instantaneous Source, Finite Aquifer Depth

 38)  Instantaneous Source. Finite Aquifer Depth —
     Average Concentration

 39)  Continuous Source Release  — Unsteady State

 40)  Continuous Source Release  — Steady  State

41)  Instantaneous Horizontal Release, Finite
     Source Length, Infinite Aquifer Depth

42)  Instantaneous Horizontal Release, Infinite
     Source Length, Infinite Aquifer Depth
 Predicts Contaminant Concentration


 Predicts Contaminant Concentration


 Predicts Contaminant Concentration


 Predicts Contaminant Concentration


 Predicts Contaminant Concentration


 Predicts Contaminant Concentration


 Predicts Contaminant Concentration


 Predicts Contaminant Concentration



 Predicts Contaminant Concentration


 Predicts Contaminant Concentration


 Predicts Contaminant Concentration
.  Pettyjohn et al.,  1982


  T.A. Prlckett and  Associates  (8)


  Van der Hetjde,  1984 (4)


  Oavandel et al., 1984


  Javandel et al., 1984


  Oavandel et al., 19S4


  Javandel et al., 1984


  Javandel et al., 1984



  vanGenuchten and Alves,  1982 (9)


  Wilson and Killer, 1978  (9)


  Wilson and Miller, 1978  (9)
Point Source Solution
Point Source Solution
Point Source Solution
Point Source Solution
Point Source Solution
Line Source Solution
Hwang, '1986
(twang, 1986
Hwang. 1986
Hwang, 1986
Hwang, 1986
Hwang, 1986
Line Source Solution
                                   Hwang, 1986
                                                           5-32

-------
                                      TABLE   5-1.      (continued)
           Analytical Solution
                                                            Ty»e
                                                                                                                   tofertnct
43)
44)
45)
44)
47)
48)
4t)
50)
SI)
52)
S3)
54)
55)
56)
57)
56)
59)
CO)
61)
62)
C3)
Instantaneous Horizontal Release. Finite
Source Length, Finite Aquifer Depth
Instantaneous Horizontal Release, Finite
Source Length, Aquifer OepthrAveraged
Concentration
Continuous Horizontal Release, Finite Source1
Length, Infinite Aquifer Depth
Instantaneous Vertical Release, Infinite
Aquifer Depth
Continuous Vertical Release, Average
Concentration — Unsteady State
Continuous Vertical Release, Average
Concentration — Steady State
Instantaneous Horizontal Source
Instantaneous Vertical Source
Continuous Horizontal Plane Source —
Unsteady State
Continuous Horizontal Plane Source --
Steady State
Continuous Vertical Plane Source —
unsteady State
Horizontal Plane Source at a Constant
Boundary Concentration
Vertical Plane Source at a Constant
Boundary Concentration — Steady State
Constant Release Rate Downward
Constant Concentration Boundary Reservoir
Voluae Source
Constant Boundary Concentration In a
Radially-Flowing Aquifer
One-Dlnenslonal Mass Transport IBM 360/91
Tw-D1aens1onal Mass Transport I6H 360/91
Three-Dlaenslonal Mass Transport: Patch IBM 360/91
Source; Finite Dimensions
Thrte-D1«ens1onal Mass Transport: B1var1ate IBH 360/91
Line Source Solution
Line Source Solution
Line Source Solution
Line Source Solution
Line Source Solution
Line Source Solution
Area! (Plane) Source Solution
.Areal {Plane} Source Solution
Areal (Plane) Source Solution
Areal (Plane) Source Solution
Areal (Plane) Source Solution
Areal (Plane) Source Solution
Areal (Plane) Source Solution
Infinite Horizontal Plane Source
Infinite Vertical Plane Source


Concentration and Flow Boundary
Conditions
Strip Boundary Condition
Finite and Infinite Width
Two-Dlwnslonal Vertical Source

Hwang, 1986
Hwang, 1966
Hwang, 19S6
Hwang, 1986
Hwang, 1986
Hwang, 19B6
Hwang, 1986
Hwang, 1986
Hwang, 1986
Hwang, 1986
Hwang, 1986
Hwang, 1986
Hwang, 1986
Hwang, 1986
Hwang, 1986
Hwang, 1986
Hwang, 1986
Cleary and Ungs, 1978
II
II
II
     Gaussian Source
U)  rrograns available  froe Janes S. Ulrlck and Associates. 2100 Los Angeles Avenue.  Berkeley. CA  94707.

(2)  Programs available  fron the National Center for Ground Water Research, Oklahoma State University,  Stillwater, OK.

(31  Calculator/Computer Type:  Osborne, Kaypro, Superbraln, IBM, Radio Shack PC-1 and'PC-2, and Sharp  PC 1250 and 1500 programs available.

(4)  Prograes available  from the International Ground Hater Modeling Center, Holconb Research Institute, Butler University. '4600 Sunset Avenue
     Ind1anapol1s, IN 46208*

(5)  Programs available  from Dr. Michael L. Voorhees of Harzyn Engineering. Inc.,  Madison. HI.

(6)  Calculator/CoBputer Type:  Apple,  Kaypro II, Victor, Vector, TRS-80, Sharp-PClSOO.

(7)  Calculator/Computer Type:  Superbraln, Osborne. Sharp-PClSOO.

(8)  Programs available  fron Thomas A.  Prlckett and Associates, Inc., 8 Montclalr  Road, Urbana, IL  61801.

(9)  See National  Council of the Paper  Industry for Air and Stream iBprovenents, Inc., Technical Bulletin Ho. 472, October, 1985.
         •- Point Source
         — Line Source
         •- Areal  Source
         •- Radial  Flow
         — x-D1mens1on
         — y-D1aens1on
         — 2-D1 tension
         - Volune
         — Dispersion In the x Direction
         - Dispersion 1n the y Direction
         — Dispersion 1n the z Direction
                                                           5-33

-------
TABLE 5-2.  ANALYTICAL SOLUTIONS WORKSHEET


t.
•^ t.
a. .«>
o- *»-
1 1
« -0
*~ £
Analytical t. •£
Solutions Ho. £ =
(see Table 1) S 3
GrounoVater
1
2
3
4
5
6
7
Con taut nant
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30-
31
32
Mounding
X
X
X
X
X
X
X
Transport
X
X
X
X
X
X X
X X
X
X X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X


10
•5
£
to
1
£

X
X
X
X
X
X
X

X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
x
X
X
X
X
X
X
X
X


[Fracture Flow




































Flow Dinenslon

R
R
R
X-Y
R
R
R

X-Y
X-Y
X-Y
X-Y
X-Z
X-Y
X-Y
X-Y
X-Y
X
R
X-Y
X-Y
X-Y
X-Y
X-Y
X-Y
X-Y
X-Y
R
X
X-Y
R
X-Y
R


o>
IS
£
OI

X
X
X
X
X
X
X

X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
U
i
1
1 Mul « -Phase
Konogeneous Hy
Parameters

X
' X
X
X
X
X
X

X
X
X
X
X
x
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
1
%
Heterogeneous
Parameters
Single-Layer

X
X
X
X
X
X
X

X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
M
M
s
&. 2=
0) P
«? 4-*
i 1

X
X
X
X
X
X
X

X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X

i
Variable Thick
[Steady State
[Transient

X
X
X
X
X
X
X

X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X.
X
X
X

1
01
c
1*8

Clrc
C1rc
Clrc
Rect
Circ
Clrc
Rect

P
P
P
P
P
P
P
P
P
P
P
P
P
P
P
P
P
P
P
P
P
LS
P
P.A
P
2
3!
0)
3
•a
+*
"E

X
X
X
X
X
X
X

X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X


X

0}
Constant Souro





























X
X
X
X
X


Dispersion

X
X
X
X
X
X
X

X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X

X


Adsorption









X
X
X
X
X
X
X
X
X
X
X



X
X
X
X
X

X
X

X
X
!
1
cs?
Degradation 1,
Radioactive









X
X
X
X
X
X
X
X
X





X
X
X
X
X

X
X



                  5-34

-------
TABLE 5-2.   (continued)


Analytical
Solutions No.
(see Table 1)
33
34
35
Transport
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
1
0)
I
I



























X
X

X
X
1
1-
f
V

c
to
X
X
X

X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
f ;.
> 































                                       3  i
                                x   p
                                x   p
                                X   P
                                   p
                                   p
                                   p
                                   p
                                   p
                                   L
                                   L
                                   L
                                   L
                                   L
                                   L
                                   L
                                   L
                                   A
                                   A
                                   A
                                   A
                                   A
                                   A
                                   A
                                X  A
                                X  A
                                X Vol
                                X P,A
                                X  P
                                X  A
                                X  A
                                X Vol
                                          +»  T-  -r-
                               *{/) 4^ O  f-  +J
pe
      g 5*
        s.ts
        •ss
        &>€
        ss
x  x  x   x
   xxx
X  X  X   X
   X  X
   X  X
   X  X
   X  X
   X  X
   X  X
   X  X
   X  X
   X  X
   X  X
   X  X
   X  X
   D7  X
   Dz  X
   Dx  X
   X  X
   X  X
xxx
xxx
xxx
xxx
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X

X
X
X
X
X
X
X
X
          5-35

-------
       TABLE  5-2.   (continued)
LEGEND
     Dimension

       X ~ X dimension
       Y -- Y dimension
       Z — Z dimension

     Dispersion

       Dx -- Dispersion in x direction
       Dy — Dispersion in y direction
       Dz -- Dispersion in z direction

     R -- Radial  Flow

     Source Type

       A — Area!  source
       L — Line  source
       P -- Point source

     Vol  — Volume
                5-36

-------
                                                         tiitxjtija        o  Ly  uj  ui  uj    oooooo        ui  g:
                                                         £  £  u?  £        £  LL  ul  a.  u_    u_  u- u-  u,  U-  u.        u_z
3AH3BCHPBH

1—
UJ
III

3=
CO
a:
o
2-

CO
I
__ I
UJ
Q
s:
_i
i — i
UJ
=>
Q
^
_J
O
P 0


LU
_J
CQ
S31S | x xx
ssau>pmi atqE^SA

SSBlDpmi 3.UB1SUO3 | X XX

aaAEi-jlinw


U8XB1-31.6U1.S | X XX
tnEJP*H snoaiaSo^lH

3H.nBjp^ snoauaBouiOH | ** x x
BSEqd-Hinwl
aSBMd-ai6ui.s | x x x





CO
CM CM" T-i
*•


MOLJ BJniDBJJI
EJ.PBH sncuod | x xx
jajinbv psuj..(uo3 | x

o
fi.

(/)


^*
s
i
VI
I i
li_ 1-

10 Rf
U U
75 jE 75 S S
5 £ 5 fe 5
XX
xxxx

ro
CM -
T-1 * CM i-l
xxxx



< '• '' •'
_1 «
°- cC
XX « X XXXXX XXXXXX
xxxx x xxx xxx
xxxx x ^ x x xxxx
XXXX 'X' XX XX X
M
,xxxx. x o!;^^ xxxx
xxxx x xxxx x
XX XXXX XXXX XXXXXX
XX X X

xxxx xxxx xxxxx xxxxxx

ro
CM •>
t-I •CMt-1 .^.T-ICMCMPO CMCMOOCMOJ COCMCMCMCOrO
1-1 _? x>
*
£
5 -
xxxx Jxxxx gxxxxx xxxxxx
^ x"Sxx xxxx
*> , •»->
3 x axx:xxx xxxxxx
to ta
CO CO
^ "** 4J
g* S- +J
6 TO o
[T ij~ CM "3 «
"~ oO -0 - 3: Q; cjo oo
ulS^3 azuZK a UJ o uj — i a: lucof-ntoto
CSCDZZ z 3 u, => U- a= CQ o u. u_ u_ i— :D >• >• 3 o
























«—
1

S
^0
£
5,
1


t
o
at
c.
g
H:

Q
U
1
x
x


CM

X
X


^


X
X
X
X

X
X
X


X


CM




X


X





UJ
1—
UJ
U-.
X
X





X


Q.


X
X
X


X


X

X







X


X






_J
0
y
UJ
Q.
                                               5-37

-------
                uoimiosi
                                                        it!
                                        es KB s    s?  e  ei  e   s
       £  K £ S. E
                                                                                                                   XX     XXX
JSp-KJ-Z'l'O
UUJ.3.1MU&PV
uojsasdsjG
*.*«« uoisu^a
asjnos ,uqsuo3
SDJIIOS an LEA iej4}ui
334 nos
^_^ IBSJV Jo 'aun ')U(Od
•a
01 ^USJSUBJJ.
.JE SWS /J>B8«
+J
c
g sssu^DHU BLqcjaBA
SSBUpmi JUBJSUOO
CO
in
li^J
CO SJaiSUBJBd
1 XXXXXX XXX XX X XX X XXXXXX
! XXXX XX XXXXX X X XXX X X X XXXXXX
«r CO ~. f° CO
_.„._, CM » CM * « CM '
PJCO»-tCM *CJ CMCMCOCM" OJ COCMCM CMCM CO CMCM«CMCOCO
*-* ^ ^4^-1 ^j. . ,_,
XXXX X XXXXXXXX X XXXXX
XXXXX XXX XX XX X
a.
XXXX XX XXXXX X X XXX X X X XXXXXX
XXXXXX XXXXXXXXXXXX X
XXXXXX XXXX XX X X XXXX
XX X XX XX
XXXXXX XX X XX X X XXXXXX
XXXX X X X
XXXXXX XXXXX XXXX X XX XXX
                        I

            snoausSodioH |
                            xxxx    xx
                                                       xxxx
                                                                         xx    xxx   xxx
                                                                                         CM   CVJ    CJ    CO
                                                                                                                  XXXXXX
                                                    >>


                                                    I X

                 snojod |
XXXX    XX     ^
                                                                        x    x    xxx   x    x    x     nxxxxxx
                                                                                                               c


                                                                                         XXXX     C)X           XX
s
TD
c


5
                            XXXX    XX
                                                    O.          CVJ
                                                    tn          a


                                                    ra    co  S g^   ^_^


                                                    H-    >  ce O   cs
                                                          UJ  UJ CO   O
                                                                        X
                                                                                      X
                                                    T     S
                                                                       £ t: w     "«

                                                                       ~"5 fe     ^
                                                                       co o- co     t.
                                                                VI O =>

                                                                t-H O —
                                                       3:  CO  CD

                                                       £  §  S
      UJ    I— (_  c\J

      az tg —i u_  a.
   LU  ^ o: 
-------
•o
 cu
 3
 £=
 o
 o
CO
 I
LO

UJ

ca
a
•z.
UJ

LU
                        re
                        o
i—     O
 re     T-

«£     re

 i     re

       en
•Z.     CU
<     id
                    to o
                    0  -r-
M U_

 I
 I
                    o T-I cvi c£ :
                                                  0
                    m  >> t=
                    o  re M-
                    cu  o  re
                    o  CD -c
                       o o

                    CD  CD  CD
                    •O  >  >

                    ^^ Ju? J--^
                        o  o
                    •o  re  re
                    coo
                    o «r- T-
                    o -o T3
                    d)  fl3  'O
                    co o; oi
                                        o
                                                         10
                                                         c:
                                                         Q
                                  re
                           p— i—  C
                           re re  o
                           C E •!-
                           o o  to
                           •i- T-  c:
                           co co  aj
                           c c  S
Q O CD

 OJ  O S-
 c  3s J=
O I— h-
                                                                   CM CO
fference
                                                                                4->
                                                                                C '
                    CD
                    U
                    c
                    o  co
                    s-  o
                    CU -r-
                    q_ +j
                    M-  CO
                    •i— -r—
                    O  S.
                        (U
                    CO 4-S
                    •P  O
                        re

                        re
                                                                                 o>'
       c

       ul ^:  co
          o .£=
       T3     O.
•p- r—   O Oi
       re      i
       s- -o  c=
       CD O  O
       cu .s: -P
                                                            OJ
                                                            •P
                                                            i=  c

                                                            1Z u.
                                            •r- C  CU
                                                                              t  I
                                                                             O LU Q «_> C£.
                                                                             ii.Li.li.dz:
                                                                                CU
                                                                                o.
                                                             CU
                                                             u


                                                             o
                                                            co
 CD     CU
 O  CO O
 I-  O i-
 3  i- 3
 O  3 O
co  o co
   CO

"re  co c
 cu  c: T-
 t T- O
< _l D-

 I   I   I
                                                                  5-39

-------
 Identified  in the  same order  as the technical selection criteria discussed in
 Section 5.1  (Figure 5-1).  The last column in the worksheet includes comments
 pertaining  to the  solution technique  (analytical model, finite difference, or
 finite element numerical model) and any other pertinent information.  A summary
 of each of  the models listed  in Table 5-3 is contained in Appendix A.  More
 detailed information, user's  manuals, and copies of many of these models can be
 obtained from the  International Ground Water Modeling Center (IGWMC) model data
 base.  IGWMC is located in Indianapolis, Indiana.
     In many instances, selection of a model that has more capabilities than
 necessary but with which the  analyst is familiar may be more cost effective than
 applying a  smaller, unfamiliar model with minimum capabilities due to the
 acquisition, testing, and learning process that would be required.
 5.4.  WASTE MANAGEMENT MODELS
     A few models have been developed by/for the EPA and others which consist of
 a methodology for tracing contaminant movement through the various environmental
media.  For the purposes of this report, these models will be called waste
management models.  Waste management models typically track the movement of
 hazardous waste from the source (point of disposal) through one or more of the
three primary environmental pathways; air, surface water,  and/or ground water.
Since this document is concerned with ground-water models, we will only discuss
those waste management models which simulate the ground-water pathway.
     Waste management models typically consist of a number of submodels to
simulate the many components of transport from a contaminant source to the point
of exposure.  These submodels usually consist of 1) source term generation,
which can include leakage through a liner(s); 2) contaminant transport through
the unsaturated zone; 3)  contaminant transport through the saturated zone; and
                                     5-40

-------
4) uptake by humans at the point of exposure.  When simulating contaminant
transport, the submodel generally simulates phenomena such as dispersion,
degradation, and adsorption.
     It is not the objective of this document to cover waste management models
in any detail.  Rather, a few such models are described briefly to make the
reader aware of them.  If a waste management model uses a ground-water transport
model that was discussed earlier, this is brought to the attention of the
reader.
5.4.1.  Risk Assessment Methodology for Regulatory Sludge Disposal Through Land
        Application
     ICF Technology, Inc., has recently developed a risk-based approach to
setting sludge criteria for land application disposal for the EPA.  The
methodology can also be applied on a site-specific basis to evaluate permit
applications.  ICF developed the modules addressing ground water, vapor, and
surface runoff pathways.  In each case, analytical models were assembled to
track the movement of contaminants from source to site of exposure.  Predicted
levels can be compared to health-based criteria, or health-based criteria can be
input to select limiting sludge contaminant concentration criteria.  A series of
representative scenarios are being evaluated with the methodology to identify
criteria thresholds and areas where best management practices should be
prescribed.
5.4.2.  Risk Assessment Methodology for Regulating Landfill Disposal of Sludge
     ICF Technology, Inc., has been responsible for the development of the
risk-based methodology for EPA upon which sludge landfill disposal regulations
and criteria are being based.  The methodology can be* applied on a
national/regional level or on a site-specific basis.  For the ground-water
                                      5-41

-------
pathway, it predicts leachate quantity and quality and subsequent impacts on
ground water by modeling contaminant movement through the unsaturated zone and
in the aquifer.  Degradation, adsorption, and geochemistry are all accounted
for.  A simple vapor-release and atmospheric-dispersion approach is taken for
the vapor pathway.  Runoff and particulate suspension are addressed through best
management practices.
     The landfill disposal methodology uses the CHAIN model and the AT123D model
to simulate contaminant transport through the unsaturated and saturated zones,
respectively.  Both of these models are discussed in the previous section and in
the appendix.
5.4.3.  RCRA Risk/Cost Policy Model (WET Model)
     The RCRA risk/cost policy model establishes a system that allows users to
investigate how trade-offs of costs and risks can be made among wastes,
environments, and technologies (W-E-Ts) in order to arrive at feasible
regulatory alternatives.  The model was developed for EPA by ICF, Inc.
     There are many components in the system.  Eighty-three hazardous waste
streams are ranked on the basis of the inherent hazard of the constituents they
typically contain.  The system assesses these waste streams in terms of the
likelihood and severity of human exposure to their hazardous constituents and
models their behavior in three media -- air, surface water, and ground water.
The system also incorporates the mechanisms by which the constituents are
affected by the environment, such as hydrolysis, biodegradation, and adsorption.
     A second integral part of the system is the definition of environments in
which the hazardous components are released.  Thirteen environments including a
special category for deep ocean waters are defined on the basis of population,
density, hydrology, and hydrogeology.  The system adjusts the exposure scores of
                                      5-42

-------
the waste streams' hazardous constituents to account for their varying effects
in the three media in each of the environments.
     The third component of the system consists of the technologies commonly
used to transport, treat, and dispose of the hazardous waste streams.  This
includes 3 types of transportation, 21 treatment technologies, and 9 disposal
technologies.  The system determines costs and release rates for each of these
technologies based on the model's existing data base.  It also incorporates
estimates of capacities of the technologies, the amount of waste to be disposed
of, and the proximity of the wastes to the available hazardous waste management
facilities.
      EPA's purpose in developing the RCRA risk/cost policy model is to assist
policy makers  in  identifying cost-effective options that minimize risks to
health and the environment.  The framework of  the system is  intended as a
screen -- to  identify situations that are of special concern  because of the
risks they pose  and  to  determine where additional controls may not be warranted
in light  of  the  high costs  involved.  The framework uses a data  base that  is too
imprecise and general to be  the  sole  basis  for regulations.   The results of  the
model will  be used  in more  detailed regulatory impact  analysis to  determine
whether  some type of regulatory  action  is warranted.
      Contaminant transport  is  not  simulated in the  WET model. Rather,  a ground-
water exposure/risk score is tallied based  on  key flow and  transport parameters
 such as  hydraulic conductivity,  depth to ground water, adsorption,  and
 hydrolysis.
 5.4.4.   The Liner Location  Risk and Cost Analysis Model
      The liner location risk and cost analysis model,  developed  for EPA by ICF,
 Inc., links a risk and a cost model.  The risk model  simulates the chronic risk
                                       5-43

-------
 to human health from  land disposal facilities  (landfills and surface
 Impoundments) with different,design technology, location, and waste stream
 combinations.  It integrates a series of submodels and algorithms that trace
 constituent releases  from landfills and surface impoundments to their movement
 through the air and ground, then to resulting human exposures,  and finally to
 the resultant human risk.  To do so, model  components predict:
 •    Releases to ground water and subsurface transport.  A failure and release
      submodel estimates facility failure and the quantity of leachate released;
      an unsaturated zone algorithm calculates the time required for the leachate
      to reach an underlying aquifer;  and a  saturated zone submodel  calculates
      the time and  concentrations of constituents reaching a  downgradient well.
 *     Releases to air and atmospheric transport.   A volatilization  algorithm
      calculates  the  quantity of constituents that volatilize over  time;  an
      atmospheric transport  algorithm calculates  the concentrations  of these
      constituents  at  the exposure points.
 •     Human  exposure.   An exposure algorithm  calculates  the exposure  from
      drinking  water and  from breathing constituents  in  air.
 •     Health risks.  A  hazard estimation  submodel calculates  expected cancer and
      noncancer risks from the exposures.
 The model calculates risk over a  400-year time horizon.   In  doing so,  it
 embodies many  assumptions:   facilities operate for 20 years with a 30-year
 post-closure period; contamination goes undetected and uncorrected; multiple
contaminants do not interact; and few constituents degrade over time, with
degradation beginning only after facility failure.  The model also assumes that
aquifers are homogeneous and isotropic.
                                     5-44

-------
     The companion cost model estimates the costs of land disposal facilities
with differing technologies and sizes.  The results of the risk model combined
with the cost data from the cost model provided EPA with the capability to
perform cost/risk and cost-effectiveness analysis.
     The liner location model uses the TRANS model (Random-Walk Particle
Transport Model) to model contaminant transport in the saturated zone.  The
TRANS model is discussed in the previous section and in the appendix.
5.4.5.  Landfill Ban Model
     The landfill ban model  is a quantitative modeling procedure used by EPA to
evaluate potential impacts on ground water and establish screening levels for
this medium.  The ground-water screening procedure involves a back calculation
from a  point  of potential exposure at  a specified distance directly  downgradient
from a  point  of release  from a land disposal unit using  a fate and transport
model.  The ground water back calculation procedure  involves the  application of
three model components:  1)  the HELP  model which  addresses performance  of
engineer controls; 2)  the  fate and transport model  (EPASMOD) which models the
behavior of constituents in  the ground-water environment; and 3)  the MINTEQ
model which models the behavior of metals  in the  ground-water environment.
     The HELP model  was developed by  EPA  specifically  to facilitate  estimation
of the  amount of  runoff, drainage,  and leachate  that may be  expected to result
 from a  hazardous  waste landfill.  The model  predicts the water  balance  by
 performing a  mass balance  between  flow into  various  components  of a  landfill  and
 water leaving these  components.   The  model  uses  climatology,  soil,  and
 design  data to produce daily estimates of water  movement across,  into,  through,
 and out of landfills.
                                       5-45

-------
     The EPASHOD model predicts the fate and transport of constituents in the
ground water.  EPSMOD is a three-dimensional, steady-state, advective dispersive
model which utilizes analytical solution procedures to predict transport through
homogeneous and isotropic porous media.  The model accounts for advection;
hydrodynamic dispersion in the longitudinal, lateral, and vertical dimensions;
absorption; and chemical degradation.
     Estimates of metal species distributions are determined using the
geochemical model MINTEQ.  MINTEQ is an equilibrium model that uses the
equilibrium constant approach in solving the chemical equilibrium problem.
     The contaminant transport model EPASMOD is not included in the appendix of
this report because it is still in the development stages and at this time is
not well documented.
                                     5-46

-------
                     6.  MODEL SELECTION EXAMPLE PROBLEMS

     Two site-specific example problems are provided below to demonstrate the
procedure for selecting the appropriate mathematical model  for a particular
application.  The first example is an application where the objective is to
perform a screening study,  while the objective of the second example is to
perform a detailed study.  The discussions of the example problems are presented
in the order that should be followed when conducting a ground-water flow and
contaminant transport modeling study, with model selection being one element of
the process.
6.1.  SCREENING ANALYSIS EXAMPLE PROBLEM
     Over 300 landfills are licensed to operate within a particular state.  The
state is interested in determining whether significant ground-water
contamination is occurring at the sites.  As a first step, the state is
interested in using models to define the scope of the problem to determine if
detailed investigations of several of the sites are required.  At this stage,
the state is not interested in detailed, site-specific assessments.
6.1.1.  Ob.iectives of the Study
     The objectives of the study are to develop a generic model of the ground-
water flow and contaminant transport system beneath a landfill and to perform a
sensitivity analysis to determine the likelihood that a significant
contamination problem might exist.  The data used in the sensitivity analysis
are generic values representative of the actual landfills.  The sensitivity
analysis consists of several model simulations using both average and
conservative values for the model parameters.
                                      6-1

-------
      The study is primarily concerned with a contaminant transport analysis of
 five contaminants commonly found in landfill leachates.  The contaminant data
 have been obtained from existing data for representative sites.
 6.1.2.  Conceptual Model of the Study Area
      Of the 300 sites in the state, 30 are chosen as a representative sample.
 The existing data for the 30 sites are .used to develop a generic conceptual
 model  of a typical landfill site.
      For this example, we will  assume the conceptual model  has been developed
 and is very simple.   It consists of a contaminant source (either point,  line,  or
 area!)  at the water table beneath the landfill;  a single layer,  homogeneous,
 porous-media aquifer; and a discharge location at a specified distance
 downgradient.   The source strengths,  ground-water velocities  (hydraulic
 conductivity and  gradient),  distance  to the discharge,  and  other pertinent  model
 inputs  have been  determined from the  existing  data for  the  initial  simulation.
     We will  also assume the five contaminants of interest  and  their  transport
 properties  have been  identified  from  the existing data.   Dispersion,  adsorption,
 and degradation are important processes  that need to be   simulated  in  the
 initial  model  run  and adjusted in the  sensitivity analysis.
 6.1.3.   Model  Selection  Process
     Having developed the conceptual model, the analyst  should now  follow the
 technical criteria to determine which model or suite  of models will be
 appropriate for this  application.  Because the objective of this study is to
 perform a generic or  screening analysis, a screening model should be selected.
Using a screening model allows for making several sensitivity runs relatively
quickly and at low cost.  Typically, such a screening model  is an analytical
solution, an analytical model,  or a simple numerical model.
                                      6-2

-------
     For this study, there is no point in using a detailed model  since the data
are not site-specific and the objective is not to predict site-specific exposure
 i
levels but rather to compare cases in a sensitivity analysis.
     Since the objective is to study the transport of 5 contaminants, a
contaminant transport model should be selected.  The ground-water flow is not
simulated in the transport model, but rather it is specified by the available
data (in terms of ground-water velocity, or hydraulic conductivity and hydraulic
gradient which can be used to solve for velocity).  Even though ground-water
flow is not simulated, the flow parameters can still be adjusted in the
sensitivity analysis.
     The first part of the technical selection criteria (Figure 5-1) relates to
ground-water flow.  Based on the conceptualization, the flow portion of the
transport model should have the capability to simulate the following conditions:
water table aquifer, porous media, steady state; single phase, single layer of
constant thickness, and homogeneous hydraulic properties.  In a contaminant
transport model, all of these properties are usually represented as a uniform
velocity down a one-dimensional flow path.
     The second part of the technical selection criteria relates to contaminant
transport.  Based on the conceptualization, the transport model should have the
capability to simulate the following conditions:  area! source (point or line
source would be sufficient), constant source term, dispersion, adsorption, and
degradation.  The transport can be simulated in either one, two, or three
dimensions.  A one-dimensional simulation is most practical since the flow is
one-dimensional.  However, a three-dimensional transport simulation could take
                                      6-3

-------
 full  advantage of the contaminant concentration reduction resulting from
 dispersion in three directions.   Density effects are not  important  in this
 application.
      At this  point,  the analyst  should  go to the model  selection worksheet  and
 select a transport model  which has all  the capabilities discussed above.  Some
 logical  choices of models are AT123D, CHAIN,  or MMT.  Most numerical  models  are
 too sophisticated for this application.   The final  selection of a single model
 should be based on the implementation criteria  (code  has  been verified and
 documented) and on the analyst's  familiarity with and access to the model.
 6.2.   DETAILED  ANALYSIS EXAMPLE  PROBLEM
 6.2.1.   Statement of the  Problem
      Benzene was  disposed of in surface  lagoons  at  an industrial site from 1960
 to 1980.  The. disposal  operation was shut  down when benzene was found in
 residential wells  downgradient (south) of  the site.  A  network of monitoring
wells  sampled in  1985  show that the benzene had migrated  up to a mile
downgradient, was  found near the surface just south of the site, was found at
depth  further south  of the site,  and concentrations ranged between 0 and 2,000
ppb.
6.2.2.  Objectives of the Study
     The objectives of the study were to select a ground-water flow and
contaminant transport model of the site and use the model  to:
1.   determine the likelihood that the shallow and deep  plumes are connected,
     and if so,  identify the reason(s)  for the plume to  migrate to a deeper
     depth;
                                      6-4

-------
2.   predict the future extent of the plume and determine if the water supply
     wells for a city south of the site will  become contaminated; and
3.   if benzene levels in the city wells could become too high,  design a
     pump-and-treat system to cleanup/contain the plume.
6.2.3.  Conceptual Model of the Study Area
     The site is located in a narrow alluvial valley about 1 mile north of a
small town (Figure 6-1).  The waste lagoons are situated just above the water
table of the regional alluvial aquifer.  The aquifer consists of uniform sands
and gravels over its entire depth, and its saturated thickness ranges from 0 ft
along the edges at the bedrock outcrops to 100 ft in the center of the valley.
The bottom of the aquifer is defined by bedrock.  South of the lagoons in the
area where the benzene plume appears to migrate to a lower depth, a creek cuts
across the valley from east to west.  Five pump tests over the study region
yielded values of hydraulic conductivity that ranged between 110 ft/day and 340
ft/day.   Recharge in the area is estimated as 10 in./yr and is uniformly
distributed over the study area.  Water levels in the valley fluctuate very
little throughout the year.  The only significant pumping in the area is the
pumping of the city wells.  The creek is located in  a low-permeability zone
where  it  cuts  across the valley.  A  single pump test in the material yielded a
hydraulic conductivity  of  25  ft/day  and geologic logs show that  the zone is
about  50-ft deep  and  500-ft wide.  The  creek is not  hydraulically connected to
the  aquifer.
      The  principal contaminant of concern  is benzene.  Monitoring data  show the
plume has migrated about  1 mile  south of the site,  is shallow (between  0 and 30
ft below  the  water table)  just south of the  site and north  of the creek, and  is
deep (between 50  and 80 ft below the water table)  south  of  the  creek.   The plume
                                       6-5

-------
                                                      Scale:   3  1/2  in.  =  5,000  ft
Bedrock
Outcrop
                      Direction of
                      Ground-Water Flow
                                                    Low
                                                    Permeability
                                                    Zone
                                                                     Bedrock
                                                                     Outcrop
          Figure 6-1.   Model  region for the assessment-level  example problem.

                                          6-6

-------
has a peak concentration of about 2,000 ppb just south of the lagoons and a peak
of about 100 ppb south of the creek.  The width of the plume appears to be
between 1,000 and 1,500 ft.
6.2.4.  Model Selection Process
     The model selection process first focuses on the capabilities of the
ground-water flow model and then on the capabilities of the contaminant
transport model.  In some instances it may be possible to simulate the flow with
one model and the transport with another.  This is often done when using
analytical solutions or simple numerical models.  In more complicated,
higher-dimensional problems, a single combined model is typically used.
     When selecting a model based on the various criteria, the analyst can
always step up to the next level but should never step back.  For example, if a
problem only requires a model that can simulate a single layer, the analyst can
always use a model that is capable of simulating multiple layers and only
simulate one.
     The conceptualization of this example problem indicates that a water table
model should be used.  However, the water table only experiences minor
fluctuations so a confined model could be used.  Also, because the water table
remains fairly constant, a steady-state model would be acceptable.
     The flow is single phase (since benzene .is soluble in water) through a
porous medium.  The flow occurs within a single aquifer with varying thickness.
Based on the above, a porous media model that has the capability to simulate
single phase flow and a single aquifer with variable thickness should be chosen.
     For this problem, a three-dimensional flow model is required as a result of
the partially-penetrating, low-permeability zone in the vicinity of the creek.
Since the aquifer appears to be fairly homogeneous, it can be simulated with a .
                                      6-7

-------
 single value of hydraulic conductivity and  porosity.   However,  since  the  low
 permeability zone exists along the  creek, a model  that allows for  an  areal
 distribution of hydraulic conductivities  should  be selected.  Also, the model
 needs  to  allow for a  vertical  variation in  hydraulic  conductivity  since the low
 permeability zone is  only partially penetrating.
     At this point the  analyst can  either choose a three-dimensional  ground-
 water  flow model  that satisfies all  the requirements  (or exceeds the
 requirements)  and select a separate transport model later, or he/she  can
 continue with  the decision process  to  select a coupled ground-water flow  and
 contaminant  transport model.   Probably the  best approach is to  determine  the
 transport model  requirements before selecting any  models.
     Benzene entered  the ground water  over  a period of 20 yrs (1960-1980) by
 leaching through  the  bottom of surface lagoons as  it was being  disposed of.
 Although disposal  stopped in 1980,  residual levels of  benzene in the  soils
 beneath the  lagoons continued  to leach  into the ground water at reduced rates.
 The length of  time required to  leach all the benzene from the soil was estimated
 based on residual  levels,  solubility,  and recharge rate.  A complete  sample of
 all monitoring wells was  completed  in  1985 to provide data with which to
 calibrate the model.
     In order to  calibrate the model,  the analyst would simulate 20 yrs of
 leaching at  full  strength  followed  by  5 yrs of leaching at a reduced rate and
 compare the  1985 model-predicted benzene concentrations to the observed values.
 In order to properly simulate the release of benzene over the calibration
period, the analyst needs to select a model  that can simulate variable leaching
rates.
                                      6-8

-------
      To predict future concentrations,  the analyst can either run the calibrated
 model  into the future or he can run an  initial  value problem where the measured
 concentration  distribution is input to  the model  in 1985 and the model  is run
 into  the future.   For this type of a simulation,  the analyst needs to select a
 model  that can simulate an initial  value  problem.
      The conceptual  model  indicates that  the  plume appears  to migrate to a lower
 depth  in the vicinity of the creek.   In order to  simulate this migration,  the
 contaminant transport would need to be  simulated  in either  two dimensions  in the
 x-z plane or in three dimensions.   A two-dimensional  x-z model  may be capable of
 assessing whether  the plumes are connected and  predicting the future  extent of
 the plume.  However,  it would not  be capable  of aiding in the design  of a
 pump-and-treat remedial  action.  A  fully  three-dimensional  model  would  be
 required to determine well  placement in the x-y plane,  number of wells  to  pump,
 pumping  rates,  and well  penetration  depths.
     Since  benzene in this  situation does  not significantly degrade,  a  model
 that simulates  degradation  is  not required.
     Since  there are  no  significant  thermal or  concentration  gradients  in  the
 study area, density effects  can  be neglected.   Therefore, a  fully-coupled  model
would not be required.
     At  this point the analyst  is ready to select a flow-and-transport model.
Since the problem is  fairly  complex,  requiring  three dimensions  for both flow
and transport,   a single combined flow-and-transport model would  be the logical
choice.  The analyst would go to the model selection worksheet and select  a flow
and transport model which has all the capabilities discussed above.  Some
logical choices of models would be SE60L,  TRUST, GROVE/GALERKIN, PINDER, CFEST,
or SWIP2.  The  final  selection of a single model would be based on the
                                      6-9

-------
Implementation criteria (has the model been verified and documented?) and on the
analyst's familiarity with and access to the model.
                                     6-10

-------
                                 7.   REFERENCES
Bachmat, Y.; Bredehoeft, J.; Andrews, B.; Holtz, D.; Sebastian S.   (1980)
     Groundwater management:  the use of numerical models.  Water Resources
     Monograph American 5, Washington, DC:  Geophysical Union.

Bear, J.  (1972)  Dynamics of Fluids in Porous Media.  New York, NY:  American
     Elsevier.

Bonn, H.; McNeal, B.; O'Connor, G.  (1979)  Soil chemistry.  New York, NY:
     John Wiley and Sons, Inc.

Bolt, G.; Bruggenwert, M.  (1978)  Soil chemistry.  A.  Basic elements.  New
     York, NY:  Elsevier Scientific Publishing Company.

Chow, Ven Te.  (1964)  Handbook of applied hydrology.  New York, NY:
     McGraw-Hill Book Company, Inc.

Cleary, R.W.; Ungs, M.J.  (1978)  Analytical models for ground-water
     pollution and hydrology.  Report 78-WR-15, Water Resources Program,
     Princeton University, NJ.

Dagan, G.; Bresler, E.  (1979)  Solute dispersion in unsaturated heterogeneous
     soil at field scale.  1.  Theory.  Soil Sci. Soc. Am. J. 43:461-467.

Dagan, G.  (1982)  Stochastic modeling of groundwater flow by unconditional and
     conditional probabilities:  2.  The solute transport.  Water Resour. Res.
     18(4):835-848.

Davis, S.N.; DeWest, R.J.M.  (1966)  Hydrogeology.  New York, NY:  John
     Wiley and Sons, Inc.

Freeze, R.A.; Cherry, J.A.  (1979)  Groundwater.  Englewood Cliffs, NJ:
     Prentice-Hall, Inc.

Freyberg, D.L.  (1986)  A natural gradient experiment on solute transport in a
     sand aquifer:  II.  spatial moments and the advection and dispersion of
     non-reactive tracers.  Water Resour. Res.

Gelhar, L.W.; Gutjarh, A.L.; Naff R.J.  (1979)  Stochastic analysis of
     macrodispersion in a stratified aquifer.  Water Resour. Res. 15:1387-1397.

Gelhar, L.W.; Axness, C.L.  (1983)  Three-dimensional stochastic analysis of
     macrodispersion in aquifers.  Water Resour. Res. 19(1):161-180.

Gelhar, L.W.; Mantoglou, A.; Welty, C.; Rehfeldt, K.R.  (1985)  A review of
     field-scale physical  solute transport processes in saturated and
     unsaturated porous media.  EPRI EA-4190, Electric Power Research
     Institute,  Palo Alto, CA.
                                      7-1

-------
Gschwend, P.M.; Wu, S-C.   (1985)  On the constancy of sediment-water partition
     coefficients of hydrophobic organic pollutants.  Environ. Sci. Techno!.
     19:90.

Gupta, S.K.  (1982)  A multi-dimensional finite element code for the analysis
     of coupled fluid, energy and solute transport.  PNL-4260, Pacific
     Northwest Laboratory, Rich!and, WA.

Hwang, S.T.  (1986)  Mathematical model selection criteria for performing
     exposure assessments:  groundwater contaminants from hazardous waste
     facilities.  U.S. Environmental Protection Agency, Exposure
     Assessment Group (RD-689), Washington, DC.  Unpublished Draft.

Javandel, I.; Doughty, C.; Tsang, C.F.  (1984)  Groundwater transport:
     handbook of mathematical models.  Water Resources Monograph 10,
     Washington, DC:  American Geophysical Union.

Jury, W.A.  (1982)  Simulation of solute transport using a transfer function
     model.  Water Resour. Res. 18(2):363-368.

Karickhoff, S.W.  (1980)   Sorption kinetics of hydrophobic pollutants in
     natural sediments.  In:  Baker, R. A., ed. Contaminants and sediments,
     Ann Arbor MI:  Ann Arbor Science Publishers Inc., pp. 193-205.

Karickhoff, S.W.  (1981)   Semi-empirical estimation of sorption of hydrophobic
     pollutants on natural sediments and soils.,  Chemosphere 10:833-846.

Karickhoff, S.W.  (1984)   Organic pollutant sorption in aquifer systems.  J.
     Hydraulic Eng. 10(6):707-735.

Kelly, W.E.  (1982)  Field reports -- ground-water dispersion calculations
     with a programmable calculator.  Ground Water 20(6).

Matheron, G.; DeMarsily, G.  (1980)  Is transport in porous media always    f
     diffusive?  Water Resour, Res. 16:901-917.

Mercer, J.W.; Faust, C.R.  (1981)  Ground-water modeling.  Ground Water
     18(2-6).

Molden, D.; Sunada, O.K.; Warner, J.W.  (1984)  Microcomputer model of  "
     artificial recharge using Glover's solution.  Ground Water 22(1).

Neely, W.B.; Blau, G.E.  (1985)  Environmental exposure from chemicals; Vol.
     I.  Boca Raton, FL:  CRC Press, Inc.

Oberlander, P.L.; Nelson,  R.W.  (1984)  An idealized ground-water flow and
     chemical transport model (S-PATHS).  Ground Water 22:(4).

Peterson, S.; Hostetler, C.; Deutsch, W.; Cowan,  C.  (1986)  MINTEQ user's
     manual.  PNL-6106, Pacific Northwest Laboratory, .Rich!and, WA,
                                      7-2

-------
 Pettyjohn, W.A.; Kent, D.C.; Prickett, T.A.; Le Grand, H.E.; Witz, F.E.
      (1982)  Methods for the prediction of leachate plume migration and
      mixing.  Draft, U.S. Environmental Protection Agency, Municipal
      Environmental Research Laboratory, Cincinnati, OH.

 Pickens,  J.F.; Grisak, 6.E.  (1981a)  Scale-dependent dispersion in a
      stratified granular aquifer.   Water Resour.  Res. 17:1191-1211.

 Pickens,  
-------
Yeh, G.T.  (1981)  AT123D:  Analytical transient one-, two-, and
     three-dimensional simulation of waste transport in the aquifer system.
     ORNL-5602, Oak Ridge National Laboratory, Oak Ridge, TN.
                                      7-4

-------
               APPENDIX A



ANALYTICAL AND NUMERICAL MODEL SUMMARIES

-------
                                     CONTENTS
 ANALYTICAL FLOW.  ......  	  ....    A-l

     PATHS	  .    A-2

 ANALYTICAL TRANSPORT ....  	    A-5

     AT123D	  .    A-6
     CHAIN	    A-8
     GETOUT	    A-10
     GWHTM1 and  GWMTM2	    A-12
     NUTRAN	    A-14
     NWFT/DVM 	  .......    A-17

 NUMERICAL FLOW  (SATURATED/UNSATURATED)  	    A-19

     FEMWATER1	    A-20
     FREEZE	    A-23
     UNSAT1	    A-24
     UNSAT2	    A-25

 NUMERICAL FLOW  (SATURATED)	    A-28

     BEWTA	    A-29
     COOLEY	;	    A-30
     FE3DGW	    A-33
     FLUMP	    A-38
     FRESURF 1 & 2	    A-39
     TERZAGI	    A-40
     USGS2D	    A-41
     USGS3D -- Modular	    A-45
     USGS3D -- Trescott	    A-46
     VTT	;	    A-49
     V3	    A-52

NUMERICAL  SOLUTE TRANSPORT  (SATURATED/UNSATURATED) 	    A-55

     FEMWASTE1	    A-56
     PERCOL	    A-60
    SATURN	    A-61
    SEGOL	    A-64
    SUMATRA-I	    A-65
    SUTRA		    A-66
    TRANUSAT	    A-68
    TRUST	-....:	    A-70

NUMERICAL SOLUTE TRANSPORT  (SATURATED)	    A-77

    CHAINT	    A-78
    DUGUID-REEVES	    A-81
    GROVE/GALERKIN 	    A-82
                                      A-ii

-------
                              CONTENTS (continued)
    ISOQUAD, ISOQUAD2. .
    KONBRED, USGS2D-MOC.
    DPCT	
    MMT	
    PINDER  	
    ROBERTSON1 	
    ROBERTSON2 	
    SWENT.  ........
    TRANS	,
    TRANSAT2 	 ,
NUMERICAL COUPLED CODES  (SOLUTE AND HEAT TRANSPORT)
    CFEST. .
    GWTHERM.
    OGRE  . .
    SHALT. .
    SWIFT. .
    SWIP2. .
A-83
A-84
A-87
A-89
A-91
A-92
A-93
A-94
A-99
A-103

A-105

A-106
A-108
A-109
A-110
A-113
A-117
                                      A-iii

-------

-------
ANALYTICAL FLOW
      A-l

-------
CODE NAME:  PATHS                            ,                       ..,-,.

PHYSICAL PROCESSES:  Analytical  flow solution with pathline and travel  time
post processors.

DIMENSIONALITY:  Two-dimensional.

SOLUTION TECHNIQUE:  Analytic.

DESCRIPTION:  PATHS provides an approximate contaminant transport evaluation
by a direct solution of the pathline equations.  The steady cases are evaluated
by holding the uniform gradient, the head in the pond, and the well strengths
constant.  Under such steady-state conditions,  only one set of flow paths,
advancing fronts, and travel times must be calculated.  In the transient cases,
each new set of fluid particles leaving the pond or wells encounters changing
velocity effects. Therefore, a range of typical departure times is selected and
the flow paths, front configurations, and travel times are calculated
successively for each selected set of fluid particles leaving the contaminant
source. The approximate equilibrium coefficient approach is used to give the
ion exchange delay effects for a single constituent.  There are, however, no
dispersion effects considered in the preliminary model.  The model can consider
as many as 35 wells at optional  locations.  Wells are represented asnumerically
solved by the code to give the paths of the fluid particles and their advance
with time toward the outflow boundary.

The LOCQUAR component completes the unit outflow rates (i.e., the water outflow
volume per unit time per unit distance, along the outflow boundary).  The unit
outflow rate is a function of location as well  as time.

The main assumptions of the code are:                        .

•    two-dimensional (horizontal plane) infinite aquifer of constant thickness;

•    confined flow;                 ,                            ,

•    homogeneous, isotropic material with constant properties;

t    uniform flow direction may include transient gradient (flow) strength;

t    round, fully penetrating wells and caverns}

t    dissipation of the well and cavern heads occurs over a specified radial
     distance;

t    diffusion and dispersion processes are neglected; and

•    contaminant adsorption is based on linear equilibrium isotherms.

CODE INPUT:  An interactive computer program actually coaches the user
through preparation of the input file for PATHS.  A worksheet is ,given as
Table 1 in Nelson and Schur and its use is recommended because it will help
ensure the use of consistent units.  The interactive program can also be
followed without the aid of a worksheet.
                                      A-2

-------
In general  the PATHS code requires the following  data:
•    stratum thickness;
•    physical  properties of wells including location,  radius,  and flow rate;
t    radial distance to remote boundary;
•    x-coordinate termination of path lines;
0    source radius;
•    steady-state or dynamic head at source;
•    hydraulic conductivity;
a    effective porosity;
•    uniform gradient;
•    starting'times and locations of pathlines;
•instructions for plots to be generated; and
•    linear distribution coefficient.
CODE OUTPUT:  A variety of outputs are generated by the three components of
the code.  There are:
1)   PATHS -- Input data file for GROUND.
2)   GROUND -- Hard copy lists and plots of
     •    the fluid or contaminant flow paths;
     •    the rate of advance and shape of the contaminant fronts moving
          through the system; and
     •    the location and time of contaminant arrival  at the outflow boundary.
3)   LOCQUAR -- The water outflow volume per unit time along the outflow
     boundary.
COMPILATION REQUIREMENTS:  The code is written entirely in FORTRAN-77.
Current versions, originally developed on a Univac 1100/44 system, have been
converted to a Digital Equipment Corporation VAX 11/780 systems. Some minor
changes were made because the two systems differ in their file operations.
The code can capture information on 55 time planes for up to 50 pathlines and
35 wells.  The wells are at arbitrary'locations and are represented as
completely penetrating, vertical line sources with steady or time-dependent
flow rates.
PATHS is constructed in a modular fashion and appear to be easily modified.
Versions exist on two hardware configurations.  The software allows someone
                                       A-3

-------
with limited computer  background to run the code.  The interrogative building
of the input data file coupled with the batch processing of the solution is a
user-friendly and computationally efficient method.

EXPERIENCE REQUIREMENTS:   Extensive.

TIME REQUIREMENTS:  Months.

CODE VERIFICATION:  As an  analytical solution, the validation is simply
required to test for errors  in formulation of the solution or in the computer
coding.  The results of PATHS have been compared to the results from both VTT
and FE3DGW.

The PATHS model was designed to provide a balance between the refinement of the
model and the limited  data usually available for initial evaluation of
subsurface contamination problems.  It allows the user to make "first cut"
evaluations inexpensively  and quickly.  Specific applications include:

•    Initial model for movement from deep underground caverns for the Advanced
     Technology Development Section, Research and Engineering Department,
     Atlantic Richfield Hanford Company.

t    Initial model of  accidental release from a fuels reprocessing facility in
     South Carolina.

§    Initial evaluation of accidental failure of earthen sewage holding ponds,
     Kennewick, Washington.

t    Initial evaluation of seepage from a copper tailings reservoir.

•    Model used in "Groundwater Engineering Short Course," sponsored by
     Agricultural Research Service, Chickasha, Oklahoma.

•    Example evaluation for State of Idaho Department of Water Administration.

•    Evaluation of a numerical generation scheme for pathlines for Atlantic
     Richfield Hanford  Company in cooperation with the Pacific Northwest
     Laboratory for the U.S. Energy Research and Development Administration.

a    Evaluation of potential hazard from subsurface reactor accidental releases
     for Sandia National Laboratories, Albuquerque, New Mexico.

DOCUMENTATION/REFERENCES:

Nelson, R.W.; Schur, O.S.  (1980)  PATHS -- groundwater hydraulic assessment
     of effectiveness  of geologic isolation systems.  PNL-3162, Pacific
     Northwest Laboratory, Richland, WA.

SOURCE:  PATHS was written by R. W. Nelson and J. A. Schur and was a direct
result of research conducted by Pacific Northwest Laboratory.  The research was
supported by the Waste  Isolation Safety Assessment Program (WISAP).
                                      A-4

-------
ANALYTICAL TRANSPORT
        A-5

-------
 CODE NAME:  AT123D (Analytical  Transient 1-,  2-,  or 3-Dimensional)
 PHYSICAL PROCESSES:  Provides analytic or Green's Function solutions  to the  •.:
 solute transport equation in 1, 2,  or 3 dimensions with a constant  uniform
 water velocity.
 DIMENSIONALITY:   One-,  two-, or three-dimensional.
 SOLUTION TECHNIQUE:  Analytic.
 DESCRIPTION:   The code  is limited to  a single component,  constant,  uniform
 parallel  velocity field,  rectangularly shaped sources  and regions of  interest,
 with releases  at a constant rate. ,  The source may be a point,  line  segment,
 rectangle,  or  rectangular prism.  The release may be instantaneous, continue
 for  a finite time period  (band  release),  or be a  step  function.  Equilibrium
 adsorption  and radioactive decay are  included;  but decay  chains are not
 treated.  Aquifers may  have finite  or infinite depth and  width.  The  program
 output is the  radionuclide concentration  in the groundwater.   AT123D  requires
 that the  water flow be  known and be approximated  by a  uniform  parallel flow.
 The  principal  simplifying assumptions are as  follows:
 t     validity  of the solute transport equation;
 •     all  boundaries are of the  no-flow type;
 •     a constant,  uniform  parallel flow velocity;
 •     the  source  is a rectangular prism,  and the rate and  duration of  release
      are  the same everywhere within the source;
 •     infinite  solubility;  and
 •     release of  contaminant at  a constant rate  over some  duration.
 CODE INPUT:  The  principal  inputs are as  follows:
 t     location  and dimensions of the source;
 •     aquifer dimensions;
 •     porosity;
 •     hydraulic conductivity;
 •     hydraulic gradient;
 •     dispersivity;
 •    distribution coefficient;  and
•    duration of release.
                                      A-6

-------
CODE OUTPUT:  The output of the code is a series of tables showing
concentration at selected points and selected times.
COMPILATION REQUIREMENTS:  AT123D is written in FORTRAN IV, and should be
adaptable to most computers.  Current problem limits are as follows:
t    15 x-locations for output;
•    10 y-locations for output;
•    10 z-locations for output;
•    1,200 time steps; and
•    1,000 eigenvalues for series evaluation.
EXPERIENCE REQUIREMENTS:  Extensive.
TIME REQUIREMENTS:  Months.
CODE VERIFICATION:  The code has been checked against hand calculations and
laboratory experiments.  The code is in the public domain.
DOCUMENTATION/REFERENCES:
Yeh, G.T.   (1981)  AT123D:  analytical transient one-, two-, and three-
     dimensional simulation of waste transport in the aquifer system.
     ORNL-5602, Oak Ridge National Laboratory, Oak Ridge, TN.
SOURCE:  AT123D was written by G. T. Yeh at Oak Ridge National Laboratory.
                                       A-7

-------
 CODE NAME:  CHAIN
 PHYSICAL PROCESSES:  Convective-dispersive solute transport,  saturated or
 unsaturated steady flow, sequential first order radioactive decay,  linear
 sorption.
 DIMENSIONALITY:  One dimensional.
 SOLUTION TECHNIQUE:  Analytical based upon LaPlace transforms.
 DESCRIPTION:  CHAIN presents analytical solutions to solute transport
 involving sequential first-order decay in saturated or unsaturated  soil
 systems.  Solutions of the simultaneous movement of up to four  member chains
 are provided for the one-dimensional convective-dispersion
 equation.  A degrading source term may be modeled by invoking the solution  to
 the Bateman equations.  The main assumptions of the code  are:
 t    steady groundwater flow;
 •    constant pore velocity and dispersivity over path  length;
 •    linear geochemical  sorption model;
 •    linear decay with up to four members;
 •    initial  value or flux source condition;  and
 •    degrading source term.
 CODE INPUT:
 §    path  lengths;
 •    pore  velocity;
 •    water content;
 t    dispersivity;
 •    decay constants;
 •    retardation  coefficients;  and
 •    time  for  which solute concentration are calculated.
 CODE  OUTPUT:   Solute  concentrations at various times and positions along path
 for up to  four members of chain.
 COMPILATION REQUIREMENTS:  CHAIN  is written in FORTRAN and may be easily
 installed  on essentially all computer systems.
 EXPERIENCE REQUIREMENTS:  Moderate.
TIME  REQUIREMENTS:  Days.
                                      A-8

-------
CODE VERIFICATION:  Numerous numerical  verification problems are provided in
the user's manual  which has been published in the open literature. Field
validation has been performed for the convective-dispersion equation in
numerous tracer experiments in the saturated and unsaturated zones.
DOCUMENTATION/REFERENCES:
found in
A full user's manual and documentation may be
VanGenuchten, M.Th.  (1985)  Convective-dispersive transport of solutes
     involved in sequential first-order decay reactions.  Con. of Geosciences
     2:129-147.

SOURCE:   M. Th. VanGenuchten
          U.S. Salinity Laboratory
          4500 Glenwood Drive
          Riverside, CA  92501
                                       A-9

-------
 CODE NAME:  GETOUT                    .                                        ;

 PHYSICAL PROCESSES:  Predicts the transport of radionuclide chains along a
 one-dimensional path.

 DIMENSIONALITY:  One dimensional.

 SOLUTION TECHNIQUE:  Analytical.

 DESCRIPTION:  The model interfaces directly with the water-dose models,  ARRG
 and FOOD,  and together they are used to predict the dose consequences  to people
 from radionuclide releases.

 The model  analyzes the transport of radionuclides by flowing groundwater
 following  a leach incident at an underground nuclear waste disposal  site.  This
 model  assumes that at some arbitrary time after the waste is deposited;  the
 contents of the site are dissolved at a specific constant rate by groundwater.
 The groundwater flows at constant velocity through a homogeneous
 one-dimensional  column of the geologic medium and discharges
 to a surface water body.  The dissolved radionuclides are assumed to be  in
 sorption equilibrium at all points in the geologic medium.   Radioactive  decay
 (including chain decay of the actinides)  is modeled both at the disposal  site
 and during migration through the geologic medium.   Trace concentrations  of the
 dissolved  nuclides are assumed and,  as a  result,  the adsorption equilibrium
 constants  are independent of concentrations.   A constant axial  dispersion
 coefficient is also assumed.  This model  is applicable to particulate  and
 fractured  media,  provided the necessary input data are obtained properly  and it
 can be  applied to heterogeneous  media if  a weighted averaging  technique  is
 properly applied to the relevant input parameters.

 CODE INPUT:  Inputs for GETOUT include:

 •    time  leaching beings;

 •    duration of leaching;

 §    path  length;

 •    pore  velocity of water; and

 t    dispersion  coefficient.

 CODE OUTPUT:   The  output of the  code  is the rate  of discharge of  each
 nuclide.   Digital  and graphic  output  is printed  and  results  are written to a
 file that  can  be read by the biosphere code FOOD.   Peak  discharge rates are
 reported for  key nuclides.

 COMPILATION REQUIREMENTS:   GETOUT  is written  in  FORTRAN  IV and  implemented on
 a UNIVAC-1100/44 EXEC-8  system.  It has been  converted to CDC equipment.

 EXPERIENCE  REQUIREMENTS:  Extensive.

TIME REQUIREMENTS:  Months.
                                      A-10

-------
CODE VERIFICATION:  Verification work by E. L. J. Rosinger at Whiteshell
(ref. 4) uncovered problems in early unpublished versions of the code and
confirmed the accuracy of the published version by comparison with analytic
solutions.  The code is in the public domain.

DOCUMENTATION/REFERENCES:

DeMier, W.V.; Cloninger, M.O.; Burkholder, H.C.; Liddell, P.O.  (1979)
     GETOUT -- a computer program for predicting radionuclide decay chain
     transport through geologic media.  PNL-2970, Pacific Northwest
     Laboratories.

Burkholder, H.C.; Cloninger, M.O.; Baker, D.A.; Jansen, 6.  (1976)  Incentives
     for partitioning high-level waste.  Nuclear Technology 31:150, also as
     BNWL-1927, Pacific NorthwestLaboratories Report BNWL-1927.

Lester,'D.-.H.; Jansen, G.; Burkholder, H.C.   (1975)  Migrationofradionucli.de
     chains through an; adsorbing medium.  AIChE Symposium Series 152,
     Adsorption and Ion Exchange 71.

Burkholder, H.C.; Rosinger, E.L.J.   (1980)  Nuclear Technology 49.

Elert, M.; Grundfelt, B.; Stenquist, C.  KBS Teknisk Rapport 79-18.

SOURCE:  GETOUT'was originally written by D. H. Lester, H. C. Burkholder, and
M. 0. Cloninger at,Pacific Northwest Laboratory.  The current FORTRAN IV
version was developed by H. C. Burkholder, M. 0. Cloninger, W. V. DeMier, and
P. J. Liddell. '.:•..

-------
 CODE NAME:   GWMTM1  and  GWMTM2  (Groundwater  Mass  Transport Model  1 Dimensional
 and 2 Dimensional)

 PHYSICAL PROCESSES:   Determines  contaminant concentrations in  the vicinity  of
 a decaying  source under steady-state  flow conditions.

 DIMENSIONALITY:  One  or two  dimensional  depending  upon code used.

 SOLUTION TECHNIQUE:   Analytical.

 DESCRIPTION:   The codes are  deterministic,  one-  or two-dimensional  analytical
 solutions of the transient convective-dispersive mass  transport  equation
 modified for first-order decay,  with  an  exponentially  decaying,  Gaussian
 boundary condition.   The one-dimensional  model is  designed to  solve for
 vertical  infiltration of wastewater through saturated  or  unsaturated soil media
 under constant vertical  seepage  velocity. Presumably it could  be used to model
 one-dimensional horizontal transport  also under  steady-state flow conditions.
 The two-dimensional code is  designed  for estimating the two-dimensional (area!
 or vertical  cross section) concentration pattern downgradient  from sanitary
 landfills,  wastewater lagoons, or other  groundwater pollution  sources.

 CODE INPUT:   GWMTM1:  Dispersion coefficient, kinetic  decay constant,
 constant  seepage velocity, and surface constant  (if surface concentration is
 not constant), user specified space,  and time positions.

 GWMTM2:   Fewer than ten  cards for parametric information  plus  space and time
 positions for desired concentration calculations.
CODE OUTPUT:
and time.
Concentrations are printed at user-specified locations in space
COMPILATION REQUIREMENTS:  The model is written in standard FORTRAN IV and
has been run on an S/360/91; it should run on any standard digital computer.
The program requires a region size of approximately 100K on the S/360/91.  One
can learn to run the model in less than a half hour and only four FORTRAN
statements need to be punched (space and time positions are specified as data
cards).  Setup time is insignificant and FORTRAN programming knowledge is
unnecessary.  It has also been run on minicomputers using less than 100K of
core.

EXPERIENCE REQUIREMENTS:  Minimal.

TIME REQUIREMENTS:  Days.

CODE VERIFICATION:  The two-dimensional  version has been "used to check the
numerical accuracy of several solution schemes of two-dimensional, numerical
models of groundwater quality.  It has been distributed widely through short
courses dealing with groundwater pollution and has been used principally to
simulate leachate plumes from landfills  and check the accuracy of
two-dimensional, numerical models."
                                      A-12

-------
DOCUMENTATION/REFERENCES:
     Professor Robert W. deary
   r  P. 0. Box 2010
     Princeton, NJ  08540     ..-• ,          ,        ;         ,
Cleary, R.W.  (1977)  Final 208 report to the Naussau-Suffolk  Regional  Planning
     Board, Hauppauge, New York.
SOURCE:  Bob Cleary, Princeton, NJ.
                                       A-13

-------
 CODE NAME:   NUTRAN
 PHYSICAL PROCESSES:   NUTRAN calculates the consequences  (in  terms  of releases
 of radioactivity or  doses to humans)  of groundwater releases of radioactivity
 from a repository.   NUTRAN evaluates  the combined effect  of  systems  of natural
 and engineered barriers;  some barriers are modeled in detail  and others are
 simply characterized by a number summarizing  their performance  (e.g.,  for  a
 canister,  lifetime).
 DIMENSIONALITY:   One,  two, or three dimensional.
 SOLUTION TECHNIQUE:   Analytical.
 DESCRIPTION:   The principal  phenomena treated by NUTRAN are:
 t     the resaturation of the repository cavity with water;
 t     leaching  of the waste matrix;
 •     dissolution of  the radioactive elements  in the waste;
 •     diffusion through clay cylinders around  waste canisters;
 •     transport of waste by groundwater through the repository,  surrounding
      strata, and adjacent aquifers  (calculated using a network  of
      one-dimensional  flow paths,  with a two-dimensional method  used  for
      aquifers  containing  wells);
 •     withdrawal  of contaminated groundwater through wells;
 a     transport of waste in surface  waters  and  associated  ecosystems; and
 •     human exposure  and dose  mechanisms.
 NUTRAN  performs  most of the  functions involved in analyzing  long-term  effects
 of  a  waste repository.  NUTRAN 'contains  only an extremely simple model  for
 groundwater flow.  In  many cases  the  results of a flow code  such as VTT must be
 used  to prepare  the  inputs to NUTRAN.  Among these cases are those in which'
 two-  or three-dimensional  effects are important or thermal convection or large
 density gradients are  present.
The code is based on representing the flow field as a network of
 one-dimensional  path segments.  Arbitrary numbers and configurations of path
 segments can be  accommodated.  Simple  models of several  of the engineered
 barriers in the  repository and waste  package are also included.
The principal  assumptions  of the code  are:
t     the one-dimensional solute transport equation is valid within  each path
      segment;
•     sorption may be represented as equilibrium adsorption;
•     once a canister has any holes  in  it, it disappears  entirely as a barrier
     to waste dissolution;
                                      A-14

-------
•    when a clay buffer cylinder is present, wastes diffuse through it in
     approximately a steady state; and
•    water in waste storage rooms is well mixed,
CODE INPUT:  NUTRAN permits the user to divide nuclides into classes with
identical sorption behavior.  It also permits "states" with different water
flow to occur over time.  The principal inputs to WASTE are as follows:
•    retardation factor of each class of nuclides in each leg in each state;
•    cross-sectional area of each Teg in each state;
•    longitudinal dispersivity of each leg;
t    hydraulic conductivity of each leg in each state;
•    heads at leg junctions in each state;
•    length of each leg;
•    effective porosity of each leg in each state;  '
•    repository dimensions and backfill characteristics;
•    age of wastes  at repository commissioning;
•    canister lifetime;
•    rates or times of transitions among states;
9    solubilities of any nuclides  (optional);
•    locations and  pumping rates of wells;  and
•    transverse dispersivity of aquifer from which wells draw.
The  inputs to PLOT, other than disk files written by ORIGEN, BIODOSE,  and
WASTE,  usually consist only of control variables.
CODE OUTPUT:  The primary output of NUTRAN  is the rate at which radioactivity
is released or the  dose to individuals or populations.  Both totals due to all
nuclides and the contributions of  any number of individual nuclides selected by
the  user are available.  Doses or  release rates are given as functions of time
with peak values identified; release rates  and population doses may also be
integrated over all time.  Both digital and graphic output may be obtained.
A variety of intermediate quantities used in the calculations may also be
output.
COMPILATION REQUIREMENTS:  The code is written in PL/I and has been run on a
number  of  IBM machines.  The user  must supply IMSL  (International Mathematical
Statistical Language) routines and, if graphical output is desired, the DISSPLA
plotting package is used.  The test cases in the User's Guide require  2 to 3
minutes CPU time on an  IBM 370/3031.  BIODOSE requires one megabyte of core;
                                      A-15

-------
 WASTE and PLOT require 512 K.   A small  amount of disk storage is required  for
 communication among the programs.

 TASC  has  not released the program to the public.

 EXPERIENCE REQUIREMENTS:  Extensive.       '

 TIME  REQUIREMENTS:   Months.              '
                        '     '     ••  -•    "    -  -         •   •           •   '"''',', 'J
 CODE  VERIFICATION:   NUTRAN has  been  compared to  GETOUT and  BIOPATH for a
 one-dimensional  problem (ref.  1),  and to a number of analytic solutions.

 DOCUMENTATION/REFERENCES:    '  '             .                                 .

 Ross,  B.;  Koplik, C.M.;  Giuffre,  M.S.;,  Hodgin, S.P.   (1981)   A computer model
      of long-term hazards =frorh  waste repositories.   Radioactive Waste
      Management  1:325-338.                                         .   '   ;,

 Ross,  B.,  Koplik, C.M.;  Giuffre,  M.S.;  Hodgin, S.P.;  Duffy,  J.J.   (1979)
      NUTRAN:   a  computer model  of long-term hazards  from  waste repositories.
      The  Analytic Science Corporation Report UCRL-15150.

 Ross,  B,;  Koplik, C.M.   (1979)   A new numerical method for  solving the solute
      transport equation.   Water Resour.  Res.  15:949-55.

 Giuffre,  M.S.; Ross,  B.  .(1979)   The effect of retardation  factors on
      radionuclide migration.   In  G.  J.  McCarthy,  ed.,  Scientific Basis  for
      Nuclear Waste  Management  1:439-442,  Plenum,  NY,  and  Longon.

 Ross,  B ;  Koplik, C.M.   (1978)   A  statistical approach to modeling transport of
      pollutants  in  ground water.   Mathematical Geology 10:657-672.

 Berman, I.E.;  Ensminger,  D.A.;  Giuffre,  M.S.; Koplik,  C.M.;  pston,  S.G.;
      Pollak,  G.D.;  Ross,  B.I.   (1978)   Analysis of some nuclear waste
     management  options.   The Analytic  Sciences Corporation  Report UCRL-13917.

 Ross,  B.;  Koplik, C.M.;  Giuffre, M.S.;  Hodgin, S.P.;  Duffy,  J.J.;  Nalbandian,
     J.Y.   (1980)   User's  guide to NUTRAN:  .a computer analysis  system  for
     long-term repository  safety.  Atomic  Energy  of Canada^  Ltd.,  Technical
     Report AECL-TR-121.

SOURCE:  The code was developed by B. Ross,  C. M. Koplik, M.  S.  Giuffre, J.
J. Duffy, S. P. Hodgih; and others at the Analytic Sciences  Corporation.
                                      A-16

-------
CODE NAME:  NWFT/DVM (Network Flow and Transport/ Distributed Velocity
Method)

PHYSICAL PROCESSES:  Predicts fluid flow and transport of radionuclide
chains.  The rate at which nuclides enter groundwater can be limited by both
leach rates and equilibrium solubility.  Water velocities can reflect density
forces caused by nonuniformities in salt concentrations.

DIMENSIONALITY:  One dimensional.

SOLUTION TECHNIQUE:  Numerical convolution of analytical solution for a
discretized flow network; utilizes the "distributed velocity method."

DESCRIPTION:  NWFT/DVM uses a network of one-dimensional flow paths.  A
steady-state water velocity is calculated in each path, with pressures and
brine densities considered.  Radionuclides enter groundwater at a rate
determined by the combined effects of kinetic leaching and equilibrium
solubility.  Transport of radionuclides can be handled by either an analytic
solution (similar to GETOUT and NUTRAN) within each path segment (or "leg") or
by the Distributed Velocity Method (DVM) which is unique to this code.
Daughter nuclides whose velocity differs from their parents' must be treated by
DVM.

NWFT/DVM is a far-field code.  It is designed to analyze repositories in well
stratified sedimentary.rocks.  NWFT/DVM is finite-difference ,or finite-element
code.                          .                •

The principal assumptions of the code are:

•    fluid flow proceeds only along a specified network of 15 path  segments;

t    Darcy's Law  is valid;

•    the one-dimensional solute transport equation  is  valid within  each path
    ... segment;
      I   ,- •   -    !  ••••••:  '.-..,•   ' -   '-'.-.•   . ' " • '    .  • . '
t    thermal convection can be neglected;

•    all water .flowing through the repository contacts the waste;

•    brine concentration and pressures do not change  over time; and

t    sorption can  be represented as equilibrium adsorption.

CODE INPUT:  Inputs for NWFT/DVM include:

•    conductivity  in each leg;

t    cross-sectional area of each leg;

•    elevation of  each node;

•    porosity of  each leg;

•    rock density  in each leg  (optional);
                                      A-17

-------
 •    brine concentration in each leg;
 t    mass of each  nuclide;
 t    half life of  each nuclide;
 t    initial  inventory of each nuclide;
 •    distribution  coefficient of each nuclide  in  each  leg;
 •    leach time;
 •    dispersion constant (same value  everywhere);
 t    solubility of each nuclide;
 •    time leaching beginnings;  and
 •    cutoff time.
 CODE  OUTPUT:   The  principal  output  is the discharge rate of each nuclide  in
 Ci/day  as a function of time.  The  total integrated discharge and the peak
 discharge rate are also given.  A variety of intermediate quantities can  be
 output  as well.
 COMPILATION REQUIREMENTS:   The code is written in FORTRAN.  The code uses one
 IMSL  (International Mathematical  Statistical Library)  routine to solve a  system
 of linear equations.   This  routine  or a substitute must be supplied by the
 user.
 EXPERIENCE REQUIREMENTS:  Extensive.
 TIME  REQUIREMENTS:  Months.
 CODE  VERIFICATION:  Results  from  NWFT/DVM have been compared with results of
 GETOUT  for a range  of  one-path-segment problems.  Also, it has been compared to
 SWIFT for a problem with a  six-member decay chain.  The code is in the public
 domain.
 DOCUMENTATION/REFERENCES:
 Campbell,  J.E.; Longsine, D.E.; Cranwell, R.M.  (1981)  Risk methodology for
      geologic  disposal  of radioactive waste:  the NWFT/DVM computer code user's
      manual.   Sandia National Laboratories Report NUREG/CR-2081.
 NWFT/DVM  lecture notes.  Draft, Sandia National Laboratories.
 Campbell,  J.E.; Longsine, D.E.; Reeves, M.   (to be published)  The distributed
     velocity method of solving the convective-dispersion equation.
SOURCE:  The NWFT/DVM model was developed at Sandia National  Laboratories.
The original NWFT model was developed at Sandia and INTERA Environmental
Consultants by Campbell, Kaestner, Langkopf, and Canty.  NWFT/DVM incorporates
the DVM, developed  by Campbell and Longsine of Sandia and Reeves of INTERA.
                                      A-18

-------
NUMERICAL FLOW (SATURATED/UNSATURATED)
                  A.19

-------
 CODE NAME:  FEMWATER1

 PHYSICAL PROCESSES:  Predicts groundwater flow through saturated or
 unsaturated porous media.

 DIMENSIONALITY:  Two dimensional  (x-y,  x-z cartesian).

 SOLUTION TECHNIQUE:  Numerical,  finite  element,  quadrilateral  element.

 DESCRIPTION:  FEMWATER1 is a revision of the Yeh and  Ward  subsurface flow
 model.   It is a flow-only  code that creates a data  file specifically for the
 operation of the FEMWASTE1 pollutant transport model.   FEMWATER1 applies to
 flow in porous media which is:

 t    transient or steady state;

 •    two dimensional  (horizontal  or vertical  cross  section); and

 •    saturated-unsaturated.

 The  model  is based on the  conservation  of mass and  momentum and  includes soil
 and  water compressibility  effects.   Water is  exchanged between the  surface and
 subsurface media by:

 t    seepage or ponding;

 •    infiltrating runoff from rainfall;

 •    artificial  recharge and  withdrawal;  and

 •    ponds,  lakes,  and  streams.

 FEMWATER1  uses  quadrilateral  bilinear and  triangular finite elements to
 represent  the two-dimensional  porous media  domain.  The Galerkin method ;of
 weighted residuals  is used to  solve the continuity equation and  Richards'
 equation.   Finite difference discretization of the time derivatives can  be
 specified  in one  of three ways:

 1)    backward difference;

 2)    central difference; or                                ;

 3)    mid-difference.

 The mid-difference  time stepping scheme was added to the previously available
 options of central  and backward differencing schemes.   The method of
mid-differencing  assumes linear variation of the unknown variable over the time
 interval such that  the computer matrices are evaluated at the midpoint of the
time  interval.                                                    "     ,

With  the three time stepping and two mass Tumping options,  six numerical      :••
solution procedures are available in FEMWATER1.  Other than the general
recommendations above, no discussion of the stability  or accuracy
characteristics of the various solutions is presented.      .:5

                                      A-20

-------
Soil properties and hydraulic conductivities in FEMWATER1 are functions of
pressure head alone.  Because pressure head is the dependent variable, the
equation set to be solved is nonlinear, requiring iterative solution
techniques.  The Newton-Raphson iteration with the option of over- and
under-relaxation is used.  For large problems, a line successive
over-relaxation technique is available.

Unlike most flow models that calculate the velocity field by taking derivatives
of the solved pressure field, FEMWATER1 formulates Darcy's Law with the finite
element solution method.  This approach has the advantage of continuous
velocity distributions at the model boundaries.

Mass balance computations are built into FEMWATER1 to monitor numerical error
generation.

The main assumptions of the code are:

•    Darcy's Law is valid and hydraulic-head gradients are the only significant
     driving mechanisms for fluid flow; and

•    the nonlinear soil properties and the hydraulic conductivity are functions
     of the pressure head only.

CODE INPUT:  Inputs to FEMWATER1 include:

•    grid geometry;

•    initial heads;

•    prescribed head and flux boundary conditions;

t    hydraulic-conductivity tensor;

t    modified storage coefficient; and

a    material nonlinearities.

CODE OUTPUT:  The output from FEMWATER1 consists of the pressure distribution
and velocity field at each time step.

COMPILATION REQUIREMENTS:  FEMWATER1 is written in FORTRAN and was originally
installed on an IBM 360 machine.  The code is generally compatible with most
all mainframe or virtual memory machines.

EXPERIENCE REQUIREMENTS:  Extensive.

TIME REQUIREMENTS:  Months.

CODE VERIFICATION:  Two sample problems to which solutions had previously
been obtained by other validated numerical models, namely 1) the seepage pond
problem and 2) the Freeze transient problem, were solved.  In addition, results
by all six alternative numerical schemes discussed below in the section on
numerical approximations were compared in both examples.  No field validation
has been performed.

                                      A-21

-------
DOCUMENTATION/REFERENCES:

Yeh, G.T.; Ward, D.S.  (1979)  FEMWATER:  a finite-element model of water flow
     through saturated-unsatureted porous media.  Oak Ridge National Laboratory
     Report ORNL-5567.

Reeves, M.; Duguid, J. 0.  (1976)  Water movement through saturated-
     unsaturated porous media:  a finite-element Galerkin model.  Oak Ridge
     National Laboratory Report ORNL-4927.
SOURCE:  FEMWATER1 was developed at Oak Ridge National Laboratory by G. T.
Yeh and D. S. Ward.  It is an extension of work done by Reeves and Duguid.
code is in the public domain.
The
                                      A-22

-------
CODE MAME:  FREEZE

PHYSICAL PROCESSES:  Predicts water flow in a groundwater basin under
transient saturated-unsaturated conditions.

DIMENSIONALITY:  Three dimensional.

SOLUTION TECHNIQUE:  Numerical, finite difference.

DESCRIPTION:  A three-dimensional finite-difference model has been developed
for saturated-unsaturated transient flow in small, nonhomogeneous, anisotropic
geologic basins.  The uniqueness of the model lies in its inclusion of the
unsaturated zone in a basin-wide model that can also handle both confined and
unconfined saturated aquifers under both natural and developed conditions.  The
integrated equation of flow  is solved by the line successive overrelaxation
technique.  The model allows any generalized region shape and any configuration
of time-variant boundary conditions.  When applied to natural flow systems, the
model provides quantitative  hydrographs of surface infiltration, groundwater
recharge, water table depth, and stream base flow.

Results of simulations for hypothetical basins provide insight into the
mechanisms involved in the development of perched water tables.  The
unsaturated basin response is identified as the controlling factor in
determining the nature of the base flow hydrograph.  Application of the model
to developed basins allows one to simulate not only the manner in which
groundwater withdrawals are  transmitted through the aquifer, but also the
changes in the rates of groundwater recharge and discharge induced by the
withdrawals.  For any proposed pumping pattern, it is possible to predict the
maximum basin yield that can be sustained by a flow system in equilibrium with
the recharge-discharge characteristics of the basin.

CODE INPUT:  Inputs consist of standard unsaturated or saturated zone
hydrologic parameters.

CODE OUTPUT:  Output consists of a pressure or head distribution.

COMPILATION REQUIREMENTS:  FREEZE is written in FORTRAN and is operational on
an IBM system.

EXPERIENCE REQUIREMENTS:  Extensive.

TIME REQUIREMENTS:  Months.

CODE VERIFICATION:  Unknown.

DOCUMENTATION/REFERENCES:

Freeze, R.A.  (1971)  Three-dimensional  transient saturated-unsaturated flow in
     a ground water basin.  Water Resour.  Res.  7(2):347-366.

SOURCE:  This code was developed by R. A.  Freeze of the University of British
Columbia.  The code is in the public domain.
                                      A-23

-------
CODE NAME:  UNSAT1
PHYSICAL PROCESSES:  Provides a simulation capability for flow in a
one-dimensional unsaturated or saturated soil column.
DIMENSIONALITY:  One dimensional.
SOLUTION TECHNIQUE:  Numerical, finite element.
DESCRIPTION:  The UNSAT1 model provides numerical solutions to the
one-dimensional flow problem of an unsaturated-saturated medium.  It employs a
Galerkin finite-element analysis of this one-dimensional physical domain.  The
hermite cubic polynomial is used as the basis for the finite-element
approximation to the continuum.
CODE INPUT:  Inputs consist of standard unsaturated or saturated zone
hydro!ogic parameters..
CODE OUTPUT:  Output consists of a pressure or head distribution.
COMPILATION REQUIREMENTS:  UNSAT1 is written in FORTRAN IV.
EXPERIENCE REQUIREMENTS:  Extensive.
TIME REQUIREMENTS:  Months.
CODE VERIFICATION:  Unknown.
DOCUMENTATION/REFERENCES:
van Genuchten, M.Th.   Numerical solutions of the one-dimensional
     saturated-unsaturated flow equation.  78-WR-09, Princeton University,
     Princeton, NJ.
SOURCE:  This code was written by M. Th. van Genuchten of the U.S. Salinity
Laboratory.  The code is in the public domain.
                                      A-24

-------
CODE NAME:  UNSAT2
PHYSICAL PROCESSES:
porous media.
    Simulates nonsteady seepage in saturated or unsaturated
DIMENSIONALITY:
cylindrical).
Two dimensional  (x-y or x-z cartesian),  Ax-isymmetric (r-z
SOLUTION TECHNIQUE:;NUmericaK finite element, triangular: elements,

DESCRIPTION:  The program can be used to investigate problems involving two
spatial dimensions in the horizontal or vertical plane. Three-dimensional
problems can be handled provided that the flow pattern retains an axial
symmetry about the vertical coordinate.  Jhe flow region may have any complex
shape and it may consist of different soil materials arranged in arbitrary
patterns.  Each soil material may exhibit an arbitrary degree of local
anisotropy with the  principal hydrauljc conductivities oriented.at any desired
angle with respect to the coordinates.                    ...--^

A wide range of time-dependent boundary conditions can be treated: prescribed
pressure head; prescribed flux normal to the boundary;, seepage faces arid  .
evaporation and infiltration boundaries where the maximum rate of flux is
prescribed by atmospheric or other external conditions while the.actual rate is
initially unknown.   In addition, the program can handle water uptake by plants
assuming that the maximum rate of transpiration is determined by atmospheric ,
conditions while the actual rate of uptake depends on atmospheric as well as
soil and plant conditions.  Internal volumetric sinks or sources of prescribed
strength can be included in the flow system at any stage of the computation:  A
special provision has been made for the.analysis of axisymmetric flow to a well
of finite radius partially penetrating an unconfined aquifer,system ind
discharging at a prescribed time-dependent rate.  The well may be partially
cased and its capacity for storing water is taken into account.  Several layers
can be tapped by the well at the same time.

The domain simulated by UNSAT2 is discretized using three-noded triangular
elements.  The time  domain is discretized by finite-difference techniques.

The main assumptions of the code are:

•    Darcy's Law is  valid and hydraulic-head gradients are the only significant
     driving mechanism for fluid flow;

•    the rate of uptake by roots is proportional to the pressure head gradient
     across the soil-root interface;

•    there is no hysteresis in the water retention or relative permeability
     curves; and

•    the relative permeability and capillary pressures are functions of
     moisture content.

CODE INPUT:   Inputs to UNSAT2 include:

t    grid geometry;

                                      A-25

-------
 •    initial  heads (total  head or pressure head);

 •    boundary conditions;

 »    rates of infiltration,  evaporation,  and transpiration;

 •    root effectiveness function;

 •    hydraulic conductivity  tensor;                                 .   ..-  i '

 •    relative permeability function;

 •    capillary pressure function;  and

 •    storage  coefficient.

 CODE OUTPUT:   The printed  output  of the program consists  of  a listing of all  "
 input information,  a  complete  description of the finite-element  network,: the
 boundary codes of all  nodes, and  the properties of each material.   During each
 time step the program prints a listing of total  head  values,  pressure head
 values,  moisture  content values,  and discharge  into or out of the  system (not
 flow through  the  system) at  all nodes.  The  rate of convergence  of the
 iterative procedure is printed during each time step  together with additional
 information pertaining to  the  particular  problem at hand.

 COMPILATION REQUIREMENTS:  UNSAT2  is written in  FORTRAN IV and was originally
 installed on  an IBM 370/165 machine.  The code  is  generally  compatible with
 most all  mainframe  or  virtual  memory machines.

 EXPERIENCE REQUIREMENTS:   Extensive.

 TIME REQUIREMENTS:  Months.

 CODE VERIFICATION:  Due  to the lack of analytic  solutions to  problems  of flow
 in the unsaturated  zone  at the time UNSAT2 was developed, verification  could
 only be  performed by comparing the results to results  obtained from other,
 previously validated,  numerical models.   The effect of vertical  flow  in  the
 presence  of evapotranspiration was simulated using  both a finite-difference
 code  and  the  finite-element code, UNSAT2.   The results were compared.

 Two  field  problems were  simulated by UNSAT2  as reported in Ref.  1.  These  are:
 1) a  field experiment  performed by Feddes  (ref.  2)  at  the groundwater  level
 experimental  field Geestmerambacht in the Netherlands, and 2) a  field
 experiment taken from  the subirrigation experimental field "De Groeve"  in  the
Netherlands (ref.  3).

DOCUMENTATION/REFERENCES:

Neuman, S.P.;  Feddes,   R.A.; Bresler,  E.    (1974)  Finite element  simulation of
     flow in saturated-unsaturated soils  considering water uptake by plants.
     Technion, Hydrodynamics, and Hydraulic Eng. Laboratory Report, July.

Feddes, R.A.   (1971)  Water,  heat, and  crop growth.  Thesis Comm. Agric. Univ.
     Wageningen 71-12, pp.  184.

                                     A-26

-------
Feddes, R.A.; van Steenbergen, M.G.  (1973)  Sub-irrigation field  'De
     Groeve'."  Mpta 735, Inst. for Land and Water Management Res., Wageningen,
     pp. 184.

Feddes, R.A.; Bresler, E.; Neuman, S.P.  (1974)  Field test of an  improved
     numerical model for water uptake by root systems.  Unpublished manuscript.

Hanks, R.J.; Klute, A.; Bresler, E.  (1969)  A numeric method for  estimating
     infiltration, redistribution, drainage, and evaporation of water from
     soil.  Water Resour. Res. 5(5):1064-1069.

Neuman, S.P.  (1972b)  Finite element computer programs for flow in
     saturated-unsaturated porous media.  Second Annual Report, Project No.
     A10-SWC-77, Hydraulic Engineering Laboratory, Technion, Haifa, Israel,
     pp. 87.

Neuman, S.P.  (1973)  Saturated seepage by finite elements.  Prpc. ASCE, J.
     Hydraul. Division, 99(HY12):2233-2250.

SOURCE:  UNSAT2 was developed at Technion,  Israel Institute of Technology, by
S. P. Neuman, R. A. Feddes,  and E. Bresler.  The code is in the public domain.
                                     A-27

-------
NUMERICAL FLOW (SATURATED),
           A-28

-------
CODE NAME:   BEWTA

PHYSICAL  PROCESSES:  Predicts the  Boussinseq  equation for a two-dimensional
water table  aquifer.

DIMENSIONALITY:  Two dimensional.

SOLUTION  TECHNIQUE:  Numerical,  finite-difference.

DESCRIPTION:  The aquifer  is given by  a  two-dimensional discrete
representation in the horizontal plane.   Its  boundary is irregular and mixed,
and it  is superimposed by  a meandering stream.  Aquifer parameters are assigned
to each nodal point and an be slightly anisptropic and nonhomogeneous.
Transient groundwater flow through a water  table aquifer is described by the
Boussinesq equation, which assumes that  the Pupuit-Forchheimer assumptions are
valid.  The  principal conductivity components are colinear with the coordinate
system X  and Y.  The water is released from storage mainly by gravity drainage
with an instantaneous decline in head.   The stream bed is semi-confined, and
the leakage  obeys Darcy's  Law.   The flow from the unsaturated zone and the
change  in fluid density over time  are  negligible.  Replenishment of the aquifer
system occurs through gravity drainage and  stream bed leakage.

The model simulates transient groundwater flow through a two-dimensional water
table aquifer using the alternative direction implicit method.  The Boussinesq
equation  is  approximated by a two-dimensional finite-difference equation of
linearized form, employing the noniterative alternat'in9 direction implicit
method.   The equation is written for two half-time steps, and is done for each
node along either a row or a column,   The resulting simultanepus equations are
solved by the Thomas algorithm.  The calculations are carried on successions!ly
row by row for the first half-time step  and column by column for the second
half-time step (where results are  approximations to the transient solution of
the problem under discussion).

CODE INPUT:  Standard saturated zone hydrologic parameters.

CODE OUTPUT:  Head distribution at  each  time  step.

COMPILATION REQUIREMENTS:  BEWTA is written in FORTRAN.

EXPERIENCE REQUIREMENTS:  Extensive.

TIME REQUIREMENTS:  Months.

CODE VERIFICATION:  Unknown.

DOCUMENTATION/REFERENCES:  Not available,

SOURCE:   BEWTA was developed  by Chang L-  Lin of the Nova Scotia Department of
the Environment.
                                      A-29

-------
CODE NAME:  COOLEY

PHYSICAL PROCESSES:  Predicts the transient or steady-state hydraulic head
distribution and velocity flow field in a confined, semi confined, or unconfined
aquifer.

DIMENSIONALITY:  Two-dimensional (x-y or x-z cartesian, r-z radial).

SOLUTION TECHNIQUE:  Numerical, finite element.

DESCRIPTION:  The code is applicable to confined, semi confined, or unconfined
flow problems which obey the generalized Boussinesq equation. Flow may be
steady or nonsteady.  The aquifer may be given an area! (plan view) or radial
description or may be cross sectional.  Aquifer parameters may be distributed
or zoned and the system may be anisotropic with the principal components
aligned with the global coordinate axes.  In the
area! description, if used, leakage from confining beds (or river bottoms,
etc.) is vertical and storage in the confining beds is neglected.  The
Dupuit-Forchheimer assumptions and delayed (or no delayed) yield constant are
used for the water table case in plan view.  The basic discretization method is
the subdomain finite-element method with the time discretized by a weighted
average technique.

COOLEY is best suited for a single layer aquifer system.  A layered aquifer
system, however, may be analyzed within using the radial coordinate system.
The code represents a two-dimensional area, whether area!, plane cross
sectional, or axisymmetric cross sectional, as a series of zones.  The zone
shapes may be triangular or quadrilateral and of nearly any convex shape except
that the boundaries must pass through, and not between, all nodes.

There are three basic versions of the program, the differences among them being
the methods used to solve the matrix equation.  The choice depends on the site.

COOLEY is most applicable to porous media.  Fractured media can  also be modeled
if the fractures are sufficiently numerous that they can be approximated by  a
porous medium.

The principal numerical approximations in COOLEY are the following:

•    Discretization of space by the finite-element method of using the
     "subdomain collocation" version of the weighted residual method (ref. 3,
     p.40).

•    Discretization of time by the explicit Euler forward difference, the
     semi-implicit Crank-Nicholson central difference, or the fully implicit
     backward difference scheme.

The main assumptions of the code are:

•    Darcy's Law is valid and hydraulic-head gradients are the only significant
     driving mechanisms for fluid flow.

•    The porosity and, hydraulic conductivity of the aquifer are  constant with
     time.

                                      A-30

-------
t    Gradients of fluid density, viscosity, and temperature do not affect the
     velocity distribution.
•    Assumptions analogous to the Dupuit-Forchheimer assumptions are used for
     unconfined aquifers.
•    The transmissivity tensor has principal axes parallel to the coordinate
     axes.
CODE INPUT:  Inputs to COOLEY include:
t    grid geometry;
•    prescribed head and flux boundary conditions;
•    transmissivity in both x and y (or r and z) directions;
0    storage coefficient or specific yield for area! problems and specific
     storage for cross-sectional or radial-flow problems;
•    hydraulic conductance for an adjacent aquitard for area! flow problems;
•    known recharge or discharge rates; and
•    initial hydraulic-head distribution.
CODE OUTPUT:  The output from COOLEY consists of the pressure distribution
and velocity field at each time step.
COMPILATION REQUIREMENTS:  The basic programs are dimensioned such that the
maximum mesh size is 50 by 50 (2500) nodes, and the maximum number of time
steps is 100.  In addition, the maximum number of iterations for program LSOR
is 100.  With these dimensions, LSOR and ADIPIT occupy about 25,000 words and
SIP occupies about 32,500 words of core on a CDC 6400 computer.  However, the
dimensions can easily be modified to accommodate other problem sizes and
smaller or larger computers.  No special library functions or subroutines are
used, and the only peripheral equipment needed is a card reader.  The final
versions of the program were tested on a CDC 64000 computer.  COOLEY is
programmed in FORTRAN IV.
EXPERIENCE REQUIREMENTS:  Extensive.
TIME REQUIREMENTS:  Months.
CODE VERIFICATION:  The code was verified for four problems to which
analytical solutions were available.
Field validation runs simulated the influence of seasonal pumping of irrigation
wells on groundwater levels in Ash Meadows, California,  and Nevada.
DOCUMENTATION/REFERENCES:
Cooley, R.L.  (1974)  Finite element solutions for the equations of groundwater
     flow.  Nevada University.
                                      A-31

-------
Bear, J.  (1972)  Dynamics of fluids in porous media.  New  York,.NY:  .American
     Elsevier.                                              •-...'.. v  ,-.-1


Zlenkiewicz, O.C.  (1971)  the finite element method of engineering1science!
     London:  McGraw-Hill.             '             '               •     ;;

SOURCE:  COOLEY was developed at the Center for Water Resources Research,
Desert Research Institute, University of Nevada, by R. L. Cooley.  The code  is
in the public domain.
                                     A-32

-------
CODE NAME:  FE3DGW

PHYSICAL PROCESSES:  Saturated groundwater flow in a homogeneous or
heterogeneous geological system.  The code provides water flow paths and travel
times.

DIMENSIONALITY:  Three-dimens'iorial, radial flow (r-z)l "'.[.,'':'._ / './",.

SOLUTION TECHNIQUE:  Numerical, finite element quadrilateral elements.

DESCRIPTION:  FE3DGW is a set of codes designed to be executed in uncoupled
stages.  This allows the user to verify the accuracy of the input data before a
full simulation is made, while storing information on disk for access by the
main program.  A plotting package is included in the data preprocessing stage
to display node and element locations and vertical logs of the hydrogeologic
strata.  Contour plots of potential surfaces and strata interfaces are produced
in three-dimensional projections to help identify obvious errors.  Once the
model parameters have been stored successfully on disk,  they need not be
processed again unless the problem configuration is changed.  This is because
boundary conditions are entered in a separate stage of the data storage
process.

The equation set is linear with spatially varying conductivities. Convergence
will be perfect if the integration of the element volumes is exact.  A Gaussian
quadrature scheme using two, three, or five integration points is used for the
numerical integration.  A necessary condition for convergence is that the basis
functions used in the Galerkin technique must be defined such that constant
values of any of the first derivatives are available throughout the element
when suitable nodal values of potential are assigned.  Backward differencing of
the time derivatives is used with the initial time step, and central
differencing is used subsequently.

The semiconfining layers, which are represented by the finite elements, are
simultaneously solved for changes in pressure by a fully three-dimensional
treatment of the nodal equations.  EQSOLV is the matrix inversion algorithm
used in FE3DGW to solve the large, sparse, nonbanded system of equations.  In
this algorithm, row pivoting is about the minimum nonzero element; whereas, the
pivot column is about the largest absolute element in the pivot row.

The main assumptions of the code are:

•    Darcy's Law is valid and hydraulic-head gradients are the only significant
     driving mechanism for fluid flow.

•    The porosity and hydraulic conductivity are constant with time.

•    Gradients of fluid density, viscosity,  and temperature do not affect the
     velocity distribution.

•    The storage term is a function of the compressibility of the fluid and
     porous medium only.

•    The medium is fully saturated.
                                      A-33

-------
•    Hydraulic conductivity principal axes are aligned parallel to the
     coordinate axes.

CODE INPUT:  Input requirements include specification of geologic
stratigraphy, liquid and media properties, positioning and rates of manmade
aquifer stresses, boundary conditions at the areal  extent of the aquifer, and
initial conditions for dynamic simulations.  Specifically, these requirements
include:

t    Geologic stratigraphy
          Number of distinct geologic layers
          Total aquifer thickness
          Elevation of geologic media interfaces
          x,y coordinates of elevation data

a    Liquid properties
          Viscosity
          Density                                 ;
          Temperature profile

•    Media properties
          Intrinsic permeability or hydraulic conductivity
          Porosity
          Storativity

a    Manmade stresses
          x,y coordinates of wells
          Screened elevation
          Rate of withdrawal or injection

t    Boundary conditions
          Water levels of hydro!ogically significant water bodies (e.g.,
          streams, lakes, oceans)
          Nodal fluxes due to pumping
          Vertical infiltration
          Lateral recharge defined by data or modeling conducted on a larger
          scale                                               ,,
          Lateral no-flow boundaries defined by groundwater divides or
          impermeable media

t    Initial conditions for dynamic simulations
          Water table elevations in unconfined zones
          Spatial distribution of pressure in confined zones.

Two- and three-dimensional plotting routines are included in the FE3DGW
technology for the purpose of visualizing any obvious errors in the input
data.  Inconsistency in the entered data is also monitored by built-in
consistency checks and diagnostics.

CODE OUTPUT:  Results are in the form of:

§    Flow field;

a    Flow paths; and

                                      A-34

-------
0    Travel times.
Various levels of output detail are available:
•    Geometric description;
•    Number of nodes;
0    Types of nodes;
•    Dimensions;
t    Surface node coordinates;
0    Input data echo; and
0    Output of all nodes and elements at selected time levels.
Written output can be directed to the line printer or disk.  Plotted output is
in the form of contour maps, grid displays, and three-dimensional,
dependent-variable surfaces.
COMPILATION REQUIREMENTS:  FE3DGW is written in FORTRAN IV and is currently
being run on a VAX 11/780.  The code is generally compatible with most all
mainframe or virtual memory machines.
Current program limits are:
0    768 surface nodes;
0    2560 system nodes;
0    20 layers per well log;
0    99 materials;
0    768 potential boundary conditions;
0    128 stream nodes;
0    20 nonzero element nodes;
0    2048 unknown nodes;
0    128 nonzero,element bandwidth;
0    768 surface elements;
0    2000 system elements; and
0    70 time steps
EXPERIENCE REQUIREMENTS:   Extensive.
TIME REQUIREMENTS:  Months.
                                      A-35

-------
 CODE VERIFICATION:  Verification analysis of the three-dimensional model was
 accomplished using two-dimensional or quasi-three-dimensional analytical
 solutions.  These include radial confineo" and leaky aquifer solutions given by
 Theis (ref. 2) and Hantush (ref. 4), respectively, and also the two-dimensional
 analytic solution PATHS (ref. 5).       ,                            ,       ......,.

 FE3DGW has been applied extensively to the grbundwater. system beneath Lpng
 Island,  New York (ref. 6).  The Long Island groundwater basin is one of the
 most intensively monitored systems in the U.S.

 The model has also been applied to the groundwater system at Sutter Basin,
 California, where it has been inferred (ref.  7) that fresh water which is
 recharging at Sutter Buttes rises through the Sutter Basin fault, creating a
 salt-water mound.

 The model has been used on several  other occasions to better understand the
 groundwater flow portion of a contaminant transport application (ref.  8, 9, and
 Ivy •

 DOCUMENTATION/REFERENCES:

 Gupta, S.K.;  Cole,  C.R.; Bond,  F.W.   (1979)   Finite element three dimensional
      ground water (FE3DGW) flow model  formulation,  program listings and user's
      manual.   Pacific  Northwest Laboratory Report PNL-2939.

 Theis, C.V.  (1935)  The relation between the lowering of the piezometric
      surface  and the rate  and duration  of discharge of a  well  using groundwater
      storage.   Trans.  Amer.  Geophys.  Union 2:519-524.

 Jacob, C.E.  (1950)  In:  Rouse,  H.,  ed.  Engineering Hydraulics.
      New York,  NY:   John Wiley and Sons,  pp.  321-386.

 Hantush,  M.S.   (1960)  Modification  of  theory of leaky aquifers.   J. Geophys
      Res.  65:3713-3725.

 Nelson,  R.W.;  Schur, J.A.   (1980)  PATHS  groundwater hydrogeological model
      Pacific  Northwest Laboratory Report  PNL-3162.

 Gupta, S.K.;  Pinder, G.F.   (1978)  Three-dimensional  finite  element model  for
      multilayered ground-water  reservoir  of Long-Island,  New York.   Department
      of  Civil  Engineering,  Princeton University,  Princeton,  NJ.

 Gupta, S.K.; Tanji,  K.K.   (1976)  A three-dimensional  Galerkin  finite  element
      solution  of flow  through multiaquifers in Sutter  Basin,  California.   Water
      Resour. Res. 12(2):155-162.

 Bond, F.W.; Eddy, C.M.   (1985)  Remedial  action modeling  assessment, Western
      Processing  Site,  Kent, Washington.   Prepared for  the U.S.  Environmental
      Protection Agency,  Region X, Seattle, WA.

Schalla,  R.; McKown, G.L.; Meuser, J.M.;  Parkhurst, R.G.; Smith,  C.M.; Bond,
     F.W.;  English, C.J.   (1984)  Source  identification contaminant transport
     simulation and remedial action analysis,  Anniston Army Depot, Anniston,
     Alabama.  Prepared  for Commander, Anniston Army Depot, Anniston, AL.

                                      A-36

-------
Cole, C.R.; Bond, F.W.; Brown, S.M.; Dawson, G.W.  (1983)  Demonstration/
     application of groundwater modeling technology for evaluation of remedial
     action alternatives.  Prepared for the U.S. Environmental Protection
     Agency, Municipal Environmental Research Laboratory, Cincinnati, OH.

SOURCE:  FE3DGW was written by S. K. Gupta, C. R. Cole, and F. W. Bond as a
result of research conducted by Pacific Northwest Laboratory and supported by
the Waste Isolation Safety Assessment Program (WISAP). FE3DGW is a derivative
of DAVIS/FE which was also written by S. K. Gupta.
                                       A-37

-------
 CODE NAME:   FLUMP

 PHYSICAL PROCESSES:   Predicts  two-dimensional groundwater  flow.

 DIMENSIONALITY:   Two  dimensional.

 SOLUTION TECHNIQUE:   Numerical,  finite element, and a mixed explicit-implicit
 point iterative  solution.

 DESCRIPTION:  The FLUMP model  evolved from the TRUST model, although it has
 been  modified considerably by  S. P. Neuman.  It is considerably easier to use
 on groundwater flow problems than is TRUST, although it is restricted to
 two-dimensional  flow  in either vertical or horizontal
 planes.   FLUMP uses a finite-element numerical scheme which has been shown to
 represent only a  small change  from the original approach used in TRUST.

 CODE  INPUT:  Standard saturated zone hydro!ogic parameters.

 CODE OUTPUT:  Head distribution at each time step.

 COMPILATION REQUIREMENTS:  FLUMP is written in FORTRAN and is operational  on
 a CDC system.                                                      :    ,

 EXPERIENCE REQUIREMENTS:  Extensive.

 TIME REQUIREMENTS:  Months.

 CODE VERIFICATION:  Unknown.

 DOCUMENTATION/REFERENCES:

 Fogg,  G.E.;  Simpson,  E.S.; Neuman,  S.P.   (1979)  Aquifer modeling  by  numerical
      methods applied  to  an Arizona  groundwater  basin.   PB  298962,  National
      Technical  Information Service, Springfield, VA,  140 pp.

 Narasimhan.  T.N.; Neuman, S.P.; Edwards, A.L.   (1977)  Mixed  explicit-implicit
      iterative  finite  element scheme  for diffusion-type  problems.   2.   Solution
      strategy and examples.   Int. J.  Numer. Methods Eng. 11:235-244.

 Neuman,  S.P.; Narasimhan, T.N.   (1977.)  Mixed-explicit-implicit  iterative
      finite  element scheme for  diffusion-type problems.  1.  Theory.  Int. J
      Numer.  Methods Eng. 11:309-323.

 Neuman,  S.P.; Narasimhan, T.N.; Witherspoon, P.A.  (1977)   Application  of mixed
      explicit finite element method to nonlinear diffusion  type problems.  In:
      Pinder, G.F.;  Gray, W.E.,  eds. Proceedings of the First International
     Conference on Finite Elements in Water Resources, Princeton, NJ, Pentech,
     pp.  1.153-1.185.

SOURCE:   This code  was developed by S. P. Neuman of the University of
Arizona,  Department of Hydrology.  The code is in  the public domain.
                                      A-38

-------
CODE NAME:  FRESURF 1 & 2

PHYSICAL PROCESSES:  Predicts two-dimensional and axisymmetric flow.

DIMENSIONALITY:  Two dimensional and axisymmetric.

SOLUTION TECHNIQUE:  Finite element.

DESCRIPTION:  In this two-dimensional model, the specifics of the
free-surface boundary as described by Neuman are incorporated into the
finite-element solution scheme used.  The free boundary is handled by the
finite-element network, expanding or contracting to accommodate the movement of
the free surface with time.  The total derivative or the potential, with
respect to time, is combined with specific free-boundary conditions that are
substituted into the generalized variational principle for the saturated flow
problem to give the expression for the free  surface.  This expression is a set
of nonlinear ordinary differential equations.  A Crank-Nicholson time-centered
scheme is used to  solve the nonsteady flow with the free surface.  The regular
finite-element approach, based on the variational principle, provides the
potential distribution below the free surface.

The above features for handling the  free surface and potential distribution are
used in the model  to solve both two-dimensional and axisymmetric flow
problems.  The program also solves for the seepage face, if  one occurs, through
a two-step iterative procedure.  It  would appear that considerable additional
development would  be required to extend the  procedure to a three-dimensional
situation.  Since  it is two dimensional, the predictions are restricted to only
vertical  sections  when the free-surface feature is used.

CODE INPUT:  Standard saturated zone hydro!ogic parameters.

CODE OUTPUT:  Head distribution at each time step.

COMPILATION REQUIREMENTS:  FRESURF 1 and 2 are written in FORTRAN and
operational on a CDC system.

EXPERIENCE REQUIREMENTS:   Extensive.

TIME REQUIREMENTS:  Months.

CODE VERIFICATION:  Unknown.

DOCUMENTATION/REFERENCES:

Neuman, S.P.; Witherspoon, P.A.   (1971)  Analysis of nonsteady flow with a free
     surface using the finite element method.  Water Resources Research,
     7(3):611-623.

SOURCE:   FRESURF  1 and 2 were written by S.  P. Neuman of the University of
Arizona,  Department  of Hydrology.  The code,is in the public domain.
                                       A-39

-------
 CODE NAME:  TERZAGI


 PHYSICAL PROCESSES:  TERZAGI solves for three-dimensional fluid flow with
 one-dimensional consolidation in saturated systems.


 DIMENSIONALITY:  Three dimensional.


 SOLUTION TECHNIQUE:  Numerical, finite difference.


 DESCRIPTION:  This model numerically simulates the movement of water in
 saturated deformable porous media.  The theoretical model considers a general
 three-dimensional  field of flow in conjunction with a one-dimensional vertical
 deformation field.  The governing partial  differential  equation expresses the
 conservation of fluid mass in an elemental volume that has a constant volume of

 Ull j'r JS^nUmu^ua1.  solution is based on the integrated finite difference
 method (IFDM), which is very convenient for handling multi-dimensional
 heterogeneous systems composed of isotropic materials.


 Tm1c?reSe.n$DnS?!puter P^grm is based on modifications  of earlier versions of
 TRUST and TRUMP.


 CODE INPUT:   Standard saturated zone hydro!ogic parameter plus  consolidation
 parameters.


 CODE OUTPUT:   Head distribution,  consolidation  information.


 COMPILATION  REQUIREMENTS:   TERZAGI was  written  in  FORTRAN IV and  implemented
 on  a CDC system.


 EXPERIENCE REQUIREMENTS:   Extensive.


 TIME REQUIREMENTS:  Months.


 CODE VERIFICATION:  Unknown.


 DOCUMENTATION/REFERENCES:                                          ..',-,•


 Narasimhan, T.N.; Witherspoon, P.A.  (1976)  An integrated finite difference
     method for analyzing fluid flow in porous media.  Water Resources Research
     i & (1).                                              .


SOURCE:  The code was developed by^T. N. Narasimhan of the University of
California.   This code is in the public domain.
                                     A-40

-------
CODE NAME:  USGS2D

PHYSICAL PROCESSES:  Saturated groundwater flow in a confined, unconfined, or
combined confined and unconfined aquifer.

DIMENSIONALITY:  Two dimensional (x-y or x-z cartesian).

SOLUTION TECHNIQUE:  Numerical, finite difference.

DESCRIPTION:  USGS2D is a finite-difference saturated flow code. ,It is
restricted to two-dimensional (areal) flow, but has many options.  USGS2D
provides the user with a variety of options for 1) groundwater flow conditions;
2) source terms; 3) numerical solution techniques; and 4) input-output.  The
options for groundwater flow conditions include:

•    confined conditions;

•    unconfined conditions; and           ,

t    combined confined and unconfined conditions.

Variations of source terms include:

•    transient leakage from confining beds;

•    steady leakage from confining beds;  .       '•...._                •

•    recharge;

«    pumping wells; and

•    evapotranspiration.

The  program is fairly general.  The  flow field  is represented  as a
two-dimensional grid.  The size of the grid blocks  is variable to allow  the
desired level of spatial detail.  Variable time steps are also allowed.

The  major variable  is hydraulic head, although drawdowns may  be computed from
the  initial head condition.

The  model was designed primarily for areal simulations  but can be used for some
cross-sectional problems.  Generally this is done using the artesian option.
An example of a cross-sectional simulation is provided  in the  documentation.

A minor problem is  that the code is  programmed  such that two  extra  columns and
two  extra rows are  required to border the gridded area.  This  results  in some
additional work when preparing input data.  It  is also  difficult to convert  the
code to run on non-IBM machines.

USGS2D is designed  for most types of geological media where two-dimensional
flow in a porous medium can be assumed.  The main assumptions  of the code are:

•    Darcy's Law is valid and hydraulic-head gradients  are the only significant
     driving mechanism for fluid flow.

                                      A-41

-------
 t    Vertically averaged properties may be used.

 t    Gradients of fluid density, viscosity, and temperature do not affect the
      velocity distribution.

 •    Transmissivity principal components are aligned along coordinate axes.

 •    Linear evapotranspiration.

 •    Maximum evapotranspiration rate may not vary spatially.       '.    ;

 •    The porosity and hydraulic conductivity are constant with time.

 CODE INPUT:  USGS2D may read single or multiple data sets.  The contents of a
 data set are described in Table A-l.

 CODE OUTPUT:  USGS2D prints the following:   1)  all  input; 2)  tinie-step
 information; 3)  mass-balance information;  4)  matrix-iteration information;  and
 5)  computed hydraulic head or drawdown.                         ;

 COMPILATION REQUIREMENTS:  USGS2D was  programmed in FORTRAN IV for use on an
 IBM machine.  It has been successfully adapted  for  use  on CDC and UNIVAC
 machines as well.   Model  results can be  presented on the line printer (rows
 should  be numbered  in the short dimension)  and  pen  plotters with  a program  that
 utilizes the graphical  display software  available from  the U.S. Geological
 Survey  Computer  Center Division.  In addition,  included in the model  are
 options for reading input data from a  disk  and  writing  intermediate results on
 a disk.

 EXPERIENCE REQUIREMENTS:   Extensive.

 TIME REQUIREMENTS:   Months.

 CODE VERIFICATION:   USGS2D  has been compared  to analytical  solutions
 including those  for leaky aquifers.  It has also been used to  simulate several
 hypothetical  problems where  detailed mass balance calculations were made.    ;

 This  code has been  the work  horse of the U.S. Geological  Survey for more than
 ten years.   It has  been applied  to  numerous sites throughout North America.
 Most  of these applications have  dealt with water supply  problems  associated
 with  relatively shallow aquifers.   Some of the  field problems  include  the
 Washington,  D.C., area (ref. 3), west-central Minnesota  (ref. 4), and  Nova
 Scotia  (ref. 5).

 DOCUMENTATION/REFERENCES:

 Documentation (refs. 1 and 2) for USGS2D is available in two government
 publications.
Trescott, P.C.; Pinder, G.F.; Larson,  S.P.  (1976)  Finite-difference model
     aquifer simulation in two dimensions with results of numerical
     experiments.  U.S. Geological  Survey, Techniques of Water-Resources
     Investigations,  Book 7,  Chapter Cl.
for
                                      A-42

-------










1

o
[•

to
3
0 0
•i- CM
S- CO
re CD
5» 00

e
o c
t»- •!-

to to to
•P E E
C S. , 0

Q) •!-'
i-  O
fV £/) QJ
O
re "o s-
•P E 3
re re1 o
Q 00
to
S_ E
O 0
•r-3-r-
re +•>
S CL
O

O E
O
«£
E E
3 O
CO O

^
i-H .
I

CD
r—
JO
£
g
,
4-*'
(0

p


(O

• j
O
O.
^
UI















.1




CD
' en

re
.E
o
iS


0)
4-^
•£. -


en-
s-
re
.E
U
,. i- W 0
4J 0) O) T-
•i- • tt^ c <*-
> »^ -NX «^
•r- 3 0 O,
4-3 "O C7" *^ ' Q^
o Q) re .C o. >>
3 ii -P to •£?
•o a> re
E CD O "O T3  "P ^^
O »i— 3 -Q ^^ t/> 
re o .1- -i- s- o i—
S- *o «i- i— o
•O t4_ to E E •!-> II
>, O O) O O  — •
S- 3= 0 O


to
s_
S- (U
{- O)
cr s_
ec re
o.
tfl
c
__ o

O 1 ^

| i *X3
C
o
o
->> '' '-'"
•r- 4->
> E '
«i- OJ
10 O '
*E cniJI
• to -«c «t- . •
E S- 
o -o
4_> 4J t|_
re re  I. 0
Q^ . 1
( •» IIJ.T J'* - ;
O) 1 3
UI O

re re i
—J S UI
s


o
•P
i.

O)
CO
s-
re
sz
o
 «/) E
O 4-> S-
.0 c a>
re  »r-
> o >-
O ••- -O
•r- 4-> 0
I— O S- •!-
3 3 Q) <*-
J- E T- O
•o o 3 a>
>, o cr o.
nr 
0.

o -o
•i™~ QJ
j^ ^^ !!••
re re «*-
> s- o
 W) E
0 4-5 S-
i^> c O
re g; 4->
E E
tO »P *r~
re to
.= *"
E -5* re
re re E
01

£
4-? 4J
•i- -P CS.
> O O
o •!- ja -p
1^ o s- .v a>
3 3 a> a» en
re "d «t- t»- 
3: «=C  O
                   O OJ
                   O O.
                      00
A-43

-------
 Larson,  S.P.   (1979)   Direct solution for the  two-dimensional  ground-water flow
     model.   U.S.  Geological Survey,  Open-File Report  78-202.

 Papadopulos,  S.S.;  Bennett,  R.R.;  Mack,  F.K.;  Trescott,  P.C.   (1984)   Water
     from  the coastal  plain  aquifers  in  the  Washington,  D.C.,  metropolitan
     area.  U.S. Geological  Survey Circular  697.

 Larson,  S.P.;  McBride,  M.S.; Wolf,  R.J.   (1975)  Digital  models  of  a glacial
     outwash  aquifer  in the  Pearl-Sallie Lakes Area.   U.S.  Geological  Survey
     Water-Resources  Investigations 40-75.

 Pinder,  G.F.;  Bredehoeft, J.D.   (1968)   Application of the  digital  computer for
     aquifer  evaluation.  Water  Resour.  Res. 4:(5):1069-1093.

SOURCE:  USGS2D was developed at the  U.S. Geological Survey and  is,described
 in Techniques  of Water-Resources Investigations, Book  7,  Chapter Cl, by P  C
Trescott, G.  F. Pinder,  and  S. P.  Larson.  The code is in the public domain
In addition, the USGS offers some  support for  certain  users and applications
                                     A-44

-------
CODE NAME:  USGS3D --Modular

PHYSICAL PROCESSES:  Predicts groundwater flow in confined/unconfined
aquifers.               ,

DIMENSIONALITY:  Three-dimensional.

SOLUTION TECHNIQUE:  Numerical, finite-difference.

DESCRIPTION:  Groundwater flow within the aquifer is simulated using a
block-centered finite-difference approach.  Layers can be simulated as
confined, unconfined, or a combination of confined,and unconfined.  Flow from
external stresses, such as flow to wells, area! recharge, evapotranspiration,
flow to drains, and flow through riverbeds, can also be simulated.  The
finite-difference equations can be solved using either the Strongly Implicit
Procedure or Slice-Successive Overtaxation.

The modular structure consists of a main program and a series of highly
independent subroutines called "modules."  The modules are grouped into
"packages," each package addresses a specific feature of the hydrologic system.

CODE INPUT:  Standard saturated zone hydrologic parameters.

CODE OUTPUT:  Head distribution at each time step.

COMPILATION REQUIREMENTS:  The program is written in FORTRAN '66 and will run
without modification on most computers which have a FORTRAN '66 compiler.  It
will also run, without modification* with most extended FORTRAN '77 compilers
and with minor modifications on standard FORTRAN  '77 compilers.

EXPERIENCE REQUIREMENTS:  Extensive.

TIME REQUIREMENTS:  Months.

CODE VERIFICATION:  Unknown.

DOCUMENTATION/REFERENCES:

McDonald, M.6.; Harbaugh, A.W.  (1984)  A modular three-dimensional
     finite-difference  groundwater flow model.  U.S. Geological Survey
     Open-File Report 83-875.

SOURCE:  The code was written by Michael McDonald and Arlen Harbaugh of the
U.S. Geological Survey.  This code is  in the public domain.
                                       A-45

-------
 CODE NAME:   USGS3D  --  Trescott

 PHYSICAL PROCESSES:   Fully three-dimensional  or quasi-three-dimensional
 saturated groundwater  flow in confined,  unconfined,  or combined  confined  and
 unconfined  aquifers.

 DIMENSIONALITY:   Fully three-dimensional,  quasi-three-dimensional.

 SOLUTION TECHNIQUE:   Numerical,  finite difference.

 DESCRIPTION:   USGS3D  simulates three-dimensional  flow in  a  porous medium
 which may be  heterogeneous and anisotropic and  have  irregular  boundaries. The
 uppermost hydrologic  unit  may have  a  free  surface.   The stresses considered are
 wells and recharge  from precipitation.

 One  or more layers  of  nodes can  be  used  to simulate  each  hydrogeologic unit.
 It it is reasonable to assume that  storage is negligible  in a  confining bed and
 that horizontal  components of flow  can be  neglected,  the  effects of vertical
 leakage  through  a confining bed  can be incorporated  into  the vertical component
 of the anisotropic  hydraulic conductivity  of  adjacent aquifers.

 A major  advantage of this  code is that it  can be  used in  a  fully
 three-dimensional mode or  it can be reduced to  a  quasi-three-dimensional  model
 1n terms of the  equations  being  solved and computer  memory  requirements.  This
 is accomplished  by  using a sequence of two-dimensional  (areal) groundwater flow
 models to represent aquifers.  These  models are coupled by  terms representing
 flow through  intervening confining  beds  to form a quasi-three-dimensional
 model.   This  latter model  converges to a solution much  faster  than the fully
 three-dimensional model  because  all equations are solved  simultaneously.  It
 should be noted,  however,  that the  leakage in this quasi-three-dimensional
 model is steady,  that  is,  it ignores  storage.   For long-term simulations which
 approach steady  state,  this  type of leakage is  adequate.

 The  flow field can be  represented as  a three-dimensional  grid  or a sequence of
 two-dimensional  grids.   The  program is fairly general  in  that  the size of the
 grid  blocks is variable  to  allow the  desired  level of spatial  detail.  Variable
 time  steps  are also allowed.

 The major variable is  hydraulic  head,  although drawdowns  may be  computed from
 the  initial head condition.  The main assumptions of  the  code  are:

 •    Darcy's  Law is valid and  hydraulic-head gradients  are the only significant
     driving mechanism for fluid flow.      '

*    The  porosity and  hydraulic  conductivity  are constant with time.

t    Gradients of fluid density, viscosity, and temperature do not affect the
     velocity distribution.

•    Hydraulic conductivity principal  components are aligned with Cartesian
     coordinate system.

t    Steady leakage from confining bed?  can be incorporated into anisotropic
     hydraulic conductivity of adjacent  aquifers.,

                                  1    A-46   "

-------
CODE INPUT:  The data required to run USGS3D includes finite-difference data,
such as spacing and physical  data.  The physical data includes:

•    initial heads;

•    boundary conditions;

•    storage coefficient distribution;

•    transmissivity distribution; and

•    recharge rate.

If the upper unit is unconfined, then hydraulic cdnductivity and the elevation
of the bottom of the water table layer is read in place of transmissivity.

CODE OUTPUT:  USGS3D prints the following:  1) all input; 2) time-step
information; 3) mass-balance information; 4) matrix-iteration information; and
5) computed hydraulic head or drawdown.

COMPILATION REQUIREMENTS:  USGS3D was programmed in FORTRAN IV for use on an
IBM machine, using some machine-dependent features.

Rows should be numbered in the short dimension, for plotting maps on the line
printer or for plotting data with an X-Y drum  plotter.  The core requirements
and computation time are proportional to the number* of nodes representing the
porous medium.

To reduce the number of cards that must be read with each run, the program
includes options to place the arrays on disk and,'on subsequent runs, read the
data from disk rather than from cards.

The documented program was designed to take advantage of certain features of an
IBM machine.  Because of this, there are difficulties in converting the code to
a non-IBM machine.

EXPERIENCE REQUIREMENTS:  Extensive.               <

TIME REQUIREMENTS;  Months.                                               .    •

CODE VERIFICATION:  The code has .been applied  to several field problems and
code results have been compared with analytical solutions.  USGS3D does have a
detailed mass balance to ensure that the solution'has converged.

This code has been applied to several field problems, including flow problems
associated with mining, hazardous waste, and radioactive waste (Columbia
Plateau).  As an example of a field application of this code to a mining
problem, see ref. 3.

DOCUMENTATION/REFERENCES:

Trescott, P.C.  (1975)  Documentation of finite-difference model for simulation
     of three-dimensional ground-water flow.   U.S.' Geological Survey Open-File
     Report 75-438.

                                      A-47

-------
Trescott, P.C.; Larson, S.P.  (1976)  Documentation of finite-difference model
     for simulation of three-dimensional ground-water flow.  U.S. Geological
     Survey Open-File Report 76-591, supplement to Open-File Report 75-438.

Weeks, J.B.; Leavesley, G.H.; Wleder, F.A.; Saulneir, G.J., Jr.  (1984)
     Simulated effects of oil-shale development on the hydrology of Piceance
     Basin, Colorado.  U.S. Geological Survey Professional Paper 908.

SOURCE:  USGS3D was developed at the U.S. Geological Survey and is described
1nu?,se!)ies of reP°rts by p- c-  Trescott and S. P. Larson.  The code is in the
public domain.   In addition, the USGS offers some support for certain users and
applications.
                                     A-48

-------
CODE NAME:  VTT

PHYSICAL PROCESSES:  Predicts the transient or steady-state hydraulic head
distribution and provides water flow paths and travel  times.

DIMENSIONALITY:  Two dimensional (x-y cartesian).

SOLUTION TECHNIQUE:  Numerical, finite difference.

DESCRIPTION:  The model is capable of calculating water flow in a
multilayered aquifer system.  The system may be confined, unconfined, or
semiconfined.  The main simplifying assumption transforms a three-dimensional
system to a layered two-dimensional system with interaquifer transfer via a
potential-driven leakage term.  The mathematical model which utilizes this set
of simplifying assumptions is the multi-aquifer formulation of the Boussinesq
equations.  VTT uses a horizontal, two-dimensional, finite-difference approach
for saturated flow in each aquifer.  The code may analyze flow in such a system
for a variety of initial and boundary conditions for steady or non-steady flow.

The velocity field in the porous medium is part of the output of VTT.  The
analysis of the flow field is the first stage in predicting the transport of
contaminants in a porous medium.  The output of the flow field could
subsequently be used to develop inputs for such transport codes as MMT or
FEMWASTE.

For numerical formulations, a horizontal x-y coordinate grid system is adopted
with uniform nodal spacing.  Standard finite-difference approximations and a
fully implicit representation of the time derivative are used.

When considering confined flow, the compressibility effects of the fluid and
matrix are incorporated, but they are neglected when considering unconfined
flow.  This assumption is quite valid as long as the specific yield is not of
the same magnitude as the specific storage.

The main assumptions of the code are:

t    Darcy's Law is valid and hydraulic-head gradients are  the only significant
     driving mechanism for fluid flow.

•    The porosity  and hydraulic conductivity are constant with time.

•    Gradients of  fluid density, viscosity, and  temperature do not affect the
     velocity distribution.

•    Hydraulic conductivity and effective  porosity can be represented by the
     vertically  averaged values and are isotropic throughout the region but may
     be  inhomogeneous.

•    The  free-surface  slope and the aquifer bottom slope are both slight  (<5o).

•    Vertical velocities are  small and can be neglected.

•    Flow in  the capillary  finge  is neglected.

•    Seepage  surfaces  cannot  be handled and are  neglected.
                                       A-49

-------
 CODE INPUT:   Inputs  to VTT include:
 t     total stress  or recharge  at  each  node;
 t     aquifer  top elevation;
 •     aquifer  bottom  elevation;
 •     initial  aquifer potential;
 •     aquifer  storage coefficient;
 •     interaquifer  transfer (leakage) coefficient; and
 t     aquifer  hydraulic conductivity  (or transmissivity) at each node.
 CODE OUTPUT:   The  output  for VTT  is  the new  spatial variation of potential
 throughout the aquifers.   From this  potential distribution in conjunction with
 other input data,  the following information  can be calculated:
 t     groundwater velocities;
 •     groundwater flow paths;
 t     travel times; and
 §     new recharge/discharge relationships along streams and rivers.
 Types  of model  output that can be produced include:
 •     Contour maps  of:
          equal potential
          equal drawdown
          equal transmissivity
 t    Three-dimensional  projection plots of:
          potential
          drawdown
          transmissivity
 §    Cross-sectional   plots showing aquifer top,  aquifer bottom,  and aquifer
     potential.
•    Flow path plots with  associated listings of travel times.
t    Numerical listings of the input data or calculated potentials.
•    Difference maps  showing the node-by-node predictions  of potential  changes.
COMPILATION REQUIREMENTS:  VTT is  written in FORTRAN IV-PLUS.  The  code can
be run on a POP 11/70 or VAX machines.   The code was converted to a CDC machine
by Intera,  Inc.
EXPERIENCE REQUIREMENTS:  Extensive.
                                      A-50

-------
TIME REQUIREMENTS:  Months.

CODE VERIFICATION:  VTT has been compared with solutions from a more general
three-dimensional model, FE3DGW, and a model which uses an analytical solution,
PATHS.

DOCUMENTATION/REFERENCES:

Reisenauer, A.E.  (1979)  Variable thickness transient groundwater flow model
     (VTT), formulation, user's manual and program listings.  Pacific Northwest
     Laboratory Report PNL-3160-1, PNL-3160-2, and PNL-3160-3.

Gupta, S.K.; Cole, C.R.; Bond, F.W.  (1979)  Finite element three dimensional
     groundwater  (FE3DGW) flow model, formulation, program listings and user's
     manual.  Pacific Northwest Laboratory Report PNL-2939.

Kellogg, O.C.   (1954)  Foundations of potential theory.  Dover, NY.

SOURCE:  VTT Was  developed at Battelle, Pacific Northwest Laboratory.  The
work was supported by the Department of Energy.  The code is  in the public
domain.
                                       A-51

-------
 CODE NAME:  V3

 PHYSICAL PROCESSES:  Predicts groundwater flow in heterogeneous anisptropic
 aquifers under a variety of flow conditions.

 DIMENSIONALITY:  Two-dimensional  (x-y cartesian).

 SOLUTION TECHNIQUE:  Numerical, finite difference.

 DESCRIPTION:   This series of computer programs simulates one-  or
 two-dimensional nonsteady-state  flow problems in  heterogeneous anisotropic
 aquifers under water table,  nonleaky,  and leaky artesian conditions.
 Multiple-aquifer problems with leakage between aquifers  can  also be treated.
 These programs cover time-varying pumpage from wells,  natural  or artificial
 recharge rates, the relationships of water exchange between  surface waters and
 the  groundwater reservoir,  the process of groundwater  evapotranspiration, the
 mechanism of  possible conversion  of storage coefficients from  artesian to water
 table conditions,  and the multiple-aquifer problem.             .

 The  program is fairly general.  One- or two-dimensional  grids,  or sequences of
 two-dimensional grids,  may be used.   The size of  the grid blocks are variable
 to allow the  desired level  of spatial  detail.  Variable time  steps are also
 allowed.   The code considers several  different flow conditions  such as
 hydraulic conductivity,  storage properties, leakage  properties,  and recharge
 properties.                                                           ,

 V3 is programmed in modular  fashion  that allows relatively convenient
 modification.   In  general,  the modules .contain logical work  tasks.  The
 modules,  however,  are not contained  in subroutines;  everything  is contained in
 the  main  program.

 The  main  assumptions of the  code  are:                            ,    ,

 •     Darcy's  Law is valid and hydraulic-head  gradients are the. only significant,
      driving  mechanism  for  fluid  flow.

 •     The  porosity  and hydraulic conductivity  are  constant with^time.

 •     Gradients  of  fluid  density,  viscosity, and temperature  do  not  affect the
      velocity  distribution.                                         ,        >   :

 •     Vertically averaged  properties  can  be  used.                    *        ;  ;

 •     Transmissivity principal components  are  aligned with  cartesian coordinate
      system.                           .              ,              ,       , •

 •     Leakage  is steady state.
                                                             •'.I.'"'  ..    >'-•'. ''''.'I
 §     Linear evapotranspiration.

 CODE INPUT:   Input  for the basic aquifer  simulation program  includes:

 •     parameter  and default value cards;

§     array data;

                                      A-52

-------
•    transmissivity;
•    storage coefficient;
0    initial hydraulic heads; and
0    pumpages.
With modifications to the program, additional data are read as follows:
         Option
                  Additional Data
Variable Pumping Rates
Leaky Artesian Conditions

Induced Infiltration

Evapotranspiration
Storage Coefficient
Conversion
Water Table Conditions
Time and pumping rate of each period
Vertical hydraulic conductivity and thickness of
confining bed; head difference across confining bed
Same as leaky artesian conditions plus areas of the
stream bed
Land surface elevation; elevations of the water
table below which ET ceases; maximum ET rate
Elevation of aquifer top; water table storage
coefficient
Water table storage coefficient; elevation of
aquifer bottom
CODE OUTPUT:  The primary outputs of V3 are hydraulic head or drawdowns.
There are options for displaying these in a readable fashion, such as
time-water level graphs.
COMPILATION REQUIREMENTS:  The computer programs were written in FORTRAN IV
for use on an IBM 360 system model 75 with a G-level compiler. However, the
programs will operate, with modifications, on other computers.  Also, these
computer programs are written so that they will operate with any consistent set
of units.
EXPERIENCE REQUIREMENTS:  Extensive.
TIME REQUIREMENTS:  Months.
CODE VERIFICATION:  V3 is one of the most widely used groundwater flow
codes.  It has been compared to several analytical solutions, including those
by Theis, Hantush, and Jacob.
DOCUMENTATION/REFERENCES:
Prickett, T.A.; Lonnquist, C.G.  (1971)  Selected Digital Computer Techniques
     for groundwater resource evaluation.  Illinois State Water Survey,
     Bulletin 55.
                                      A-53

-------
McDonald, M.G.; Fleck, W.B.  (1978)  Model analysis of the impact on '
     groundwater conditions of the Muskegon County waste-water disposal system,
     Michigan.  U.S. Geological Survey Open-File Report 78-99.

SOURCE:  V3 was developed at Illinois State Water Survey.  The code is in the
public domain.
                                      A-54

-------
NUMERICAL SOLUTE TRANSPORT
  (SATURATED/UNSATURATED)
            A-55

-------
 CODE NAME:   FEMWASTE1  (Finite-Element Model  of  Waste Transport)
 PHYSICAL PROCESSES:   Predicts  waste  transport through  saturated-unsaturated
 porous media under dynamic  groundwater conditions.
 DIMENSIONALITY:   Two-dimensional  (x-y,  x-z  cartesian).
 SOLUTION TECHNIQUE:   Numerical,  finite-element.
 DESCRIPTION:  FEMWASTE1  is  an  upgraded version  of the  FEMWASTE code developed
 by Yeh and  Ward  for  subsurface transport.   It is a  transport-only code  that
 requires fully specified hydrodynamics as part  of the  input data set.
 FEMWATER1,  a flow model  using  an  identical  numerical representation of  the
 problem domain,  creates  a geometric  and hydrodynamic data file expressly  for
 FEMWASTE1.
 FEMWASTE1 is capable of  transient, two-dimensional  simulations of pollutant
 transport in saturated and  unsaturated porous media.   The model includes  the
 following transport  processes:
 •     advection;
 •     hydrodynamic dispersion;
 •     sorption;
 •     first-order decay;  and;
 •     source/sink.
 FEMWASTE1 uses quadrilateral bilinear elements  for  spatial discretization of
 the  porous  media.  Solution  is by the finite element weighted residual  method.
 Two  finite  element weighting techniques are  available:  Galerkin and upwind.
 The  option  of "lumping"  the  finite element mass matrix into a type of finite
 difference  unit  mass matrix  by scaling  is also  available.  In some situations
 the  lumping technique  has been more  accurate.   Included in the FEMWASTE1
 formulation is the option of three time-stepping techniques:
 1)    central  difference;
 2)    backward difference; and
 3)    mid-difference.
 Regardless  of the  technique selected, Gaussian elimination is used to invert
 the  resulting matrices.
 With  the  options of weighting,  lumping, and time stepping, 12 different
 computational approaches are possible with FEMWASTE1.
 FEMWASTE1 is designed to be applied uncoupled from the flow field
 calculations.  This design implicitly assumes that transport processes do not
 affect the  fluid transport;   The model is applied to a single chemical species
without considering .the effects of other chemicals that may be present in the
                                     , A-56

-------
porous media.  The only irreversible reaction in FEMWASTE1 is in the
first-order decay process where degradation is assumed to be directly
proportional to the total pollutant concentration (including dissolved and
adsorbed phases).  FEMWASTE1 does not explicitly account for biological  uptake
or other degradation mechanisms, although these effects can be included
approximately by adjusting the decay constant. Transformation and loss
mechanisms can be represented by the source/sink term local to each element.
Chemical sorption is assumed to be a reversible fast-exchange reaction that
attains local equilibrium within a single time interval according to a linear
isotherm.  The distribution coefficient, Kd, is moisture independent in
FEMWASTE1.  This assumption restricts the model application to cases where all
soil grains are effective in the adsorption process.  The calculated
retardation factor, however, is an inversely dependent function of the moisture
content.

CODE INPUT:  FEMWASTE1 requires standard material and aquifer properties:

t    adsorption distribution coefficient;

t    bulk density;

•    longitudinal dispersivity;

•    transverse dispersivity;

•    decay  constant;

•    porosity; and

•    modified coefficient of compressibility.

Aquifer  characteristics  can be  entered  on a regional basis or can be specific
to  a given  element.

Three types of boundary  conditions are  possible with FEMWASTE1:

1)   Dirichlet,  specifications  of time-varying concentrations at a particular
     element;

2)   Neuman, specification of time-varying waste fluxes at a particular
     element; and

3)   Cauchy, waste-flux  boundary conditions that are specific to inflow
     boundaries.

Initial  conditions  are concentrations of pollutant  in the porous media.

An  error checker  is present in  FEMWASTE1 to ensure  that the  input data are
correct.  When errors are detected, execution is stopped.

CODE OUTPUT:  Output from FEMWASTE1 consists of formatted line-printer
listings  of concentrations at specified locations and times.  An auxiliary
storage  device file is created  for post processing  on local  hardware.
                                      A-57

-------
COMPILATION REQUIREMENTS:
on the program are:
FEMWASTE1 is written  in FORTRAN.   Current limits
•    595 nodes;
•    528 elements;
•    500 time steps;
•    29 water boundary  conditions;
•    199 boundary elements;
t    200 boundary nodes;
•    99 rainfall seepage  element  sides; and
a    100 rainfall seepage element sides.
FEMWASTE1 is designed to  be  applied  in a batch mode.  The model is constructed
in a modular fashion with one main program and 15 subroutines.  Efficient
storage of the banded matrix arrays  are used in FEMWASTE1.  Annotation of the
source listing is good  with  short descriptions and identified computations.
Conversion to the VAX 11/780 is straightforward.
EXPERIENCE REQUIREMENTS:   Extensive.
TIME REQUIREMENTS  Months.
CODE VERIFICATION:  A sample problem of transport from a seepage pond
reported by Duguid and  Reeves was used to compare the simulation by the
original computer codes with that by the new waste-transport code coupled with
the revised water-flow  code.  The code is in the public domain.
DOCUMENTATION/REFERENCES:
Bear, J.  (1972)  Dynamics of fluids in porous media.
     Elsevier.
                            New York,  NY:  American
Duguid, J.O.; Reeves, M.   (1976)  Material transport in porous media:  a finite
     element Galerkin model.  ORNL-4928, Oak Ridge National Laboratory, Oak
     Ridge, TN.
Reeves, M.; Duguid, J.O.   (1975)  Water movement through saturated- unsaturated
     porous media:  a finite-element Galerkin model.  ORNL-4927, Oak Ridge
     National Laboratory,  Oak Ridge, TN.
Yeh, G.T.  (1982)  Training course no. 2:  the implementation of FEMWASTE
     (ORNL-5601) computer  program.  ORNL/TM-8328, NUREG/CR-2706, U.S. Nuclear
     Regulatory Commission, Washington, DC.
Yeh, G.T.; Ward, D.S.  (1981)  FEMWASTE:  a finite-element model of waste
     transport through saturated-unsaturated porous media.  ORNL-5601, Oak
     Ridge, National Laboratory, Oak Ridge, TN.
                                       A-58

-------
Yeh, 6.T.; Ward, D.S.  (1980)  FEMWATER:  a finite-element model of water flow
     through saturated-unsaturated porous media.  ORNL-5567, Oak Ridge National
     Laboratory, Oak Ridge, TN.

SOURCE:  FEMWASTE was developed at Oak Ridge National Laboratory and is
described in Report No. ORNL-5601 by G. T. Yeh and D. S. Ward.  It is an
extension of work done by Duguid and Reeves.
                                     A-59

-------
CODE NAME:  PERCOL
PHYSICAL PROCESSES:  Predicts the movement of a solute through a soil column.
DIMENSIONALITY:  Unknown.
SOLUTION TECHNIQUE:  Numerical, Newton-Raphson method.
DESCRIPTION:  PERCOL has been developed to simulate the movement of
radionuclides through porous media as a function of measurable chemical
parameters of the media.  System parameters include soil type, radionuclide
type, waste composition, flow rate, column length, and soil saturation.
CODE INPUT:  Standard transport parameters.
CODE OUTPUT:  Predictions of concentrations.
COMPILATION REQUIREMENTS:  PERCOL is written in FORTRAN and implemented on an
IBM 360 machine.
EXPERIENCE REQUIREMENTS:  Extensive.
TIME REQUIREMENTS  Months.
CODE VERIFICATION:  Laboratory column studies have been conducted to verify
the model.
DOCUMENTATION/REFERENCES:
Owen, P.T.  An inventory of environmental  impact models relating to energy
     technologies.  ORNL/EIS-147, Oak Ridge National  Laboratory, Oak Ridge, TN.
Routson, R.C.; Seme, R.J.   (1972)  One-dimensional model of the movement of
     tracer radioactive solute through soil columns:   the PERCOL model.
     BNWL-1718, Battelle, Pacific Northwest Laboratories, Richland, WA.
SOURCE:  PERCOL was written by R. C. Routson of Argonne National Laboratory
and R. 0. Serne of Battelle, Pacific Northwest Laboratories. The code is in the
public domain.
                                      A-60

-------
CODE NAME:  SATURN

PHYSICAL PROCESSES:  Predicts saturated-unsaturated flow and radioactive
radionuclide transport.

DIMENSIONALITY:  Two dimensional.

SOLUTION TECHNIQUE:  Numerical,  finite element.

DESCRIPTION:  SATURN is a two-dimensional finite.-element model developed to
simulate fluid flow and solute transport processes in variably saturated porous
media.  The model solves the flow and transport equations separately.
Transport mechanisms considered include advection, hydrodynamic dispersion,
adsorption, and first-order decay.

The flow equation is discretized using the Galerkin finite-element method.
Nonlinearity is treated using either Picard or Newton-Raphson iterations. The
transport equation is discretized using an upstream-weighted finite-element
method designed to alleviate the problem of numerical oscillations.  Simple
rectangular and triangular elements are used.  The combination of such elements
enables flow regions of complex geometry to be modeled accurately.  A highly
efficient "influence coefficient" technique is used to generate element
matrices.  This technique avoids numerical integration and leads to a reduction
in CPU time required for element matrix generation.  For rectangular elements,
the saving of CPU time

•    prescribed values of nodal  fluid flux;

•    longitudinal dispersivity;

•    transverse dispersivity;

•    molecular diffusion coefficients;

0    decay coefficient;

•    retardation coefficient;

•    initial inventory of solute;

0    leach duration;

•    prescribed values of concentration; and

t    prescribed values of solute flux.

CODE OUTPUT:  The primary line  printer output from the flow model of SATURN
includes  nodal values  (at various time levels) of pressure head and element
centroidal values of Darcy velocity components and saturation at various time
levels.   SATURN allows the user to select the fluid mass balance calculation
option.   This budget contains information about the net flow rate of fluid due
to boundary fluxes, sources, and sinks; the rate of fluid accumulation in the
entire flow domain; the mass balance error; and the cumulative fluid storage up
to the current time.

                                      A-61

-------
 The primary line printer output  of the  transport  model  of SATURN  includes  nodal
 values (at  various time levels)  of solute  concentration.   SATURN  also  allows
 the user to select the solute  mass balance calculation  option.   If the solute
 mass balance budget is activated, information  about  the total dispersive and
 advective fluxes;  the net rate of material  accumulation taking  into account
 storage,  adsorption,  and decay;  the mass balance  error; the  cumulative mass of
 solute still  remaining in the  porous medium at the current time;  and the
 cumulative  mass  decay up to  the  current time value will be printed at  the  end
 of each time step.

 COMPILATION REQUIREMENTS: SATURN is written in FORTRAN with versions
 existing  on the  following computer systems:

 •     PRIME;

 •     DEC  VAX 11/780;  and

 t     CDC  7600.

 The code  is  written in double  precision (eight-byte  decimal  words).  The code
 treats  the  transport  processes as an uncoupled  phenomenon  from  the  flow
 processes;  thus, two  successive  applications of SATURN  are necessary to
 simulate  flow and  transport.

 Current problem  limits:

 •     500  nodal points;

 •     450  elements;

 •     20 materials;

 •     99 Dirichlet boundary conditions;

 •     99 flux  boundary  conditions;

 •     40 semi-bandwidth for global matrix;

 §     10 time-dependent Dirichlet boundary nodes;

 t     10 time-dependent flux boundary nodes;

 •     20 entry pairs for relative permeability versus saturation; and

 •     20 entry pairs for pressure head versus saturation.

 Due to the manner in which these limits are prescribed  in the code, changes in
 dimension can require the modification  of common blocks, in almost every
 subroutine.

 EXPERIENCE REQUIREMENTS:  Extensive.
                    i                                                           (i

TIME REQUIREMENTS  Months.
                                      A-62

-------
CODE VERIFICATION:  The code has been verified against several test problems
and against UNSAT2.              .                          :

DOCUMENTATION/REFERENCES:

Huyakorn, P.S.; Thomas, S.D.; Mercer, J.W.; Lester, B.H..  (1983)  SATURN:  a
     finite element model for simulating saturated-unsaturated flow and
     radioactive radionuclide transport.  Prepared by Geotrans for the Electric
     Power Research Institute, Palo Alto, ,CA.

SOURCE:  SATURN is a proprietary code.  It was developed by P. S. Huyakorn,
S. D. Thomas, J. W. Mercer, and B. H. Lester, all of Geotrans, Incorporated.
                                       A-63

-------
CODE NAME:  SEGOL

PHYSICAL PROCESSES:  Predicts contaminant transport and flow for combined
saturated and partially-saturated flow systems.

DIMENSIONALITY:  Three dimensional.

SOLUTION TECHNIQUE:  Numerical, finite element,

DESCRIPTION:  The transport equations are solved using a Galerkin
finite-element approach.  The model solves for three-dimensional combined
partially-saturated and saturated flow.  Since solution for both saturated and
partially-saturated flow is possible, the free surface is very nicely handled.
At the same time, a more realistic representation of flow from ponds or lakes
is provided by the model, rather than requiring the assumptions involved when
only saturated models are used.  The SEGOL model uses finite elements for the
numerical reduction with isoparametric elements.  It is an operational model
and has been tested, but has not yet been extensively used with field
problems.  A severe test case indicated that a spatial discretization of a few
centimeters may be necessary for some problems.

CODE INPUT:  Standard transport parameters.

CODE OUTPUT:  Predictions of head and concentration.

COMPILATION REQUIREMENTS:  SEGOL is written in FORTRAN.

EXPERIENCE REQUIREMENTS:  Extensive.

TIME REQUIREMENTS  Months.

CODE VERIFICATION:  SEGOL has been verified by comparison to field data.

DOCUMENTATION/REFERENCES:

Segol, G.A.  (1976)  Three-dimensional Galerkin finite element model for the
     analysis of contaminant transport in variably saturated porous
     media -- user's guide.  Department of Earth Sciences, University of
     Waterloo, Waterloo, Ontario, Canada, June.

Segol, G.A.  (1977)  A three-dimensional Galerkin-finite element model for the
     analysis of contaminant transport in saturated-unsaturated porous media.
     In:  Gray, W.G.; Pinder, G.F., eds., Finite elements in water resources,
     (Proceedings of the First International Conference, July, 1976).

SOURCE:  SEGOL was developed by G. A. Segol of Bechtel Corporation.  The code
is in the public domain.
                                      A-64

-------
CODE NAME:  SUMATRA-I

PHYSICAL PROCESSES:  Predicts the simultaneous flow of water and solutes
transport in a vertical soil profile under transient saturated-unsaturated
conditions.

DIMENSIONALITY:  One dimensional.

SOLUTION TECHNIQUE:  Numerical, finite element.

DESCRIPTION:  SUMATRA-I is based on a Hermitian (cubic) finite-element
solution of the governing transport equations.  The model includes such
processes as linear equilibrium adsorption and zero- and first-order decay.

Flow in the saturated-unsaturated medium,is posed and solved with a pressure
head dependent variable  The convection dispersion equation is amended to
include retardation by adsorption and decay mechanis.ms.  These continuum models
are driven by complete physical properties and boundary condition information
including:  transient data for soil surface, boundary conditions, hydraulic
functions relating moisture content and,hydraulic conductivity, and physical or
chemical parameters (e.g., density, dispersivity, adsorption, and decay
parameters).

CODE INPUT:  Standard groundwater flow and contaminant transport parameters.

CODE OUTPUT:  Predictions of head and concentration.

COMPILATION REQUIREMENTS:  Predictions of head and concentration.

EXPERIENCE REQUIREMENTS:  SUMATRA-I is written in FORTRAN IV.  It consists of
the main program and nine subprograms.

TIME REQUIREMENTS  Extensive.

CODE VERIFICATION:  Months.

DOCUMENTATION/REFERENCES:

van Genuchten, M.Th.   (1978a)  Mass transport in saturated-unsaturated media:
     one-dimensional solutions.  Research Report 78-WR-ll, Water Resources
     Program, Department of Civil Engineering, Princeton University, Princeton,
     NO.

van Genuchten, M.Th.   (1978b)  Numerical solutions of the one-dimensional
     saturated-unsaturated flow equation.  Research Report 78-WR-9, Water
     Resources Program, Department of Civil Engineering, Princeton University,
     Princeton, NJ.

SOURCE:  SUMATRA-I was written by van Genuchten of the U.S. Salinity
Laboratory.  The code  is in the public domain.
                                       A-65

-------
 CODE NAME:  SUTRA (Saturated-Unsaturated Transport)

 PHYSICAL PROCESSES:   Predicts  fluid movement and the transport  of either
 energy or dissolved  substances in a subsurface environment.

 DIMENSIONALITY:   Two dimensional.
 SOLUTION TECHNIQUE:
 finite-element.
Numerical, finite-element, and integrated
 DESCRIPTION:   SUTRA flow simulation may be employed for areal  and
 cross-sectional  modeling of saturated groundwater flow systems,  and  for
 cross-sectional  modeling of unsaturated zone flow.   Solute  transport simulation
 using SUTRA maybe employed to  model  natural  or man-induced  chemical  species
 transport including processes  of solute sorption,  production,  and  decay,  and
 may be applied to analyze groundwater contaminant transport problems and
 aquifer restoration designs.   In addition, solute transport simulation with
 SUTRA may be  used for modeling of variable density leachate movement and  for
 cross-sectional  modeling of salt-water intrusion  in aquifers at  near-well or
 regional  scales,  with either dispersed or relatively sharp  transition zones
 between fresh water and  salt water.   SUTRA energy transport simulation may be
 employed to model  thermal  regimes in aquifers,  subsurface heat conduction,
 aquifer thermal  energy storage systems,  geothermal  reservoirs, thermal
 pollution of  aquifers, and natural  hydrogeological  convection  systems.

 Mesh construction is quite flexible  for arbitrary geometries employing
 quadrilateral  finite elements  in Cartesian or radial-cylindrical coordinate
 systems.   The mesh may be  coarsened  employing "pinch  nodes"  in areas where
 transport is  unimportant.   Permeabilities may be  anisotropic and may vary both
 in  direction  and  magnitude throughout the system  as may most other aquifer and
 fluid  properties.   Boundary conditions,  sources,  and  sinks  may be
 time-dependent.   A number  of input data  checks  are  made in  order to  verify the
 input  data set.   An option  is  available  for  storing the intermediate results
 and restarting simulation  at the intermediate time. An  option  to plot results
 produces  output which may  be contoured directly on  the  printer paper.  Options
 are also  available to  print fluid velocities  in the system  and to make temporal
 observations  at points in  the  system.

 CODE INPUT:   Standard groundwater flow and contaminant  transport parameters.

 CODE OUTPUT:  SUTRA provides,  as the  primary  calculated result, fluid
 pressures  and either solute concentrations or temperatures,   as they  vary with
 time,  everywhere  in  the  simulated subsurface  system.

 COMPILATION REQUIREMENTS:  SUTRA was written  in FORTRAN 77.

 EXPERIENCE REQUIREMENTS:   Extensive.

TIME REQUIREMENTS   Months.

CODE VERIFICATION:   Unknown.
                                      A-66

-------
DOCUMENTATION/REFERENCES:

Voss, C.I.  A finite-element simulation model for saturated-unsaturated,
     fluid-density-dependent groundwater flow with energy transport or
     chemically-reactive single species solute transport,  USGS Water Resources
     Investigations Report 84-4369.

SOURCE:  The code was developed by C. I. Voss of the USGS Water Resources
Department.  The code is in the public domain.
                                       A-67

-------
 CODE KAHE:   TRANUSAT
 PHYSICAL  PROCESSES:   Predicts the transient migration  of water  and
 contamination  in  unsaturated and saturated geologic media.
 DIMENSIONALITY:   One  and two dimensional.
 SOLUTION  TECHNIQUE:   Numerical, finite element.
 DESCRIPTION:   The fundamental equations  solved  are the pressure head
 formulation  of unsaturated water flow, Darcy's  equation, and  contaminant mass
 conservation.  The principal assumptions are  as  follows:
 t     fluid flow is described by Darcy's  equation;
 t     functional relationships exist for  pressure head  versus  moisture  content
      and  hydraulic conductivity versus moisture  content;
 t     solute  transport may include advection,  dispersion, diffusion, adsorption,
      and  first-order  reactions; and
 •     system  properties may vary spatially.
 CODE INPUT:  Standard groundwater flow and contaminant transport parameters.
 CODE OUTPUT:   Predictions of head and concentration.
 COMPILATION  REQUIREMENTS:  This one- and two-dimensional cartesian model
 includes  the following boundary conditions:
 •     specified pressure head;
 •     specified fluid flux;
 •     specified concentration;
 •     zero concentration gradient;
 •     cauchy boundary condition;  and
t     boundary conditions may vary temporally.
EXPERIENCE REQUIREMENTS:  Extensive.
TIME REQUIREMENTS  Months.
CODE VERIFICATION:  Unknown.
DOCUMENTATION/REFERENCES:
Pickens, J.F.;  Gillham,  R.W.; Cameron,  D.R.   (1979)   Finite-element analysis of
     the transport of water and  solutes in tile-drained soils.  Journal of
     Hydrology 40:243-264.
                                      A-68

-------
Pickens, J.F.; Gillham, R.W.  (1980)  Finite element analysis of solute
     transport under hysteretic unsaturated flow conditions."  Water Resour.
     Res. 16(6):1071-1078.

SOURCE:  TRANUSAT is a Geologic Testing Consultants, Ltd.,  proprietary code
developed at GTC.
                                       A-69

-------
t

§
 CODE NAME:  TRUST

 PHYSICAL PROCESSES:  Predicts transient fluid movement in a
 multi-dimensional, partially saturated or saturated, deformable porous media.

 DIMENSIONALITY:  Multi-dimensional.

 SOLUTION TECHNIQUE:  Numerical, finite difference.

 DESCRIPTION:  TRUST is based on a volumetric expression of the continuity
 equation applied to a flow region of finite volume.  The method of integral
 finite differences is used to represent the problem. Essentially, the model
 domain is a conglomeration of arbitrarily shaped volumes linked by connectors
 representing the porous media between volume nodal points.  Changes in pressure
 at each node are computed for each time step by a finite-difference algorithm
 that uses both explicit (point-by-point) and implicit (simultaneous) solution
 schemes.  Darcy velocities are calculated through finite-difference gradients
 of computed pressure heads.

 It features the processes of:

      pore desaturation;

      hysteresis in permeability and saturation behavior;

 •    fluid compressibility; and

 •    one-dimensional  (vertical)  deformation  of the soil  skeleton  due to water
      withdrawal.

 The discretization of the continuity equation  features  a  variable weighting
 scheme  that allows the solution  to range between  central  and  backward
 differencing.   The weighting  factor is  computed  by TRUST  at every time  step
 based on the rate  of pressure change.   Slowly  changing  phenomena  are weighted
 toward  central  differencing while  rapid changes  are weighted  toward  backward
 differencing.   Regardless  of  the weighting scheme,  the  discretized equation
 contains both  explicit  and implicit  parts.   The explicit  portion  of  the
 equation is  used to solve  for the  pressure change  over  the entire  domain.   At
 locations where the explicit  stability  constraint  for time step size  is
 violated, the  implicit  part of the equation  is applied  and added  to  the
 existing explicit  solution.   Thus, the  final matrix of  equation coefficients
 created  by the mixed explicit-implicit  algorithm can be partitioned  into those
 submatrices  that require simultaneous solution.

 The volumetric equation of mass  conservation balances external fluxes and
 internal fluid generation  with changes  in mass storage.  Nonlinearity in the
 formulation  arises  in the  pressure-dependent permeability  and fluid density in
 the external flux calculation, and pressure-dependent processes of fluid
 compressibility.  Nonlinearity also arises from soil skeleton deformation and
 pore desaturation in the mass storage calculation.  TRUST circumvents an
 iterative solution by using a point-slope prediction of pressure at the new
time level to calculate the dependent parameters.  Pressure is then solved with
the nonlinear equation set.
                                     A-70

-------
TRUST uses an iterative over-relaxation matrix solver to compute the pressure
changes at the new time level.  For smaller matrices, iterative solvers are
more efficient than direct solution techniques.  However,  convergence in highly
nonlinear problems can be poor.

Selecting an appropriate time step is crucial to the success of the
quasi-linearization approximation used in TRUST.  For this reason, time step
size is computed internally, based on the following criteria:

•    the maximum pressure change is maintained at an average of a specified
     value;

•    time- and pressure-dependent functions are not allowed to change more than
     an average of 1%; and

•    convergence of the matrix solution must be satisfied in under 80
     iterations.

Results of time steps that exceed any of these limits are discarded and the
time step is halved.

Mass balance error in TRUST  is monitored for each node and for the entire
system being modeled.  Model errors are discussed in the following categories:

•    inaccurate specifications or interpolation;

•    time truncation;

•    pressure truncation;

•    convergence;  and

t    machine roundoff.

Assumptions and simplifications  for TRUST  are:

t    parameters can  be  functions of time,  space, or  pressure;

•    hysteresis is modeled  approximately by  the  use  of scanning  curves;

•    fluid density,  volume,  void ratio, and  saturation are  functions of
     pressure only;

•    flow region  deforms  with  time;

•    deformation  of  the media  structure is one dimensional  according to
     Terzaghi's theory  of consolidation;

•    for  shallow  reservoirs,  deformation is  expressed at  the ground surface  as
     subsidence;

•    deep reservoirs  with overburden  are not resolved accurately by the
     consolidation computation;

•    soil  structure  has a constant volume  of incompressible  solids;
                                       A-71

-------
 •    during  the  time  interval,  the  elevation of the matrix is unchanged;
 t    time effects  of  consolidation  are  ignored;
 •    media properties are  isotropic; and
 t    the model is  best suited to soils  of moderate to high saturation.
 CODE INPUT:  System properties  required by TRUST are standard soil data
 usually included in a field sampling program:
 •    initial drying or wetting  state;
      elevation to  land  surface  from zero datum;
      average specific  gravity of flow region material;
      fracture length  or characteristic length;
      flow rate from well per unit aquifer thickness;
      permeability,  analytical,  or tabulated function  of pressure;
      specific storage, analytical,  or tabulated function of pressure;
 •    saturation,  tabulation function of pressure;
 •    reference void ratio;
 •    reference effective stress;
 t    void ratio as  a function  of effective stress;
 •    deformability  of matrix;
 t    swelling index;
 •     compression  index; and
 •     slope of the void ratio versus  log  permeability relationship.
 Fluid properties  required by TRUST can be  found  in  handbooks:
 •    viscosity;
 •    compressibility coefficient;
 •    density  at atmospheric pressure; and
 t    gravitational  constant.
 Initial  concentrations  for  TRUST are spatially distributed pressure heads and
 fluid generation rates.  An application of the steady-state option of the model
will generate a physically  meaningful initial condition. Boundary conditions
fall into three categories:
                                      A-72

-------
1)   flux;

2)   pressure head; and                                             '

3)   flux or pressure head, when controlling conditions are unknown.,

These boundary conditions can be constant, time-dependent, or sinusoidal.  Flux
boundary conditions occur at nodes with very small fluid mass capacity where a
fixed or variable fluid generation rate is specified.  These "artificial  nodes
are then connected to the actual surface nodes of the problem.  For fluid flux
from the problem domain, external connectors with constant or tabulated
functions of pressure or time are used to transfer fluid.  Seepage faces are
possible in TRUST with the assumptions of atmospheric pressure and one-way
efflux.

The time-dependent boundary conditions can be  input  on a  regional basis and can
be incremented by  a constant change.  Regional input of parameter data  is also
available.  Changes in the problem geometry can  be done by using the
model-scale factor to regionally reduce or measure the length scales of the
original  problem.

During  execution,  TRUST  performs a check  of the  input data for consistency.
Upon  detection of  an  error, diagnostic statements  are printed, which enable the
user  to trace  and  correct  the mistake.  The user's manual presents guidelines
for the input  of data tables  and possible actions  to correct  inaccurate
results.

CODE  OUTPUT:   TRUST  has  the  following output  results:

 t    input data  echo

 a    input error summary

 t    results of the first, second,  and last time steps are always printed

 •    system results
           net fluid flow into system
           average pressure change
           average fluid flow rates
           fluid mass capacity
           moisture content
           fluid generation rate                           ,             ,
           fluid generation amount

  •    levels of  output to  select from
           fluid pressure
           elevation
           pressure change during time step
           estimated  time  derivatives of  pressure
           fluid generation                                 ,
           total fluid content
           change  in  fluid content                          ,
           net fluid  transported into node by  internal and external
           information

                                       A-7 3

-------
 t    boundary information
           pressure
           net fluid flow into system from boundary node
           average fluid flow rate
           flow into node
           flow rate across connection

 •    nodal
           type
           volume
           density
           fluid capacity
           permeability
           conductance from node,  N
           time constant
           void ratio
           saturation
           preconsolidation stress

 §    connection
           area
           overall  conductance
           net  fluid  flow
           average  rate of fluid flow
           flow into  node
           flow rate  across  a connection
           fluid  transfer coefficient (for external connection)

 •     diagnostics
           flags  node changing to  a special node
           nonconvergence
           repeat of  a  time  step due  to criteria violation

 The user has control over the frequency and level of detail for the output.

 COMPILATION REQUIREMENTS:   TRUST  is  written in FORTRAN with versions existing
 on the following computer systems:

 t     UNIVAC

 t     CDC 6400/6000/7000

 •     DEC VAX 11/780.

 Single- and double-precision versions exist on the VAX.  The double-precision
 version has an accuracy  of  eight byte words,  which is standard on the UNIVAC
 and CDC machines.  In addition,  the VAX versions return run-time information to
 the terminal from which  they are run.

 Current problem limits are:

•    10 materials;

§    1 fluid property;

                                      A-74

-------
•    300 nodes;
•    600 internal connections;
t    20 external connections;
•    20 boundary nodes;
•    100 fluid generation tables;
t    300 initial conditions; and
•    100 table lengths.
EXPERIENCE REQUIREMENTS:  Extensive.
TIME REQUIREMENTS  Months.
CODE VERIFICATION:  TRUST has been verified against analytical  solutions of
flow problems,  in particular:
•    Theis solution for radial flow to a well; and
•    Carslaw and Jaeger solution  for a continuous point  source  in  an  isotropic
     three-dimensional medium.
Validation of  the consolidation mechanism of the code  has  been  performed in  the
following problem areas:
•    saturated flow;
•    unsaturated flow;
•    saturated-unsaturated  flow;  and
•    liquefaction.
Most of the validation tests are  based on results observed with laboratory
models,  and encompass  one-,  two-,  and three-dimensional  problems on both
rectangular and radial coordinate systems.
DOCUMENTATION/REFERENCES:
McKeon,  T.J.;  Tyler,  S.W.;  Mayer,  D.W.;  Reisenauer, A.E.  (1983)  TRUST-II
     utility  package:  partially  saturated  soil  characterization,  grid
     generation,  and  advective  transport analysis.  NUREG/CR-3443, U.S.  Nuclear
     Regulatory Commission,  Washington,  DC.
 Narasimhan, T.N.   (1975)   A unified  numerical  model for saturated- unsaturated
     groundwater flow.   LBL-8862,  Lawrence  Berkeley Laboratory, Berkeley,  CA.
 Narasimhan. T.N.;  Witherspoon,  P.A.   (1977)   Numerical model for
      saturated-unsaturated flow in deformable porous  media, 1.   theory.    Water
      Resour.  Res.  13(3):657-664.
                                       A-75

-------
Narasimhan, T.N.; Witherspoon, P.A.; Edwards, A.L.  (1978)  Numerical model for
     saturated-unsaturated flow in deformable porous media, 2. The algorithm
     Water Resour. Res. 14(2):255-261.

Narasimhan, T.N.; Witherspoon, P.A.  (1978)  Numerical model for saturated-
     unsaturated flow in deformable porous media, 3. Applications.  Water
     Resour. Res. 14(6):1017-1034.

Reisenauer, A.E.; Key, K.T.; Narasimhan, T.N.; Nelson, R.W.  (1982)  TRUST:  a
     computer program for variably saturated flow in multidimensional
     deformable media.  PNL-3975, Pacific Northwest Laboratory, Rich!and, WA.

SOURCE:  TRUST was developed by T. N. Narasimhan at the Lawrence Berkeley
Laboratory.  The code is in the public domain.
                                     A-76

-------
NUMERICAL SOLUTE TRANSPORT (SATURATED)
                 A-77

-------
CODE NAME:  CHAIN!
PHYSICAL  PROCESSES:  Simulates the transport of radionuclides in a fractured
porous medium.
DIMENSIONALITY:  Two dimensional  (x,y cartesian).
SOLUTION  TECHNIQUE:  Numerical, finite element.
DESCRIPTION:  The CHAINT model simulates multicomponent nuclide transport in
a fractured-porous medium.  The processes modeled  include advection,
dispersion/diffusion,  sorption, chain decay coupling, and mass release. The
computational method is based  on  a finite-element  solution of the system of
equations.  Continuum  portions of the medium are modeled as a single porosity
system using two-dimensional isoparametric elements.'Discrete fractures are
modeled using isoparametric line  elements embedded along the sides of the
two-dimensional elements.  Principal input to the  code is the groundwater flow
calculation obtained with the  MAGNUM2D code (or a  comparable nonisothermal flow
model).
The principal assumptions of the  code are:
•    The  diffusive flux Jri is assumed to be Fickian.
•    Radionuclide transport occurs only in the fractures.
•    Sorption may be represented  by equilibrium adsorption.
In addition, the assumptions incorporated in MAGNUM2D must also be incorporated
in CHAINT, these being:
•    The  fractured-porous medium  is nondeformable.
•    The  fluid is slightly compressible.
t    Flow is laminar (Darcian).
•    Macroscale (REV)  hydraulic gradients are independent of fracture
     orientation or geometry.
•    The  fluid system  is single phase.
•    The medium is fully saturated.     ,
•    Moisture is stored in both primary, apd secondary pores.
•    Flow in fractures is governed by a nonisothermal version of Darcy's Law.
•    Flow between primary and secondary pores depends on the  difference between
     primary and secondary heads..                   ,
•    Heat flux is governed by the convection-diffusion equation.
                                      A-78

-------
•    Conservation of mass applies separately in the primary and secondary
     storage systems, but conservation of energy applies in the system as a
     whole.
CODE INPUT:  As previously mentioned, principal input to the code is the
groundwater flow calculation obtained with the MAGNUM2D code (or a comparable
nonisothermal flow model).  Other necessary inputs include:
t    mesh geometry;
•    half lives of the radionuclide contaminants;
•    retardation factor of each contaminant;
•    velocity field;
t    mass dispersion tensor;
•    nuclide splitting;
•    decay constants;
•    secondary porosity;
•    mass  source term;                  ;
t    fluid density;                         ••.•,,
t    initial parent  concentrations;  and
•    initial daughter  product  concentrations
                               i,   • .• •  /     <    •   , •
CODE OUTPUT:  Output for  CHAIN! consists  of the  concentration of each
radionuclide in  the  fractures  at  each  time stepJ
COMPILATION  REQUIREMENTS:  Unknown.
EXPERIENCE REQUIREMENTS:   Extensive.
TIME REQUIREMENTS:   Months.
CODE VERIFICATION:   The code CHAINT  was tested against  an  analytical solution
based  on  the uranium decay series.   The code  is  in the  public domain.
DOCUMENTATION/REFERENCES:
King,  I.P.;  Mclaughlin, D.B.;  Norton,  W.R.; Baca,  R.G.; Arnett, R.'C.   (1981)
     Parametric  and  sensitivity analysis  of waste  isolation  in  a basalt
     medium.  Rockwell Hanford Operations Report RHO-BWI-C-94.
Baca,  R.G.;  Arnett,  R.C.;  King,  I.P.   (1981)   Numerical modeling of flow  and
     transport  in  a  fractured-porous rock system.   Rockwell  Hanford Operations
     Report  RHO-BWI-SA-113.
                                       A-79

-------
Neretnieks, I.  (1980)  Diffusion in the rock matrix:  an important factor of
     radionuclide retardation?  J. Geophys. Res. 84(B8):4379-4397.

SOURCE:  CHAIN! was developed by*Resource Management Associates, Lafayette,
California, for Rockwell Hahfbrd Operations/
                                      A-80

-------
CODE NAME:  DUGUID-REEVES

PHYSICAL PROCESSES:  Predicts contaminant transport for use on the flow model
results produced by the REEVES-DUGUID flow model.

DIMENSIONALITY:  Two dimensional.

SOLUTION TECHNIQUE:  Numerical, finite element.

DESCRIPTION:  The model considers advection, dispersion, and exchange of one
chemical constituent which may involve radioactive decay under partially
saturated flow conditions.  The single chemical component exchange reaction is
handled through an equilibrium exchange coefficient modified by the moisture
content when partially-saturated conditions exist.  The spatial integration is
accomplished through the Galerkin finite-element approach using linear basis
functions.  The time integration uses a modified Crank-Nicholson
finite-difference form.

A special, rather standard numbering scheme for element nodes is used to reduce
the matrix bandwidth that must be stored and operated upon for solution.
Special treatment of various boundary condition is also used to maintain the
desired matrix form.  The actual solution is standard Gaussian elimination by
decomposition into the product of upper and lower triangular matrices.  The
lower triangular matrix is used to modify the right-hand side for
back-substitution into the upper triangular matrix to obtain the solution.

CODE INPUT:  Input for DUGUID-REEVES is the saturated/unsaturated flow model
results from REEVES-DUGUID.

CODE OUTPUT:  Predicts contaminant concentrations.

COMPILATION REQUIREMENTS:  DUGUID-REEVES is written in  FORTRAN and is
operational on IBM systems.

EXPERIENCE REQUIREMENTS:  Extensive.

TIME REQUIREMENTS:  Months.

CODE VERIFICATION:  Unknown.

DOCUMENTATION/REFERENCES:

Duguid, J.O.; Reeves, M.  (1976)  Material transport through porous media:  a
     finite element Galerkin model.  ORNL-4928, Oak Ridge National Laboratory,
     Oak  Ridge, TN, March.

Reeves, M.; Duguid, J.O.  (1975)  Water movement through saturated-unsaturated
     porous media:  a finite-element Galerkin model.  ORNL-4927, Oak Ridge
     National Laboratory, Oak Ridge, TN.

SOURCE:   The code was developed by J. 0. Duguid of Battelle Memorial
Institute and M. Reeves of Oak Ridge National Laboratory.  The code is in the
public domain.
                                      A-81

-------
CODE NAME:  GROVE/GALERKIN

PHYSICAL PROCESSES:  This model  employs Galerkin finite-element methods to
solve mass transport equations.  The model successfully simulates solute
transport for an unreactive  conservative solute chloride, a solute with a
first-order irreversible rate reaction, radioactive decay, and a solute with
equilibrium controlled  ion exchange.

DIMENSIONALITY:  Three  dimensional.

SOLUTION TECHNIQUE:  Numerical,  finite element.

DESCRIPTION:  The partial differential equation that describes the transport
and reaction of chemical solutes in porous media was solved using the Galerkin
finite-element technique.  These finite elements were superimposed over finite
difference cells used to solve the flow equation.  Both convective and flow due
to hydraulic dispersion were considered. Linear and Hermite cubic
approximations (basis functions) provided satisfactory results; however, the
linear functions were found to be computationally more efficient for
two-dimensional problems.  Successive overrelaxation (SOR) and iteration
techniques using Tchebyschef polynomials were used to solve the space matrices
generated using the linear and Hermite cubic functions, respectively.
Comparisons of the finite-element methods to the finite-difference models arid
to analytical results indicate that a high degree of accuracy may be obtained
using the method outlined  The technique was applied to a field problem
involving an aquifer contaminated with chloride, tritium, and 90Sr.

CODE INPUT:  Unknown.

CODE OUTPUT:  Unknown.

COMPILATION REQUIREMENTS:  GROVE/GALERKIN is written in FORTRAN and is
operational on an IBM 360 system.

EXPERIENCE REQUIREMENTS:  Extensive.

TIME REQUIREMENTS:  Months.

CODE VERIFICATION:  Unknown.

DOCUMENTATION/REFERENCES:

Grove, D.B.  The use of Galerkin finite-element methods to solve mass-transport
     equations.  Water Resources Investigation 77-49, U.S. Geological Survey,
     Water Resources Division, Denver, CO.

SOURCE:  This code is written by D.  B. Grove of the U.S.  Geological  Survey,
Water Resources Division.  The code is in the public domain.
                                      A-82

-------
CODE NAME:  1SOQUAD,  ISOQUAD2
PHYSICAL PROCESSES:  Predicts contaminant transport.
DIMENSIONALITY:  Two dimensional.
SOLUTION TECHNIQUE:  Numerical,  finite element.
DESCRIPTION:  This model uses a Galerkin approximation with various basis
functions, with a finite-element integration scheme to solve the conservative
transport equation.  The time integration is performed through a backward
difference time scheme.
CODE INPUT:  Unknown.
CODE OUTPUT:  Unknown.
COMPILATION REQUIREMENTS:  ISOQUAD and ISOQUAD2 are written in FORTRAN and
implemented on an IBM 360/91 machine.
EXPERIENCE REQUIREMENTS:  Extensive.
TIME REQUIREMENTS:  Months.
CODE VERIFICATION:  Unknown.
DOCUMENTATION/REFERENCES:
Pinder, 6.F.   (1973)  A Galerkin-finite-element simulation of groundwater
     contamination of Long Island, New York.   Water Resour. Res.
     9(6):1657-1669.
SOURCE:   The codes were written by George Pinder of Princeton University and
Emil Frind of  the University  of Waterloo.  The code is  in the public  domain.
                                       A-83,

-------
 CODE NAME:  KONBRED, USGS2D-MOC

 PHYSICAL PROCESSES:  KONBRED simulates groundwater flow and solute transport
 in one or two dimensions.  Radioactive decay is not included in the program as
 originally published.  A modified version has been prepared which incorporates
 decay of single species but omits formation of radioactive daughter products.

 DIMENSIONALITY:  One or two dimensional.

 SOLUTION TECHNIQUE:  Groundwater flow --  numerical,  finite difference.
 Transport -- random walk.

 DESCRIPTION:  KONBRED solves the groundwater flow equation by a
 finite-difference method.  It then computes solute transport in the calculated
 flow field by the method of characteristics.   Both steady-state and transient
 flows can be calculated, and the aquifer  may be heterogeneous and anisotropic.
 Forces resulting from differences in  temperature or  concentrations of dissolved
 solids are not considered.

 Advective transport is computed by tracking particles,  and a finite-difference
 method is used after each step to treat dispersion,  fluid  sources and sinks,
 and  velocity divergence.  The code can accommodate injection and  withdrawal
 wells, diffuse leakage,  and a variety of  boundary and  initial  conditions.   The
 modified  version includes radioactive decay (but not formation  of radioactive
 daughters)  and equilibrium  sorption.   The code  represents  a  two-dimensional
 area as a rectangular network of equally-spaced  nodes.  As presently written,
 there can be no  more than 20  rows and 20 columns  of nodes.   The principle
 assumptions  of the  code  include:

 §     Darcy's  Law is  valid and hydraulic-head gradients  are the  only  significant
      driving  mechanism for  fluid  flow.

 •     The  porosity and  hydraulic conductivity of  the aquifer  are constant with
      time, and porosity  is  uniform  in space,                                ,

 •     Gradients of fluid  density,  viscosity, and  temperature  do  not affect the
      velocity  distribution.

 t     The  two-dimensional  solute transport equation is valid.

 •     Sorption may be represented  as equilibrium adsorption.

 •     Vertical variations  in head  and concentration are negligible.

 •     The  aquifer is  homogeneous and isotropic with respect to the coefficients
      of longitudinal and  transverse dispersivity.

 CODE  INPUT:  The principal  inputs to the original versions are  as, follows:

 t    transmissivity tensor;

•    aquifer thickness at each node;

•    diffuse recharge and discharge at each node;

                                      A-84

-------
•    initial head at each node;
•    initial solute concentration at each node;
•    storage coefficient;
•    location of no-flow boundaries;
t    effective porosity;
•    longitudinal dispersivity;
•    lateral dispersivity;
•    locations of wells;
•    pumping rate of each well;
•    solute concentrations  at  each  injection well; and
•    pumping period.
The  revised version requires additional  inputs describing  sorption and
radioactive decay.
CODE OUTPUT:  The principal output  are the heads and concentrations. These
can  be  printed out either  after  each  time step at  up to  5  "observation wells"
or at all  nodes  after  each  50  time  steps.
COMPILATION REQUIREMENTS:   The original  program is written in  FORTRAN  IV  and
 is compatible with many computers.   It has been run successfully  on  Honeywell,
 IBM, DEC,  Univac,  and  CDC  computers.  The revised  program  is written in  FORTRAN
 77 and, apparently,  has run on IBM  and CDC computers.
 EXPERIENCE REQUIREMENTS:  Extensive.
 TIME REQUIREMENTS:   Months.
 CODE VERIFICATION:   KONBRED has  been tested  by comparison  with several
 analytical solutions.   Results from these  comparisons  are  included  in  the
 documentation  (ref.  1) and include  both  one-dimensional  steady-state flow and
 plane radial  steady-state flow.
 This code (or earlier versions of it) has  been applied to  a wide  variety of
 field problems.   These include 1) chloride movement  at the Rocky  Mountain
 Arsenal (ref.  2);  2)  chloride buildup in a stream-aquifer  system  (ref.  3);  and
 3) radionuclide transport at  INEL (ref.  4).
 DOCUMENTATION/REFERENCES:
 Konikow, L.F.;  Bredehoeft, J.D.   (1978)   Computer model  of two-dimensional
      solute transport and dispersion in  groundwater.   Techniques  of
      Water-Resources Investigations of the United States Geological  Survey,
      Book 7,  Chapter C2.
                                       A-85

-------
Konikow, L.F.   (1977)  Modeling  chloride movement  in the alluvial aquifer at
     the Rocky  Mountain Arsenal,  Colorado.  U.S. Geological Survey Water-Supply
     Paper 2044.

Konikow, L.F.;  Bredehoeft, J.D.   (1974)  Modeling  flow and chemical quality
     changes in an irrigated stream-aquifer system.  Water Resour. Res.
     10(3):546-562.

Robertson, J.B.  (1974)  Digital  modeling of radioactive and chemical waste
     transport  in the Snake River plain aquifer at the National Reactor Testing
     Station, Idaho.  U.S. Geological Survey Open-File Report  IDO-22054.

Tracy, J.V.  (1982)  User's guide and documentation for adsorption and decay
     modifications to the U.S.G.S. solute transport model.  U.S. Nuclear
     Regulatory Commission Report NUREG/CR-2502.

SOURCE:  The model was developed  by L. F. Konikow  and J. D. Bredehoeft of the
U.S. Geological Survey.  The modifications were made by J. V. Tracy of ERTEC.
The code is in the public domain.
                                      A-86

-------
 CODE NAME:   DPCT (Deterministic-Probabilistic Contaminant Transport)
 PHYSICAL PROCESSES:   Predicts  groundwater flow and  contaminant transport
 accounting  for  advection,  dispersion,  radioactive decay, and equilibrium
 sorption for a  single contaminant.
 DIMENSIONALITY:   Two  dimensional.
 SOLUTION TECHNIQUE:   Flow  distribution -- numerical,  finite element. Solute
 transport -- particle-tracking method.
 DESCRIPTION: The code treats  a two-dimensional  vertical cross-section.
 Almost  any  water table and geologic configuration is  permissible, and there are
 a variety of allowable boundary conditions.   Water  flow is steady state.
 The  cross section is  divided into a rectangular  array of cells.  The head
 distribution is  found by the finite-element method.   Solute transport is then
 treated  by  tracking the motion of individual  particles.
 DPCT will calculate the long-term effects of  a repository for specified
 scenarios,  if used in conjunction with  a  biosphere  transport code (e.g.,
 PABLM).   The code solves an inherently deterministic  problem -- solute
 transport with known  velocity  and dispersion  --  in  a  probabilistic manner.  It
 does not treat any probabilistic problems.
 The  principal assumptions  of the code  are:
 •     A treatment in a two-dimensional  cross section is acceptable.
 •     The solute  transport  equation is  valid.
 •     Sorption may be  represented as equilibrium  adsorption with the
      distribution coefficient  given by  Equation  8.25.
 •     Principal axes of the transmissivity tensor are  parallel to coordinate
      axes everywhere.
 •     Groundwater flows  are steady state.
 CODE INPUT:   Inputs for DPCT include:
 •     hydraulic conductivity (horizontal and vertical) at each node;
 t     porosity at  each  node;
 •     longitudinal  dispersivity at each node;
 t     ion exchange  capacity at  each node;
t     location of water  table;
•     boundary conditions;   and
•    contaminant  input  rates and locations.
                                      A-87

-------
CODE OUTPUT:  The principal outputs are maps of velocity or head and of
contaminant concentration at any times selected by the user.  A wide variety of
other optional outputs are available.

COMPILATION REQUIREMENTS:  The program is written in FORTRAN IV and has .been
run on an Amdahl 470V/7 computer.  The code is in the public domain.

EXPERIENCE REQUIREMENTS:  Extensive.

TIME REQUIREMENTS:  Months.

CODE VERIFICATION:
                            (1980)  A deterministic-probabilistic model for
                           ,  U.S. Nuclear Regulatory Commission Report
DOCUMENTATION/REFERENCES:

Schwartz, F.W.; Crowe, A.
     contaminant transport
     NUREG/CR-1609, August

CGS, Inc.  (1980)  Scenario development and evaluation related to the risk
     assessment of high level radioactive waste repositories.  U.S. Nuclear
     Regulatory Commission Report NUREG/CR-1608, August.

Schwartz, F.W.  (1978)  Application of probabilistic-deterministic modeling to
     problems of mass transport in groundwater system.  Third International
     Hydrology Symposium, Ft. Collins, pp. 281-296.

Detailed derivations are given in:

Ahlstrom, S.W.; Foote, H.P.; Arnett, R.C.; Cole, C.R.; Seme, R.J.  (1977)
     Multicomponent mass transport model:  theory  and numerical implementation
     (discrete-parcel-random walk version).  Battelle, Pacific Northwest
     Laboratory Report PNL-2127.

SOURCE:  DPCT was developed by Franklin Schwartz and A. Crowe of CGS, Inc.
                                       A-88

-------
CODE NAME:  MMT (Multicomponent Mass Transport)

PHYSICAL PROCESSES:  Predicts the transport profile of dissolved contaminants
in groundwater.

DIMENSIONALITY:  One dimensional.

SOLUTION TECHNIQUE:  Discrete-parcel random-walk method.

DESCRIPTION:  In the formulation of a mathematical model for simulating
transport processes in the environment, the system of interest can be viewed as
a continuum of matter and energy or as a large set of small discrete parcels of
mass and energy.  The latter approach is used to formulate the discrete-parcel
random-walk transport model.  Each parcel has associated with it a set of
spatial coordinates, as well as a set of discrete quantities of mass and
energy.  A parcel's movement is assumed to be independent of any other parcel
in the system.  A Lagrangian scheme is used to compute the parcel advection,
and a Markov random-walk concept is used to simulate the parcel diffusion and
dispersion.  The random-walk technique is not subject to numerical dispersion,
and it can be applied to three-dimensional cases with only a linear increase in
computation time. A wide variety of complex source-sink terms can be included
in the model with relative ease.  Examples of the model's application include
the areas of oil spill drift forecasting, coastal power plant effluent
analysis, and solute transport in groundwater systems.

The principal assumptions of the code are:

•    The effect of changing atmospheric pressure is negligible

•    Flow patterns are independent of the chemical composition or temperature
     of the groundwater solution.

•    Hydrodynamic dispersion processes can be included with molecular
     diffusion.

•    Relative mass flux can be adequately described by Pick's First Law.

0    Darcy's Law holds for description of saturated groundwater flow.

•    Total mass density of the mixture is constant.

•    The number or type of particles does not significantly alter the flow
     properties of the host medium.

•    The one-dimensional solute transport equation is valid.

•    Sorption may be represented by equilibrium adsorption.

CODE INPUT:  Inputs for MMT include:

•    retardation coefficient;

t    dispersion;
                                    '  A-89

-------
 t     half lives  for  all  nuclides;
 •     path length;
 t     groundwater velocity;
 •     flow tube size;
 •     initial  inventory;
 •     time after  repository closure when the breach occurs;
 •     leach  information to control entry of waste into groundwater system; and
 •     a mapping illustrating the parent-daughter relationships.
 CODE  OUTPUT:  The principal outputs of MMT are the release rates of the
 contaminants.  Both  printed and graphic output are available and the output can
 be communicated  to codes which calculate doses to humans.
 COMPILATION REQUIREMENTS:  MMT is written in FLECS, a higher-order language
 which compiles into  FORTRAN, and is operational on a VAX machine.  The code is
 in the public domain.
 EXPERIENCE REQUIREMENTS:  Extensive.
 TIME REQUIREMENTS:  Months.
 CODE VERIFICATION:  MMT was compared to analytic results from GETOUT for a
 variety of problems.
 DOCUMENTATION/REFERENCES:
Ahlstrom, S.W.; Foote, H.P.; Arnett, R.C.;  Cole,  C.R.; Seme, R.J.   (1977)
     Multicomponent mass transport model:   theory and numerical  implementation
     (discrete-parcel-random-walk version).   Pacific Northwest Laboratory
     Report BNWL-2127, May.
Washburn, J.F.; Kaszeta,  F.E.;  Simmons,  C.S.;  Cole,  C.R.  (1980)
     Multicomponent mass transport model:   a model  for simulating migration  of
     radionuclides in groundwater.   Pacific  Northwest Laboratory Report
     PNL-3179, July.
SOURCE:   MMT was  developed at Battelle,  Pacific Northwest Laboratories.
                                      A-90

-------
CODE NAME:  PINDER

PHYSICAL PROCESSES:  Predicts the movement of groundwater contaminants.

DIMENSIONALITY:  Three dimensional.

SOLUTION TECHNIQUE:  Numerical,  finite element.

DESCRIPTION:  The Galerkin method of approximation in conjunction with the
finite-element method of analysis is used to simulate the movement of
groundwater contaminants.  In solving groundwater flow and mass transport
equations, this approach allows  a functional representation of the dispersion
tensor, transmissivity tensor, and fluid velocity, as well as an accurate
representation of boundaries of irregular geometry.

CODE INPUT:  Unknown.

CODE OUTPUT:  Unknown.

COMPILATION REQUIREMENTS:  PINDER is written in FORTRAN and implemented on an
IBM 360 or 390 machine.

EXPERIENCE REQUIREMENTS:  Extensive.

TIME REQUIREMENTS:  Months.

CODE VERIFICATION:  A field application of the method to chromium
contamination on Long Island, New York, shows that accurate simulations can be
obtained.

DOCUMENTATION/REFERENCES:

Pinder, G.F.  (1973)  A Galerkin-finite element simulation of groundwater
     contamination on Long Island, New York.  Water Resour. Res.
     9(6):1657-1669.

SOURCE:  PINDER was developed by G. F. Pinder at Princeton University. The
code is in the public domain.
                                      A-91

-------
 CODE NAME:   ROBERTSON1

 PHYSICAL PROCESSES:   Predicts the'movement  of radionuclides  by groundwater
 and  soil  transport.

 DIMENSIONALITY:   Two  dimensional.

 SOLUTION TECHNIQUE:   Analytical  and  numerical,  finite difference.

 DESCRIPTION:  Aqueous chemical and low-level  radioactive effluents have  been
 disposed  to  seepage ponds since  1952 at  the Idaho National Engineering
 Laboratory.  The  solutions percolate toward the Snake River  Plain aquifer
 (9,135 m  below) through  interlayered basalts  and unconsolidated sediments and
 an extensive zone of  groundwater perched on a sedimentary layer about 40 m
 beneath the  ponds.  A three-segment  numerical  model was developed to simulate
 the  system,  including effects of convectional  hydrodynamic dispersion,
 radioactive  decay, and adsorption. The first  segment uses an analytical
 solution  to  simulate  transport from  the  ponds to the 25-m thick perched water
 lens, assuming steady vertical flow  through a 15-m long saturated homogeneous
 column.   The second segment simulates two-dimensional horizontal transport  in
 the  perched  water body using finite-difference methods, assuming complete
 vertical  mixing with  vertical leakage from  the bottom.  The  third segment
 simulates simulates vertical solute  transport from the perched water body
 toward the aquifer by assuming unsatu'rated,  but steady water flow in a series
 of contiguous, nonhomogeneous independent vertical columns.  The transport
 equation  is  solved by a  "hop-scotch"  finite-difference scheme for each column.
 Simulated hydraulics  and solute  migration patterns for all segments agree
 adequately with the available fteld  data.   The model can be  used to project
 subsurface distributions of waste solutes under a variety of assumed conditions
 for  the  future.

 CODE INPUT:  Standard hydrologic and  transport parameters.

 CODE OUTPUT:  Predicts head and  concentration.

 COMPILATION  REQUIREMENTS:  ROBERTSONl is  written in FORTRAN and implemented
 on a CDC  machine.

 EXPERIENCE REQUIREMENTS:  Extensive.

TIME REQUIREMENTS:  Months.

CODE VERIFICATION:  The code has been verified by the Idaho site application.

DOCUMENTATION/REFERENCES:

Robertson, J.B.   (1974)   Digital  modeling of radioactive and chemical  waste
     transport in the Snake River plain aquifer at the National  Reactor Testing
     Station, Idaho.   U.S.  Geological Survey Open-File Report,  AEC No.
      IDO-22054, 41 pp.

SOURCE:   ROBERTSONl was  developed by J.  B. Robertson of the U.S.  Geological
Survey,  National  Center.   The code is in the public domain.
                                      A-92

-------
CODE NAME:  ROBERTSON2

PHYSICAL PROCESSES:  Predicts groundwater transport of radioisotopes.

DIMENSIONALITY:  Two dimensional.

SOLUTION TECHNIQUE:  Numerical, finite-difference, and
method-of-characteristics solution technique.

DESCRIPTION:  ROBERTSON2 was developed to predict radioisotope migration at
the National Reactor Testing Station.  The influences on migration include
space and time variations in groundwater flow, hydraulic dispersion,
radioactive decay, ion exchange, and other chemical reactions.  These processes
are included in problems of movement of radioactive wastes in groundwater;
design and analysis of tracer tests in groundwater systems; and analysis of
natural isotope distributions in groundwater. The model is composed of two
coupled phases:  the first simulates the hydraulics, and the second simulates
solute transport.  The hydraulic phase solves the transient, two-dimensional,
partial-differential equation of groundwater flow for a bounded,
two-dimensional, one-layer aquifer, using finite-difference techniques
(iterative, alternating direction, implicit  scheme).  This method is described
by  Bredehoeft  and  Pinder. Groundwater velocity vectors are computed  by this
equation, for  every grid point  at  any finite time step.  The velocities are
transferred to the solute transport phase of the model that solves the
transient particle differential  equation of  t.he transport, using the method  of
characteristics.   The method  characterizes the dissolved nuclides by
characteristic imaginary particle.  Considered terms of the solute transport
phase are:  hydraulic dispersion,  convective.transport, aquifer compression
factors,  sources  and  sinks,  radioactive  decay, and  sorption.

CODE INPUT:   Input for  the model  includes transmissivity,  storage
coefficient,  boundaries,  hydraulic dispersion  coefficients (transverse  and
longitudinal),  initial  concentration  distributions,  ion exchange, distribution
coefficient,  radioactive  decay constant,  and source-sink .inputs.

CODE OUTPUT:   Predicts  head  and concentration.

 COMPILATION REQUIREMENTS:   ROBERTSON2 is written  in FORTRAN  and  implemented
 on a CDC machine.

 EXPERIENCE REQUIREMENTS:   Extensive.

 TIME REQUIREMENTS:  Months.

 CODE VERIFICATION:  Unknown.

 DOCUMENTATION/REFERENCES:

 Bredehoeft, J.D.; Pinder, G.F.  (1973)  Mass transport in flowing
      groundwater.  Water Resour. Res. 9(1):194-210.

 SOURCE:  ROBERTSON2 was developed by J.  B.  Robertson of the U.S.  Geological
 Survey, National  Center.   The code is in the public domain.
                                       A-93

-------
 CODE NAME:  SWENT

 PHYSICAL PROCESSES:  Predicts fluid, energy, and solute radionuclide
 transport.

 DIMENSIONALITY:  One dimensional, two dimensional, axisymmetric (r-z), and
 three dimensional.             •                 ,

 SOLUTION TECHNIQUE:  Numerical, finite difference.

 DESCRIPTION:  SWENT is based on the coupled, transport solution of the fluid,
 energy,  and solute transport equations.  Fluid density and viscosity in this
 code are treated as functions of pressure,  temperature, and solute
 concentration; thus, the transport equation set being solved is nonlinear.
 SWENT can be applied in one-dimensional,  two-dimensional  (x-y,  x-z),
 axisymmetric (r-z), and three-dimensional  (x-y-z)  heterogeneous geologic
 systems.  The code also features a comprehensive radionuclide transport model
 that includes a radionuclide data base and  the ability to account  for the
 generation and transport of daughter products in straight or branched decay
 chains.   The user has the option to solve  for any  or all  of the dependent
 variables.   Basic processes addressed by  SWENT include:

 •     confined flow of a two-component,  single-phase fluid;

 •     pressure-dependent aquifer porosity;

 •     vertical  recharge;

 •     detailed well-bore modeling based on well  characteristics;

 §     convective and conductive heat transport;

 •     solute advection;

 •     hydrodynamic dispersion;

 t     first-order decay  reactions;

 •     equilibrium isothermal  sorption;  and

 •     salt dissolution.

 SWENT has an  aquifer reservoir model  that is coupled  to a  well-bore model.   The
 well-bore model  offers  a  more  detailed  account  of  the well-bore  physics  than is
 possible with  the finite  difference  mesh of  the general problem.   Results from
 the  modeling  of the well  are then  applied as boundary conditions to the  aquifer
 reservoir.  The radionuclide modeling  is not coupled  to the  flow,  energy, and
 solute computations;  consequently, this part of the simulation is  performed
 independently  after the pressure field  has been  established.

 SWENT has an option  that  allows the  effect of the  surrounding aquifer to be
 incorporated into the aquifer boundary conditions without  actually modeling the
 regional  problem.  Aquifer influence functions  allow  subregional modeling by
 posing boundary  conditions that respond as if a  larger  aquifer were being
modeled.  Generally, these influence functions are necessary when the

                                      A-94

-------
simulation time is long enough for calculated changes to occur at peripheral
grid blocks.
The well-bore model uses a direct simultaneous solution of the energy equation
to calculate the pressure and temperature changes over the well-bore depth.  On
the other hand, the reservoir model is based on a semi-implicit
finite-difference scheme of the flow, energy, and solute equations.  The
dependent variables in this formulation, pressure, temperature, and
concentration appear in the space derivatives at the new time level.
An iterative procedure is used for the problem solution.  Changes in the
pressure, temperature, and concentration are computed by applying each equation
sequentially with updated values for all dependent variables.
Transmissibilities and dispersion, on the contrary, are always treated at the
old iterate level.  The iterative procedure ceases when the fractional density
change falls below an internal tolerance.
Solution of the simultaneous equation is performed by a direct, reduced-band
Gaussian elimination technique or an iterative, two-line, successive,
overrelaxation method.  Each method has an optimal range of applicability which
is presented in the model documentation.
The principal assumptions of the SWENT code are:
•    only a single-phase fluid exists;
•    porous medium is saturated with fluid;
•    flow is laminar and governed by Darcy's Law;
•    kinetic energy is negligible in the energy balance;
•    fluid viscosity is an exponential function of temperature, or  a power law
     function of concentration;
•    salt dissolution is a first-order reaction;
•    linear equilibrium, sorption;
•    hydrodynamic  dispersion  is a linear function of velocity; and
•    effects of hydrodynamic  dispersion and molecular diffusivity  are additive.
CODE INPUT:  The  input requirements  of SWENT are  as follows:
t    fluid  compressibility;
•    rock compressibility;
•    fluid  thermal expansion  factor;
•    heat capacity of rock;
•    resident  and  injection  fluid densities;
                                      A-95

-------
 •    resident and injection fluid viscosities;

 t    thermal conductivity of porous medium;

 §    hydrodynamic dispersivities;

 §    molecular diffusivity;

 §    porosity;

 §    hydraulic conductivity;

 t    well  data
           depth
           diameter
           roughness
           heat transfer coefficient
           pressure conditions;  and

 •    adsorption distribution  coefficient.

 The specified boundary conditions of SWENT  are as  follows:

 •    constant pressure,  temperature,  concentration (Dirichlet);

 t    radiation boundary condition (Cauchy);

 t    steady-state  or transient  aquifer  influence functions;

 •    heat  loss to  overburden  and  underburden;

 •    well  specification  of  pressure,  temperature;  and concentration;

 •    radioactive sources; and

 •     recharge.
SWENT has many features that reduce the tedium of data entry:  1) grid-block
characteristics can be entered on a regional basis; 2) boundary conditions are
entered only when an update occurs; 3) any system of units can be used with the
proper entry of conversion factors; 4) variable timestepping can be performed
by a user-defined function; and 5) inclusion of a radionuclide data library
There is an input data checker in SWENT that alerts the user to up to 69 error
conditions.

CODE OUTPUT:  SWENT produces a very readable line printer output of computed
results.  The detail and frequency of the output is at the option of the user
Specifically, the available output include:
•

t

•
input data echo;

Darcy velocities;

flow, heat, diffusive transmissibilities; viscosity, enthalpy,
dispersivities; thermal conductivities in all grid blocks;

                                 A-96

-------
•    fluid, energy, component, and nuclide balance;

•    maximum pressure, temperature, concentration changes, and the
     corresponding grid blocks;

•    well performance summary (pressure, concentration, temperature) at well
     head and bottom hole, and water, heat, and inert component production
     rates and their integrated values over time;

t    aquifer influx rates for water, heat, and inert components;

•    nuclide discharge rates at biosphere grid blocks;

•    integrated nuclide discharge rates to different regions;

•    pressure, temperature, component concentrations, and nuclide
     concentrations at all grid blocks;

•    two-dimensional  area! contour maps for pressure, temperature,  brine,
     concentration, and nuclide concentration; and

•    plots of pressure, temperature, and  brine concentration  for observed  and
     calculated values at wells.
                                                         *                -
COMPILATION  REQUIREMENTS:  SWENT  is  written in FORTRAN with versions  existing
on  the  following  computer systems:                  >

•    CDC 7600; and                                 ,

9    DEC VAX 11/780.

The code was designed to  be executed on a CDC  7600 and  includes external
 references to CDC-specific functions Conversion  to the  VAX  11/780 was hampered
 by  these references.   The VAX version of  the code has  dispensed with  the
dynamic allocation of core storage available on CDC hardware  and  replaced  other
 CDC routines with function subroutines  coded into the  software.   To equal  the
 level  of accuracy and range of the CDC  7600, real  variables  in the code were
 double-precisioned from  four- to  eight-byte decimal  words with an  extended
 range  option.

 Current limits  to the code are:

 •     20 wells;                             '                             •

 •     7 overburden layers;

 e     7 underburden layers;

 t     50 aquifer influence functions; and

 t     10 entries in the viscosity and temperature tables.

 EXPERIENCE REQUIREMENTS:  Extensive.

 TIME REQUIREMENTS:  Months.                 '
                                       A-97

-------
CODE VERIFICATION:  SWENT has been verified against analytical solutions of
fluid flow, heat flow, inert component transport, and radionuclide transport.
Good agreement between model results and analytical solutions was found in each
of the 11 cases cited.  Three field applications of SWENT are included in the
code documentation.  The predicted SWENT results are in reasonable agreement
with field observations.

DOCUMENTATION/REFERENCES:

INTERA Environmental Consultants, Inc.  (1983)  SWENT:  a three- dimensional
     finite-difference code for the simulation of fluids, energy, and solute
     radionuclide transport.  ONWI-457, Prepared for Battelle Memorial
     Institute, Office of Nuclear Waste Isolation, Columbus, OH.

SOURCE:  SWENT was developed by S. B. Pahwa, R. B. Lantz, and B. S. Ramaro of
INTERA Environmental Consultants, Inc.
                                      A-98

-------
CODE NAME:  TRANS

PHYSICAL PROCESSES:  Predicts groundwater pollution problems.

DIMENSIONALITY:  Two dimensional.

SOLUTION TECHNIQUE:  Random walk.

DESCRIPTION-  TRANS provides a generalized computer code that can simulate a
large class of problems involving convection and dispersion of chemica
contaminants associated with fertilizer applications  hazardous waste leachate
from landfilled  and other sources, and injection of chemical waste into the
subsurface using disposal wells.  TRANS does not address density-induced
convection.  Concentration distribution in the aquifer represents a
vertically-averaged value over  the saturated thickness of the aquifer.

TRANS  is  capable of considering:

.   Saturated groundwater flow in a single  confined  or  unconfined aquifer
     where wate? flow  is  typically horizontal.   (The  code addresses  temporal
     variations  in two-dimensional  (x-y)  flow  for  a variety of  boundary
     conditions  and  arbitrary  x-y geometry.)

 •   Advection of  a chemical  contaminant  in  a  saturated  groundwater  system
      released from a variety of typical  sources.

 •    Hydrodynamic  dispersion (both  lateral  and transverse)  and  diffusion of a
      chemical  contaminant in a saturated groundwater  system.

 •    Retardation of a chemical contaminant when it can be characterized by  a
      constant Kd and the assumptions of instantaneous and reversible adsorption
      are adequate.

 •    Radioactive decay of a chemical contaminant.

 TRANS addresses only a single aquifer.  Spatial and temporal distribution of
 head in the aquifer can be calculated by four methods:

  1)   analytic (HSOLV2) solution for a uniform 1-ft/d flow  in the x direction;

  2)   analytic (HSOLV4) solution to the Theis formula centered at node (15,  15);

  3)   numerical  finite difference solution (HSOLVE) to the  two-dimensional (x-y)
      vertically-averaged groundwater flow equation*  (this  solution  is for
      transient  or steady-state flow); and

  41    user-supplied subroutine  for reading or calculating head on the
       finite-difference grid used in the TRANS transport model.

  The transport model portion of TRANS uses a direct simulation technique  The
  SJcenSatlon of  a chemical constituent  in  a groundwater Astern is  assumed to
  be represented  by a finite  number of discrete particles.   Each of these
  particles  is  moved  according  to  the advective velocity  and dispersed according
  ?o random-walk  theory.   The mass assigned to each partic e represents a
  fraction of the total mass  of chemical constituents  involved.  In the limit, as

                                       A-99

-------
 the number of particle approaches the molecular level,  an exact solution  to  the
 actual  situation is obtained.   This kind of transport model  is  inherently mass
 conservative.  Convergence can be checked by increasing the,number of
 particles.  There are restrictions, as with any numerical  method,  which limit
 the size of time step that can be taken for both a time-dependent  and
 spatially-dependent problem.   Time steps for particles  are limited such that
 advective plus dispersive movement is no greater than the spacing  between
 velocity (head) nodes.

 The principal assumptions regarding flow are:

 §     Darcian flow is assumed;

 •     flow in the aquifer is horizontal  and  controlled only by hydraulic head
      gradients;

 •     leakage between the simulated aquifer,  rivers,  lakes,  other aquifers, and
      springs is a linear function of head difference with  the slope of this
      relationship determined from the leakage  parameter,  K/m, where K is  the
      permeability of the aquitard (or stream bed)  and m is  the  thickness;  and

 •     storage in the stream, lake,  or river  beds and  aquitards is ignored.

 The principal assumptions regarding contaminant transport  are:

 •     the advection-diffusion equation for solute transport  is assumed valid;

 •     dispersion in porous media is random process; and

 •     retention of a contaminant (or retardation of a concentration front)  may
      be represented by an instantaneous and  reversible  sorption process.

 CODE INPUT:   Input requirements for the code are those  typically available
 from standard field or laboratory measurements.   For the  flow portion of  the
 model they include:

 •     a  variable finite-difference grid  description;

 •     time step and number of time  steps to  be  run;

 t     area! distributions  of
           permeability
           source aquifer  potential  for  leaky artesian simulations
           aquifer bottom  elevations;
           aquifer top  elevations;
           head  (initial conditions)
           aquitard thickness and  permeability  for leaky artesian aquifers
           simulations
           artesian and water table  storage coefficients;

•    pumping  and  recharge well   locations and temporal rates;

t    stream  (river or lake) node locations,   surface-water elevations,  stream or
     lake bed thickness and permeability; fraction of node area  available  for
     transfer;

                                      A-100

-------
•    constant head node locations and elevation for held head;
a    location of springs, elevation at which spring flow begins,  and slope of
     the spring flow versus groundwater head for the spring production line;
     and
a    locations of nodes where evapotranspiration from the water table is to be
     considered and the slope of the rate versus head line and the water-table
     elevation at which evapotranspiration effects are to be ignored.
For the transport model, additional input requirements include:
a    longitudinal dispersivity;
a    lateral dispersivity;
a    effective porosity;
a    retardation factor or Kd;
a    bulk mass density of porous medium;
a    location and concentration of sources, description of source geometry, and
     selection of method for release of particles; and
a    sink locations and groupings of sink locations for summarizing outflow
     versus time results.
The model contains no checking of input for consistency and automatic
termination for faulty or inconsistent inputs.
CODE OUTPUT:  Results are printed in a 132-character format with a concise
and readable output layout.  The code echoes input parameters and produces line
printer plots of head, numbers of particles, and concentrations.  The code also
reports the concentration of water entering sink nodes and groups of sink nodes
versus time.  The code produces no contour maps or output fields that can be
passed on to other computer system programs for plotting and produces no mass
balance summaries for water flow or transport.
COMPILATION REQUIREMENTS:  The TRANS code  is written in FORTRAN and run on a
CDC CYBER-175 machine.  The code has also  been brought up on a Digital
Equipment Corporation VAX 11/780.
EXPERIENCE REQUIREMENTS:  Extensive.
TIME REQUIREMENTS:  Months.
CODE VERIFICATION:  The code has been compared to analytic and
hand-calculated examples.
DOCUMENTATION/REFERENCES:
McDonald, M.G.; Fleck, W.B.  (1978)  Model analysis of the impact on
     groundwater conditions of the Muskegon County waste-water disposal system,
     Michigan.  U.S. Geological Survey Open-File Report 78-79.
                                      A-101

-------
Prickett, T.A.; Lonnquist, C.G.   (1971)  Selected Digital Computer techniques
     for groundwater resource evaluation.  Illinois State Water Survey Bulletin
     55.

Prickett, T.A.; Namik, T.G.; Lonnquist, C.S.  (1981)  A random-walk solute
     transport model for selected groundwater quality evaluations.  Illinois
     State Water Survey Bulletin 65.

SOURCE:  The program was written by Thomas A. Prickett of Thomas A. Prickett
and Associates, and Thomas 6. Naymik and Carl 6. Lonnquist of Illinois Water
Survey.
                                     A-102

-------
CODE NAME:  TRANSAT2
PHYSICAL PROCESSES:  Predicts groundwater flow and contaminant transport in
saturated geologic media.
DIMENSIONALITY:  Multi-dimensional.
SOLUTION TECHNIQUE:  Numerical,  finite element.
DESCRIPTION:  The fundamental equations solved are those of fluid mass
conservation, Darcy's equation,  and contaminant mass conservation.  The
principal assumptions are as follows:
•    fluid flow may be described by Darcy's equation;
t    steady-state groundwater flow exists;
•    solute transport includes advection, dispersion, diffusion, adsorption,
     and first-order reactions (i.e., radioactive decay); and
•    system properties may vary spatially.
TRANSAT2 utilizes the Galerkin finite-element  technique with linear triangular
elements.  The solution technique is a Gaussian elimination method.   This
multi-dimensional code includes the following  boundary conditions:
•    specified hydraulic head;
t    specified fluid  flux;
•    zero concentration gradient;
•    cauchy  boundary  condition; and
•    temporally  varying solute boundary  conditions.
CODE INPUT:   Standard groundwater flow and  contaminant  transport.
CODE OUTPUT:   Predicts head  and concentration.
COMPILATION REQUIREMENTS:  Unknown.
 EXPERIENCE  REQUIREMENTS:   Extensive.
TIME REQUIREMENTS:  Months.
 CODE VERIFICATION:  Unknown.
 DOCUMENTATION/REFERENCES:
 Pickens, J.F.;  Lennox, W.C.   (1976)   Numerical simulation of waste movement in
      steady ground-water  flow systems.   Water Resour.  Res.  12(2):171-180.
                                      A-103

-------
Lee, D.R.; Cherry, J.A.; Pickens, J.F.  (1980)  Groundwater transport of a salt
     tracer through a sandy lake bed.  Limnology and Oceanography 25(1):45-61.

Grisak, G.E.; Pickens, J.F.  (1980)  Solute transport through fractured
     media -- I.  the effect of matrix diffusion.  Water Resour. Res.
     26(4):719-730.

SOURCE:  TRANSAT2 is a GTC proprietary code developed at Geologic Testing
Consultants in Ottawa, Ontario, Canada.

-------
  NUMERICAL COUPLED CODES
(SOLUTE AND HEAT TRANSPORT)
            A-105

-------
 CODE NAME:   CFEST --  Coupled  Fluid,  Energy and  Solute Transport

 PHYSICAL  PROCESSES:   Fluid, energy,  and  solute  transport  in a confined,
 saturated aquifer.

 DIMENSIONALITY:   Up to three  dimensional.

 SOLUTION  TECHNIQUE:   Finite element.

 DESCRIPTION:  CFEST was developed for the  analysis of a confined aquifer's
 response  to  thermal energy storage.  This  model employs a standard Galerkin
 finite-element method in the  solution of the coupled equations of mass  energy
 and solute-mass conservation.   A sequential solution algorithm bearing
 resemblance  to the SWIP model  is used to solve the flow of water, the transport
 of energy, and finally the transport of solute. The finite-element
 approximation to the continuum is made with a bilinear,  two-dimensional,
 quadrilateral element that is  simply expanded to the trilinear,  eight-node
 brick when three-dimensional  analysis is warranted.  CFEST can be used to
 analyze two-dimensional  vertical  or horizontal  planes,  two-dimensional
 axisymmetric cross sections,  and fully three-dimensional  aquifer situations
 Verifications against analytical  and semianalytical solutions  have been made
 and are given in documentation on the model.  CFEST has  been  applied to the
 analysis of solid waste  landfills for the U.S.  Environmental  Protection Agency.

 CFEST simulates  confined,  saturated aquifer systems.  Unconfined,  saturated
 aquifers can be  simulated  by assuming the top  elevation  of the aquifer is  at
 the water table  and assigning  piezometric heads  based on  the water table
 elevation.

 CODE INPUT:   Finite-element grid  describing soil  profile  geometry  (two or
 three dimensional),  soil hydraulic  characteristics, recharge,  solute
 characteristics  (Kds), and time step  size.

 CODE OUTPUT:   Output consists  of  line-print listings (output data  files) and
 plots  (output plot files) of potentials,  temperatures, and solute
 concentrations at  user-specified  nodes  (grid points).

 COMPILATION  REQUIREMENTS:  CFEST  is available in FORTRAN and is currently
 being  used on DEC  VAX  11/780 and  MICRO-VAXs.

 EXPERIENCE REQUIREMENTS:  Extensive.

TIME REQUIREMENTS:  Months.

CODE VERIFICATION:  CFEST has  been verified  against analytical  and
semianalytical solutions.  Results of the verification are included in the
model documentation.

DOCUMENTATION/REFERENCES:

Gupta, S.K.;  Kincaid, C.T.;  Meyer, P.R.; Newbill, C.A.;  Cole,  C.R.   (1982)   "A
     multidimensional finite element code for the analysis of  coupled fluid,
     energy,  and  solute transport  (CFEST).  PNL-4260,  Pacific  Northwest
     Laboratory,  Richland,  WA.

                                     A-106

-------
SOURCE:
     S. K. Gupta
     Office of Nuclear
     Batten e Memorial
     505 King Avenue
     Columbus, OH  43201
Waste Isolation
Institute
     C. T. Kincaid or C. R. Cole
     Battelle, Pacific Northwest Laboratory
     Battelle Boulevard
     Richland, WA  99352
                                      A-107

-------
 CODE NAME:  GWTHERM
 PHYSICAL PROCESSES:  Predicts fluid flow and transport in a heated porous
 medium.                                                            r
 DIMENSIONALITY:  Two dimensional .
 SOLUTION TECHNIQUE:  Numerical,  finite difference.
 DESCRIPTION:  GWTHERM is a two.-dimensional  model  based on the equations for
 nomsothermal  single-phase flow and solute  transport in porous media.  It allows
 for amsotropy of hydraulic conductivities  and inhomogeneous density and
 temperature-dependent fluid properties,  and it uses  the alternating-direction
 implicit technique with an integrated finite difference scheme to  provide an
 ?S?S?DM ^"nli?,,!?^16.50]^10]  Procedure-   Recent development has coupled
 GWTHERM to DAMSWEL to include dependence of permeability on  effective  stress
 changes.  In this sense the connection between these models  is sequential
 code^keaSTEALTH/HARTUd'  ^ U  represents a "Polity available only in a
 CODE INPUT:   Unknown.
 CODE OUTPUT:   Unknown.
   ?E(*UIREMENTS:
system.
                                    is written in FORTRAN and implemented on a
 EXPERIENCE REQUIREMENTS:  Extensive.
 TIME REQUIREMENTS:  Months.
 CODE VERIFICATION:  Unknown.
 DOCUMENTATION/REFERENCES:
 Runchal, A.jTreger, D.; Segal, G.  (1979)  Program EP21 GWTHERM:
     Sr/TMmP«Si?nal/luid flow' heat and mass transport in porous media.
     ATG/TN-LA-34, Advanced Technology Group,  Los Angeles,  CA,  April.
SOURCE:  GWTHERM was developed by A.  Runchal  of Dames & Moore's Advanced
Technology Group.  The code is in the public  domain.
                                     A-108

-------
CODE NAME:  OGRE
PHYSICAL PROCESSES:  Predicts groundwater flow and nuclide transport in a
heated porous medium.
DIMENSIONALITY:  One or two dimensional.
SOLUTION TECHNIQUE:  Numerical, finite element.
DESCRIPTION:  OGRE can be used for the time simulation of fluid flow and mass
transport through porous media, Using an implicit (backward Euler)
finite-difference scheme.  Specifically,  OGRE has been used to simulate the
time-dependent flow of groundwater into or out of underground openings and the
mass transport of radionuclides under the influence of a pressure gradient.
Parameters may be either time and space dependent or fixed in either time or
space.  Initial and boundary conditions may also vary with time and space.
Zoning of the grid must be constant in both directions, but zoning is dynamic
and set at execution time.
CODE  INPUT:  Unknown.
CODE  OUTPUT:  Unknown.
COMPILATION  REQUIREMENTS:  OGRE is written in  FORTRAN and  implemented on  a
CDC 7600  machine.
EXPERIENCE REQUIREMENTS:   Extensive.
TIME  REQUIREMENTS:   Months.
CODE  VERIFICATION:   Unknown.
DOCUMENTATION/REFERENCES:
Korver, J.A.   (1970)   UCRL-50820,  Lawrence Livermore  Laboratory,  Liyermore,  CA,
      February.
SOURCE:   OGRE  was  written  by J. A. Korver of  the  Lawrence Livermore
 Laboratory.  The  code  is  in  the public domain.
                                      A-109

-------
 CODE NAME:  SHALT  (Solute, Heat, and Liquid Transport)

 PHYSICAL PROCESSES:  Predicts liquid flow, heat transport, and solute
 transport in a regional groundwater flow system.

 DIMENSIONALITY:  Two dimensional (x-y or x-z cartesian).

 SOLUTION TECHNIQUE:  Numerical, finite element.

 DESCRIPTION:  SHALT performs the two-dimensional simulation of fluid flow,
 solute transport, and heat transport in a porous medium.  The spatial domain is
 discretized using three-noded triangular elements and the time domain by a
 fully implicit backward difference scheme'.  The aquifer parameters may be
 distributed or zoned and the system may be anisotropic. The viscosity of the
 liquid phase and the diffusion coefficient of the solute are functions of
 temperature.  The density of the liquid phase is represented as a function of
 temperature and total solute concentration. The equations describing the fluid
 flow,  energy transport, and solute transport are fully coupled with the
 dependent parameters upgraded after each time step.   Fractured media may be
 modeled by treating the fractured rock as a continuum.

 SHALT may be considered both a near-field and far-field code as
 temperature-dependent parameters have been implemented in this code.   It would
 be considered more as a far-field code,  however,  as  deformation and the
 stress/strain relationships are not considered  in  this model.

 The main assumptions of the code are:

 •    Darcy's Law is valid;

     the compressibility and  heat capacity of the  liquid phase are  constant;

     k,   , o,  and QP can vary spatially  but do  not vary with time and are not
     dependent  on the concentration or the  temperature;

 •    the thermal  dispersion tensor  for the  liquid phase includes the  effects of
     mechanical dispersion  and  thermal conduction and  is a  function of
     velocity;

 t    the exchange of heat and solute  is  instantaneous  between  the liquid  and
     solid phases at  the same point; and

 *    the hydrodynamic dispersion tensor  for the liquid  phases  includes  the
     effects of mechanical  dispersion and thermal conduction and is a function
     of  velocity.

 CODE INPUT:  Inputs to SHALT  include:

 •    mesh geometry;

•    initial  heads;

•    initial  concentrations and initial temperatures;

•    fluid density;
                                     A-110
•

t

-------
•    porosity;
•    compressibility of the fluid and porous medium;
•    the permeability tensor;
t    viscosity;
•    solute density;
•    heat capacity of both the solid and liquid phases;
•    the thermal conductivity tensor;
•    the thermal dispersivity tensor;
0    the hydrodynamic dispersivity tensor;
•    the first-order reaction constant;
•    distribution coefficient; and !
•    the bulk density.
CODE OUTPUT:  The output  of  SHALT consists of the pressure, concentration,
and temperature distribution at  each time step.'
COMPILATION REQUIREMENTS:  SHALT is written  in  FORTRAN  IV.
EXPERIENCE REQUIREMENTS:   Extensive.               '
TIME REQUIREMENTS:  Months.
                                         >  ,
CODE VERIFICATION:  The  liquid flow portion  of  the model was tested  by
calculating steady-state  pressure and  hydraulic head distributions for various
flux inputs.  Values of  system parameters were  chosen to be constant.  The
calculated hydraulic head gradient for steady-state conditions was correct.
The heat transport  portion of the model  was  tested by comparison  with results
of the  analytical solution of Bredehoeft and Papadopulos (ref. 2) for
one-dimensional steady-state transport.
The solute transport portion of  the model was tested by comparison with  results
of the  analytical solution by Ogata and Banks (ref. 3)  for one-dimensional
advection-dispersion with a  step input in concentration.
SHALT was used  successfully  to model results of pressure testing  in  fractured
rock at Chalk River (ref.  4).
DOCUMENTATION/REFERENCES:
Pickens, J.F.;  Grisak, G.E.   (1979)  Finite  element analysis of liquid flow,
     heat transport, and  solute  transport in a  ground-water flow  system:
     governing  equations  and model formulation.  Atomic Energy of Canada, Ltd.,
     Report TR-81,  September.
                                      A-I'll

-------
Bredehoeft, J.D.;  Papadopulos,  S.S.   (1965)  Rates of vertical groundwater
     movement estimated  from the  earth's thermal profile.  Water  Resour. Res
     l(2):325-328.

Ogata, A.; Banks,  R.B.   (1961)  A solution of the differential equation of
     longitudinal  dispersion in porous media.  U.S. Geological Survey
     Professional  Paper  411-A.

Davison, C.C.  (1981)  Physical hydrogeologic measurements in fractured
     crystalline rock:   summary of 1979 research program at WNRE  and CRNL
     Atomic Energy of Canada, Ltd., Technical Record 161.

SOURCE:  SHALT was developed by J. F.' Pickens and G. E. Grisak of the Inland
Waters Directorate, Environment Canada for Atomic Energy of Canada, Ltd.  The
code is in the public domain.
                                     A-112

-------
CODE NAME:  SWIFT (Sandia Waste Isolation Flow and Transport)

PHYSICAL PROCESSES:  Predicts flow, solute, and heat migration from the
repository through the groundwater system.  ,

DIMENSIONALITY:  The pressure, temperature, and concentration field is
represented by a series of three-dimensional rectangular cartesian grid
points.  In addition, a two-dimensional (r,z) grid system is also
provided.

SOLUTION TECHNIQUE:  Numerical, finite difference.

DESCRIPTION:  The code simulates the flow and transport of energy, solute,
and radionuclides in a geologic media.  SWIFT is a three-dimensional,
finite-difference, groundwater flow and nuclide transport code.  The model
takes  into account saturated flow  in an isothermal or heated porous medium as
well as sorption and desorption mechanisms.  In addition, the code takes into
explicit account nuclide decay and the creation of daughter products.  For the
nuclide decays, the code considers conservation of dissolved contaminants,
energy, and total liquid mass.  The fluid density can be a function of
pressure, temperature, and concentration.  Viscosity can also be a function of
temperature and concentration.  Aquifer properties can vary  spatially.
Hydrodynamic dispersion  is described as a function of velocity.  Boundary
conditions allow natural water movement in the aquifer, heat losses to the
adjacent  formation and location of injection, production, and observation
points anywhere  in the system.

SWIFT  solves four  coupled differential equations, together with  a number of
submodels describing  the nonlinearities,  in  a sequential manner. Options
include:

•    steady-state  or transient  flow;

•    solute transport;

•    eat transport;

•    well  bore;

t    heterogeneous and/or anisotropic  media;

•    confined  and/or water table conditions; and

•    recharge  and/or wells.

 SWIFT is a descendant of the code SWIP (Survey Waste .Injection Program,  ref.  3)
developed for  the U.S.  Geological Survey.  SWIP was originally put together,  in
 part,  from oil industry codes.

The main assumptions of the code are:

 •    Flow follows Darcy's Law.  -
                                       A-113

-------
 •
 t

 •
 •
CODE
t
t
t
§
 Fluid density can be a function of  pressure, temperature, and
 concentration of the inert component.   Fluid viscosity can be a function
 of temperature and concentration.
 Injection wastes are miscible with  the  in-place fluids.
 Aquifer properties vary with position (i.e., porosity, permeability,
 thickness, and elevation can be specified for each grid block in the
 model).
 Hydrodynamic dispersion 1s described as a function of fluid velocity.
 Radioactive constituents are present in trace quantities only,  that is
 fluid properties are independent of the concentrations of these
 contaminants.
 The energy equation can be described as "enthalpy in - enthalpy out =
 change in internal  energy of the system."  This is rigorous except for
 kinetic and potential  energy which have been neglected.
 Boundary conditions allow natural  water movement in the aquifer,  heat
 losses to the adjacent formations, and the location of injection,
 production, and observation points anywhere within the system.
 INPUT:  Inputs for  SWIFT include:
 half-life of each nuclide;
 distribution coefficient  of each nuclide on each rock type;
 fluid compressibility;
 porous medium compressibility;
 coefficient of thermal  expansion of  fluid;
 fluid heat  capacity;
 rock  heat capacity;
 thermal conductivity of rock-fluid mixture in each direction for each rock
 type;
 longitudinal  and transverse dispersivities for each rock type;
 molecular diffusivity in porous medium;
 rock density;
fluid density;
pressure and temperature of injected or produced fluids in each  well;
thermal diffusivity  of rock surrounding well  bores;
                                A-114

-------
t    fluid viscosity as a function of temperature and brine concentration;

0    hydraulic conductivity of each rock type in each direction;

t    porosity of each rock type;

•    heat capacity of each rock type;

•    boundary conditions;

•    initial velocities and concentrations;

•    salt dissolution rate in each rock type;

•    size, placement, and contents of waste canisters;

•    solubility limits;

•    production rate of each well;

t    location, angle, and depth of each well;

•    diameter  and  pipe roughness  of  each well;  and

a    leaching  time of wastes.

CODE OUTPUT:   Output  for SWIFT  consists of the  pressure,  temperature,  solute
concentration, and the concentration of each  radioactive  isotope.  These  are
given  at every grid point  after each time  step  as required.

 COMPILATION REQUIREMENTS:   SWIFT  is  written in  FORTRAN  IV for use  on a CDC
 6600 machine.  With minor  modifications,  it can be  used on other machines.
 This modification primarily involves the  real-time  dimensioning feature  of
 SWIFT.

 EXPERIENCE REQUIREMENTS:   Extensive.

 TIME REQUIREMENTS:  Months.

 CODE VERIFICATION:  To evaluate the  effect of numerical truncation errors
 arising due to isotope decay terms  and to develop a set of criteria to delete
 components in numerical  simulations  without losing  any  accuracy in the results,
 SWIFT was compared against the  results from ORIGEN  (ref.  2).  The ORIGEN  model
 is a matrix exponential  solution  of the Bateman equations for radioactive
 decay.  The comparison was considered excellent. Other parts of SWIFT that
 solve for flow and transport have been tested against both analytical  and
 laboratory results.

 DOCUMENTATION/REFERENCES:

 Dillon, R.T.; Lantz, R.B.; Pahwa, S.B.  (1978)   Risk methodology for geologic
      disposal of radioactive waste:   the Sandia waste isolation flow and
      transport (SWIFT) model.   Sandia National  Laboratories Report SAND
      78-1267.

                                      A-115

-------
Bell, M.J.  (1973)  ORIGEN -- the ORNL Isotope generation and depletion code.
     Oak Ridge National Laboratory Report ORNL-4628.

Reeves, H.; Cranwell, R.M.  (1981)  User's manual for the Sandia waste-
     in2;^lSn«flow and transP°rt model-  Sandia National Laboratories Report
     NUREG/CR-2324, November.                                            K

Papadopulos, S.S.; Larson, S.P.  (1978)  Aquifer storage of heated water:   part
     II -- numerical simulation of field results.  Ground Water 16:242-48.

SOURCE:  SWIFT was developed by R. T. Dillon at Sandia Laboratories and by R
B. Lantz and S. B. Pahwa at Intera, Incorporated.
                                     A-116

-------
CODE NAME:  SWIP2
                      .-,
PHYSICAL PROCESSES:  Predicts the effects of liquid waste disposal in deep
saline aquifers.

DIMENSIONALITY:  Three dimensional (cartesian or radial}	  --
SOLUTION TECHNIQUE:  Numerical, finite difference.

DESCRIPTION:  SWIP2 is a transient, three-dimensional subsurface waste
disposal model to provide methodology to design and test waste disposal
systems.  The model is a finite-difference solution to the pressure, energy,
and mass-transport equations.  Equation parameters such as viscosity and
density are allowed to be functions of the equations' dependent variables.
Multiple user options allow the choice of x, y, and z cartesian or r and z
radial coordinates, various finite-difference methods, iterative and direct
matrix solution techniques,,restartroptions, |and various provisions for output
display.  Well-bore heat and pressure-loss calculation capabilities are also
available.                                     /

The 1979 update of the SWIP model  involved additions and modifications to
include free water surface, vertical recharge, equilibrium controlled linear
adsorption, and a first-order  irreversible rate reaction.  These modifications
make this model more adaptable to  general hydro!ogic problems and those
involving waste disposal with  simple chemical reactions.

CODE INPUT:  Unknown.

CODE OUTPUT:  Unknown.

COMPILATION REQUIREMENTS:   SWIP2  is written in FORTRAN and implemented on a
CDC machine.

EXPERIENCE REQUIREMENTS:  Extensive.

TIME REQUIREMENTS:  Months.

CODE VERIFICATION:  Unknown.
  /
DOCUMENTATION/REFERENCES:

INTERA Environmental Consultants,  Inc.  \1976)  A model for calculating effects
 r   of liquid waste disposal/in deep saline aquifer, part I -- development,
     part II -- documentation'. Houston, TX, June.

INTERA Environmental Consultants,  Inc.  (1979)  Revision of the documentation
     for a model for calculating effects of liquid waste disposal in deep
     saline aquifers.  U.S. Geological Survey, Water Resources Investigations
     79-96.

SOURCE:  SWIP2 was developed by INTERA Environmental Consultants,
Incorporated.  This code is in the public domain.
                                      A-117

-------

-------

-------
                                                                   S ">
                                                                   CO
                                                                   o
                                                                   o
O
O

oo

CO
CO

o
•-J
tn
§«!
J 2 2
w O CD
.  < O
  3 -S

  0 3
  5«

  CD O
  D, X
  IS
  — m
  SP
                                                                                > ml
                                                                                CD 3 '
                                                                                CD <.

                                                                                II

                                                                                  II

                                                                                  i

                                                                                   §
                                                                                n> -I
                                                                                -• o

                                                                                O3


                                                                                £>
                                                                                01
                                                                                NJ
                                                                                05
                                                                                00

-------